chemical transformations of polymers

INTERNATIONAL UNION OF PURE AND APPLIED CHEMISTRY

MACROMOLECULAR DIVISION in conjunction with the

POLYMER INSTITUTE OF THE CZECHOSLOVAK ACADEMY OF SCIENCES,

THE CZECHOSLOVAK CHEMICAL SOCIETY, THE CZECHOSLOVAK TECHNICAL UNIVERSITY

and SLOVCHEMIA

CHEMICAL TRANSFORMATIONS OF POLYMERS

Plenary and main lectures presented at the INTERNATIONAL CONFERENCE ON CHEMICAL

TRANSFORMATIONS OF POLYMERS held in Bratislava, Czechoslovakia

22-24 June 1971

Conference Editor R. RADO

Polymer Institute, Bratislava

LONDON BUTTERWORTHS

ENGLAND : BUTTERWORTH & CO. (PUBLISHERS) LTD. LONDON: 88 Kingsway, WC2B 6AB

AUSTRALIA : BUTTERWORTH & CO. (AUSTRALIA) LTD. SYDNEY: 586 Pacific Highway, Chatswood, NSW 2067 MELBOURNE: 343 Little Collins Street, 3000 BRISBANE: 240 Queen Street, 4000

CANADA : BUTTERWORTH & CO (CANADA) LTD. TORONTO: 14 Curity Avenue, 374

NEW ZEALAND: BUTTERWORTH & CO. (NEW ZEALAND) LTD. WELLINGTON: 26-28 Waring Taylor Street. 1

SOUTH AFRICA: BUTTERWORTH & CO. (SOUTH AFRICA) (PTY) LTD. DURBAN: 152-154 Gale Street

The contents of this book appear in

Pure and Applied Chemistry, Vol. 30, Nos. 1-2 (1972) Suggested U.D.C, number 541-64 (063)

International Union of Pure and Applied Chemistry 1972

ISBN: 0 408 70310 5

Printed in Great Britain by Page Bros. (Norwich) Ltd., Norwich

ORGANIZING COMMITTEE Chairman:P. HRDLOVIC

Secretary: M. CAPLA

Members: E. BORSIG V. OURD'OVIÎ 0 . HOFFMAN 1. LUKAC J. MYNAftiK V. POLLAK N . SCHUBERTOVA E. SPIRK I. ZVARA

vi

PHOTOCHROMIC BEHAVIOUR OF POLYMERIC SYSTEMS AND RELATED PHENOMENA

G. SMETS

University of Louvain, K.U.L. Leuven/Belgium

ABSTRACT The different photochromic systems which have been used in polymer chemistry are mainly based on cis-trans isomerization of azomethines and azocompounds which were built into the polymeric backbone, or attached as sidegroups. These systems are reviewed and briefly commented upon. Special attention is paid to spirobenzopyran derivatives of which the photochromism is very important and characterized by a strong negative solvatochromism. Different spirobenzopyran photochrome-containing polymers are considered : (1) copolymers with vinyl monomers, e.g. methyl methacrylate-styrene. (2) polypeptides such as polytyrosine and polylysine. (3) polycondensation products. Their behaviour in solution is discussed and the existence of different isomerie mesoxy amines demonstrated.

In the solid state when these photochromes are used in a polymeric matrix, they permit detection of secondary glass transition phenomena; by contrast, when attached on a copolymer, they permit accurate determination of Tg. In crosslinked systems the copolymers show a reversible photochromic behaviour. Finally a comparison of these photochromic phenomena with

thermal racemization of some new optically active polymers is presented.

INTRODUCTION Substances which undergo reversible colour formation under irradiation

with light are called photochromic compounds. The fundamental processes which depend on this phenomenon can be very different. For inorganic species such as metal oxides and halides, they are usually related with the presence of some impurities or crystal defects which interact with the electrons liberated under the influence of light. For organic compounds, photochromism is mostly linked with molecular structural modifications, e.g. valence isomerization, tautomerism, bond scission (homolytic and heterolytic), cis-trans isomerization and dimerization reactions ; sometimes even chemical reactions may be involved, for example, redox reactions. It is the purpose of the present lecture to discuss some systems, which have been used recently in the domain of polymer chemistry, and to concentrate more on details in the use of spirobenzopyran derivatives which were the basis of our own experiments. As a general rule, it can be assumed that the incorporation of photochromic groups into a polymer molecule may deeply affect the photochromic behaviour on account of polar and steric effects

1

P.A.C—30/1—B

G. SMETS

due to the proximity of the polymer chain and on account of internal viscosity effects which may restrict segmental motions. On the other hand, in order to be conclusive, the results obtained with photochromic polymers should be compared with those of mixtures of the corresponding polymers with the homologous low molecular weight photochromes.

There is therefore a first experimental condition, which should be fulfilled, if one expects a strong difference between the photochromic behaviour of the systems to be compared, namely the phenomenon which is involved should include an appreciable change of configuration of the photochromic group, for example an important movement of one moiety of the molecule with respect to the other one; valence tautomerism would therefore a priori be much less interesting than cis-trans isomerization.

Thus, most data in the literature on photochromic polymers are concerned with cis-trans isomerization phenomena either in the case of azomethines, or that of azo-compounds, such groups being attached to a polymeric chain as sidegroups, or incorporated in the main chain. These data will be now discussed successively.

1. PHOTOCHROMIC POL Y AZOMETHINES The photochromism and thermochromism of some hydroxylated poly-

azomethines has been studied recently by Laverty and Gardlund1. The reaction scheme is represented by equation 1, where the hydrogen bonded anil (I) can be photoisomerized into the trans-ksio derivative (III). This irans-keto form isomerizes thermally to the ds-keto isomer (II), which itself can return to the ds-enol (I) by hydrogen jump through the equilibrium I ^ II (AH ^ 1 kcal).

polar a i ) (1 )

The authors prepared different polyazomethines, e.g. from methylene-bis-salicylaldehyde and p-phenylenediamine (IV) and p.p'-méthylène dianiline (V). The polymers were insoluble in most organic solvents, but soluble in concentrated sulphuric and concentrated formic acid.

2

PHOTOCHROMIC BEHAVIOUR OF POLYMERIC SYSTEMS

HO

CH?

Ox OH

(IV)

Figure 2

Although not photochromic, these polymers behave thermochromically on account of the easy proton shift; the bathochromic effect between 25° and 100°C is, however, very small. The movements and energy required to produce the trans-form of the polymer backbone were assumed to be prohibitive.

Cis-trans isomerization of an azomethine group followed by a rate-determining proton shift is also proposed for the orange-blue photochromism of mercurydithizonate, and was applied very recently to photochromic polymers by Kamogawa2. Following Meriwether et ai3,4 reaction scheme 2 can be written :

Ph

' I I * -N=N

H-wr At'

Ph ^ ^ N - N - H . , ^Hg | ; N — P h

^Ph

S = C - N ^

(VII)

(2)

(VI)

The copolymers contained p-amino phenyl-mercuric-diphenylthio-carbazonate side groups (VIII), with styrene or methylacrylate as comonomers :

-NH—C6H4—Hg—S—C //

N—NH—R

\ (VIII)

N=N—R

The spectral recovery occurs thermally. On the basis of decoloration halflife-times, the reaction is about a hundred times slower in a film, than in solution ; moreover, a higher glass transition of the copolymer (styrene comonomer) causes a longer decoloration time than that for copolymer with methylacrylate.

3

G. SMETS

2. PHOTOCHROMIC AZOBENZENES Cis-trans isomerization of azo compounds as the photochromic principle

has been used by several authors and first as sidegroups by Lovrien and Waddington5. Indeed, they synthesized photochromic polyelectrolytes by

CH3 CH,

-CH 2 —C- -CH2—C—CH2

I I COOH CO

CH3

I COOH

50000 < M„< 200000

NH-(( ) V N

I N-\0) ocH3

0CH3

98.8 1.2%

(IX)

frans -^ eis 23 kcal/mole

Figure 3

copolymerization of acrylic-methacrylic acid with a few per cent of azo-dye acrylamide monomers (1.2 to 3.6 per cent). Under the influence of light trans-azo compounds are transformed into their ds-isomer, the energy of

1.00

0.80

Q O

>> % 060 c ω

TJ

"o υ Έ.0Μ0 o

0.20

-

-

-

/

- / N-

/0

— i

hv

N

1

* '^ q

1 1 1

\a \ J

dark \

\ -j

I—l 1 2800 3200 3600

Wavelength, A £000

Figure 4. Photochromism of /?-acetamino-ira«s-azobenzene (0.01M) at pH 8 [R. Lovrien and J. C. B. Waddington, J. Amer. Chem. Soc. 86, 2315 (1964)]

4


activation of the reverse reaction (eis -> trans) being about 23 kcal/mole, while the enthalpy difference is about 5.7 kcal/mole. The photochromism of the model compound, p-acetamino-azobenzene is illustrated by Figure 4.

In the case of covalently bound dyes, irradiation response and dark recovery are linked to the charge and conformations of the polyacid chain ; thus the 'isosbestic' crossing point wavelengths depend on the pH, and decrease markedly when the intrinsic viscosity increases on account of the uncoiling of the charged macromolecule, i.e. around pH = 6.

Kamogawa and co-workers6 synthesized copolymers of vinylamino azobenzene and styrene (X), polyvinylhydroxy-azobenzenes (XI), copolymers of 4-acrylamidomethyI-aminoazobenzene (XII) with styrene, butylacrylate and methyl methacrylate.

ί2=ΟΗ-(Γ^^Ν=Ν-^ (X)

rN-0~C H '

2 - N H V ( ^ V N = N V ( ^ > (XII) CH ^=CH—CO-NH-CH

The authors measured the decoloration recovery halftime values of the polymers and compared them with those of the model compounds. Sometimes more favourable results were obtained for the polymers; they were explained on the basis of the aggregation of azobenzene molecules in a film or solution which is prevented by incorporation of an azobenzene component in a copolymer chain by separation with inert comonomers. On the other hand, incorporation in the polymeric chain causes steric hindrance for cis-trans isomerization ; these two effects may compensate each other, the total behaviour resulting from the balance between them.

The authors also compared the photochromic behaviour of the copolymers in solution and in film ; as expected, the cis-trans isomerization is slower in the film state, especially if the comonomer constitutes a rigid chain component (styrene and methyl methacrylate compared to butylacrylate).

Tabak and Morawetz7 examined the cis-trans conformational transitions in solution of polyamides containing azobenzene residues in the backbone, and compared these isomerizations with those of corresponding model compounds, namely compounds XIII/XIV and XV/XVI.

s

G. SMETS

0-"-0 -CO—(CH2)4—CO—NH-

NH—CO—(CH2)4—CO—NH—(CH2)6—NH (XIII)

:6H 13-^-NH-/Q\-N=N- vT^/ N H _ C ° ~ C ò H ! 3 (XIV)

-NH-<CH2)6-NH-CO- . / ( ^ V - N = N - a ^ V

CO—NH—(CH2)6—NH—CO—(CH2)4—CO— (XV)

C6H, 3 - N H ^ o V Q y N = = N - / Q V c O - N H - C 6 H l (XVI)

For XIII/XIV the isomerization rate of the polymer X is only 15 per cent lower than that of XIV ; it is independent of the concentration of the azo-benzene residues in the chain and corresponds strictly to a first order reaction.

Copolymer XV did not obey first order kinetics and is best represented by the following equation :

C/C0 = a exp (-&!*) + (1 - a)exp(-/c2i)

in which C and C0 are concentrations of azo-ds-form at time t and zero time respectively.

The two rate constants kx and k2 do not vary with the azobenzene content ; the fraction of the azobenzene units isomerizing at the faster rate do not change with a change in the reaction temperature, k1 being equal to about three times k2\ kjk2, however, decreases with increasing acidity of the medium.

The model compound XVI decolorizes with a rate constant corresponding to the slow azo-groups of the copolymer XV.

In order to explain these effects, Tabak and Morawetz admit that the time required for the change of chain conformation to a form which corresponds to the transition state of the azo-group must be comparable to the relaxation time of the cis-trans isomerization in the absence of such restraints.

It must, however, be pointed out that the experiments of these authors were all carried out in formic acid as solvent, and only in dilute solutions.

3. PHOTOCHROMIC POLYTHIAZINE REDOX SYSTEM Very interesting results were also described recently by Kamogawa in the

domain of the redox photochromism, i.e. the reversible cplour changes associated with reversible photochemical oxidation-reduction reactions8. It is well known that thiazine derivatives such as méthylène blue are photo-bleached in the presence of reducing agents like ferrous ions ; thus méthylène

6


blue can be reduced into the semi-quinone, which disproportionates rapidly into a colourless leuco derivative and méthylène blue.

The reverse reaction consists of slow oxidation of the leuco compound into semi-quinone, and consequently into the dye, under the influence of the iron(in) ions. This reaction principle has been applied for the synthesis of photochromic polymers9, as shown by reaction scheme 3.

M B ^ M B ' ^ M B H Fe2+ Fe3+

semi-quinone (3) I I

thiazine polymer disproport. leuco polymer t I I Ü U

rapid

Thus several compounds were obtained from N-hydroxy methylacryl-amide sidegroups of acrylamide copolymers by reaction with thiazine compounds ; the systems include various reduction agents, such as poly vinyl alcohol, ferrous ions, or mercapto derivatives and their photochromism depends strongly on the moisture content of the polymer.

4. PHOTOCHROMIC POLYSPIROBENZOPYRANES We have concentrated our efforts on spirobenzopyran derivatives10, where

the photochromism consists of scission of the C—O-pyran bond followed by a rotation of one part of the molecule so as to approach coplanarity (reaction 4)11.

Y n-d X

R A»-'.r ^C^^X /r^C 0) (4) N O — \ [ ')) hv.T

I R

when X = CMe2, Y = H; when X = S, Y = Me; hV : ultra-violet light ; hv : visible light.

The open ring merocyanines are characterized by a very strong solvato-chromism12 affecting their maximum absorption wavelength as well as their decoloration rates1 3 - 1 5 ; moreover, they are very sensitive to steric effects, especially in the benzthiazol series16 '17. These facts were the main reason for choosing these benzopyrylspirans for incorporation in macromolecules17'18. From the synthetic point of view, photochromic spiropyran polymers can be obtained either by copolymerization of unsaturated photochromic monomers with vinyl comonomers, or by polycondensation of bifunctional compounds.

Photochromic sidegroups can also be attached to a polypeptide chain, by reaction of a haloalkyl- or haloacylphotochrome with a hydroxyl- or amino-

7

G. SMETS

group-containing polymer ; thus by condensation of the free phenol groups of polytyrosine with N(ô-iodo-butyl) spirobenzopyran, a photochrome polypeptide was obtained. In these first cases, the photochromes are attached as sidegroups on the main chain, of which the influence depends on the distance of the photochrome to the chain, on the nature of the comonomer, and on the site of attachment in the photochrome itself. If the comonomer is acrylic or methacrylic acid, irradiation response and dark recovery will probably be linked to the degree of neutralization and the conformation of the polyelectrolyte.

In the polycondensates, where the photochromes are incorporated into the main chain, the synthetic approach is much more difficult, from the additional point of view of the synthesis of bifunctional dimers as well as that of the poly condensation. Indeed the insertion of substituents appropriate for further condensation is often tedious, and their choice necessarily limited on account of their influence on photochromic behaviour. On the other hand, the polycyclic structure of the photochromes causes a decrease of solubility of the condensation products, which precipitate frequently before attaining a high molecular weight; in this respect copolycondensation seems to be more likely to be satisfactory.

We will consider successively different types of such photochromic polymers.

(a)Photochromic spirobenzopyran copolymers Several substances were synthesized by copolymerization of methyl

methacrylate (90 molar per cent), e.g. with N-(ß-methacryloxyethyl)-l,3,3-trimethyl-ó'-nitro-indoleninobenzopyrylospiran (a), 1,3,3-trimethyl-6'-nitro-

R—CO—N N 0 2

(c) (d)


8'-methacryloxymethylene-indoleninobenzopyrylospiran (b), l,3'-dimethyl-5-methacrylamino-6'-nitro-benzothiazolinobenzopyrylospiran (c), and l-methyl-S'-phenyl-ó'-nitro-S'-methacryl-oxymethylene-benzothiazolino-benzopyrylspiran (d).

In these formulae R is isopropenyl (methacrylic copolymers) or isopropyl (isobutyric model substances). The photochromic behaviours of these copolymers and models in acetone solution are reported in Table 7, and illustrated by Figure 5 for the copolymer —A.

Table 1. Photochromic behaviour of spirobenzopyran copolymers (cop.) and their model compounds (mod.) in acetone solution

Structure

^■max

kl 20°C x l 0 3 . s e c _ 1

Ea

Mn x IO"3

mod.

577

8.2

22.3

a

cop.

577

9.9

22.4

213

b

mod.

567

1.3

24.3

cop.

567

1.0

24.5

379

mod.

526 (506)

1.7 (4.8)

(22.1)

c

cop.

478 (504)*

1.7Î (3.8)

(25.6)

64

d

mod.

569

2.0

21.6

cop.

569

1.05

23.0

426

* values in dimethylformamide. X apparent initial rate constants.

1 1 1 L-5 10 15 20

t, min

Figure 5. Decoloration of copolymer-A in acetone

9

G. SMETS

It can be seen from these data that the decoloration rate constants are about equivalent for the copolymers and corresponding models, except for the series d with a 3'-phenyl group, where the copolymer's rate is only half that of the model ; it shows already the influence of steric hindrance of the polymeric chain on the rotation possibility around the 3'-4'-bond. While the maximum absorption wavelengths for copolymers and models are identical for compounds a, b and d in acetone solution, for compound c this only applies in highly polar solvents, e.g. dimethylformamide. In such conditions, the decoloration kinetics correspond to a first order reaction rate up to high degrees of conversion.

Copolymer c has been submitted to further detailed study by P. H. Vandewijer on account of its peculiar behaviour in relatively low polarity solvents like acetone, dichloromethane and tetrahydrofuran18. By following the variation of optical intensity as a function of time, a strong deviation from a first order relationship was observed ; the decoloration curve can be represented by the equation

D = aexpi-X^) + fcexp(-X2i)

This equation assumes the existence of two photochrome isomers of which the rate of interconversion is low compared to that of decoloration, and of which a and b represent their contribution to the optical density at zero time.

Table 2. Decoloration kinetics of a methylmethacrylate-5-methacrylamino nitro spirobenzopyran copolymer in solution

Solvent

Acetone Dichloromethane Tetrahydrofuran

Temp. °C

20.5 20.0 22.2

k* x 103

sec - 1

1.7 3.4

38.0Î

Xi x 103

sec - 1

1.7 5.7 4.6

X2 x 103

sec" 1

0.27 0.48 0.70

%

a

36 66.7 68

isomers

b

64 33.3 32

* rate constant for the model compound. ί extrapolated value.

It must be noticed that /lmax shifts towards shorter wavelengths during decoloration, and that the optical density becomes a function of the wavelength and the duration of irradiation. The kinetics of decoloration in low polarity solvents permit demonstration in this way the existence of two (or more) isomerie merocyanines, and confirm the results generally obtained only at low temperatures by Fischer, Wippler, Lashkow et al.19.

It is also understandable that it will be in low polarity solvents that copolymers of the 5-methacrylamino photochrome (compound c) with other comonomers as styrene, methacrylonitrile, a- and ß-vinylnaphthalene will behave differently, the differences being most pronounced with the most polar comonomers.

10


1.0

Q en o

+

0.5 10 20 30 Time,min

Figure 6. Influence of f-butanol on kinetics of decoloration reaction of copolymer 2 in chloro-benzene at 15°C: (1) pure chlorobenzene; (2) chlorobenzene containing 10 vol. % i-butanol

The solution photochromism of spiropyran (c) built into a methyl metha-crylate copolymer was also examined in the presence of a non-solvent ; it was indeed assumed that it could be affected by coiling up of the macromolecule. In fact, the behaviour is essentially dependent on the polarity of the medium on account of the very strong negative solvatochromism of these compounds. This effect can be illustrated by Figure· 6, which represents the influence of the addition of i-butanol on the decoloration kinetics in chlorobenzene at 15°C. It can be seen that first order kinetics are restored by the addition of the precipitant. On the other hand, as expected, the behaviour of the model substance as well as that of the copolymer are strongly affected if one changes the overall viscosity of the medium; an increase of viscosity provokes a strong deviation from first order kinetics. All these effects have been analysed and discussed previously in detail18.

The departure of compound (c) from first order kinetics in solvents of relatively low polarity seems to be general where the photochromic group is linked to the polymeric chain through a methacrylaminophenyl group. Thus a copolymer of methyl methacrylate and N-methyl-benzthiazol(2-2')spiro-3'(p-methacrylamino)phenyl-6' nitro benzopyran decolorizes not only in at least two successive steps, but where the rate constant k2 is much slower than that of the model. It should, however, be pointed out that the activation energies of decoloration are similar (£modei : 25 kcal/mole ; for the polymer the first and second step have respectively values of Eal: 24.3 and £ a 2 : 26 kcal/mole). The monomer (a), l-(ß-methacryloxyethyl)-3,3-dimethyl-6-nitro-indolinospiropyran has also been copolymerized with acrylic acid. In dimethylformamide solution the rates of decoloration of the copolymer are practically identical with those of the model ; in dioxane, however, the rates are only one-half of them. On neutralization of the acrylic acid with piperidine in dimethylformamide solution, the rate decreases progressively with an increasing degree of neutralization (up to 40 per cent) very likely on account of a stabilization of the merocyanine open form.

11

G. SMETS

(b ) Photochromic spirobenzopyranpolypeptides18

It seems interesting to investigate the photochromic behaviour of spiro-pyran sidegroups when they are attached to a polypeptide chain. Indeed it could be expectea that accumulation of charged merocyanine groups along the chain would alter the chain conformation, and conversely that this conformation, if respected, will influence the photochromism itself. Thus two photochromic polytyrosines Px and P2 containing 54.7 and 72.3 weight per cent photochromes, were prepared following the reaction scheme 5.

(5)

H3C CH3

H3C CH

Their molecular weights were 10000 (PJ and 4000 (P2), i.e. 10 and 5.5 photochromic units per chain. Their behaviour was compared to that of the N-acetyl-tyrosine methyl ester derivative. Very strong differences have indeed been found: (i) the polypeptides showed a much less pronounced solvatochromism than their model; moreover, their absorption spectra always presented two absorption maxima instead of one for the model (see Table 3); (ii) in acetone and tetrahydrofuran, the decoloration kinetics Table 3. Absorption maxima of photochromic polytyrosines (P1 and P2) and model substance

(M) after ultra-violet irradiation

Solvent

Ethylene glycol Methanol Dimethylformamide Acetone Pyridine Tetrahydrofuran

E? 56.3 55.5 43.7 42.2 40.2 37.4

M

530 534 567 578 590 591

Absorption maxima, nm

Pi

520, 553 t t Ins§

526, 560 522, 558J 530, 568 529, 563Î

p 2

Ins Ins

527. 560 525. 560Î 530, 568 529, 563Î

* Et = Dimroth's solvent polarity values. t Àmax determined on the non-irradiated solution. Î Saturated solution (cone. < 0.1 g/1.). § Ins = insoluble.

12


of model and polymers deviate from a first order relationship ; the variation of the optical density with the time is given by

D = ae" + be~ (see Table 4)

Table 4. Decoloration rate constants Xl and X2 of M, Pl and P2 in acetone and tetrahydrofuran at 20°C

X, x IO2, min io2, min - 1

Photochrome Acetone Tetrahydrofuran Acetone Tetrahydrofuran

[M] = 3.3 x 10~4 mole/1. 89.0 ± 0.7 39.8 ± 0.4 17.36 ± 0.02 23.9 ± 0.5 P1( saturated soin* 21.9 ± 0.7 18.9 ± 0.1 3.32 ± 0.03 5.43 ± 0.07 P2, saturated soin* 28.6 ± 0.1 22.6 ± 0.2 4.02 ± 0.02 5.96 ± 0.60

* In acetone. <0.1 g/l.; in THF. ca. 0.2 g/f.

The model discolours much more rapidly than the polymer; and the existence of two isomers, which are not convertible into each other, must therefore be admitted.

A photochromic polylysine has also been prepared recently by reaction of polylysine with l-ß[3,3-dimethyl-6'-nitro-spirobenzopyran indolenino] pro-pionic acid in the presence of carbodiimide reagent (reaction scheme 6).

H3C CH3

(6)

Ν θ 2

H3C CH3

Its photochromism will be described in the near future and corresponds to that of polytyrosine.

i(c) Photochromic condensation polymers It was assumed that the incorporation of the photochromic groups into a

polymeric backbone should affect most strongly their photochromism by lack of segmental mobility in these highly aromatic systems.

13

G. SMETS

With this purpose in mind, several benzopyrylspiran dimers have been prepared by condensation of a bisindolenine with two molecules of substituted monosalicylaldehyde and inversely by condensation of a bissalicylaldehyde with two molecules of monoindolenines. Similarly the condensation of bis-indolenines with bissalicylaldehydes should give polycondensation products ; however, only oligomers were obtained on account of the low solubility of these polyaromatic compounds. The structures of these condensation products are given in Figures 7 and 8; the values of n varies from five to ten.

H 3 C^ / C H 3

Model spiropyran pol. 1

Pol. 1


(M2)

Figure 7

14


H3C CH


Pol. 3

H3C CH, \ / 3

CH,

(CH2) 2>2 J

O 2


W4)

Pol. 4

Figure 8

As a result the differences dimers/oligomers were much less pronounced, as could be expected. The rate of decoloration of the oligomers is nevertheless about half that of the corresponding dimers. Even in a solvent as polar as dimethylformamide their reaction kinetics deviate from a first order relationship, while the dimers behave linearly up to complete decoloration.

As to how far the existence of two or three isomers must be admitted in order to explain these kinetic deviations, is still questionable. A different behaviour of the photochrome endgroups compared to those inside the oligomeric chain could also be assumed.

In order to avoid these difficulties copolyesters have been prepared starting from a bismethylol photochrome, bisphenol-A and a mixture of iso-and terephthalylchloride following the reaction scheme 7.

15

G. SMETS

1 U L n î Cl—CO—C6H4—CO—0--CH2—(Phot)2—CH2—O—CO—C6H4—CO—Cl i HO—C6H4—C C6H4—OH

Me Me —O—C6H4—C--C6H4—O—CO—C6H4—CO—[O—CH2—

Me Me (Phot)2—CH2—O—CO—C6H4—CO]—

Bis-phenol polyester 9.5 w% 90.5%

This copolyester contained 9.5 per cent weight photochrome, i.e. one unit for twenty bisphenolphthalate units. In cyclohexanone solution its decoloration rate at 29°C must again be explained by two successive rate constants which are equal to kl = 30.6 χ 10-3sec-1and/c2 = 8.3 x 10" 3 sec-1, while for the model compound k is equal to 19.5 x 10" 3 sec- *.

On account of its polyester content, this copolyester can be cast easily into a film from its dichloromethane solution. When completely dry, these films no longer show photochromic behaviour ; only on heating above the glass transition (Tg ~ 194°C) do they become thermochromic. By contrast, swelling in the presence of dichloromethane vapours is already sufficient to restore their reversible photochromism and complete recovery.

(d) Photochromic copolymers in the solid state Some very interesting results have been obtained from comparison of the

photochromic copolymers with the corresponding mixture of their model substances.

(i) First of all, as had already been described by Gardlund20, it must be remembered that the rate of decoloration of a spirobenzopyran is 400 to 500 times smaller in a polymethylmethacrylate matrix than in the homologous solvent, i.e. methylpivalate. In a polymeric matrix, the existence of three isomers, each characterized by a different rate constant, must be admitted, while in the pivalate due to rapid interconversion of the isomers, the rate constant corresponds only to the slowest isomer.

16


(ii) The nature of the polymeric matrix plays a very important role, as can be seen from Figure 9. The rate of decoloration is much higher in polystyrene than in polymethylmethacrylate due to the negative solvatochromism. The activation parameters of the second and third decoloration steps for both matrices are the following :

Ea2 Ea> ASt AS?

polystyrene 16.3 17.6

-18.8 -14.6

PMMA 22.5 24.4

+ 1.8 + 3.9

10/ T [ °K ;

Figure 9. Influence of the polarity of the film

Not only the activation energies differ considerably, but the activation entropies have opposite signs in the two matrices. Similar results were found by Flannery13 with l,3,3-trimethyl-6'-nitroindolinospiropyran solution, the entropy of activation being positive in polar solvents (ethanol, acetone) and negative in apolar, non-hydrogen-bonding solvents.

(iii) When a photochrome is dissolved (five per cent) in a given polymeric matrix, the behaviour is dependent on the glass transition of the polymeric

17

G. SMETS

g» 2.0

1.0

5% in PMMA

5% in PI BMA

3.00 3.10 3.20 10VM°K

Figure 10. Influence of T(l on activation energy

substrate. This can be seen from the Figure 10. In polyisobutylmethacrylate (Tg ~ 45°C) the rate is two to three times higher than that in polymethyl-methacrylate, and the Arrhenius diagram is linear over the whole domain of temperature measurements. On the contrary, in polymethylmethacrylate (Tg ~ 124°C) a 40 per cent increase of rate of decoloration can be found at 56°C though the diagrams remain parallel. This acceleration corresponds to a secondary transition temperature of the polymer, at which the mobility inside the matrix increases. Several photochromes have been used in the same way for different polymers and confirm the reproducibility of the experiment and the validity of the conclusion. Thus a photochrome can be used for the detection of local motions in a polymer molecule, and can act as an indicator for the determination of secondary glass transition temperatures.

(iv) On the contrary, if the photochrome is bonded to the polymer chain, the behaviour is completely different (Figures 11 and 12). In this case, the Arrhenius plot of the decoloration reaction shows a marked kink at the glass

18


30r

25l·

en o

20l·

1 51-

Δ Copolymer CA [k3 )

o Mixture {k3)

1 0 J / n ° K ] Figure 11. Arrhenius plot of decoloration reaction

3.0r ChU ChL

.2 5r

2.0

3.0 31 3.2 103/7- [°K ]

Figure 12. Arrhenius plot: copolymer 6 (k3)

19

G. SMETS

transition of the copolymer, respectively at 61° and 53°C for the isobutyl-and n-propylmethacrylate copolymers. These Tg were controlled by differential scanning calorimetry. The activation parameters for the second and third decoloration steps can be summarized in the following Table 5.

Table 5. Activation parameters of decoloration kinetics of photochromic copolymers

E AS? Eai ASj

Isobutylmethacrylate

below Tg

12.6 -28.8

15.1 -25.5

copolymer

above Tg

26.4 + 12.4

32.3 + 25.8

tt-Propylmethacrylc

below Tg

15.4 -20.1

17.2 - 1 9

ite copolymer

above Tq

31.5 + 27.7

36.9 + 41

The increase of overall activation energy above Tg is due to the additional activation energy for viscous flow. Very striking is the change of activation entropy values, which vary from strongly negative to highly positive values, by passing from below to above the glass transition temperatures respectively.

(v) Another very stimulating and new phenomenon which was observed recently with spirobenzopyran copolymers is the photomechanical behaviour of a crosslinked copolymer (Figure 13) of a bismethacryldiphotochrome

Ο,Ν

CH,

CH 2 =C—CO—O—CH 2—<Phot)2—CH 2—O— Me w%

I 5 CO—C=CH 2

C H 2 = C H - C O O E t 95 H 2 C = C — C O — O

CH3

Figure 13

20


(five per cent) with ethylacrylate (95 per cent). It was found that on irradiation with ultra-violet light a dry thread of this polymer stretches with a relative elongation of about two per cent, and recontracts reversibly in the dark after a few minutes.

This phenomenon must be related to the observation described by Agolini and Gay21 in the case of aromatic imides containing an azo link ; these compounds show a reversible photo and thermal contractile behaviour that may be associated with cis-trans isomerization of the dicarboxy azo-benzene units. A similar observation was made recently also by Prins22 on ß-hydroxyethylmethacrylate methylmethacrylate copolymers containing some sulphonic azodyestuff.

PHOTORACEMIZATION OF OPTICALLY ACTIVE POLYMERS A phenomenon related to photochromism is the photoracemization of

optically active polymers. Such experiments were undertaken by Schulz and Jung23 who compared racemization kinetics between 100°C and 120°C of the polyvinyl and methyl/ethyl esters of the ( + ) 2-methyl-6-nitro-biphenyl-2'-carboxylic acid in dioxane solution.

o RO^H

R=CH3—. C2H5—. (—CH2—CH—)„

N0 2

While the low molecular ester shows a linear first order relationship on plotting (log α0/α, versus time), the polymer racemizes much more slowly ; after some twenty per cent conversion, the racemization proceeds apparently in a second step the rate of which is more than ten times slower than the first one.

Similarly in our laboratory, P. Hermans has prepared polyesters and copolyesters of bisphenol-A with optically active ( + ) 2-bromo-dibenzo (a.e) cyclooctatetraene-6,11-dicarboxylie acid (BCT), of which the interconversion of enantiomers becomes important at 120°C. The principle of the racemization is based on the interconversion of both cyclooctatetraene conformations24

(Figure 14). The racemization behaviour of (BCT) esters and polyesters is summarized

in Table 6. As can be seen from these data, the rates of racemization in dioxane solution are very similar for the diphenylester and for homo- and copolyester of bisphenol-A. In the solid state, in polystyrene and polycarbonate matrices, of which the glass transition temperatures are below the temperatures of measurement, the racemization is independent of the nature of the matrix. For the amorphous oligomer (Tg > 200°C) the rate is about twenty times slower than for the mixtures ; the influence of the glass transition on such phenomena is thus again taken in evidence. For both mixtures c, d and for the oligomer the racemization (log [α]0/[α]) proceeds

OHO

21

G. SMETS

HOOC HO OC

HOOC HOOC

Figure 14. Racemization of 2-bromo-dibenzo (a.e.) cyclo octatetraene-6.11-dicarboxylic acid

0.2

o 0.1

• Copolymer o Homopolymer

Temp. 180°C

50 100 150 200 t, min

Figure 15. Rate of racemization of BCT-polymers

22


Table 6. Racemization of 2-bromodibenzo-(a.e.)-cyclooctatetraene-6,11-dicarboxylic esters (BCT)

^ 1 5 0 X 104sec" £flact kcal/mole log ,4 AS* AF*

in dioxane solution

diphenyl

1 7.6 28.8 11.7

-7 .2 31

homo* polyester

6.3 27 10.7

-11.9 31

cot polyester

5.4 27.8 11

-10.5 31.2

Pstyrj: matrix

5.6 33.3 13.9

+ 2.5 31.4

in solid state (film)

Polycarbonate§ matrix

3.5 33.6 13.9 2.5

31.7

oligomer|| amorphous

0.15 52.1 22

+ 39.9 33.8

* Polycondensate of bisphenol-Awith BCT-diacid chloride; molecular weight M„ = 8 300 with Mechrolab vapour-pressure osmometry.

t Copolycondensate of bisphenol-A with BCT-diacid chloride (15 per cent) and COCl2; molecular weight M„ = 8 160 with Mechrolab vapour-pressure osmometry.

X 15 per cent weight model in polystyrene (Tg\ 82° to 105°C). § 15 per cent weight model in polybisphenol-A carbonate. TSln)t. 101° to 113°; 116° to 126° after racemization. II — [ C O — B C T - C O — 0 - C 6H 4- C C 6H 4— 0 ] 4— [ C O — 0 - C 6H 4 C C6H4 - 0 ] 3 composition; M„: 3500. / \ / \

Me Et Me Et

as expected for a first order reaction, linearly with time, and was observed up to 70 per cent conversion. By contrast, for the homopolyester a and copolyester b the racemization (Figure 15) curves level off rapidly and become very slow at conversions of 11 and 30 per cent respectively as was the case in Schulz's experiments.

CONCLUSION Polymeric photochromic systems compared to low molecular weight

homologous types show kinetic differences in their photochromic properties. These differences are especially noticeable in the solid state, and depend strongly on the glass transition temperature of the polymer. On the other hand, photochromic groups make it possible to determine secondary transition temperatures of the polymeric matrices, in which they are incorporated.

ACKNOWLEDGEMENT The author expresses his sincere thanks to Dr P. Vandewijer, Dr J. Verborgt,

Dr Hoefnagels, Dr Hiatt, drs G. Evens, drs F. De Blauwe and drs P. Hermans for their valued collaboration, as well as to the IRSIA, Belgium and Agfa-Gevaert, Antwerp, Belgium, for supporting these researches.

REFERENCES 1 J. J. Laverty and Z. G. Gardlund. Polvmer Letters, 7. 161 (1969). 2 H. Kamogawa, J.Polym. Sci. A-1, 9, 335 (1971). 3 L. S. Meri wether, E. C. Breitner and C. L. Sloan, J. Amer. Chem. Soc. 87. 4441 (1965). 4 L. S. Meriwether, E. C. Breitner and N. B. Colthup, J. Amer. Chem. Soc. 87. 4448 (1965). 5 R. Lovrien and J. C. B. Waddington, J. Amer. Chem. Soc. 86. 2315 (1964). 6 H. Kamogawa, M. Kato and H. Sugiyama, J. Polym. Sci. A-1. 6, 2967 (1968).

23

G. SMETS

7 D. Tabak and H. Morawetz, Macromolecules, 3, 403 (1970). 8 S. Dähne, Z. wiss. Photographie, 62, 205 (1968). 9 H. Kamogawa, J. Appi. Polym. Sci. 13, 1883 (1969).

10 G. Smets, Proceedings of the Hungarian Academy of Sciences-IUPAC Symposium. Budapest 1969, in press.

11 (a) R. Dessauer and J. Paris, Advanc. Photochem. Vol. I, pp 275-321. Interscience: New York (1963).

(b) P. Douzou and C. Wippler, J. Chim. Phys. 60, 1409 (1963). (c) W. Luck and H. Sand, Angew. Chem. Intern. Ed. 3, 570 (1964). (d) R. Exelby and R. Grinter, Chem. Rev. 65, 247 (1965). (e) E. Fischer, Fortsch. Chem. Forsch. 7, 605 (1967). (0 T. Bercovici, R. Heiligmann-Rim and E. Fischer, Molec. Photochem. 1. 23 (1969). (g) S. Dähne, Z. wiss. Photographie, 62, 189 (1968). (h) R. Lovrien, American Chemical Society Division of Polymer Chemistry Preprints, 4. 715

(1963). 12 (a) K. Dimroth, C. Reichardt, T. Siepmann and F. Bohlmann, Ann. Chem. 661. 1 (1963).

(b) L. G. S. Brooker, G. H. Keyes and D. W. Heseltine, J. Amer. Chem. Soc. 73. 5350 (1951). (c) A. I. Kiprianov and W. J. Petrukin, Zh. Obshch. Khim. 10, 613 (1940). (d) A. I. Kiprianov and E. S. Timoshenko, Zh. Obshch. Khim. 17, 1468 (1947).

13 J. B. Flannery Jr, J. Amer. Chem. Soc. 90, 5660 (1968). 14 A. Hinnen, C. Audic and R. Gautron, Bull. Soc. Chim. France, 3190 (1968). 15 J. Hoefnagels, Ph.D. thesis, Louvain University (1969). 16 P. H. Vandewijer, J. Hoefnagels and G. Smets, Tetrahedron, 25, 3251 (1969). 17 P. H. Vandewijer and G. Smets, J. Polym. Sci. C 22, 231 (1968). 18 (a) P. H. Vandewijer and G. Smets, J. Polym. Sci. A-1, 8, 2361 (1970).

(b) G. Smets and P. H. Vandewijer, American Chemical Society Division of Polymer Chemistry Preprints, 9, 211 (1968).

19 (a) Y. Hirshberg and E. Fischer, J. Chem. Soc. 297, 3129 (1954). (b) R. Heiligmann-Rim, Y. Hirshberg and E. Fischer, J. Chem. Soc. 156 (1961);

R. Heiligmann-Rim, Y. Hirshberg and E. Fischer, J. Phys. Chem. 66, 2465 and 2470 (1962). (c) J. C. Metras, M. Mosse and C. Wippler, J. Chim. Phys. 62, 660 (1965). (d) J. Arnaud, C. Wippler and F. B. d'Angeres, J. Chim. Phys. 64, 1165 (1967). (e) C. I. Lashkow and A. V. Shablya, Opt. i. Spektr. 19, 455 and 821 (1965). (0 C. Balny, P. Douzou, T. Bercovici and E. Fischer, Molec. Photochem. 1. 225 (1969).

20 Z. G. Gardlund, Polymer Letters, 6, 57 (1968); 7, 719 (1969). 21 F. Agolini and F. P. Gay, Macromolecules, 3, 349 (1970). 22 W. Prins, private communication. 2 3 R. C. Schulz and R. H. Jung, Makromol. Chem. 96, 295 (1966). 2 4 K. Mislow and H. D. Perlmutter, J. Amer. Chem. Soc. 84, 3591 (1962).

24

NETWORK FORMATION IN POLYMERIC MEDIA AND SOME NETWORK PROPERTIES

J. ASHWORTH, C. H. BAMFORD and E. G. SMITH

Department of Inorganic, Physical and Industrial Chemistry, Donnan Laboratories, University of Liverpool, UK

ABSTRACT The following network copolymers have been synthesized and examined: PVTCA-PMMA, PVTCA-PSt, PVBA-PMMA, PVTCA-PCP (PVTCA-polyvinyl trichloroacetate, PMMA-polymethyl methacrylate, PSt-polysty-rene, PVBA-polyvinyl bromacetate, PCP-polychloroprene). The synthetic routes utilized the radical-forming reactions between organic halides and organometallic derivatives [Mn2(CO)10, photochemical initiation (λ = 435.8 nm) at 25°C, Mo(CO)6, thermal initiation at 80°C] ; network properties were investigated by dilatometry, n.m.r., torsion pendulum and electron microscopy.

The networks show two-phase behaviour. In the case of PVTCA-PMMA systems one phase consists of an intimate mixture of PVTCA and PMMA chains, while with PVTCA-PSt phase separation into the pure components is more nearly complete.

It is concluded that in the PVTCA-PCP copolymers the size and separation of polychloroprene domains depend on the mean chain length of the PCP segments and the density of crosslinking, respectively.

The two-phase structure of PVBA-PMMA copolymers may be demonstrated directly by electron microscopy without recourse to staining.

The morphology of these networks appears to be determined by the opposition between thermodynamic incompatibility of the component homopolymers

and the geometrical constraints arising from the network structure.

1. INTRODUCTION

The properties and morphologies of multicomponent polymer systems are currently of considerable interest, and block copolymers based on styrene and butadiene of the AB and ABA types have attracted particular attention1. In this paper we shall be concerned with the synthesis and properties of more complex copolymers, namely networks composed of blocks of different molecular species.

Investigation of the relations between structure and properties requires ideally substances of which the structures are fully determinable—a situation which, unfortunately, is never encountered with synthetic high polymers.

Anionic polymerization has been widely employed in the synthesis of AB and ABA block copolymers, since the relatively narrow (ideally, Poisson) molecular weight distribution obtainable by this technique in favourable

25

J. ASHWORTH, C. H. BAMFORD AND E. G. SMITH

circumstances allows unusually precise control of block length. However, its application is limited, being confined, with few exceptions, to hydrocarbon monomers. On the other hand, free-radical polymerization offers choice from a wider range of components, but suffers from the disadvantage of producing broader distributions. Nevertheless, we believe that interesting correlations between structure and properties can be established by studies of copolymers prepared by free-radical reactions, and we have recently been exploring the possibilities of this approach.

For present purposes, a synthetic route to block and graft copolymers and networks is only of value if it satisfies the following minimal requirements : (i) it must not produce significant quantities of homopolymers, (ii) it should allow calculation of average block lengths and (iii) it should permit calculation of average crosslink or graft densities. In free-radical synthesis of block and network copolymers, a radical attached to a polymer molecule of type A must at some stage propagate with another type of monomer to generate ultimately a block of B, or to form a junction point with a block of B, or it must combine with a radical attached directly or remotely to a polymer chain of type B. The basic problem is therefore to produce 'attached' radicals under conditions which minimize undesirable combination and transfer processes leading to homopolymers. Polymerizing systems can be envisaged which approach this situation when conventional azo or peroxide initiators are used ; the crosslinking of dienes by free-radical propagation through their double bonds, or grafting on to preformed polymer by chain transfer fall into this category, provided the relevant mean kinetic chain lengths are sufficiently large. Such systems are special cases, however, and the most general free-radical syntheses of the types of polymer under discussion involve direct production of radicals attached to prepolymer chains in the initiation step. Several methods of doing this have been used, e.g. homolysis of suitable linkages, such as peroxy, incorporated in the backbones or mechanical rupture of the backbones themselves, and redox reactions of hydroperoxide groups attached to polymer chains. Redox reactions are particularly suitable for our purposes since they often give no unattached radicals.

We have utilized the radical-forming reactions between organic halides and complexes of metals in low oxidation states2 in block copolymer and network syntheses3-6, and summarize some of our results in the present paper.

2. NETWORK SYNTHESIS AND CHARACTERIZATION The chemistry of the radical-generating processes has been described

elsewhere2 and will not be reproduced in detail here. A wide variety of organo-metallic derivatives has been studied, but most of the preparative work has been carried out with molybdenum carbonyl Mo(CO)6, which shows convenient activity at 80°C, and manganese carbonyl Mn2(CO)10, which is an effective photoinitiator (λ = 435.8 nm) at 25°C. Radical formation is a redox process, following, for example, equations la, b, which, however, do not represent reaction mechanisms.

Mo° + Cl—R - Mo1—Cl + R· (a) Mn2(CO)10 + Cl—R ^ Mn'iCO^Cl + R· + iMn2(CO)10 (b)

26

NETWORK SYNTHESIS AND PROPERTIES

Further oxidation of Mo1 up to Mov may occur, with additional radical formation7. When R is part of a polymer chain, initiation leads directly and entirely to attached radicals. Thus, if the reaction is carried out in the presence of a polymerizable monomer M, block formation occurs, and if the termination reaction in the polymerization of M involves radical combination, a crosslink between two polymer chains is formed, as shown in 2.

ecu Μ°Γ6 ) -ca,M

(2)

C C I 2 M A A / W V MCI 2c-

Clearly, if the prepolymer chains carry a number of halide groups, network formation is possible3, while if the halide groups are situated only at the ends of prepolymer molecules the product is a block copolymer. The prepolymer chains will carry branches as well as crosslinks if the termination reaction occurs partly by disproportionation. Addition of a chain transfer agent, or a monomeric halide such as ethyl trichloracetate, will lead to the formation of unattached radicals, and hence ultimately to branches on the prepolymer. Some homopolymer of M will also be produced in these circumstances.

This type of synthesis is versatile ; the only restriction on the prepolymer is that it should be soluble and carry the required halide groups, and M may be any monomer polymerizable by free-radicals.

The initial rate of initiation /0 may be derived from observed rates of polymerization of the vinyl monomer under comparable conditions, except that a chemically equivalent monomeric halide is substituted for the prepolymer. This procedure eliminates effects of variation in the termination coefficient arising from gelation or attachment of growing chains to prepolymer molecules. There is considerable evidence6 that 10 does not depend on the nature of M. Mean crosslink lengths may be calculated from 70, [M] and the familiar kinetic parameter kpk~*. When the conditions are such that kt is not constant (e.g. at high crosslink density) it is simplest to estimate the mean crosslink length from the conversion of M and the total number of initiating radicals generated. The crosslink density is calculable from the gel-time of the system and the reaction time, since at the gel point the crosslink density corresponds to one crosslinked unit per weight average prepolymer chain. In simple cases, branch: crosslink ratios may be obtained from a knowledge of the relative importance of disproportionation and combination in the termination reaction (i.e. ktc/ktd)', when chain transfer is significant, or when unattached radicals are generated the calculation is more cumbersome, although conventional. Allowance must be made for initiator consumption during reaction, if this is significant. The relevant procedure has been described in earlier papers4'5; unfortunately, available data do not always allow corrections for consumption to be made with certainty, so that uncertainties in some network parameters are inevitable at present (cf. Table 1).

27


3. PROPERTIES OF NETWORKS

Most of our work has been carried out with polyvinyl trichloracetate (PVTCA) as prepolymer; other prepolymers studied include polyvinyl bromacetate (PVBA), the polycarbonate (I) and cellulose acetate containing a proportion of trichloracetate groups.

(I)

Methyl methacrylate (MMA), styrene (St) and chloroprene (CP) have been used as vinyl monomers.

3.1 PVTCA-PMMA and PVTCA PSt networks

3.1.1 Dilatometrie studies8

Nine networks were prepared by the methods outlined. After polymerization, excess monomer was removed under vacuum for 15 h, approximately.

I 1 1 1 1 L _ 20 40 60 80 100 120

T,°C

Figure 1. Dilatometrie data for networks IV, V, VI {Table 1) showing the presence of two glass-transition temperatures (except for VI). Q represent experimental data ; φ represent experimental data for network V after holding the specimen at the indicated temperatures for several hours.

Curves below TgX and above Tg2 calculated from equation 3.

28


Polymers were annealed at 100°C in the dilatometer bulbs in high vacuum, before the confining liquid, mercury, was added. Differential dilatometry was used, to eliminate errors arising from temperature fluctuations in the thermostat. Some preparative details and structural characteristics of the networks are given in Table 1. Dilatometrie observations showed the existence of the

Table 1. Network preparation and structures (PVTCA-PMMA, PVTCA-PSt systems)

Specimen

I II

III IV v

VI VII

VIII IX

Vinyl monomer

MMA MMA MMA MMA MMA MMA St St St

Mean units

no. of crosslinked per weight average

prepolymer chain

3 9

16-18 1 3-5 7 1 3 6

Pn crosslinks

4560-4740 5060-5440 6900-8000

11820 12700 16760 2880 4750 5800

Branch : crosslink ratio

4 4 4

10 10 10

~ 0 ~ 0 ~ 0

I—III, VII-IX polymerized at 25°C; Mn2(CO)10 as photoinitiator (A = 435.8 nm); IV-VI polymerized at 80°C; thermal initiation by Mo(CO)6 ; PW(PVTCA) = 2600.

two glass transitions (temperatures TgU Tg2), except for network VI, which contained only 3 % w/w of PVTCA. Figure 1 illustrates some typical results, as well as those with network VI. Under our conditions, therefore, all the networks, with the possible exception of VI, exist as two-phase systems.

Analysis of the slopes in volume/temperature diagrams such as Figure 1 provides results of interest. We define Su S3 as the slopes below Tgl and above Tg2, respectively, and S2 as that in the intermediate region. If the coefficients of cubical expansion are additive we have :

S l = α·ΡΡΚΒΦ + agp»*gp» + «HgAPHg

S3 = «lpp^lpp + «lpm^lpm + « H g A ^ (3)

in which the as are the coefficients of cubical expansion and the Vs the volumes of the components at the appropriate temperature, and in the appropriate states, which are denoted by g (below glass transition) and 1 (above glass transition), respectively. Subscripts pp, pm refer to PVTCA and PMMA, respectively and AVHg is the difference between the volumes of mercury in the two dilatometers. Values of Si calculated from equations 3, with the aid of the measured coefficients of expansion, agree very closely with those observed. This is illustrated in Figure /, in which lines with the calculated slopes are superimposed on the experimental points. In the temperature region just above Tg2 the calculated values of S3 also appear to be consistent with the experimental points, although there is some divergence above 115°C, especially with polymers relatively rich in PVTCA. This is attributable to thermal decomposition of the prepolymer, which has been shown in independent experiments to set in above 115°C.

Values of S2 have been calculated from equation 4, analogous to 3, and are compared with the observed slopes in Table 2.

29


+ agpm gpm + aHgAKHg (4) It will be seen that for the PVTCA-PMMA networks the observed slopes are appreciably higher than those calculated (except for network VI). The assumption of independent behaviour of the two network components is therefore invalid in this region. We believe this implies that the two phases do not consist of the pure components, and that the phase relatively rich in PVTCA (phase 1) also contains PMMA which partakes in the transition at Tgl. We assume then that phase 1 contains all the PVTCA together with a fraction a of the PMMA in the network, while phase 2 consists of pure PMMA. This assumption leads to equation 5.

S2 = ^ l p p ^ l p p ~^~ ^ l p m ^ l p m

+ (1 - a) agpmFgpm + aHgAFHg (5) Values of a may be calculated from equation 5 by equating S2 with the observed slope and are shown in Table 2. The composition of phase 1 may readily be deduced from a and the overall composition of the network ; we present in Table 2 values of W, the weight of PMMA associated with unit weight of PVTCA in phase 1. It appears that this phase contains large proportions of PMMA, especially when the content of PMMA in the network is high. If phase 2 is not pure PMMA as assumed, but contains also some PVTCA which does not take part in the transition at TgU our calculations give underestimates of W.

The values of Tgi in networks I-V are perceptibly higher than Tgl for pure PVTCA (Table 2), and therefore support the view that phase 1 contains PMMA. On the other hand, Tg2 is always close to the value for PMMA, as would be expected if phase 2 consists wholly of this polymer.

A blend of PVTCA and PMMA (52:48 w/w) behaves differently from the networks. It shows two transitions located at temperatures indistinguishable from the 7 s of the pure components, and the observed slope 52 agrees with that calculated from equation 4 to within four per cent. Separation of the components is therefore much more nearly complete in this case.

The PVTCA-PSt networks VIMX also show two glass transitions. As

Table 2. Network properties (PVTCA-PMMA, PVTCA-PSt systems)

Sample

I II

III IV V

VI VII

VIII IX

PVTCA PSt PMMA

10452 obs.

3.75 5.23 3.83 4.78 5.06 3.41 3.03 3.64 4.00

(mlK"1) cale.

2.64 4.66 3.57 2.34 3.42 3.51 2.89 3.52 3.85

a

0-326 0.15 0.071 0.87 0.37 —

0.57 0.083 0.061

W

1.2 1.7 2.1 2.8 4.4 —

0.17 0.12 0.22

PVTCA in network (%w/w)

21.5 8 3.5

21 7.5 3

77 40.5 21.5

r9rc 74 71 69 74 71 — 55 60 60 59

7,2 °C

97.5 102 101 99

102 103 97 97

100

100 103

30


with the other networks, the observed slopes Sl5 S3 agree with those calculated from equation 3 with the appropriate parameters for polystyrene (agps, Vps etc.). Values of S2 similarly estimated from equation 4 are again

-150 -100 -50 0 50 Temperature ;°C

100 150

Temperature,°C Figure 2. Dependence of spin-lattice relaxation time Tx on temperature at 45 MHz. (a) Δ PVTCA homopolymer. OPMMA homopolymer; (b) O PMMA homopolymer, ■ PMMA/PVTCA network (88% w/w PMMA), D PMMA/PVTCA network (83% w/w PMMA), #

PMMA/PVTCA network (50% w/w PMMA).

31


higher than those observed (Table 2), but the differences are much smaller than with the PVTCA-PMMA networks. Table 2 presents the derived values of a and W, which are also relatively low, showing that only small amounts of polystyrene are incorporated into phase 1. In agreement with this, we find that Tgl, Tg2 for the networks are close to the transition temperatures of the pure components.

3.1.2 N.m.r. studies Pulsed n.m.r. experiments on the PVTCA-PMMA system have been

carried out to measure the relaxation times Tx (spin-lattice) and T2 (spin-spin). Figures 2 (a, b) show the dependence of Τγ on temperature for PVTCA, PMMA and networks of PVTCA-PMMA. The rapid decrease in Tx for PVTCA at temperatures exceeding 60°C, approximately (Figure 2a) is associated with the glass transition region, and demonstrates the sensitivity of the n.m.r. techniques to changes in molecular motion. At much higher temperatures, 7\ would be expected to pass through a minimum as the characteristic frequency of the motion passes through the experimental n.m.r. frequency (cf. Slichter9). A similar onset of chain motion occurs with the PMMA sample (Figure 2a) when the temperature reaches 100° to 110°C; in this case, however, the overall behaviour of Tx is more complicated, and relaxation below this temperature region is mainly determined by a spin diffusion process through the rotating methyl groups.

Corresponding increases in T2 for PVTCA and PMMA occur at 60°C and 110°C, respectively.

The variation of 7Ì with temperature for the networks (Figure 2b) resembles that for pure PMMA, but the onset of mobility occurs at temperatures appreciably lower than that for MMA homopolymer. Since PVTCA protons make a negligible contribution to the n.m.r. signals, and hence to the measured Tt in these networks, it follows that the decrease in Tv arises from the onset of motion in the PMMA component. Thus the presence of PVTCA has induced motion in some PMMA chains at temperatures below the Tg of polymethyl methacrylate. Clearly this conclusion is in accord with dilatometrie evidence indicating the existence of a composite phase 1 with a relatively low Tg. Figure 2b shows that the turning points in 7\ occur at successively higher temperatures in the network with increasing PMMA content. This also is consistent with the dilatometrie results since, according to Table 2, the weight of PMMA in phase 1 of a given type of network (e.g. I, II, III) decreases with increasing PMMA content.

3.1.3 Mechanical properties A few determinations are available of the glass transition temperatures

derived from torsion pendulum measurements of mechanical loss tangents. The technique is not suitable for measuring both Tg values in these systems. Two PVTCA-PMMA networks and one PVTCA-PSt network were prepared, with structures summarized in Table 3. The glass-transition temperatures are shown in Table 3, together with values for the homopolymers and blends of the latter. Tg is higher for the PVTCA-PMMA networks than for the blend ; the latter has a Tg identical with that of PVTCA. According to Table 3, the glass transition of the network containing 7 % w/w PVTCA

32

N E T W O R K SYNTHESIS A N D PROPERTIES

Table 3. Properties of networks and blends (PVTCA-PMMA, PVTCA-PSt systems): glass transition temperatures from torsion pendulum observations

PVTCA Mean no. of Pn Branch: in sample, crosslinked crosslinks crosslink Tg

CC % w/w units per ratio

weight-average PVTCA chain

PVTCA P M M A P V T C A - P M M A network P V T C A - P M M A network P V T C A - P M M A blend PVTCA-PSt network PVTCA-PSt blend

100 0 7

34 33 25 26

— — 11 3

— 8

—

— —

2850 1690 —

1400 —

— —

4 4

— ~ 0 —

60 (0.35 Hz) 105 (0.25 Hz) 76 (0.54 Hz) 73 (0.28 Hz) 60 (0.20 Hz) 60 (0.20 Hz) 60 (0.20 Hz)

is 16°C higher than the value for the blend. On the other hand, PVTCA-PSt networks and blends of the two homopolymers have similar values of Tg (60°C). These findings therefore reinforce the view that in our PVTCA-PMMA networks phase 1 contains PMMA in substantial amounts whereas in the PVTCA-PSt systems separation of the components is more nearly complete.

3.2 Networks containing polyvinyl bromacetate

Electron microscopy is capable of providing much direct information about the morphology of copolymers. Attainment of adequate contrast often necessitates use of a selective staining technique to create suitable variations of electron density in the specimen. Attachment of heavy atoms such as osmium and silver may be employed ; thus most morphological studies of copolymers containing polydienes have involved the osmium tetroxide fixation treatment first used for these systems by Kato10 .

The networks we have considered so far do not respond to attempted staining by osmium tetroxide, since they do not contain suitable acceptor sites. Although it might be possible to introduce these by performing chemical reactions on the copolymers we feel that such treatments could alter the structures and introduce other complications. Preparation of a prepolymer containing bromine such as polyvinyl bromacetate therefore appeared to be of interest ; the high reactivity of the bromine makes PVBA suitable for incorporation into networks by the technique described in § 2 and it seemed likely that domains of PVBA would be directly visible in the electron microscope by virtue of the high electron density associated with the bromine atoms.

In preliminary experiments carried out to investigate this aspect PVBA was crosslinked by PMMA; the gel time was 28 min, and specimens were prepared for examination after a reaction time of 10 min. These materials, which contained 60% w/w PMMA, therefore had a rather low crosslink

33

P.A.C—30/1—C


Figure 3. Electron micrographs of film (thickness - 1 000 nm) of PVBA PMMA network (60% w/w PMMA) cast from M MA monomer.

Figure 4. Electron micrograph of film (thickness ~~ 1000 nm) of PVBA-PMMA blend (76% w/w PMMA) cast from MM A monomer.

density, one crosslinked unit being present in three weight-average PVBA chains, approximately. A high proportion of H-shaped molecules, and simple grafts of PMMA on to PVBA, would therefore be present.

Figures 3 and 4 show electron micrographs of films of the copolymer and a blend of the homopolymers cast from methyl methacrylate solution. Figures 5 and 6 refer to specimens prepared by addition of PMMA and PVBA, respectively, to the crosslinked sample of Figure 3 before casting. It is clear from these results that phase-separation in these systems can be observed directly in the electron microscope, without recourse to staining,

34


Figure 5. Electron micrograph of a film (thickness ^ 1 000 nm) of a blend of PVBA-PMMA network (60% w/w PMMA) and PMMA (44% w/w of whole specimen) cast from MM A

monomer.

Figure 6. Electron micrograph of film (thickness ~ 1 000 nm) of blend of PVBA PMMA network (60% w/w PMMA) and PVBA (44% w/w of whole specimen) cast from MMA monomer.

the PVBA component being, in effect, 'self-staining'. Further investigation of networks containing this polymer is in progress.

3.3 Networks containing chloroprene11

One of the main objectives in our programme of work on networks is the investigation of materials containing components with widely differing properties in the same network. Combination of hard and rubbery polymers

35


is obviously an attractive possibility, particularly since such systems might provide information relevant to the fields of thermoplastic elastomers and impact-modified plastics. Neither butadiene nor isoprene is suitable for forming the rubbery segments, since these monomers polymerize only slowly in solution on account of high termination and low propagation velocity coefficients. Chloroprene (CP) appeared to be the most suitable monomer. Unfortunately, little quantitative information on the free-radical solution polymerization of chloroprene is available in the literature, and it was necessary to determine the kinetic parameter kpk~* and the incidence of combination in the termination reaction. These data are necessary for derivation of the structural characteristics of the networks as described in § 2. This aspect of the work will only be summarized briefly here.

Polymerization of chloroprene (80% v/v in benzene) at 25°C was photo-initiated (λ = 435.8 nm) by manganese carbonyl in the presence of ethyl trichloracetate, rates of initiation being determined by comparable experiments with methyl methacrylate (compare § 2). The rate of polymerization was found to be directly proportional to the monomer concentration and the square root of the rate of initiation, showing that the reaction follows conventional free-radical polymerization kinetics. At 25 °C, it was found that jkpfc-i = 0.012 ± 0.001 mol"* 1* s"*.

The nature of the termination reaction was investigated by the method devised by Bamford, Dyson and Eastmond6, depending on determination of gel times in systems photoinitiated by PVTCA + Mn2(CO)10. It turns out that under our conditions termination occurs exclusively by radical combination. Networks prepared with chloroprene by our technique will therefore be substantially free from branches.

Two networks containing chloroprene were prepared with PVTCA as prepolymer. Some preparative details and structural parameters are given in

Figure 7. Electron micrograph of film (thickness - 100 nm) of PCP-PVTCA network I {Table 4) stained by osmium tetroxide.

36


Table 4. Figure 7 is an electron micrograph of a thin film of network I (Table 4) stained by exposure to osmium tetroxide vapour, and shows a relatively regular distribution of polychloroprene domains. The film thickness is probably of the order of the domain size and less than the average inter-domain separation, so that Figure 7 is likely to represent a single layer of domains. Figure 8 is an electron micrograph of a thin section of network II

Figure 8. Electron micrograph of a section (thickness ~ 80 nm) of PCP-PVTCA network II {Table 4) stained by osmium tetroxide.

(Table4) cut from a specimen previously stained in aqueous osmium tetroxide. The distribution of domains appears less regular than that in Figure 7, but this may, in fact, be the result of damage during sectioning.

The mean interdomain separation in Figure 7, 200 nm, approximately, is greater than even the fully-extended length of an average PVTCA chain. It follows that the PCP domains cannot be connected by single PVTCA

Table 4. Network preparation and structure (PVTCA-PCP system)

Polymerization

Initial [Mn2(CO)10],(mol l"1) [PVTCA] (gl"1) [M] (mol 1_1) Observed gel time (min) Reaction time (min) 109/0(mol l ^ s " 1 ) 102 [ΔΜ] (mol r 1 ) % chloroprene in network (w/w) Mean no. of crosslinked units per

prepolymer chain Fn crosslinks

weight-average

I

5.14xl0" 4

10 8.16

44 25

7.5 1.29 (cale.)

10.4

0.57 2290

II

6.4 x 10"4

12.5 7.65 3-4

35 110 43 75.5

11.7-8.8 3720

Polymerization temperature: 25°C. Diluent-benzene.

37


molecules. The crosslink density in this polymer is low (Table 4) and the material must contain much unreacted PVTCA. Indeed, it is easy to show from the data in Table 3 that the probability of a given PVTCA unit reacting during the network preparation is only 2 x 10"4, approximately; consequently the majority of the copolymer molecules are H-shaped, with only minor amounts of more complex structures. We propose that the domains are formed by agglomeration of the PCP crosslinks of the polymer, with exclusion of most of the attached PVTCA, which therefore assumes the form of external branches and loops. The result is a PCP particle with attached PVTCA chains which confer stability when the particle is dispersed in a PVTCA matrix (consisting mainly of the unreacted polymer). From Figure 7 we see that the average domain diameter is of the order 150 nm. This is similar in magnitude to the unperturbed mean end-to-end distance for a polychloroprene molecule with Pn = 2290, so that most molecules could have their terminal groups on the surface of a domain, in agreement with our proposed model. A schematic drawing of the morphology is presented in Figure 9 a.

Figure 9. (a) Schematic representation of network I (Table 4). (b) Schematic representation of network II (Table 4. X, Y represent PCP, PVTCA chains trapped in the 'wrong' phases).

~ PCP chains; PVTCA chains.

The situation is naturally more complicated with network II {Table 4\ in which the crosslink density is much higher. Figure 8 suggests that the domain size is not very different (perhaps a little smaller), but the inter-domain separation is greatly reduced. We believe that the basic domain structure is similar to that in Figure 9a, but that the domains must now be linked together, predominantly by PVTCA (see Figure 9b), but probably by some polychloroprene in addition (X, Figure 9b). Geometrical constraints may necessitate the trapping of some PVTCA in PCP domains (Y, Figure 9b).

If these views are correct, it follows that the domain size in these systems should be mainly determined by the mean crosslink length, while the inter-domain separation should depend on the density of crosslinking.

Figure 10 shows the structure of a film cast from a blend of the homo-polymers; it resembles electron micrographs of many blends of incompatible polymers12. Comparison of Figures 7,8 and 10 shows that only in the blend are large PCP domains (~2000 nm) in evidence. Growth of a domain in

38


Figure 10. Electron micrograph of film (thickness 2000 nm) of blend of PVTCA and PCP (14% w/w) stained by osmium tetroxide.

the crosslinked polymer beyond the stage represented in Figure 9a would require further incorporation of PVTCA, and would therefore be thermo-dynamically unfavourable.

4. GENERAL

The evidence we have presented suggests that the existence of two phases in networks of the type described is of common occurrence. The factors which influence the morphology of these materials are naturally more complex than those operative in simple block copolymer systems. In the networks, natural incompatibility of polymers is opposed by geometrical constraints which prevent complete separation of the components in the absence of chain rupture. The opposition gives rise to the formation of phases containing both components intimately dispersed, which, at least in some systems (e.g. PVTCA-PMMA) behave like a pure component and exhibit a single glass transition. Such phases are less commonly encountered with mixtures of homopolymers and their appearance in networks may be attributed to 'enforced compatibility'. Their composition would be expected to be determined inter alia by the degree of incompatibility of the homo-polymers, the density of crosslinking and the overall composition of the system. Thus we believe that PVTCA and PMMA are less incompatible than PVTCA and PSt, and that this is one factor accounting for the differences in behaviour of the networks. The importance of differences in crosslink length and the presence of branches is not yet clear, and further work in this area is required.

The behaviour of the systems containing chloroprene indicates the way in which morphology may be influenced by the density of crosslinking, and we hope that further studies will reveal and elucidate a corresponding influence on mechanical properties.

39


REFERENCES 1 See e.g. P. R. Lewis and C. Price, Polymer, Lond. 12, 258 (1971). 2 C. H. Bamford, European Polymer Journal-Supplement, 1969, p 1 and references quoted

therein. 3 C. H. Bamford, R. W. Dyson and G. C. Eastmond, J. Polym. Sci. (C), 16, 2425 (1967). 4 C. H. Bamford, R. W. Dyson, G. C. Eastmond and D. Whittle, Polymer, Lond. 10, 759 (1969). 5 C. H. Bamford, R. W. Dyson and D. Whittle, Polymer, Lond. 10, 771 (1969). 6 C. H. Bamford, R. W. Dyson and G. C. Eastmond, Polymer, Lond. 10, 885 (1969). 7 C. H. Bamford, G. C. Eastmond and F. J. T. Fildes—in course of publication. 8 C. H. Bamford, G. C. Eastmond and D. Whittle, Polymer, Lond. 12, 247 (1971). 9 W. P. Slichter, J. Polym. Sci. (C), No. 14, 33 (1966).

10 K. J. Kato, Electron Microscopy, 14, (3), 219 (1965); Polymer Letters, 4, 35 (1966). 1 * J. Ashworth, C. H. Bamford and E. G. Smith—in course of publication. 12 M. Matsuo, C. Nozaki and Y. Iyo, Polym. Engng Sci. 9, No. 3, 197 (1969).

40

DARSTELLUNG, EIGENSCHAFTEN UND ANWENDUNGEN VON STABILEN POLYMEREN MIT

UNGEPAARTEN ELEKTRONEN DIETRICH BRAUN

Deutsches Kunststoff-Institut, D 6100 Darmstadt, West Germany

ABSTRACT There are three known types of stable polymers with unpaired electrons: neutral polyradicals, polyradical ions and polymeric charge transfer complexes. The preparation, properties and reactions of the most important stable polyradicals are described and their e.s.r. spectra are discussed. By spin-labelling of polymers conclusions on their mobility and the chain entanglement can be drawn. Polyradical ions can be used as initiators for graft copolymerizations.

I. EINLEITUNG Während neutrale Makromoleküle und geladene Polymere (Polyionen)

seit langem gut untersucht sind, ist über stabile makromolekulare Stoffe mit ungepaarten Elektronen bisher relativ wenig bekannt. Sogenannte Makroradikale1 mit normalerweise nur einer radikalischen Endgruppe spielen als aktive Zwischenstufen und Träger der kinetischen Ketten bei radikalischen Polymerisationen eine wichtige Rolle. Makroradikale mit einem oder wenigen ungepaarten Elektronen pro Molekül treten ferner bei der proliferierenden Polymerisation2, bei der Pyrolyse von Polymeren oder beim mechanischen Abbau von Makromolekülen auf, ebenso bei der Einwirkung energiereicher Strahlung auf Polymere. Derartige Radikale sind meist sehr kurzlebig. Nur in besonderen Fällen (z.B. unterhalb der Glastemperatur in festen Polymeren und unter Ausschluss von Radikalfängern wie Sauerstoff) sind sie für mehr oder weniger lange Zeit beständig ('trapped radicals') ; entstehende Radikale lassen sich dann zwar vielfach ESR-spektro-skopisch oder durch chemische Reaktionen nachweisen, doch kann ihre Zahl und ihre Lokalisierung im einzelnen Makromolekül nur selten genau angegeben werden.

Erst in neuerer Zeit gelang es, auf chemisch eindeutigen Wegen Makromoleküle herzustellen, die entlang der Polymerketten eine grosse Zahl von Gruppen mit ungepaarten Elektronen tragen, wobei die Struktur durch den Syntheseweg genau festgelegt ist. Für solche Polymere wurde in Analogie zu den Polyionen (mit vielen ionischen Gruppen pro Makromolekül) der Ausdruck Polyradikale vorgeschlagen3; der Begriff Makroradikal soll dagegen Polymeren mit endständigen ungepaarten Elektronen vorbehalten bleiben.

41

DIETRICH BRAUN

Makromolekulare Stoffe mit ungepaarten Elektronen lassen sich in drei Klassen einteilen :

(a) neutrale Polyradikale (b) Polyradikalionen (c) polymère Donator-Acceptorkomplexe mit Elektronenübergang und

Radikalionenbildung.

II. NEUTRALE POLYRADIKALE Praktisch alle bisher erhaltenen Polyradikale wurden durch chemische

Umsetzungen an Makromolekülen dargestellt3,4; die Synthese von stabilen Polyradikalen aus ungesättigten monomeren Radikalen ist bisher nur in wenigen Fällen möglich gewesen, da derartige Monomere im allgemeinen Inhibitoren radikalischer Polymerisationen sind und mit ionischen Initiatoren infolge Nebenreaktionen nicht oder nur schlecht polymerisiert werden können.

1. Polymere Hydrazyle Henglein und Boysen5 erhielten bei der y-Bestrahlung von Polymeren in

Gegenwart von Diphenylpicrylhydrazyl (DPPH) gelb gefärbte Produkte, die leicht oxydierbare NH-Bindungen besitzen. Sie schlössen daraus, dass sich die bei der Bestrahlung gebildeten Polyradikale nicht an das ß-Stickstoff-atom des DPPH, sondern an einen der Benzolkerne am α-Stickstofî anlagern, wie dies auch bei anderen Reaktionen des DPPH beobachtet wurde6. Über die Zahl der Radikalstellen, ihren Abstand und die sonstigen Eigenschaften liegen keine Angaben vor.

Das erste definierte Polyradikal überhaupt war das von Braun et al.1 auf folgendem Wege erhaltene Polymere :

—CH2—CH

42

STABILE POLYMERE MIT UNGEPAARTEN ELEKTRONEN

Das Polyhydrazyl oxydiert Hydrazobenzol zu Azobenzol und inhibiert die radikalische Styrolpolymerisation ; die Induktionszeit ist etwa proportional der Konzentration an Polyradikalen7.

Aus Polyvinylphenyl-diarylmethylchloriden oder Polyvinylphenyltetra-phenylcyclopentadienylchlorid erhält man in Tetrahydrofuran mit α,α-Diphenylhydrazin folgende Polyhydrazine8 :

- C H - C H , - CH— CU— -

C6H5

C6H,'

N - N ( C 6 H 5 ) 2

Q H 5

R1 = HundC^Hs R2 = H undC6H5

Sie lassen sich in benzolischer Lösung mit aktivem Bleidioxyd zu Poly-hydrazylen oxydieren, die in trockenem Petroläther ausgefällt werden können. Die gelbbraunen polymeren Hydrazyle sind in fester Form stabil und paramagnetisch. Bemerkenswerterweise ist die Lebensdauer dieser Polyradikale in Lösung grosser als die ihrer niedermolekularen Modelle. Im festen Zustand behalten sie jahrelang ihren radikalischen Charakter.

2. Polyradikale mit Verdazyl-Gruppen Kinoshita und Schulz9 stellten aus p- Vinylbenzaldehyd über das Formazan

l,5-Diphenyl-3-(p-viny1phenyl)-verdazyl dar.

CH =CH

Die Vinylverbindung lässt sich jedoch radikalisch nicht polymerisieren, da Verdazyle starke Radikalfanger s ind 1 0 ' n . Auch die radikalische Copolyme-risation mit Styrol gelingt nicht. Dagegen wurden Polymere mit Verdazyl-gruppen durch stufenweise Umsetzung von Homo- und Copolymeren des p-Vinylbenzaldehyds mit Phenylhydrazin, Phenyldiazoniumsalz und an-schliessende Methylierung mit Methyljodid, Dimethylsulfat oder Formaldehyd in Dimethylformamid erhalten9. Der Umsatz kann spektroskopisch aus

43

DIETRICH BRAUN der Extinktion bei 719 ιημ oder den ESR-Spektren bestimmt werden; danach enthält etwa jeder zweite Grundbaustein ein ungepaartes Elektron.

Oligomere mit Verdazyl- und Verdazyliumgruppen wurden durch Umsetzung von l,4-Bis[N,iV'-diphenylformazyl-(C)]-benzol mit 1,4-Dijodbutan, 4,4'-Dichlormethyldiphenyläther oder p-Xylylendibromid in Dimethyl-formamid mit BaO oder Ba(OH)2 als Katalysator als grüne pulvrige Produkte erhalten12.

3. Poly-Stickoxyd-Radikale Bisher wurden drei verschiedene Typen von Makromolekülen mit

Stickoxyd-Gruppen hergestellt. Griffith et al.13 synthetisierten aus Meth-acrylsäurechlorid und 2,2,6,6-Tetramethyl-4-piperidinol-l-oxyl in Pyridin bei 20 °C folgendes Monomere :

CH2:=C— C O - ( N—O ·

Bei dessen Polymerisation mit Phenylmagnesiumbromid in einem Gemisch aus Äther und Toluol bei 25 °C entstanden jedoch nur recht niedermolekulare orangefarbene Polymere (osmotisch bestimmtes Molekulargewicht 1000-2000).

Ein hochmolekulares Produkt mit dem gleichen Stickoxyd-Rest konnten sie durch Umsetzung des genannten Alkohols mit einem Copolymeren aus Maleinsäureanhydrid und Vinylmethyläther (1:1) (Molekulargewicht etwa 500000) in Gegenwart von Toluolsulfonsäure erhalten. In dem braunen Copolymeren hatte etwa die Hälfte der Anhydridgruppen unter Esterbildung reagiert.

Ein Polyradikal vom Diphenylstickoxyd-Typ wurde von Drefahl et al. dargestellt. Sie setzten ein aus Polystyrol erhaltenes Polymères mit etwa 24 Mol- % /7-Nitrosostyrolgrundbausteinen mit Phenylmagnesiumbromid zum polymeren Diarylhydroxylamin um, das mit Silberoxyd bei Zimmertemperatur in Tetrahydrofuran dehydriert werden kann :

Das rotbraune Polyradikal inhibiert die mit Azoisobutyronitril gestartete Styrolpolymerisation ; bei 50 °C in Benzol steigt die Induktionszeit fast linear mit der Polyradikal-Konzentration.

44


Ein anderes Polyradikal mit Diphenylstickoxydresten entsteht bei der Oxydation von Poly-iV-(4-diphenylamino)-acrylamid mit Bleidioxyd15 :

N—o

Das Polyradikal ist gegen Luft unempfindlich. Die Dehydrierung des Polyamins zu einem polymeren Diphenylstickstoff-Radikal war bisher nicht möglich15.

Schliesslich sei auf den Einbau von Stickoxyd-Radikal-Gruppen in Biopolymere hingewiesen160. So kann Poly-L-lysin an der ε-Aminogruppe mit einem Stickoxydderivat mit Isocyanatrest umgesetzt werden, wobei etwa 1.2 NO-Gruppen pro Polymermolekül (Molekulargewicht 50000) eingebaut werden :

— CH—Nf^-CO— CH— NH— CO-CH—

(CH2)4 I

NH2

(CH2)4 I

NH,

I (CH2)4 H

NH— CO-NH H,C H3C |

O

-H CH3

CH,

Eine ähnliche Markierung wurde auch bei Rinder-Serum-Albumin und anderen Biopolymeren erreicht160.

Calvin et al verwendeten Di- und Poly-Stickoxyde aus Polyiminen und Peptiden als eine geometrische Sonde zur Untersuchung der Konformation gelöster Makromoleküle durch 'spin labelling'. Die Wechselwirkung der mit ungepaarten Elektronen markierten Grundbausteine in den Polymerketten erlaubt aus den ESR-Spektren Schlüsse auf deren Gestalt, was möglicherweise zu einer neuen interessanten Methode zur Strukturanalyse werden kann160. Die Einführung der Stickoxydgruppe erfolgt durch Um-acylierung mit l-(l-Oxyl-2,2,5,5-Tetramethylpyrrolin-3-carboxyl)imidazol :

45

DIETRICH BRAUN

—CH,—CH — N H — CH,— CH— N —

CO

JC

-.σ^ I o

OH

—NH—CH—CO-

H N U

(CH2)4

NH2

-NH—CH—CO-I CH,

-NH—CH—CO—

(CH,)4

I NH —CO-

o-co-

TD< α Ύ o

T o

4. Polymere Kohlenstoffradikale Im Gegensatz zu den bisher in Polymerketten eingebauten Stickstoff-

Radikal-Gruppen sind alle bekannten stabilen Kohlenstoff-Radikale empfindlich gegen Luftsauerstoff, was ihre Herstellung und Untersuchung sehr erschwert. Die meisten der unter inerten Bedingungen relativ stabilen Triarylmethyle stehen ausserdem im Gleichgewicht mit ihren Dimeren, wodurch bei intermolekularer Kombination von Radikalstellen aus zwei verschiedenen Ketten Vernetzungen entstehen können.

Polytriarylmethylradikale lassen sich bisher nur aus Polyvinyltriaryl-methylchloriden durch polymeranaloge Umsetzungen darstellen17,18. Allerdings können Vinyltriarylmethylchloride nicht direkt polymerisiert oder copolymerisiert werden, da die Triarylmethylchloridgruppen stark kettenübertragend wirken19, wobei Tritylradikale entstehen, deren retardierender Einfluss auf Radikalpolymerisationen bekannt ist20. Man muss daher von Vinyltriarylcarbinolen21 ausgehen, die gut radikalisch homo- und copoly-merisierbar sind17. Die dabei entstehenden Polymeren lassen sich mit Acetylchlorid polymeranalog und nahezu vollständig in die entsprechenden Chloride überführen, die dann mit Zink oder Alkalimetallen Polyradikale bilden :

-CH—CH- -CH—CH —


Bei der Umsetzung mit Zink ist auf absoluten Wasserausschluss zu achten, da sonst mit der aus ZnCl2 und Wasser gebildeten Zinkaquosäure Poly-triarylmethylkationen an Stelle der Radikale entstehen19. Die Polyradikal-darstellung mit Zink verläuft wahrscheinlich über primär gebildete Hexa-aryläthanreste, die gemäss der Lage des Gleichgewichts erst langsam in Tritylradikale dissoziieren.

Beim Poly-4-vinyl-4\4''-diphenyltriphenylmethylradikal mit dem Polymerisationsgrad P = 350 konnten bei Zimmertemperatur bis 100 Radikale pro Molekül nachgewiesen werden, d.h., jeder dritte bis vierte Grundbaustein trägt eine Radikalstelle. Bei den Poly-4-vinyl-4'-phenyltriphenylmethyl-radikalen (P = 325) wurden etwa 40 Radikalstellen pro Makromolekül nachgewiesen, und bei den Poly-4-vinyltriphenylmethylradikalen (P = 300) waren maximal 10 Radikalstellen pro Makromolekül bei Zimmertemperatur vorhanden.

Die Stabilität der Poly-4-vinyltriarylmethylradikale ist, verglichen mit niedermolekularen Modellen, relativ gross. Während die als Modellsubstanzen herangezogenen 4-Isopropyltriarylmethylradikale beim Stehen am Tageslicht in zwei bis drei Wochen völlig zu diamagnetischen Verbindungen disproportionieren, konnte beispielsweise beim Poly-4-vinyl-4',4"-diphenyl-triphenylmethylradikal bei ungehindertem Lichtzutritt nach einem halben Jahr noch etwa die Hälfte der ursprünglichen Radikalkonzentration nachgewiesen werden.

Aus Untersuchungen von Marvel und Mitarbeitern ist bekannt, dass alle p- und o-alkylsubstituierten Triarylmethylradikale, die am α-C-Atom des Alkylrestes ein H-Atom tragen, rasch disproportionieren22 :

R — CH— R R — C — R

C 6 H 5 - C · Q H 5 - C

C6H5 C6H5

R — CH— R2

C 6 H 5 - C H I

C6H<

Bei der wesentlich langsameren Disproportionierung der Poly-4-vinyl-triarylmethylradikale entstehen gelb gefärbte Polymere mit Chinodimethan-Strukturen, die ähnliche Absorptionsspektren wie auf anderem Wege erhaltene Modellverbindungen18 besitzen ; sie entfärben Bromlösungen, wobei sie selbst ihre gelbe Farbe verlieren. Mit überschüssigem Alkalimetall entstehen aus Polyvinyltriarylmethylchlorid über die Radikalstufe Polyanionen, die tief gefärbt sind:

- C H , - C H — —CH,—CH—

Me

— CH,—CH—

Me

47

DIETRICH BRAUN

Polyvinyltriarylmethylchloride, in denen Ar = Biphenyl ist, können aus-serdem mit Alkalimetallen Biphenylradikalanionengruppen bilden, deren überraschend gut aufgelöste ESR-Spektren19 neun Liniengruppen aufweisen, die das sehr breite Spektrum der Polytriarylmethylradikale überlagern.

Bei der Reaktion von Poly-4-vinyltriarylmethylradikalen mit Sauerstoff erhält man je nach der Konzentration, in der das Polyradikal vorliegt, mehr oder weniger stark intra- und intermolekular vernetzte Polymere. Als Zwischenstufen werden auf Grund von kinetischen Untersuchungen an niedermolekularen Modellen23,24 Triarylmethylperoxy-Radikale formuliert, die jedoch nicht fassbar sind ; derartige äusserst reaktive Peroxy-Radikale lassen sich aber unter bestimmten Bedingungen ESR-spektroskopisch nachweisen.

Bei der ESR-spektroskopischen Verfolgung der Umsetzung von Poly-4-vinyltriarylmethylradikalen mit Sauerstoff sieht man, dass das breite Signal des Triarylmethylradikals verschwindet und statt dessen ein Signal mit wesentlich kleinerer Linienbreite erscheint. Dass man bei den Polymeren im Gegensatz zu niedermolekularen Triarylmethylverbindungen die entstehenden Peroxy-Radikale relativ gut spektroskopisch nachweisen kann, Hegt an der sterischen Behinderung der Peroxidbildung. Bei der Umsetzung der Polyradikale und auch der Polyanionen mit Sauerstoff entstehen aus einem Teil der Radikalstellen sofort Triarylmethylperoxid-Gruppen, die eine starke Vernetzung der Polymeren verursachen. In diesen vernetzten Polymeren sind weitere Triarylmethylperoxy-Radikale eingeschlossen und dadurch in ihrer Bewegung so behindert, dass sie nur sehr viel langsamer als die entsprechenden niedermolekularen Triarylmethylperoxy-Radikale weiterreagieren können. Damit zeigt sich, dass die äusserst reaktionsfähigen Peroxy-Radikale nicht nur in eingefrorenem Zustand in festen Polymeren ESR-spektroskopischen Untersuchungen zugänglich sind, sondern auch in vernetzten, gequollenen Gelen, wo zwar Rotationsbewegungen einzelner Gruppen in den Makromolekülen noch relativ wenig behindert sind, Translationsbewegungen von Kettensegmenten jedoch stark eingeschränkt sind.

Wegen der bereits erwähnten Gleichgewichte zwischen den Triarylmethyl-radikalen und ihren Dimeren (zur Struktur der Dimeren s. Lit. 25, 26) entstehen bei den in den Arylresten unsubstituierten Polyvinyltriarylmethyl-radikalen je nach Molekulargewicht und Radikalgehalt mehr oder weniger weitgehend vernetzte Polymere; nur bei hohen Verdünnungen lassen sich in einigen Fällen auch lösliche Polyradikale fassen. Es wurde daher versucht, solche polymère Tritylradikale zu erhalten, bei denen durch entsprechende sterische Hinderung auf Grund des Verhaltens ihrer niedermolekularen Modelle nur eine schwache oder keine Vernetzung durch Kombination von Radikalstellen zu erwarten war27. Die folgende Zusammenstellung zeigt verschiedene solche Polymere, die analog den oben genannten Polyradikalen aus den monomeren Carbinolen durch Polymerisation, Chlorierung und Umsetzung mit Metall gewonnen wurden.

Einen anderen Typ von z.T. recht stabilen Kohlenstoff-Radikalen erhält man aus Fluorenderivaten. Während z.B. das 9-Phenylfluorenylradikal bei Zimmertemperatur vollständig dimerisiert ist und erst bei 100 °C eine merkliche Dissoziation eintritt, bewirkt der Ersatz der Phenylgruppe durch den Mesitylrest eine vollständige Dissoziation28. Trotzdem sind bei dem aus

48


— C H - C H , — ■—CH-CH —· — C H - C H , —

0 OCH,

H,CO IQ) -tj " ^ CH3

Fluorenon und p-Vinylphenylmagnesiumchlorid nach der Polymerisation zugänglichen Polymeren nach der Umsetzung analog wie bei den Poly-tritylradikalen bereits bei Zimmertemperatur Radikale nachweisbar, was auf die bei den Polymeren stärkere sterische Hinderung der Kombination zurückzuführen ist27.

Stärker dissoziiert und damit noch weniger vernetzt sind die aus dem aus 1,3-Diphenylfluorenon und p-Vinylphenylmagnesiumchlorid erhältlichen Monomeren durch Polymerisation und entsprechende polymeranaloge Umsetzung gewonnenen Polyradikale :

Zu 100% dissoziiert und damit völlig un vernetzt zugänglich ist folgendes Polyradikal27 :

CH3

49

DIETRICH BRAUN

Stabile Polyradikale lassen sich ferner von den bereits lange bekannten29 '30

Pentaphenylcyclopentadienylradikalen ableiten. Infolge der vielen Meso-meriemöglichkeiten und der sterischen Hinderung sind die aus Tetraphenyl-cyclopentanon und p-Vinylphenylmagnesiumchlorid als Ausgangsprodukte zugänglichen Polyradikale selbst im Licht monatelang haltbar31.

—CH—CH2—

Ö C6H5J Lc6H5

5. Polymere Sauerstoff-Radikale Die durch Sauerstoffeinwirkung auf Polyolefine erhältlichen polymeren

Peroxyradikale wurden mehrfach untersucht (s. z.B. Lit. 32), und auch in Lignin und seinen Oxydationsprodukten auftretende freie Radikale konnten ESR-spektroskopisch nachgewiesen werden33, doch handelt es sich hier nicht um chemisch einheitlich aufgebaute Poly-Sauerstoff-Radikale. Dagegen sind verschiedene stabile niedermolekulare Oxyradikale bereits länger zugänglich ; besonders stabil sind die von E. Müller34 sowie von K. Dimroth35

untersuchten Aroxyle. Die ersten stabilen Polyphenoxyradikale wurden von Braun und Meier

aus Vinyl-2,6-di-tert-butylphenol und aus 4-Isopropenyl-2,6-di-tert-butyl-phenol erhalten36.

Beide Monomere sind wegen ihrer phenolischen Gruppen nicht radikalisch polymerisierbar, können aber mit kationischen Initiatoren homo- und copolymerisiert werden. Bei der Homopolymerisation der Vinylverbindung entstehen jedoch nur unlösliche vernetzte Polymere, was vermutlich auf Übertragungsreaktionen mit den phenolischen Hydroxylgruppen zurückzuführen ist. Dagegen sind lineare Copolymere, z.B. mit Styrol oder a-Methyl-styrol, mit nicht zu hohen Anteilen an Vinylphenol und bei niedrigen Polymerisationsumsätzen zugänglich. Aus der Isopropenylverbindung erhält man auch bei der Homopolymerisation unvernetzte lösliche Polymere. Die entsprechenden Polyphenoxyle entstehen durch Oxydation der Polymeren mit Bleidioxyd in benzolischer Lösung unter Stickstoff. Die Polyisopropenyl-phenoxyle sind unter inerten Bedingungen ebenso stabil wie Tri-tert-butyl-phenoxyl ; die aus der Vinylverbindung erhaltenen Polyradikale besitzen je

- C H 2 - C H - - C H 2 - C -

O O

50


nach dem Radikalgehalt Halbwertszeiten von einigen Stunden und sind damit stabiler als ihr niedermolekulares Modell, wo die auch beim Polymeren beobachtete Disproportionierung sehr viel schneller geht :

- C H — CH— —CH—CH CH,—C —

OH O

6. ESR-Spektren der Polyradikale Die ESR-Spektroskopie liefert den direkten Nachweis für das Vorliegen von Polyradikalen. Am Beispiel der Polyhydrazyle wurde zuerst gezeigt, dass die Auflösung der Spektren sehr stark von der Beweglichkeit der Polymerketten und von Wechselwirkungen zwischen den radikaltragenden Gruppen abhängt17. Es ist bekannt, dass bei höheren Radikalkonzentrationen eine Begrenzung der Auflösung durch die Spin-Spin-Wechselwirkung benachbarter Radikalstellen (exchange interaction) verursacht wird37. Im festen Zustand ist die Auflösung der Spektren daher i.a. gering; in vielen Fällen erhält man nur eine einzelne Linie, deren Breite sich kaum mit der Temperatur ändert13. Auch in Lösung ist die Auflösung bei den Polyradikalen i.a. schlechter als bei ihren niedermolekularen Modellen ; sie wird grosser, je geringer die Radikalkonzentration ist. Der gleiche Effekt tritt ein, wenn in Copolymeren der Radikalgehalt durch Änderung der Zusammensetzung verringert wird.

Da die Verringerung der Beweglichkeit eines Radikals ebenfalls eine Verbreiterung der Linien im ESR-Spektrum bewirkt38, steigt bei der Untersuchung von gelösten Polyradikalen i.a. die Auflösung mit Erhöhung der Messtemperatur7.

Die Knäuel Struktur der gelösten Makromoleküle bedingt eine hohe lokale Radikalkonzentration und Wechselwirkungen zwischen den unge-paarten Elektronen schon bei molaren Konzentrationen, bei denen bei niedermolekularen Radikalen noch keine gegenseitige Beeinflussung beobachtet wird. Steigt die Radikalkonzentration noch weiter an, so findet man asymmetrische Spektren, wie sie für polykristalline Proben von niedermolekularen Radikalen39, für eingefrorene Lösungen40 oder für Polyradikale in festem Zustand erhalten werden. Dies beruht auf der Asymmetrie des -Faktors und ist nicht auf eine Überlagerung der Spektren verschiedener Radikalsorten in der gleichen Probe zurückzuführen.

III. POLYRADIKALIONEN 1. Darstellung

Durch Addition von Alkalimetallen an Doppelbindungen können Radikalanionen sowie Dianionen entstehen :

Me + A = B -+ -A - Β θ | Me® 2 Me + A=B - Me® |ΑΘ - Β θ | Me®

51

α

DIETRICH BRAUN

Als A = B können hierbei u.a. folgende Gruppen dienen: C = C , C = 0 , G=N, N = N , N = N O . Durch Anlagerung von Alkalimetallen an entsprechende Makromoleküle mit derartigen Resten erhält man Polyradi-kalionen.

Aus Polyvinylbiphenyl und Polyvinylnaphthalin bilden sich farbige Polyradikalionen41. Aus Copolymeren mit p-Vinyl-trans-stilben-Grund-bausteinen42 und Natrium entstehen in Tetrahydrofuran braune Polyradikalionen ; bei weiterer Zugabe von Natrium bilden sich Polydianionen :

C H = C H

Na Na

-CH2—CH-

e C - C e

Dagegen bilden sich aus Copolymeren mit p-Vinyl-a,oc-diphenyl-grund-bausteinen mit Natrium sofort Dianionen, da die primär entstehenden Radikalionen hier sofort dimerisieren43 :

C = CH2 C H 2 = C I

C6H5 CeH,

2Na

e |C-CH 2 -CH 2 -C |©

C6H5 C6H5

Je nach Polymerkonzentration und -Zusammensetzung treten dabei bevorzugt inter- oder intramolekulare Verknüpfungen ein, was Rückschlüsse auf die Knäuelgestalt und -durchdringung in Lösung gestattet44.

Bei der Anlagerung von Alkalimetallen an die Carbonylgruppe in Poly-vinylbenzophenon (erhalten durch Polymerisation des Monomeren17 oder durch Benzoylierung von Polystyrol) entstehen polymère Metallketyle45'46.

Als Beispiel für Polyradikalionen aus Polymeren mit C=N-Bindungen seien Alkalimetalladdukte polymerer Schiffscher Basen (Poly-p-vinyl-benzophenonanil)47 genannt :

-CH2—CH—

Na

C = N—C6H5

C6H5

•C—N— C6H5 I θ

C6H5

52


2. Polymerisationsauslösung durch Polyradikalionen Radikalionen können die Polymerisation von ungesättigten Monomeren

entweder durch Elektronenübertragung oder durch Anlagerung starten :

A=B + -CH2-CH|© θ y i

•A-B| + C H 2 = Ç H ^ i R ·Α—Β—CH2— CH|0

R

Entscheidend ist die Orbital-Energie des Initiators; niedrige Orbital-Energie bewirkt eine hohe Elektronen-Affinität und damit eine geringere Neigung des Radikalanions zur Elektronenübertragung48. In beiden Fällen verläuft die nachfolgende Polymerisation nach anionischem Mechanismus ; eine radikalische Polymerisationsauslösung durch Radikalionen wurde bisher nicht beobachtet.

An niedermolekularen Mono- und Dinatriumaddukten von Verbindungen mit C=€, C = 0 , C=N, N = N und N=NO-Gruppen wurde untersucht, wieweit sie zur Polymerisationsauslösung von Acrylnitril, Methylmethacrylat und Styrol geeignet sind und ob der Start dabei durch Elektronenübertragung oder unter Ausbildung einer Bindung zwischen Initiator und Monomerem erfolgt49.

Polyradikalionen aus Polymeren mit C=C-Doppelbindungen (trans-Stilben-Resten) starten die Polymerisation der drei Monomeren im wesentlichen durch Elektronenübertragung und nicht über Pfropfung ; mit Styrol entstehen dabei 'lebende' Polymere42.

Rembaum et al erhielten aus Poly-4-vinylbiphenyl und aus Poly-2-vinylnaphthalin mit Cäsium in Tetrahydrofuran bei — 80 °C Polyradikalionen, die bei 0 °C die Polymerisation von Äthylenoxyd durch anionische Pfropfung auslösen41.

Makromolekulare Radikalanionen aus Polymeren mit p-Vinylbenzo-phenon- oder p-Vinylbenzophenonanilgrundbausteinen starten die Pfropf-copolymerisation von Acrylnitril und Methylmethacrylat47, während sie in Übereinstimmung mit dem Verhalten ihrer niedermolekularen Modelle49

die Styrolpolymerisation nicht auslösen. Vom Mononatriumaddukt des Copolymeren aus p-Vinyl-p'-dimethylaminoazobenzol und Styrol wird nur Acrylnitril unter Aufpfropfung polymerisiert47. In den genannten Fällen enthalten die Pfropfprodukte praktisch keine ungepfropften Rückgratpolymeren und keine freien Homopolymeren, wie durch Lösungsversuche gezeigt werden konnte.

Allerdings erfolgt die Metalladdition nicht gleichmässig an allen dazu befähigten Gruppen entlang der Polymerketten47. Vielmehr ist aus den Ergebnissen von Pfropfcopolymerisationsversuchen mit makromolekularen Mononatriumaddukten der Copolymeren aus Vinylbenzophenon sowie Vinylbenzophenonanil und Styrol zu schliessen, dass besonders bei höheren Metallgehalten neben Monometalladdukten und noch nicht mit Metall umgesetzten Grundbausteinen auch zweifach umgesetzte Gruppen vorkommen; dadurch enthalten die Rückgratpolymeren verschieden aktive Initiatorstellen.

53

DIETRICH BRAUN

Schliesslich startet schon bei den entsprechenden niedermolekularen Initiatoren nicht jedes Molekül tatsächlich eine Polymerkette49 ; deshalb muss man auch bei makromolekularen Initiatoren damit rechnen, dass bei Zugabe des aufzupfropfenden Monomeren nicht jede potentielle Startstelle wirklich mit einem Seitenzweig besetzt wird, zumal im Makromolekülknäuel diese Stellen sicher nicht immer frei zugänglich sind. Ausserdem ist zu erwarten, dass dadurch die Seitenzweige recht unterschiedlich lang werden. Aus diesen Gründen ist bisher bei anionischen Pfropfcopolymerisationen die Bildung sehr einheitlich aufgebauter Pfropfcopolymerer, an denen physikalische Messeungen sinnvoll wären, nicht möglich gewesen.

III. POLYMERE ELEKTRONEN-DONATOR-ACCEPTOR-KOMPLEXE

Aus Molekülen mit niedriger Ionisierungsenergie, sogenannten Elektronendonatoren (D), und solchen mit hoher Elektronenaffinität, Elektronen-acceptoren (E), können Molekülverbindungen entstehen, die als EDA-Komplexe bezeichnet werden :

D + A ^ ( D . . . A < - - D + . . . A " · )

Im Grundzustand liegt dabei Mesomerie zwischen einer neutralen und einer ionischen Grenzstruktur vor, wobei erstere normalerweise überwiegt. In manchen Fällen, insbesondere bei Komplexen aus starken Donatoren und starken Acceptoren oder bei Anregung durch Bestrahlung, kann der Elektronenübergang so weitgehend sein, dass solche Komplexe teilweise auch im Grundzustand schon aus Radikalionenpaären bestehen und dann ESR-Signale zeigen.

Es gibt bisher nur relativ wenige Untersuchungen über hochmolekulare EDA-Komplexe mit weitgehendem Elektronenübergang. Beispiele sind Komplexe aus Poly-4-vinyl-4'-dimethylatninoazobenzol und 2,3-Dichlor-5,6-dicyanochinon (DDQ), die schon im Dunkeln bei Raumtemperatur ein Einlinien-ESR-Spektrum liefern50. Genauere Untersuchungen liegen über Komplexe aus homopolymerem Dünethylaminostyrol bzw. Copolymeren mit Styrol vor51. Hier zeigen die Komplexe mit DDQ als Acceptor in Cyclohexanon schon im Dunkeln ËSR-Signale; durch Bestrahlen mit sichtbarem Licht kann auch bei Komplexen ìnit Chlòranil die Radikalionenkonzentration soweit gesteigert werden, dass ein ESR-Spektrum auftritt. Die Komplexe mit t)DQ in Cyclohexanon zeigen eine Hyperfeinstruktur (fünf äquidistante Linien mit den Intensitätsverhältnissen 1:2:3:2:1), was auf die Kopplung des Radikalkations mit den beiden Stickstoffkernen des DDQ-Radikalanions zurückzuführen ist.

In allen bisher bekannten Fällen sind die Radikalkonzentrationen sehr niedrig; je nach Art des Polymeren und des Acceptors sind weniger als 0,01 bis 0,1 Prozent der Komplexe in Radikale dissoziiert, was jedoch auch von der Solvatation der Koni^lexe durch das verwendete Lösungsmittel abhängt.

LITERATUR 1 S. E. Bresler und E. N. Kazbekov, Fortschr. Hochpolym.-Forschg. 3, 688 (1964). 2 G. H. Miller, D. Chock und E. P. Chock, J. Polym. Sei. A, 3, 3353 (1965).

54


3 D. Braun, J. Polym. Sci. C, 24, 7 (1968). 4 R. C. Schulz, Angew. Makromol. Chem. 4/5, 1 (1968). 5 A. Henglein und M. Boysen, Makromol. Chem. 20, 83 (1956). 6 A. Henglein, Makromol. Chem. 15, 188 (1955); Z. Naturforsch. 10b, 616 (1955). 7 D. Braun, I. Löflund und H. Fischer, J. Polym. Sci. 58, 667 (1962). 8 D. Braun und G. Peschk, Angew. Chem. 80, 1002 (1968). 9 M. Kinoshita und R. C. Schulz, Makromol. Chem. I l l , 137 (1968).

10 F. A. Neugebauer, M h. Chem. 97, 853 (1966). 11 M. Kinoshita und Y. Miura, Makromol. Chem. 124, 211 (1969). 12 Y. Kurusu, H. Yoshida et al. J. Chem. Soc. Japan, Industr. Chem. Sect. 72, 1402 (1969). 13 O. H. Griffith, J. F. W. Keana, S. Rottschaefer und T. A. Warlick, J. Amer. Chem. Soc. 89,

5072(1967). 14 G. Drefahl, H.-H. Hörhold und K. D. Hofmann, J. Prakt. Chem. 37, 91 und 137 (1968). 15 D. Braun und S. Hauge, Makromol. Chem. 150, 57 (1971). 16 (a) J. D. ingham, J. Macromol. Sci. C,l, 279 (1968). 16 (b) P. Ferruti, D. Gill, M. P. Klein, H. H. Wang, G. Entine und M. Calvin, J. Amer. Chem.

Soc. 92, 3704(1970). 17 D. Braun, W. Neumann und J. Faust, Makromol. Chem. 85, 143 (1965). 18 D. Braun und R. J. Faust, M h. Chem. 100, 968 (1968). 19 D. Braun und R. J. Faust, Makromol. Chem. 121, 205 (1969). 20 R. F. Mayo und R. A. Gregg, J. Amer. Chem. Soc. 70, 1284 (1948). 21 D. Braun, G. Arcache, R. J. Faust und W. Neumann, Makromol. Chem. 114, 51 (1968). 22 C. S. Marvel, M. B. Mueller, C. M. Himel und J. F. Kaplan, J. Amer. Chem. Soc. 61, 2771

(1939). 23 J. R. Thomas, J. Amer. Chem. Soc. 85, 591 (1963). 24 W. Schlenk und E. Markus, Ber. Dtsch. Chem. Ges. 47,1666 (1914). 25 H. Lankamp, W. Th. Nauta und C. MacLean, Tetrahedron Letters (London), 249 (1968). 26 H. A. Staab, H. Brettschneider und H. Brunner, Chem. Ber. 103, 1101 (1970). 27 D. Braun und W. Euler, unveröffentlichte Versuche. 28 K. Ziegler, Angew. Chem. 6Î, 168 (1949). 29 K. Ziegler und B. Schnell, Liebigs Ann. Chem. 445, 226 (1925). 30 E. Müller und J. Müller-Rodloff, Ber. Dtsch. Chem. Ges. 69, 665 (1936). 31 D. Braun, W. Euler, R. J. Faust und G. Peschk, unveröffentlichte Versuche. 32 J. C. W. Chien und C. R. Boss, J. Amer. Chem. Soc. 89, 571 (1967). 33 C. Steelink, 'Stable free radicals in lignin and lignin oxidation products', in J. Marion, ed.,

Lignin Structure and Reaction, Advances in Chemistry Series No. 59, p 51. American Chemical Society: Washington, D.C. (1966).

34 E. Müller und K. Ley, Chemiker-Ztg, 80, 618 (1956); C. D. Cook und R. C. Woodworth, J. Amer. Chem. Soc. 75, 6242 (1953).

35 K. Dimroth, F. Kalk und G. Neubauer, Chem. Ber. 90, 2057 (1957). 36 D. Braun und B. Meier, unveröffentlichte Versuche. 37 D. J. E. Ingram, Free Radicals as Studied by ESR, Butterworths: London (1958). 38 S. I. Weissman, J. Chem. Phys. 25, 890 (1956). 39 J. S Chalmers und Y. W. Kim, f. Chem. Phys. 44, 112 (1966). 4 0 P. D. Bartlett und J. M. McBride, IUPAC-Symposium 1966, Pure Appi. Chem. 15(1), 102

(1967). 4 1 A. Rembaum, J. Moacanin und E. Cuddihy, J. Polym. Sci. C, 4, 529 (1963). 42 D. Braun, F.-J. Quesada Lucas und W. Neumann, Makromol. Chem. 127, 253 (1969 4 3 D. Braun, M. H. Tio und W. Neumann, Makromol. Chem. 123, 29 (1969). 4 4 D. Braun und F.-J. Quesada Lucas, Makromol. Chem. 142, 313 (1971). 45 D. Braun und I. Löflund, Makromol. Chem. 53, 219 (1962). 4 6 G. Greber und E. Egle, Makromol. Chem. 54, 136 (1962). 47 D. Braun, W. Neumann und G Arcache, Makromol. Chem. 112, 97 (1968). 48 A. V. Tobolsky und D. B. Hartley, J. Amer. Chem. Soc. 84, 1391 (1962). 4 9 D. Braun und W. Neumann, Makromol. Chem. 92, 180 (1966). 50 D. Braun und G. Arcache, unveröffentlichte Versuche. 51 D. Braun und J. Sterzel, unveröffentlichte Versuche.

55

THE DECAY OF FREE RADICALS IN POLYMER MEDIA

P. JU. BUTIAGIN

Institute of Chemical Physics, Academy of Sciences of the USSR, Moscow, USSR

ABSTRACT Kinetic laws and particularities of migration mechanisms of free valence in the processes of decay of free radicals in polymers are dealt with. The basic examples correspond to the results for polyethylene and polymethyl methacry-late. The kinetics of the decay reactions are discussed from the viewpoint of Waite's cage diffusion theory and Levedev's model. From the kinetic data of various authors the cage radius and diffusion coefficient are calculated. With increasing temperature the cage radius in polyethylene increases from 5.3À (120°K) to 40-60 Â (360°K) and it has to be looked upon as an effective kinetic parameter. The diffusion coefficient corresponding to free valence displacement at long distances is in the range 10"16 to 10"18 cm2 s"1.

In amorphous polymers at temperatures exceeding the glass temperature diffusion proves to be of major importance for long-distance migration. In a series of other examples, chemical reactions play a more important role, such as hydrogen atom transfer and radical decomposition. Results of measurement of the corresponding elementary reaction compared with the rate constants of decay are mentioned. Under real conditions, migration is affected simul

taneously by both diffusion and chemical mechanisms.

The investigation of free radical decay proves to be a complicated problem closely associated with several fields of the chemistry and physics of polymers, reactivity theory and reaction kinetics in the solid state etc. The radicals are formed in chemical, irradiation, mechanical, photochemical and other effects on the polymer. Each of these processes provokes deep changes in the material. Therefore the overall complex of investigations of the laws of decay should consist of a great many phenomena :

The central problem of free radicals in polymers (their structure, conformation, intrinsic mobility, radical ractivity in fundamental kinds of reactions at low temperatures etc.);

The supermolecular structure of polymers (transition and relaxation effects, intrinsic mobility, diffusion and the like);

Radiation and photo chemistry (main types of defects and distortions, radical and ion-molecular reactions, distribution of distortion centres according to their distances and depths of 'traps', structural changes in polymers and so on);

57

P. JU. BUTIAGIN

Mechano-chemical and mechano-emission phenomena in polymers (diffusion, amorphization and recrystallization and so forth);

Kinetics of low-temperature reactions in the solid phase (diffusion, activation energy and 'compensation' effect, and oxidation).

From this incomplete array the present lecture will deal briefly with a few laws of kinetic decay and the migration mechanism of the free valence, only.

FORMAL KINETIC LAWS OF THE PROCESS The investigation of the kinetics of chemical reactions in solids and

viscous liquids1-4 starts from the cage model. To react with one another the particles should meet in a single cage, i.e. in one elementary structural volume of the solid. The rate of the bimolecular reaction is determined by the correlation between the following parameters: r0—radius of the elementary cell-cage in cm ; D—diffusion coefficient of reacting particles, cm2/s; S = KJKD where Kr is the reaction probability of the particles in the cage, and KD is the probability of the particle leaving the cage to enter the reaction.

Waite1 '2 solved the differential equation of the rate of bimolecular reaction for the irregular motion of reacting particles and their random starting distribution in the volume of the system. The general form of the solution is fairly complicated except in extreme limiting cases where individual values may well be neglected.

Thus, at low diffusion rate and a high reaction probability when the formation of each pair of particles in the cage leads to their interaction, i.e. S 5> 1, the rate of the process is limited by diffusion :

W = - dC/dt = KD(1 +T±/f±)C2 (1)

where KD = 4nr0D is the diffusion rate constant at t ρ τ and τ = r^/nD the lifetime of particles in the cage.

This extreme limiting case is most interesting for analysis of the reaction kinetics of radical decay. The elementary recombination step proceeds here practically without activation energy and the reaction probability in the cage may well be presumed to be very high. For calculation purposes the mechanism of displacement of particles in the volume (diffusion or chemical migration) does not appear to be a decisive one.

Equation 1 is of second order according to concentration but the proportionality coefficient is time-dependent :

KD,t = 4nr0D[l + J(r20/nD) x l/Ji] (2)

The relatively high starting reaction rate is caused by the fact that at t <ζτ (observation time is comparable with that of transmission of particles from one cage into another), the particles which appeared in pairs due to random distribution, i.e. at distances less than r0, react with one another initially.

Another equation form of diffusion kinetics was obtained in quite a simple way by Lebedev :

W = - dC/dt = 2KD In (1 - C/r30)C (3)

At C -* l/rl the rate is relatively high because all the cages are occupied

58


by radicals. Provided that C <ζ 1/r , the expression 3 can be reduced to the ordinary second order equation.

In coordinates of the second order equation (related to the abscissa axis ί/τ, and related to the ordinate axis CJC\ the kinetic curve of the diffusion-controlled bimolecular reaction in agreement with Waite's and Lebedev's equation will have the form shown in Figure 1. The deviation from linearity at low values of t/τ appears to be a characteristic feature of reaction kinetics in the condensed phase. From the shape of the kinetic curve on this section the importance of the parameters r0 and D may be deduced.

^ 3

2

1

50 100 /, min

Figure 1. (a) Theoretical curve according to equation 3. (b) Dole's experimental data of allyl radical decay in polyethylene.

S >1

Figure 2. Theoretical curves according to Waite's equation.

Figure 2 illustrates the calculation method used. From the general form of the rate equation 1 it is seen that the proportionality coefficient KD t given b y KD,t = (dC/di)(l/C2) (4)

has to decrease in the course of reaction to approach the constant KD value.

59

P. JU. BUTIAGIN

1

10

8

2=6 0) t

1 J 2

IK 1 V x _ o 1 h \ ^ " ^ " ô " "

0" *>' U"·" "

Y ; T i i ? , ι f

♦ J 2 *

î X| Ί 0 1 2 3 4 5 6

Time,h Figure 3. Time-dependent rate constants as a function of time for allyl radical decay in poly

ethylene5.

The dependence of KD t(t) is shown in Figure 2. At tjx > 5 the proportionality coefficient becomes practically constant with the reaction evidently following a bimolecular law. In the linear section the value KD t has to be linearly dependent on i/y/t, whereas from equation 2 it follows that in the figure with coordinates KD t and 1/y/t (see Figure 4) the segment cut on the ordinate axis equals 4nr0D and the slope is 4y/nr^D. Thus, by a detailed analysis of the form of the starting section of kinetic curves the radius of the elementary cell as well as the effective diffusion coefficient may well be calculated.

The presumption that radical decay should follow the second order reaction is so persuasive that several authors acquainted with the cage model involuntarily divide the one kinetic curve into two sections ('fast' and 'slow'

a or

Polyethylene, 70 °C, 0.057 Mrad

KDft=KD[i+ V777)

rn=6.7 x 10"7cm, D = 1.2 x 10"16cm2/sec

1/ VUh-1/2] Figure 4. Calculation of the data of ref. 5 according to Waite's equation.

60


o -

6 1-υ100 <υ Q. ω ι_' «Λ 0»

c to "c. • —

>

Q^

-o -A

T

—^%^-D-Q.... „ „ \ V D· · B

^ V*-VV'D° \ ^—-û—A^^ >t -g

t > ^ ^ \ T \ 5k

• o / • - • s

î \ -,. ·, _ ^ I " i * -

100 150 200 250 emperature of heating, °K

(a)

% «

50

300

p

I

120

v" t i

^ s

\ \ \

\

\ \

ι I | | |

200 280 Temperature, °K

(b)

Figure 5. Decay curves of free radicals formed in polyethylene by γ-irradiation (a) and by mechanical degradation (b).

reactions) with their own rate constants. Auerbach5 measured the kinetics of allyl radical decay in polyethylene over the temperature range 70° to 135°C. Initial concentrations of radicals are not referred to in the paper, but knowing both the dose and radiation yield they may be evaluated in quite a revealing way. In Figure 3 the kinetic curve of decay is shown in coordinates according to the second order equation. The curve shows the change of coefficient KD t in the course of the reaction. The shape of the curves on both the figures corresponds to Waite's and Lebedev's equations. In Figure 4 the same data are plotted in the coordinates KD t and i/yjt ; from the position of the straight line the parameters r0, D and τ have been calculated. The results of the calculation are summarized in Table 1.

Table Li

Dose M rad

0.057 27.4

Parameters r0, D and τ of the decay reaction of allyl ra polyethylene at 70°C (evaluations according to ref. 5)

C0 x 10"1 8

c m - 3

0.3 8

Â

150 50

Ì 67 43

D cm 2 s _ 1

1 x 10~16

2 x 10~18

dicals in

τ s

130 2600

Thus, in polyethylene, the effective kinetic cage radius in the premelting region appears to be quite high and reaches 40 to 60 Â. At the same time, the initial concentration of radicals results in the average distance between them amounting to between 50 and 150 Â respectively, which is comparable with the cage radius. The comparison of the values / and r0 clearly illustrates the role of radical recombination in the cage.

Equation 1, on the basis of which the values of r0 have been calculated, corresponds to the limiting case when S > 1. Consequently, the effective cage

61

P. JU. BUTIAGIN

radius may be determined as some distance between the radicals where the probability of recombination is considerably higher when compared with the probability of their separation without any interaction. For this reason, the kinetic radius of the cage cannot always be identified with such parameters as the principal dimension of the elementary cell of a crystal or, for instance, the average distance between adjacent chains.

PECULIARITIES OF VARIOUS METHODS OF RADICAL GENERATION

From the viewpoint of the cage model the kinetic analysis of decay of radicals exhibits some peculiarities. In real systems the requirement of a random particle distribution in the volume of the system is not always satisfied and the actual or local radical concentration frequently does not correspond to the average concentration measured, for example, by integration of the line of the e.s.r. absorption spectrum.

The character of particle distribution in the volume depends on the method of radical generation and on the supermolecular structure of the polymer. In the action of low-energy particles (e.g. in the treatment of films by a high-frequency charge), bombardment by hydrogen atoms and by other chemically active particles, ultra-violet irradiation of massive specimens and grinding crystals or glasses, the radicals get concentrated, particularly in the thickest surface layers. In individual cases the thickness of the layer in which the radicals are formed amounts to several hundred Angstrom units (bombardment by hydrogen atoms6, grinding particles7, treatment by gas discharge8) and consequently, in the case of a specific material surface about 1 m2/cm3 (characteristic particle dimension ~ 5 χ IO"4 cm) the local concentration of radicals exceeds the average one by a hundredfold and the effective rate constant of decay differs from the actual one by a hundredfold also.

The volume distribution by mechanical radical generation depends on the character of the forces acting. The stretching tensions in amorphous crystalline polymers are localized in amorphous intermediate layers where the radicals also concentrate. The treatment of powders in mills is accompanied by intensive amorphization even to complete loss of crystallinity. The tension provoking both the displacement of chains and amorphization activates the diffusion. The curve of radical accumulation exhibits a limit, whereas the coefficient of forced diffusion which may be calculated from this curve is comparatively high equalling 10~18cm2/s. Thus, in long-term milling the polymer may be expected to be amorphous and the radical distribution random.

The decay rate and its dependence on temperature are related to the kind of particle distribution in the volume given by the method of radical generation. In Figure 510 curves of alkyl radical (—CH2CHCH2—) concentration in polyethylene as a function of temperature are shown. Each temperature was maintained for five to ten minutes. The left-hand curve has been obtained after irradiation of the specimen with a dose of 3 Mrad (the initial radical concentration being 3 x 1919cm~3), after treatment in a vibration mill (C0 = 1 x 1018cm"3). In the milled amorphized polyethylene the decay of

62


radicals takes place over a relatively narrow temperature range between 240° and 300° K. In the γ-irradiated polymer three temperature regions may well be differentiated: 120°, 190° and 240°K.

Figure 6. The e.s.r. spectra of individual radicals (1,3) and pairs (2,4) in (a) polyethylene (1,2) and (b) polyoxy ethylene (3,4) taken from ref. 11.

Close to 120°K the temperature coefficient of the decay rate appears to be very small (less than 1 kcal/mole) and it may well be presumed to correspond to the activation energy of the elementary recombination step of radicals located one beside the other, i.e. the radical pairs.

The pairs, i.e. the radicals located at a distance of a few Angstrom units from one another, have been found in some organic crystals and polymers ι ί Λ 2

following irradiation (y or u.v.) at 77°K. The e.s.r. spectra of radical pairs in polymers are shown in Figure 611. The unpaired electron interacts with

CH? CH CHo^

(e) E ~ 4 kcal/mole

10 20 30 6.4 6.8 7.2 7.Ì Time,min (1/7") x 103

Figure 7. Transformation of—CH2—ÙH2 radicals into —CH2—CH—CH2— radicals in polyethylene: 1—130°K; 2—146°K; 3—157°K (taken from ref. 13).

63

P. JU. BUTIAGIN

the protons of both the radicals; as a result the value of the hyperfine splitting decreases and the number of hyperfine system components increases. According to qualitative evaluations the concentration of pairs may reach a tenth of the overall radical concentration. In agreement with these data12

the distance between the radicals in pairs, decaying in a low temperature region, equals 5.3 Â. This value may evidently be accepted as the cage dimension in polyethylene at 120°K when the defrosting of the inner motions of the chain is just starting.

■12

-H

-16

High density

E^10 + 1 kcal/mole

5.0 1000/7"

6.0

Figure 8. Alkyl radical decay in polyethylene (170°-215°K). Calculation of the data of ref. 10.

Over the same temperature range in polyethylene also another reaction, i.e. the transformation of end radicals into inner ones which may quite as well be denoted as a cage reaction, has been observed. The kinetic reaction curves are shown in Figure 713. The activation energy of 4 kcal/mole well characterizes the elementary process of hydrogen atom transfer.

The next step of radical concentration decrease is in the region of 190°K. In the irradiated polymer the average distances among radicals and among vinyl groups do not exceed 20 to 30 Â and for their interaction there is no need of gradual displacement of the free valence to longer distances (chemical migration or diffusion). The partial radical decay in the region of 190°K is associated, first of all, with a further increase of cage radius as a result of the defrosting of inner motions. The corresponding decrease of alkyl radical concentration is due to the cage reactions, i.e. recombination or interaction with weak C—H bonds around vinyl groups (transformation of alkyl radicals into allyl ones14).

The effective activation energy of decay in this range is close to 10 kcal/mole. The dependence of the rate constant on temperature is shown in Figure 8; the data on the amorphous and crystalline polyethylene specimens are taken from Figures 2 and 3 in reference 10.

The elucidation of the steps of decay at 190°K by the increase of the effective cage radius (presumably up to 10 to 20 Â) is confirmed by comparison of the curves taken from Figure 5a and 5b. In milled amorphized polyethylene

64


(Figure 5b) the starting radical concentration equals 1 x 1018 c m - 3 and the average distance among them is — 100Â. The decay starts over the temperature range 230° to 240°K (i.e. at a temperature corresponding to the last decrease of the curves shown in Figure 5a) with the temperature dependence lacking any steps whatsoever.

_J I I L_ 50 100 150 200

Time, min

Figure 9. Decay of peroxy radicals in polymethyl methacrylate at high vacuum and 0°C. Concentrations of radicals {[ROO*] + [R*]}—1, [ROO*]—2 and [R*]—3 from ref. 20.

Evidently, starting from 240°K, in both the polyethylene specimens the motion of free valence becomes possible over relatively long distances. At a constant temperature the kinetic reaction law depends on whether the recombination (equation 1) or the reactions with 'weak' C—H bonds appear to be prevailing.

Thus, at the temperature increase the effective kinetic cage radius increases from a few Angstrom units (120°K) to tens of Angstrom units (pre-melting region, r0 % 40 to 60 Â).

POSSIBLE MECHANISMS OF FREE VALENCE MIGRATION Thermal kinetic analysis of the decay reaction dealt with above is in fact

based on the presumption that active centres participate in two types of motions: (1) Inner motions of macromolecular chain segments; the comparatively

intense motion in a cage volume appears to be a distant analogy of rotation diffusion in low molecular compounds ;

(2) Displacements in polymer volume at relatively long distances (translation diffusion or chemical migration). Reasonable division of the motion into two parts is, of course, possible

in the cases when the frequencies of both the motion forms differ considerably from one another (vrot vtz). The kinetic (effective) cage radius depends, to a great extent, on the correlation of the frequency and amplitude of the two motion forms.

65 P.A.C—30/1—D

P. JU. BUTIAGIN

In many systems, marked with great local concentrations of active centres and intense inner segmental motion the decay processes may be elucidated by the course of reactions in the cage without the need to apply the conception of long-distance free valence migration.

In another extreme case, when the inner motions are hindered (crystalline phase, low temperature) the effective cage radius decreases up to the geometric dimension of the elementary crystalline cell, the basic role in the decay processes then being played by free valence migration.

Table 2. Temperature of decay of polymer radicals

Polymer

Polyethylene Polypropylene

Polyisobutylene Polystyrene Polymethyl methacrylate Polyvinyl alcohol Polyformaldehyde Polycaprolactam

r * ° K

188 256

200 370 380-390 358 250 320

Stability temperati R - C H 2

120-140 < 90

< 90 120-140

< 9 0 120-140 250-270

120

R—CXY

120-140

200-220 330-350 330-350

—

—

ure, °K inner radicals

260-280 250-290

200-220

— 250-270 330-360 350-370 280-290

ROO·

260 280 300

300-340 200-220 260-280 260-280 200-220 200-220

In most of the intermediate cases, the contribution of both motion types may be expected to be comparable and the laws of decay not to be attributed to any single mechanism of free valence displacement.

The long-distance free valence migration may be associated with both diffusion and chemical reactions. The migration mechanism depends on many factors, namely on radical reactivity, structure and chemical properties of macromolecules, supermolecular polymer structure, medium properties etc. The first image on the nature of the processes of free valence displacements is presented by analysis of the conditions of various types of radical decay in the given polymer. The simplest criterion is the comparison of the temperature of structural transition in the given polymer with the temperature range in which the radical decay proceeds at a measurable rate. In Table 21 5

the temperature ranges of the registration of decay reaction for two types of end, inner and peroxy radicals in various polymers are shown; here, fifteen authors' data are summarized.

Diffusion Alkyl and peroxy radicals in polyisobutylene have been generated by the

method of mechanical dispersion at 80°K in a vacuum. The initial concentration is about 1018cm~3. Three types of radical decay over the narrow temperature range from 200° to 220°K coincide with the glass transition temperature of polyisobutylene (200°K). On this basis diffusion may well be presumed to be the fundamental mechanism of free valence migration.

66


The rate constant of peroxy radical decay in polyisobutylene in air depends on the temperature according to the equation

KD = 3.2 x l ( T 4 e x p ( - 18000/RT)cm3 s"1

In the diffusion mechanism KD = 4nr0D and provided that r0 — 10"7 cm, we obtain

D % 0.3 x lO^xpi- lSOOO/RTJcn^s- 1 (5)

In polyisobutylene the coefficients of diffusion of hydrocarbons of paraffinic order C22 to C2 5 have been measured16 and for them we obtain

D = 3 x Ι Ο ^ χ ρ ί - Ι δ Ο Ο Ο / Κ η α η ^ " 1 (6)

With regard to the discrepancy between the dimensions of the segment in polyisobutylene and that in the hydrocarbons C2 2 to C2 5 as well as the possibility of the molecular weight effect on the diffusion coefficient, the coincidence of equation 5 and 6 may be found quite satisfactory. Similar evaluations confirm the presumption of the prevalence of the diffusion mechanism of radical decay in polyisobutylene.

Both lower and upper limits of diffusion rate assuring measurable rates of macroradical decay may be evaluated by very general considerations. In the measurement of decay kinetics, the radical concentration is usually 10 1 8 ± 1 c m - 3 and the duration of the reaction observation is 103 to 104 s. Provided that during the observation period the radical concentration decreases by twofold, then

τ± Ä 1/KDC0 % l/4nr0DC0 s

Provided that r0 ~ 10"7 cm

Dx IO7IO x 1 0 3 5 ± 0 · 5 x 10 1 8 ± 1 % between 10"15 and 10"1 8 cm 2 s" 1

Should the diffusion coefficient be less than 10"1 8 , the process may last some tens of hours and should it exceed 10"15 , the reaction will terminate within one minute, i.e. at the moment of establishment of thermal equilibrium.

Table 3. The diffusion coefficient, cm2/sec.

Polymer

Natural rubber

Butylstyrene rubber

Ethylene-propylene (1 : rubber

1)

M

1 x 104

3 x IO5

1 x IO4

2 x IO5

1 x IO4

T°K

220

220 220

220 220

cm2

D sec

4.5 x IO - 1 7

2.4 x IO - 1 7

5.0 x IO"15

4.2 x IO"17

5.0 x IO"15

Ref.

17 (Bresler)

17 (Bresler)

18 (Skewis)

67

P. JU. BUTIAGIN

Table 4. The rate constants (l./mol. sec) of the reaction

^ ^ . radical

compound ^ \ ^

C6H5CH7—-H C6H5

CH—H H3C ' C6H5NH—H C6H5S—H

R'o + RH ^ R0H + R*

—CH(C6H5)ÔH2

IO - 6

7 x IO"6

IO"6

^ Ι Ο " 4

ROO*

—

l O ^ t o l O " 5

^ 1 0 ~ 4

The diffusion rates have been measured by direct methods in some elastomers, such as natural and butadiene-styrene rubbers, and ethylene-propylene copolymer. In Table 3 values of diffusion coefficients extrapolated to the temperature range of radical decay are shown. In all cases the D values are in the mentioned ranges. Consequently, in these elastomers as well as in polyisobutylene the diffusion rate proves to be quite sufficient for the elucidation of the laws of radical decay.

Evidently, in amorphous samples of many polymers at temperatures close to the glass temperature, the diffusion decay mechanism prevails and the measurement of macroradical kinetics may under these conditions be considered as a method of measuring diffusion coefficients. It has to be stated that chemical migration does not play any role here.

The reaction of hydrogen atom transfer For many polymer radicals the possibility of hydrogen atom transfer at

low temperatures starting with room temperature or less has now been established. The corresponding experimental results are summarized in Tables 4 and 5 respectively.

In Table 419 the values of the rate constant of the reaction of end [ —CH(C6H5)—CH2] and peroxy (ROO') radicals of polystyrene with some aromatic compounds in frozen solutions at 100°K are shown. In all cases

Table5. The reaction ROO* -► R' in some polymers (the radicals have been produced by grinding)

Polymer

Polyethylene Polypropylene Polycaprolactam Polyacrylonitrile Polystyrene

Polymethyl methacrylate

T°C

273 273 250 273 273

250-350

Time of reaction, sec

- 104

10 4 to l0 5

103 to 104

104

- 5 x 103

A + 1 / 12000 V 1. \ Λ — iu expi — H i

*\ RT J\mol.sec./

68


as well as in that of toluene, the reaction of hydrogen atom abstraction proceeds at a measurable rate.

Analogous reactions have also been found in solid polymers when the radical abstracts the hydrogen atom from the adjacent macromolecule (Table 6). In polyethylene, the transformation of end radicals into inner ones at 120°K has been mentioned13. The activation energy of this reaction is 4 kcal/mole and generally half that of the analogous reaction of low molecular radicals in the gas phase. In just the same way, the end radicals in polyformaldehyde and polypropylene change into inner ones. The interaction of the inner radicals with adjacent macromolecules has only been proved in polyethylene where, at room temperature, alkyl radicals are transformed into allyl radicals.

The reactions of peroxy radicals with macromolecules have been observed in many polymers (Table 5). Only in capron, end products of the reaction ROO' -» R# are alkyl radicals with a free valence inside the chain20. In polyethylene allyl radicals14 and in polymethyl methacrylate the end radicals20 are formed. Evidently, in the latter case abstraction of the hydrogen atom is succeeded by further transformation of inner radicals including decomposition reactions.

Table 6. The reaction ft0 + HR -+ R H + ft in polyolefins

Polymer

~ C H 2 -> - C H -(polyethylene)

- C H ( C H 3 ) ^ -C(CH 3 ) -(polypropylene)

- C H - - - H Ò — C H = C H ~ (polyethylene)

Γ°Κ

120

140

297

Time of reaction, sec

10 to l00(£ = 4kcal/mol) (Radtzig)

10 2 to l0 3

(Radtzig)

1 x 105

(Waterman, Dole)

The reaction of hydrogen atom transfer (Table 6) is always accompanied by a marked decrease of the total concentration of radicals. The corresponding kinetic curves for the reaction ROO* -> R' in polymethyl methacrylate are plotted in Figure 920.

From the viewpoint of the cage model the single hydrogen atom transfer appears to be a simple model of the elementary process of transition of the free valence from one cage to another. It is evident that the radicals occurring in one cage following such a transition, recombine. The fraction of radicals decaying in the single free valence transfer

ω = (C0 - C)/C0 (7)

should be dependent on both the cage dimensions and radical concentration. The ω values for some reactions of free valence transfer are shown in Table 720. These values are preliminary, because both the kinetics and mechanism of corresponding reactions have not been investigated completely.

6°

100 150 20 40 60 80 Time, min

i A.o l

3.0

2.0

i.ol-

30h

20l·

(c)

/ /

10h . /

V 10

10

20

15

A 1 o 2 * 3 a U o 5

30

20

Figure 10. Decay of peroxy radicals in polymethyl methacrylate: (a)—22°C; (b)—0°C; (c)—15°C. The oxygen pressure was: 10 torr—1 ; air—2; 3 atm—3; 4 atm—4 and 10 atm—5 [ref. 20].


Table 7. The values of ω = (C0 - C)/C0 (reaction ROO* -* R')

Polymer T°C ω

Polycaprolactam — 20 0.7 Polystyrene 0 0.4 Polyacrylonitrile 0 0.75 Polyethylene 0 0.96 Polymethyl methacrylate - 22 0.4 to 0.6

Thus, for the active end radicals as well as for peroxy radicals in a vacuum, the decay process terminates in just a single free-valence transfer into the adjacent cage, which is accompanied by the formation of radicals more stable at the given temperature and by partial recombination. For long-distance free valence displacement due to the mechanism of hydrogen atom transfer, the development of the chain reaction is to be admitted. This migration course has been postulated by several authors for polyethylene (see e.g. ref. 21), though it has strictly been proved only in the case of radical decay in the presence of oxygen. Quite an important characteristic of the chain process consisting in the alternation of reactions

(I) R· + 0 2 -* ROO', (II) ROO* -» R*

proves to be the chain length which may be determined by measurement of the oxygen consumption or of the fraction of recombining radicals in one step of the chain growth. Actually, should ω not change considerably in the course of the process, following the first cycle of the reactions I and II in polymer, the sequence (1 — ω) following to the second, (1 — ω)2, following to the third, (1 — ω)3 etc. of radicals is preserved. Consequently, the chain length equals

v = [O2] /C0 = (1 - ω) + (1 - ω)2 + (1 - ω)3 + . . . = (1 - ω)/ω

The results of chain length measurement in peroxy radical decay in air are shown in Table 8. In all cases, the chain length is relatively small not exceeding a few units. This is in agreement with the notion requiring a comparatively large cage radius. The kinetic curves of peroxy radical decay in polymethyl methacrylate in air at 0° and — 22°C respectively are shown in Figure 1020.

Table 8. The length of chain during the decay of peroxy radicals

Polymer

Polyethylene

Polypropylene

Polyvinylchloride

Polymethyl methacrylate

Chain length

5 to 12

3 to 6

3 to 7

1.5

Ref.

Lawton

Fischer

Loy

71

P. JU. BUTIAGIN

02

0.1

-4 2x10 2 1 ^L3 Λ fR]2 sec

\ -22°C

\ o

I . 1 1

o

/,sec

2.0

1.0

[ R ] 2

2x103 4x103 6x103 500 1000

Γ72Χ1021

r ^ ^

1

< ^ r 0 = 45 Ä D = 9 x 10~18

1 -v? -7=-. sec z

1

0.02 0.04 0.06 0.05 0.1

PMMA, R00#- radicals Figure 11. Calculation of data taken from Figure 10 according to Waite's equation.

By elaboration of these results in agreement with equations 1 and 2 (Figure 11) the values r0 - 40-50 Â and D - 10"16 to 5 x 10"18 cm2 s"1 have been obtained. Therefrom, at 0°C the value τ = rl/πϋ or the time of free valence transition from one cage to another equals 650 s. When transition from one cage to another takes place due as a reaction to hydrogen atom transfer, the characteristic reaction time should be comparable with the value of τ.

Figure 12. Formation of low molecular products in polyvinyl acetate (a) and polystyrene (b) according to ref. 20.

72


In polymethyl methacrylate, as mentioned above, the rate constant of reaction ROO' -► R* is 20

KROO-+R = 106 ± 1exp [ - ( 1 2 0 0 0 + 1000)/RT] 1. mole" 1 s"1

and the characteristic reaction time at 0°

TROO--R- - V ^ R O O ^ R C R H ] * 10"76χρ(12000/^Τ) Ä IO2·5 S

(in the calculation the value [RH] » 10 mole l"1 has been assumed). The characteristic reaction time (TR O O ._R . ) is of the same order of magnitude as the previously obtained time of free valence transition from one cage to another (650 s).

It should be stated that in calculating the decay rate irrespective of large cage radius, i.e. presuming every reaction step ROO* -► R# to lead to free valence displacement only at 'inter-chain' distances (3 to 5Â), the decay rate would appear to be less by a hundred times and the expected chain length more (and oxygen consumption) by ten to a hundredfold, which does not correspond with experimental data available.

The above reaction of peroxy radical decay in air is quite a good example of migration from one cage to another as well as of intensive inner motions within the cage itself. Thus, every reaction step leads to long-distance free valence displacement (up to 50 Â), the chain length being small. The nature of 'inner' motions leading to such a high value of an effective cage radius will be discussed below.

The fact of long-distance chemical migration in alkyl radicals in the absence of air has so far not been unambiguously proved because there does not appear to be any experimental idea confirming the —CH2—CH—CH2— radical transformation within the chain. There are some supporting grounds for assuming th is 1 4 , 2 1 , 2 2 :

(1) The reaction rate of hydrogen atom transfer evaluated by extrapolation of the results referring to gas-phase reactions to room temperature is quite sufficient to account for the kinetic laws of alkyl radical decay in polyethylene;

(2) The decay is accelerated in the atmosphere of hydrogen and methane being accompanied by H-D exchange;

(3) The reaction of radical decomposition in polyethylene at room temperature does not take place.

The major 'negative' ground lies in the fact that all the measurements of decay kinetics have been carried out with samples containing up to 2-3 x 1019 cm"3 'weak' C—H bonds (around vinyl and other groups). The average distance between the mobile hydrogen atoms amounts altogether to 1/^/(3 x 1019) %30Â, i.e. may be comparable with the cage radius. The quantitative transformation of alkyl radicals into allyl ones, i.e. abstraction of mobile hydrogen atoms, has been demonstrated sufficiently14.

Thus, the nature of free valence migration in polyethylene has so far not been established.

73

P. JU. BUTIAGIN

Reactions of radical decomposition The strength of individual chemical bonds in radicals is markedly lower

than in molecules and the decomposition appears to be quite a widespread group of radical reactions.

The activation energy of end radical decomposition according to the depolymerization scheme

—CH2—CHX—CH2—CHX -> — C H 2 - C H X + CH 2 =CHX

equals the sum of the heat and the activation energies of polymerization. For the end radicals of polyethylene

Ed = 22.3 + 5.5 % 28 kcal mole" l

and for those of polymethyl methacrylate Ed = 13.0 + 4.7 » 18 kcal mole - 1

The end radicals in polymethyl methacrylate decay at a measurable rate at 50° to 60°C and higher. At these temperatures the rate constant oi depolymerization is

K = 101 2exp(-18000/KT) ^ l s " 1

At a single depolymerization step the free valence displaces to one chain element (i.e. irrespective of inner motions) distant 3 to 4 Â. The diffusion coefficient D = (1/6) l2 (dn/dt) ^ 1 0 ~ 1 6 c m 2 s _ 1 corresponds to the displacement rate of 3 Â/s, i.e. end radical decay may under these conditions be elucidated by free valence migration according to the depolymerization-polymerization mechanism. Actually, radical decay in polymethyl methacrylate is always accompanied by monomer release, whereas in a vacuum a few tens of molecules for each pair of radicals decayed may be found. The decomposition rate of inner radicals should not differ substantially from the depolymerization rate. Random heat motion of segments in amorphous samples may provoke the increase of the probability of decomposition reactions accompanied by the formation of low molecular compounds. The analysis of light low molecular products in the process of radical decay appears to be quite a simple and sensitive method of registration of the decomposition reaction. .

In Figure 12 kinetic curves of volatile product release in the decay of alky 1 and peroxy radicals in both polystyrene and polyvinyl acetate in a vacuum are given. (In polyethylene and polypropylene the release of such volatile products is negligible.) In polyvinyl acetate the decay is accompanied by acetone release and in polystyrene, toluene, ethylbenzene and eumene are released in an amount comparable with the radical concentration.

Evidently, in the decomposition reaction either the monomer or the light' radical may split away from the macroradicals. In polystyrene either styrene or ethylbenzyl radicals are formed in this way. The diffusion of low molecular radicals appears to be a fairly efficient mechanism of free valence displacement.

The formation of low molecular products was observed in the process of peroxy radical decay when the reaction of hydrogen atom transfer (ROO' -> R") is the primary step. To elucidate the results obtained the

74


chemical mechanism of free valence migration may be supposed to be composed of a sequence of elementary steps. In the decay of peroxy radicals in polystyrene the following reactions may be expected : (1) r\ R' + 0 2 -> ROO* free valence displacement at the

p. bond distance with possible re-I turn along the axis

(2) ROO' + —R ► ROOH + — R'— valence transmission to the adjacent chain, i.e. displacement from one cage to another

(3) —R" > R' + P— shortening of the ends of a ruptured chain and high mobility of the end radicals

(4) —R* -> —R" + m displacement along the bond length and 'plasticization' of the cage by monomer

(5) —R' -» —P -fr" Ì diffusion of low molecular (6) r' + —RH > rH + — R'— J radical (7) Recombination of r\ R' and ROO'.

In the presence of oxygen or with sufficient activity of radicals r* and —R* these cycles may repeat in a manifold way. Almost all of the processes presumed are accompanied by free valence displacement and the large kinetic cage radius alluded to above (Figure 11, r0 — 50 Â) may be supposed to be bonded not only to inner segmental motion but also to the mobility depending on the reaction course 1 to 5.

At present, this scheme is a hypothetical one, the correlation of the constants of individual steps still not being established and the distances at which the free valence can be displaced not being known. On the other hand, the possibility of the participation of the reaction of hydrogen atom transfer as well as of radical decomposition in the processes of free valence migration may be considered established.

At the close of this short survey it must be pointed out that in polymer systems there are, actually, rare cases of free valence migration proceeding according to an arbitrary single definite mechanism. In the diffusion mechanism the inner motion in the cage may be more important than the cage-to-cage displacement. The segment mobility characteristic for the polymers is often associated with chemical mechanisms. 'Chemical' migration may be composed of a series of elementary steps, viz. following the reactions of free valence transfer the radical decomposes with release of low molecular compounds etc.

REFERENCES 1 T. R. Waite, J. Chem. Phys. 28, 103 (1958). 2 T. R. Waite, Phys. Rev. 107, 463 (1957). 3 Ja. S. Lebedev, Kinetika i Kataliz, 8, 245 (1967). 4 J. Frank and E. Rabinowitsch, Trans. Faraday Soc. 70, 120 (1934).

75

P. JU. BUTIAGIN

5 J. Auerbach and L. H. Sanders, Polymer, Lond., 10, 579 (1969). 6 A. M. Dubinskaya, Preprints of the Conference 'Chemical Transformations of Polymers',

Bratislava (1971). 7 P. Ju. Butiagin, Vysokomol. Soedin. 9-A, 136 (1967). 8 A. I. Michailov and Ja. S. Lebedev, Kinetika i Kataliz, 6, 48 (1965). 9 Ju. D. Tsvetkov and others, Khimia vysok. energiy, 4, 180 and 369 (1970).

10 S. Nara, S. Shimada, H. Kashiwabara and I. Sohma, J. Polym. Sci. A-2, 6, 1435 (1968). 11 M. Iwasaky, T. Schikawa and T. Ohmori, J. Chem. Phys. 50, 1984 (1969). 12 T. Fujimura, N. Nayakawa and N. Tamura, Preprints of the Conference 'Chemical Trans

formations of Polymers', Bratislava (1971). 13 V. A. Radtzig and P. Ju. Butiagin, Vysokomol. Soedin, 9-A, 2549 (1967). 14 D. C. Waterman and M. Dole, J. Phys. Chem. 74, 1913 (1970). 15 P. Ju. Butiagin, A. M. Dubinskaya and V. A. Radtzig, Uspekhi Chimii, 38, 593 (1969). 16 V. K. Gromov, A. E. Galych, R. M. Vasenin and S. S. Voyutski, Vysokomol. Soedin. 5,

802 (1965). 17 S. E. Bresler, G. M. Zakharov and S. V. Kirilov, Vysokomol. Soedin. 3, 1072 (1961). 18 J. D. Skewis, Rubber Chem. Technol. 39, 217 (1966). 19 A. M. Dubinskaya and P. Ju. Butiagin, Kinetika i Kataliz, 9, 1016 (1968). 20 P. Ju. Butiagin, I. V. Kolbanev, A. M. Dubinskaya and M. U. Kislyuk, Vysokomol. Soedin.

10-A, 2265 (1968). 21 V. V. Voevodsky, Fizika i khimia elementarnich khimitcheskikh protsesov, Nauka: Moscow

(1969). 22 D. O. Geymer and C. D. Wagner, Nature Lond. 208 (5005), 72 (1965).

76

CONTROLLED PROPAGATION IN ASSOCIATED MONOMER AGGREGATES

A D O L P H E CHAPIRO

Laboratoire de Chimie des Radiations du CNRS, 92-Bellevue, France

ABSTRACT The polymerization of acrylic acid in bulk proceeds with a very high rate and gives rise to a syndiotactic polymer. This result is attributed to the association of monomer molecules by hydrogen bonds into linear aggregates in which a stereospecific propagation is favoured. In order to verify this assumption the polymerization of acrylic acid was investigated in various solvents. It was found that the addition of methanol, dioxane and water did not significantly affect the reaction rates nor the tacticity of the polymer. In these solvents the viscosity of the monomer remains high, suggesting that the linear aggregates are not dissociated. In contrast, the addition of toluene or n-hexane sharply reduces the polymerization rate and the fraction of syndiotactic polymer. The viscosity of acrylic acid also drops in the presence of these solvents. These results are in agreement with the assumption that the polymerization of acrylic acid is strongly controlled by linear monomer aggregates.

A similar situation is believed to apply to acrylamide. The propagation rate constant of this monomer was found to vary widely depending on the solvent used. This could indicate that the propagation rate is governed by monomer aggregates, the extent of aggregation being a function of the nature of the solvent. Methacrylic acid behaves in a different manner. This monomer forms the same types of associations as acrylic acid but the polymerization kinetics do not reflect any peculiarities related to such aggregates.

A very strict control of chain propagation is met when 4-vinylpyridine is polymerized in the presence of polycarboxylic acids. A considerable rate increase was observed when vinylpyridine was grafted into polytetrafluo-roethylene films which contained poly(acrylic acid) branches. This effect is accounted for by assuming that the pyridine substituents associate with the carboxylic groups, thereby providing a very favourable orientation of the

vinyl groups for chain propagation.

Several polymerization studies conducted in this laboratory have led to unusual kinetic features which are correlated with the association of monomer molecules resulting in ordered aggregates. The purpose of the present communication is to summarize these results and to stress some of the similarities or differences observed in the various systems.

I. BULK AND SOLUTION POLYMERIZATION OF ACRYLIC ACID

Earlier work in this laboratory1 has shown that when pure acrylic acid is

77

ADOLPHE CHAPIRO

subjected to gamma radiation at room temperature a very rapid polymerization ensues which leads to a syndiotactic polymer of extremely high molecular weight (Mv = 1 to 3 x 107). The stereoregular polymer is easily separated from the atactic isomer (virtually absent in bulk polymerization) by solvent extraction : the atactic polymer being soluble in anhydrous dioxane while the syndiotactic material only dissolves when 20 per cent water is added to this solvent. An insoluble gel is left as a residue after extraction. The stereo-regular polymer readily crystallizes to fairly large single crystals. It is noteworthy that the syndiotacticity remains unaltered for polymerizations carried out at 20°, 40°, 60° or 76°.

(1) Monomer aggregates in acrylic acid In order to explain these findings it was assumed that these peculiar

features of the polymerization of acrylic acid are related to the fact that this monomer forms associated structures via hydrogen bonding. It is well established2 that carboxylic acids form cyclic dimers. Originally, the dimeric structures such as I were believed to be responsible for the stereospecific propagation. Such structures can be assumed to exhibit planar symmetry and one can indeed expect that the simultaneous reaction of two monomer molecules having this configuration would lead to a syndiotactic diad.

Alternatively, one can assume that the monomer forms linear aggregates of type II. In such an event, the double bonds lie in a narrow zone and on encounter with a growing chain the addition of the first monomer molecule

CH2 CH2

'/ W CH 0 _ . . H 0 CU CH=CH2 CH=CH, CH=CH2

\ // \ / I I I c c c c c

\ / / / / \ / / \ / / \ O H — O — O ΟΉ--0 OH—O OH—

(I) (ID

could be followed by a fczip' reaction involving all monomer units of the aggregate. This would account for the very high rate of propagation found in this system and eventually for the syndiotacticity of the polymer, provided the double bonds are properly oriented.

Very little information is available in the literature on such associations in carboxylic acids. Dimeric structures of type I are characterized in dilute solutions2 ; linear structures of type II have also been reported3,4.

(2) Viscosity of acrylic acid solutions The presence of linear aggregates of type II should give rise to an increased

viscosity. A study was therefore carried out on the viscosity of acrylic acid in bulk and in various solvents which were believed to interfere with hydrogen bonding5-7. Some of the results are presented in Figure 1 which shows the flow times of various binary mixtures as a function of the concentration of acrylic acid.

78


, 200

100

» /I Water

/ / /

/^^^

: ^

*f'' B, Toluene

I

Q \ o x < ^ \χ

^ —^ χ

·-" "^· ^ S \ X

Acetic acid ,._, η • • • • ' ■ ■ " ' ^ ''

nzene ...^:>£^'

^,.··^"

1 1 0 0.25 Solvent

0.50 0.75 1.00 Acrylic acid

Figure 1. Flow times of acrylic acid in various solvents as a function of monomer content6

The pure monomer exhibits high viscosity, higher than that of water which is known to have a polymeric structure. The addition of methanol, dioxane or water leads to an increase in tlow time and a maximum appears which is particularly pronounced in the case of water. Acetic acid has approximately the same flow time as acrylic acid and in the mixtures the viscosity remains constant. In toluene, ether and n-hexane, the viscosity drops and all experimental points lie below the linear relationship which corresponds to a simple dilution effect. These results are further discussed below.

(3) Polymerization kinetics of acrylic acid in solutions Poly(acrylic acid) is insoluble in its monomer ; consequently, the polymeri

zation in bulk occurs in a precipitating medium and exhibits the usual kinetic anomalies associated with such reactions. Water, methanol and dioxane dissolve the polymer and, hence, upon addition of these compounds the reaction medium gradually changes from a heterogeneous to a homogeneous one. The kinetic features associated with this change in the case of methanol have been described earlier5. Figure 2 shows the rate data obtained with water solutions. A pronounced maximum appears at a concentration of ca. 50 per cent monomer. The molecular weights of the polymer exhibit a maximum for the same concentration (Figure 3) which indicates a particularly

79

ADOLPHE CHAPIRO

Acrylic ac id -wate r

25 50 75 100 % Acrylic acid

Figure 2. Rate of polymerization of acrylic acid in aqueous solutions as a function of monomer concentration. Initiation by gamma rays (16 rads/min) at 20°C7

20h

10r-

Acrylic acid-water

s* ^ N # Syndiotactic

25 50 7o Acrylic acid

Figure 3. Limiting viscosity indexes of syndiotactic and atactic fractions of poly(acrylic acid) obtained in aqueous solutions at different monomer concentrations7

favourable propagation to termination ratio in this mixture, in which the polymer forms a viscous solution (gel effect).

It should be noted that the polymerization rate hardly changes upon the addition of up to 20 per cent solvent. A similar result is obtained with methanol, dioxane and acetic acid. In contrast, in the presence of small amounts of toluene and n-hexane the rates and molecular weights are drastically reduced. Some of the data are presented in Figure 4.

The above results show a striking correlation between the polymerization rates and the viscosities of the initial solutions : only those solvents which reduce the viscosity of acrylic acid also reduce the reaction rates.

80


25 50 °/e Acrylic acid

Figure 4. Rate of polymerization of acrylic acid in toluene and n-hexane solutions as a function of monomer concentration. Initiation by gamma rays (12 rads/min) at 20°C

(4) Tacticity of the resulting polymer In these experiments all reaction products were fractionated by solvent

extraction as described above in order to determine the relative amounts of atactic and syndiotactic polymer.

100

c o

σ <

i

1 - MeOH 2 - Dioxane 3 - Acetic acid

V^V

\

\

2

i. - Water 5 - n - Hexane 6 - Toluene

4 \

0\3

i

\ 6

25 50 % Acrylic acid

75 100

Figure 5. Amount of atactic fraction in the poly(acrylic acid) obtained from various solutions as a function of monomer content in polymerizing mixture7. 1 Methanol, 2 Dioxane, 3 Acetic acid,

4 Water, 5 n-Hexane, 6 Toluene

Figure 5 shows the results obtained in the various solutions investigated as a function of monomer content. It can be seen that syndiotactic polymer is formed in methanol, dioxane acetic acid and water solutions up to fairly

81

ADOLPHE CHAPIRO

high dilutions (50 to 70 per cent solvent) whereas in toluene and n-hexane only atactic polymer arises when more than 20 per cent solvent is added to the monomer.

The various results presented above conform with the assumption that the polymerization of acrylic acid in concentrated media is governed by linear monomer aggregates of type II. These are responsible for the high rate of the reaction and for the formation of syndiotactic poly(acrylic acid). Methanol, dioxane and water do not seem to destroy these linear structures and the fact that the viscosities of these solutions exhibit maxima suggests that these solvents superimpose their own associated structure on to that of acrylic acid and perhaps link several linear aggregates together. This new structure neither significantly affects the rates nor the tacticity of the polymer.

In contrast, non-polar solvents such as toluene and n-hexane destroy the linear associations of acrylic acid as indicated by the drop in viscosity (Figure 1). Infra-red data7 suggest that in these solutions the linear structures of type II are converted to the dimeric form I. Under such conditions the rate of propagation is strongly reduced and the resulting polymer is atactic. In order to account for the high viscosity of acrylic acid and for the peculiar kinetic features of its polymerization, the linear association aggregates should contain on the average at least 20 to 50 monomer molecules and the lifetime of such aggregates should be reasonably long. Experiments are in progress to gain direct evidence for the presence and of the characteristics of such linear aggregates in acrylic acid.

II. OTHER SYSTEMS (1) Acrylamide

The free radical polymerization of acrylamide has been studied by several authors8-11 . Most experiments were carried out in aqueous solutions and extremely long kinetic chains were observed. On the other hand, the determination of the rate constants for chain propagation (kp) and termination (kt) led to the surprising result that both kp and kt strongly depended on the solvents used for the reaction medium11. Some of the results are summarized in Table L

Table 1. Rate constants in the polymerization of acrylamide at 19'C11

moles 1 1 sec l kcal mole 1 moles 1 l sec" ' kcal mole l

Water (pH = 7) 9.4 2.7 7.2 2.7 Water (pH = 13) 4.3 3.6 1.7 1.0 Formamide 0.9 5.0 2.1 0.8 Dimethylsulphoxide 0.27 5.4 2.5 0

In order to interpret the very high kp value observed in aqueous media the authors11 assumed that the double bond of the monomer became protonated in water, thus increasing the reactivity of the monomer in this solvent.

82


Alternatively, the results can be accounted for, as with acrylic acid, by assuming the formation of linear aggregates of type I. The length and stability of these aggregates would depend on solvent-monomer interactions and could thus explain the observed results, if indeed such aggregates occur preferentially in water. Experiments are in progress to verify these assumptions12.

(2) Methacrylic acid Methacrylic acid, like acrylic acid, forms linear aggregates in the bulk

liquid as indicated by its high viscosity. Dilution of the monomer in various solvents leads to a pattern similar to that shown in Figure 1 for acrylic acid13,14. The polymerization of methacrylic acid occurs, however, much more slowly than that of acrylic acid under similar experimental conditions. Figure 6 shows a comparison of the rates observed with both monomers at

100l·

D o i_ Φ a c o äi

c o o

1 100 10 Rad /min

Figure 6. Rates of polymerization of acrylic (AA) and methacrylic acids (AMA) at 20°C as a function of dose rate of gamma rays13

Methacrylic acid-toluene 32 rad/min

0 25 50 75 100 %> Methacrylic acid

Figure 7. Rate of polymerization of methacrylic acid in toluene solutions as a function of monomer concentration. Initiation by gamma rays (32 rads/min) at 20 C1 4

83

ADOLPHE CHAPIRO

different dose rates13. Moreover, the resulting poly(methacrylic acid) does not exhibit any striking stereoregularity. These results indicate that although linear aggregates of type I are present in this monomer they do not play a major role in controlling the polymerization. This conclusion is substantiated by experiments carried out in toluene solutions14. Figure 7 shows that the addition of toluene to methacrylic acid results in a linear decrease of the polymerization rate, in striking contrast to the results obtained with acrylic acid (Figure 4). It thus appears that the destruction of linear aggregates by toluene (as shown by the drop in the viscosity of these solutions) does not significantly affect the polymerization kinetics of methacrylic acid.

The reason for such different behaviour of two very similar monomers is not clear. It is possible that the methyl group in the methacrylic derivative prevents by some steric or 'hydrophobic' action the alignment of the monomer molecules to form the favourable configuration for fast chain propagation and stereospecific control. This point requires further experimental support.

(3) Polymerization of vinyl pyridine in the presence of polyacids Kabanov et al.15 have shown that the polymerization of vinyl pyridine in

the presence of polycarboxylic or polysulphonic acids is controlled by a regular arrangement of the monomer molecules which can be schematically represented by structure III.

COOH COOH COOH COOH

N N N N

Q Q Q y CH=CH2 CH=CH 2 CH=CH 2 CH=CH2

(III)

Here again the double bonds of the monomer appear to come into close contact thus favouring a fast propagation reaction involving properly oriented monomer molecules in the aggregate.

We have met a similar situation in our work on the preparation of semi-permeable membranes containing both carboxylic and pyridine groups. The technique involved the grafting in a first step of acrylic acid into poly-tetrafluoroethylene films by gamma irradiation; the resulting carboxylic membranes are thereafter neutralized by potassium hydroxide and grafted in a second step with 4-vinylpyridine16. The treatment by KOH is required in order to prevent the spontaneous polymerization of vinylpyridine, presumably initiated by proton transfer15.

In these experiments it was found that in the presence of poly(potassium acrylate) branches, vinylpyridine polymerized several orders of magnitude faster than in the ungiafted PTFE films17. Some results are summarized in Table 2.

84


Table 2. Rates of grafting of 4-vinylpyridine into PTFE films containing various amounts of grafted acrylic acid; gamma-rays of 48 rads/min at 20°C17

Acrylic acid in initial PTFE film

(%) Time required to reach Total dose

100% grafting (rads)

0 48 hours 138000 16 21 min 1000 23 14 min 670 35 less than 6 min less than 300

It can be seen that the increase in rate in the presence of polycarboxylic branches is considerable. An extremely small radiation dose leads to very high grafting ratios, thus indicating very long kinetic chains. From the observation that no polymerization is observed in the absence of irradiation, provided the carboxylic groups are converted to the potassium salt, it can be concluded that the increase in rate is entirely due to faster chain propagation, presumably due to a 4zip' propagation process similar to that postulated in the bulk polymerization of acrylic acid.

It should be noted that the complexed aggregates of structure III seem to be much more efficient in increasing the polymerization rate than the linear structure of type II in acrylic acid. If one assumes that the rate of polymerization of acrylic acid in toluene or n-hexane solutions is that of the 'normal' reaction not involving any ordered aggregate, one can estimate this 'normal' rate by extrapolating the low rate region of the curves to the pure monomer as shown in Figure 4. This procedure leads to values which suggest that the linear aggregates are responsible for an increase in rate not exceeding a factor of seven to ten. In contrast, the data presented in Table 2 show that the structure of type III is responsible for much larger increases in rate. One reason for the difference in order of magnitude of these effects lies presumably in the fact that structures of type III are permanent and presumably involve most monomer molecules present in the system, whereas structures of type II are transient. The linear aggregates are in dynamic equilibrium with the dimeric structure I, thus their effective lifetime is presumably short. Moreover, only a small fraction of the monomer molecules is likely to be involved in structures such as II, most carboxylic groups being associated as cyclic dimers.

CONCLUSION The results discussed above indicate that associated monomer aggregates

may affect polymerization kinetics in many different ways. Linear structures in which the monomer is associated by hydrogen bonds may either accelerate the polymerization as with acrylic acid and perhaps acrylamide or have no significant effect like in methacrylic acid. It is noteworthy that in acrylic acid such structures are presumably responsible for a stereospecific propagation. On the other hand, complexed structures such as schematically shown in III are permanent matrices which lead to much higher rate increases. Steric control of propagation is also expected to arise in such matrices provided

85

ADOLPHE CHAPIRO

the proper configuration of the double bonds is reached in the corresponding complex.

REFERENCES 1 A. Chapiro and T. Sommerlatte. Europ. Polym. J. 5, 707 and 725 (1969). 2 G. C. Pimentel and A. L. McClellan, The Hydrogen Bond, Freeman: (I960). 3 E. Constant and A. Lebrun, J. Chim. Phys. 61, 163 (1964). 4 J. Lascombe, M. Haurie and M. L. Josien, J. Chim. Phys. 59, 1233 (1962). 5 A. Chapiro and F. Laborie, C.R. Acad. Sci., Paris, 267," 1110 (1968). " A. Chapiro, Plastiques Modernes et Elastomeres 23, No. 5, 154 (1971). 7 A. Chapiro. J. Dulieu and F. Laborie. 3rd Internat. Symp. Radiation Chemistry. Tdiany (\9Ί\). 8 C. E. Schildknecht, Vinyl and Related Polymers, pp 317-322. Wiley : New York (1952). 9 D. G. Currie, F. S. Dainton and W. S. Watt, Polymer (London), 6, 451 (1965).

10 V. F. Gromov, A. V. Matveeva, A. D. Abkin. P. M. Khomikovskii and E. I. Mirokhina. Dokl. Akad. Nauk SSSR, 179, 374 (1967).

11 V. F. Gromov, P. M. Khomikovskii and A. D. Abkin, Vysokomol. Soedin. 12. 767 (1970). 12 L. Perec. Unpublished results. 13 A. Chapiro and Nguyen-Thi Tuyet-Hao, J. Chim. Phys. 68, 1070 (1971). 14 Le Thi-Nhi, Unpublished results. 15 V. A. Kabanov, V. A. Petrovskaya and V. A. Kargin, Vysokomol. Soedin. Ser. A, 10,925 (1968). lb A. Chapiro, G. Bex, A. M. Jendrychowska-Bonamour and T. O'Neill. Advcmc. Chem. Ser.

91,560(1969). 17 A. Chapiro and A. M. Jendrychowska-Bonamour, C.R. Acad. Sci., Paris, 270. 27 (1970).

86

PHOTOPHYSICAL PROCESSES AND THEIR ROLE IN POLYMER PHOTOCHEMISTRY

ROBERT B. F O X

Naval Research Laboratory, Washington, DC 20390, USA

ABSTRACT Elementary photophysical processes, well defined in the field of small molecules. are discussed in a polymer context. Examples of most of these processes have been adduced for specific polymers. Intramolecular energy migration is of particular importance in linear vinyl-ring polymers. The relationships between photophysical processes and photochemical processes such as polymer

photodegradation are stressed.

INTRODUCTION With the first plastic material came the first plastic material degradation

problem. For many purposes, it was immediately clear that a major causative agent for many kinds of degradation was light. Complicating the problem is the involvement in degradation of other environmental constituents—water, oxygen, atmospheric 'pollutants', as well as constituents of the plastic material itself, added for some specific reason or present as impurities.

Gathered together, these causes have been pointed to the effect called 'weathering'. And generally, 'weathering' has been approached like weather : to avoid sunburn, one keeps his hat on. Likewise, to avoid light-generated degradation in a plastic, one adds something that absorbs the light in preference to the plastic. Occasionally, it was a disconcerting fact that the absorber did more harm than good, and almost always the absorber seemed to be consumed by some photoreaction as it did its work.

In recent years, it has become apparent that the empirical approach to polymer photostabilization is not necessarily the best one. Coupled with the need for improved photostabilizers has come the utilization of photo-reactions of polymers in other contexts. Clearly, a knowledge of the fundamental processes involved in the photochemistry of polymers will lead to a more efficient application of photophenomena in the polymer field. Much of this knowledge is already available from studies of the photochemistry and photophysics of non-polymeric organic compounds1-6, and it is generally extendable to organic polymers.

Many processes can be and are involved in the interaction between light and an organic material such as a polymer. The ultimately observed photochemical reaction products, both kind and amount, should in principle be controllable through control of the intervening steps. These may be generally

87

ROBERT B. FOX

grouped as shown in Figure 2. It is convenient to think of such steps as photophysical and photochemical processes, with the dividing line between them set by the first bond break occurring following absorption. The photo-physical processes are those taking place without chemical change in the system. It is the purpose of this paper to describe in a polymer context those processes that occur between the event of photon absorption and the event of bond dissociation. Particular emphasis will be placed on the photophysical processes that are expected to be more prominent in long chainlike molecules, such as linear organic high polymers, than in 'small' molecules whose dimensions are of the same order of magnitude in all directions.

Excited polymer and impurities

Photophysical processes

Photon absorption

(-f)

Bond dissociation

Photochemical processes

(-£)

Undegraded polymer

and impurities Degraded polymer

Figure L General steps in the overall photochemical process

PHOTOPHYSICAL PROCESSES AND EXCITED STATES IN ORGANIC MOLECULES

Unimolecular processes Most organic molecules contain an even number of electrons. An unexcited

molecule in which the electron spins are paired is said to be in the ground electronic singlet state, *M, where the term 'singlet' and the superscript refer to the spin multiplicity of the electronic state. Absorption of a photon by the molecule results in excitation of the molecule into a new state of higher energy. If the transition occurs from a ground singlet state without change of spin, the excited electronic state will also be a singlet, lM*. Higher excited singlets, lM** etc., will also be formed if energy of appropriate frequency is absorbed. A plot of the intensity of absorption of the incident light against its frequency or wavelength is called the electronic absorption spectrum. If in an excited state two electron spins become unpaired, a triplet state, 3Af*, will result. Transitions between states of unlike multiplicity are spin-forbidden and occur to a limited extent only because of spin-orbit coupling between the singlet and triplet states. Absorption spectra for the 3M* <- lM transition in principle can also be measured, but the optical densities are usually of the order of 10~6 of those of the 1M* <- lM transition.

88

PHOTOPHYSICAL PROCESSES ROLE IN POLYMER PHOTOCHEMISTRY

Absorption of light energy by a molecule to yield an excited state is the first of a series of photophysical processes that may occur before the molecule returns to the lM state or it dissociates; Birks2 has discussed the kinetics of these processes in detail. The excited electronic states with their associated manifolds of vibrational levels are shown as a function of energy in Figure 2. The energy of a triplet electronic state is always lower than that of the corresponding singlet state (Hund's rule).

1M*

a» c

3/u**

Absorption-:

Fluorescence- Phosphorescence

Figure 2. Unimolecular photophysical processes. Solid lines: radiative processes; wavy lines: non-radiative processes

De-excitation of an isolated molecule can take place by radiative or radiationless transitions, shown by solid and wavy lines, respectively, in Figure 2. Fluorescence is radiative emission resulting from a transition between energy levels of like multiplicity ; this emission has a short lifetime (1CT9 to 10"5 s) and usually takes place from the 1M* level, although 3M** -* 3M* fluorescence has been observed in a few compounds. Phosphorescence is the radiation from a transition between states of different multiplicity, almost always the 3M*-► *M transition; phosphorescence lifetimes are about 10"3 to 102 s. The relationships among absorption, fluorescence, excimer fluorescence (see below) and phosphorescence spectra are shown in Figure 3. Fine structure present in the spectra of model compounds is often absent from the spectra of polymers.

Radiationless internal conversion via vibronic coupling from an upper excited level to the lowest excited level (XM* or 3M*) is usually a rapid (10~12 s) process, and for this reason emission from an upper level is only rarely seen. The same process with a slower rate may take place between the lowest excited state and the ground state, and this competes with the radiative processes of fluorescence and phosphorescence. Intersystem crossing from lM* to 3M* is a spin-forbidden radiationless transition that also competes

89

ROBERT B. FOX

with fluorescence ; for practical purposes, this is the most important route through which triplets are formed.

The probabilities that these processes will occur in an isolated molecule are dependent on the kind of electronic transition that takes place, i.e. whether the transition is spin, symmetry, and parity allowed, and whether vibronic and spin-orbit coupling are important in the system. In a carbocyclic aromatic ring, as found in polystyrene, the longest wavelength transition is π* <- π. In an aliphatic ester, such as poly(methyl methacrylate), ketone, acid,

Prompt and delayed

fluorescence Phosphorescence

Excimer fluorescence

-*—Wavenumber Wavelength +-

Figure 3. Spectral relationships in absorption and emission bands

or amide, the longest wavelength absorption band will involve a non-bonding electron and the transition will be π* <- n (symmetry forbidden) with other transitions at much shorter wavelengths. Poly(N-vinylcarbazole) and other compounds containing N-heterocyclic rings will show both π* <- π and π* <- n as low energy transitions with the energy levels quite close together and sometimes easily transposable by environmental influences or substitu-ents. As a first approximation, spin-orbit coupling is not operative between states of the same configuration7. Where a coupling between, say, a singlet π* <- n and a triplet π* <- π state is allowed and the intersystem crossing between them takes place typically in 10 ~9 sec, the corresponding transition to a triplet π* <- n state might require 10"5 sec. Thus, it would be expected that the ratio of phosphorescence to fluorescence would be greater in poly(l-vinylquinoline) than in poly(l-vinylnaphthalene) because in quinoline there is a π* <- n triplet lying between the lowest π* <- π singlet and π* <- π triplet levels, while in naphthalene intersystem crossing must be predominantly between π* <- π states8.

Other unimolecular processes are less frequently encountered but in certain circumstances they may prove to be quite important. One such is Έ-type' (for eosin) delayed fluorescence, which results from the thermal activation of 3M* followed by 3M* -» lM* back intersystem crossing, the reverse of the process shown in Figure 2 ; the emission will have the spectrum

90


of normal fluorescence but a relatively long lifetime (hence 'delayed') depending on the triplet lifetime. This and previously mentioned unimolecular processes are the result of excitation by a single photon.

Under very intense excitation from a light flash or a laser beam, biphotonic processes may also occur. Absorption of a second photon by an excited molecule will raise the energy of the molecule to an upper excited level and could ultimately lead to ionization if the photons are sufficiently energetic. In a typical flash experiment, solutions of polymers and copolymers of 1-vinylnaphthalene are exposed to a 'main' flash to give a triplet population, and this flash is followed almost immediately by a second 'spectral' flash through which the 3A/** <- 3M* spectrum and its intensity are measured to determine the extent to which triplets are formed in the polymers under specific conditions9.

Bimolecular processes While each of the aforementioned processes can take place to a greater or

lesser extent in any Isolated' molecule and certainly can and do take place in organic polymers, it is evident that in practical terms the situation is more complicated. Aside from the problems of chain entanglement even in highly dilute solution, which may lead to special cases of inter-chain interactions, it is clear that intra-chain phenomena may well dominate the photophysical picture due to the close proximity (3-4 Â) of adjacent chromophores on a chain. For the present purpose, interactions between chromophores, whether on the same or different chains, can be considered in the class of bimolecular processes.

Perhaps the most ubiquitous of the intermolecular photophysical processes is non-radiative energy transfer, which can involve both singlets and triplets. Transfer of electronic energy occurs during the lifetime of the excited donor molecule, before emission can take place. The act of transfer can be the result of collisional interaction between donor and acceptor, involving a wide variety of mechanisms and excited state intermediates ; it may be due to electron exchange interaction over distances of about 6-15 Â; or it can take place by dipole-dipole resonance (Coulombic) interaction over distances between 20 and 60 Â, that are large compared to molecular diameters. Experimentally, transfer is readily observed through the quenching of emission from a donor molecule and/or a sensitization of emission from an acceptor having a generally lower energy level than the donor. The phenomenon can be seen in gases, liquids, or solids, although collisional processes are inhibited in rigid systems. In polymers, energy transfer has usually been studied in fluid solution, but it has been observed in a few instances in solid polymers in such systems as poly(l-vinylnaphthalene)-benzophenone10.

Energy migration between molecules of the same kind can also take place. Again, the migration may be via either singlets or triplets. It has been studied in greatest detail in organic crystals, but does occur in liquids as well. In crystals, the migration is usually by way of exciton hopping or a coherent exciton wave motion. Since most of the molecules in the crystal lattice are energetically equivalent, absorbed energy is rapidly delocalized throughout the crystal as an exciton wave that migrates until it degrades either radiatively or non-radiatively or is trapped by transfer to a crystal defect or

91

ROBERT B. FOX

Figure 4. Energy relationships for excimer emission

an impurity. In the latter case, emission from the impurity or defect site is frequently observed. The same kinds of energy migration can also take place in linear polymers, which might be viewed as unidimensional crystals. This will be discussed in detail below.

Of the excited state intermediates formed by bimolecular processes, the best known are the excimers2, n , formed by the interaction of an excited singlet molecule with a ground state molecule of the same species. This excited state intermediate is dissociated in the ground state and therefore cannot be detected in the absorption spectrum. Its fluorescence consists of a broad structureless band shifted 4000-6000 c m - 1 to the low-energy side of the normal fluorescence (see Figure 3). In a very few cases, excimer phosphorescence has also been observed. Some of the energetic relationships are shown in Figure 4.

300

Poly(/V-vinylcarbazole) N- isopropylcarbazole 1,3-Dicarbazolylpropane

Solution

350 400 Wavelength, nm

Figure 5. Absorption and emission spectra for poly(N-vinylcarbazole) and model compounds

92


Intermodular excimer formation is concentration-dependent, since it is a diffusion-controlled interaction. Intramolecular excimer formation can take place independently of concentration by the interaction of an excited chromophore with another chromophore in the same molecule if the two are in proper geometrical configuration. HirayamaJ 2 observed excimer formation in diphenylalkanes only when the phenyl groups were separated by three carbon atoms. Excimer fluorescence is the emission usually observed from room-temperature solutions of vinyl-ring polymers such as polystyrene. An example of an excimer spectrum is shown in Figure 5 for poly(N-vinyl-carbazole)13, together with the spectra for the model compounds 1,3-dicarbazolylpropane14 and N-isopropylcarbazole15.

Excited state intermediates can also be formed between chromophores of different species ; in this case, they are termed exciplexes, but their characteristics are similar to those of excimers. Although exciplexes have received less attention than excimers, they may well be more common. An excited state proton transfer interaction in 4-methylumbelliferone, for example, has been made the basis for an organic laser16. Exciplexes formed between different aromatic hydrocarbons have been observed«, and donor-acceptor excited state complexes, which are a kind of exciplex formed, for example, between aromatic hydrocarbons and tertiary amines, have been studied.

There are many other types of bimolecular photophysical interactions that might be considered2. Excited states may be perturbed in the presence of paramagnetic molecules such as oxygen or molecules containing 'heavy' atoms, such as ethyl iodide4. The general effect is to enhance spin-orbit coupling in the perturbed molecule and thereby increase the rate of inter-system crossing processes. A particularly important process is that of triplet-triplet annihilation, an interaction between two excited states that may lead to excimer formation, the dissociation of which gives an excited singlet state and a ground state molecule :

3M* -f 3M* -► f1/)*] -► lM* + lM

The radiative emission from the lM* is identical to normal fluorescence but with a lifetime proportional to the square of the triplet lifetime. This emission is called P-type (for pyrene) delayed fluorescence, and it has been observed in dilute solutions of certain aromatic polymers such as poly(l-vinyl-naphthalene)17.

Each of these excited state intermediates has its own unimolecular photo-physical processes. Each, in one form or another, will probably eventually be observed in some polymer system, and each will be as important to the ultimate photochemistry of the polymer as it is in the analogous model compound system.

PHOTOPHYSICAL PROCESSES IN POLYMER MOLECULES The development of polymer photochemistry had its genesis in a need to

stabilize plastic materials toward weathering. Consequently, much of the research has been of an applied nature and has been oriented toward improvement of material properties. In recent years, however, investigations in polymer photochemistry have been turning toward an understanding of

93

ROBERT B. FOX

photochemical pathways as a means of controlling material properties. Necessarily, this has involved the photophysical processes delineated in the preceding section. In general, these processes have been invoked to rationalize empirical results, but, increasingly, an insight into the processes themselves in polymers is being gained.

Establishment of the nexus between polymer photochemistry and photo-physics is receiving emphasis in spite of difficult problems of purification in polymer systems. Pivovarov and co-workers18 '19 related 'energy transfer' in polystyrene and polypropylene systems to the stabilizing effect of certain additives over and above that resulting from light shielding. In their studies of the photodegradation of ethylene-carbon monoxide copolymers, Heskins and Guillet20 implicated the triplet state of the polymer by relating the reduction in quantum yield for chain scission to the concentration of an added triplet quencher, cyclooctadiene. Similarly, the rate of photo-degradation of poly(methyl methacrylate) containing an unknown impurity could be roughly correlated with the triplet levels of a variety of inhibitors and accelerators21. A necessary background to further research in these directions is information on the most prominent photophysical processes in such systems. It is therefore of interest to examine the detailed emission spectra of highly purified prototype polymers with the object of learning something of energy transfer and migration processes and excited states relative to the phenomena seen in non-polymer molecules.

Intermolecular electronic energy transfer in polymers There is abundant evidence pointing to intermolecular electronic energy

transfer from polymers to solutes from work in the scintillator field22'23. Polystyrene is the most frequently used solvent, and transfer is by way of the polymer excimer (see below) or singlet excited styrene monomer present as an impurity ; a wide variety of acceptors such as terphenyl, anthracene, and other common scintillator molecules have been employed. Transfer usually takes place by single-step long-range dipole-dipole interaction.

Lashkov and Ermolaev24 investigated solid films of poly(N-vinyl-phthalimide) containing additives, such as benzil or anthracene, having triplet energy levels lower than that suggested by phosphorescence of the polymer. Both a sensitization of the additive phosphorescence and a quenching of the polymer phosphorescence were observed, clearly demonstrating an intermolecular transfer of triplet energy from polymer to additive ; transfer among singlet levels was not observed. It was proposed that the act of transfer was preceded by triplet migration along the polymer chains and that the additive molecules acted as traps in a manner similar to energy trapping in doped organic crystals.

Intermolecular energy transfer may also be a biphotonic process. The naphthalene-sensitized photolysis of poly(dimethylsiloxane) at 77°K has been shown25 by an electron paramagnetic resonance study of the decay of naphthalene triplets and the formation of radicals from the polymer (no naphthalene-derived radicals were observed) to proceed by way of absorption of a photon by the lowest excited triplet state of naphthalene, followed by energy transfer from the upper triplet to the polymer.

As noted above, David and co-workers10 have studied the quenching of 94


excimer emission at 77°K from solid poly(l-vinylnaphthalene) by benzo-phenone and by anthracene ; sensitized emission from benzophenone was not observed, but anthracene fluorescence was emitted from the system. In vinylnaphthalene-vinylbenzophenone copolymers, no emission was observed from either donor or acceptor. Application of the Forster theory for long-range dipole-dipole singlet interaction showed good agreement between experiment and theory with a critical transfer radius, R0, of about 15 Â.

David and co-workers have also investigated triplet energy transfer from poly(vinylbenzophenone)26 and from poly(phenyl vinyl ketone)27 to naphthalene as acceptor in solid films at 77°K. In both of these systems, triplet formation is a highly efficient process in the polymers as well as in the corresponding model compounds. The systems were excited with 366 nm radiation, which is absorbed by the polymer but not by naphthalene. Triplet energy transfer took place by the mechanism

3M* + 1Y-> 1M + 3y* where Y is the acceptor naphthalene, the sensitized phosphorescence of which was followed as a function of concentration. For poly(vinylbenzo-phenone) and poly(phenyl vinyl ketone), R0 was 36 Â and 26 Â, respectively, which may be compared to the value of 13 Â reported28 for the benzo-phenone-naphthalene system. The difference between the polymers and the model system was attributed to energy migration in the polymer prior to transfer. The nature of the migrating exciton and the question of inter-versus intra-molecular migration were not considered. Naphthalene was observed to be an inhibitor of the photodegradation of both polymers at room temperature, but it should be noted that degradation mechanisms at room temperature and at 77°K need not be identical. Lukâc and co-workers29, investigating the inhibition of the main-chain scission photolysis of poly(phenyl vinyl ketone) by triplet quenchers in benzene solution at 30°C, found that naphthalene as part of the polymer chain in phenyl vinyl ketone-vinylnaphthalene copolymers was 21 times more efficient than free naphthalene in quenching the photolysis. This would appear to suggest the idea of some form of intramolecular migration and transfer under these conditions.

Many additional examples of intermolecular electronic energy transfer involving polymers have been reported. In our own laboratory, for example, we have observed in dilute glasses at 77°K the quenching of both poly(l-vinylnaphthalene)17 and polystyrene30 phosphorescence by triplet quenchers such as piperylene, and we have sensitized the phosphorescence of poly(l-vinylnaphthalene) with benzophenone17 by processes analogous to those cited for the poly(vinylbenzophenone)-naphthalene system26.

Intramolecular electronic energy transfer in polymers Linear organic polymers such as polystyrene differ from their low molecular

weight analogues in at least two significant respects : adjacent chromophores on a polymer chain are typically 3-4 Â apart, and polymer chains may have long sequences of such closely spaced chromophores, but the sequences may have internal orientations. Thus orbital overlap between chromophores is easily visualized, and it is a simple step to the concept of some form of energy

95

ROBERT B. FOX

migration through chains of interacting chromophores. Such migration would be crucial in the photochemical fate of a polymer chain, for absorbed energy would seek out an 'energy trap'. If the trap itself dissociated, a main chain break might take place ; if the trapped energy were dissipated harmlessly, a stabilization of the chain would result. The latter appears to apply to the phenyl vinyl ketone-vinylnaphthalene copolymers investigated by Lukâc and co-workers29.

Delayed fluorescence has been utilized to demonstrate intramolecular triplet migration at 77°K in dilute glasses containing poly(l-vinylnaph-thalene)17, poly(L-tyrosine) or poly(adenylic acid)31. Where the polymer emits

(PVN) x10* bM □ 1 Δ 2.5 o 5

5X10" 4 M 1-Ethylnaphthalene

0.01 0.02 0.03 0.04 Piperylene

0.05

Figure 6. Quenching of phosphorescence of poly(l-vinylnaphthalene) and 1-ethylnaphihalene by piperylene

delayed fluorescence, the low molecular weight model compound [1-ethylnaphthalene in the case of poly(l-vinylnaphthalene)] at an equivalent concentration of chromophores shows only phosphorescence in the delayed emission spectrum. The delayed fluorescence was shown to be the result of triplet-triplet annihilation by the finding that the process was second order with respect to the intensity of the exciting light. Intramolecular triplet migration was also supported by the observation that the phosphorescence of the polymer but not that of the model compound was quenched by piperylene17. These results are shown in Figure 6.

The intramolecular migration of excitation energy at 77°K can be shown in polymers such as polystyrene by incorporating traps as part of the chain, i.e. by utilizing copolymers in which one sequence acts as a trap for energy initially absorbed by a second sequence. Thus, intramolecular migration is

96


followed by intramolecular transfer. In Figure 7 is shown an example of such a process in which sequences derived from styrene transfer triplet energy to sequences derived from 1-vinylnaphthalene32. Here, the phosphorescence from the styrene sequences is quenched and that from the naphthalene-containing sequences is sensitized relative to an equivalent mixture of the homopolymers in a dilute glass. The same experiment with monitoring of fluorescence quenching and sensitization shows that singlet energy also migrates and is transferred intramolecularly to the naphthalene sequences in copolymers from styrene and 1-vinylnaphthalene9.

in Έ Z3

>% L. Ö 1_

15 o

-4-·

C

A -1.02 Mole-per cent VN in PVN/S B-1.02 Mole-per cent PVN in mixture with PS Aex 260 nm,

-

Delayed j fluorescence /

(PVN) /

77°K, (VN) = 10 _ 3 M

/ V

/ Polystyrene \ (phosphorescence^

A

I

/ \ / \A

\ / / V /Λ ^ Polyvinylnaphthalene ^ - ^ i

phosphorescence >J 1 I I

350 400 450 500 550 Wavelength, nm

Figure 7. Intramolecular energy migration in styrene sequences and transfer to 1-vinylnaphthalene sequences in a 1-vinylnaphthalene-styrene copolymer

In certain cases where donor and trap segment emission occur at nearly the same wavelength, time-resolved phosphorescence measurements can be used to demonstrate intramolecular triplet migration and transfer. For example, at 77°K in a rigid glass, τρ for polystyrene and poly(4-chlorostyrene) are 3.0 and 0.004 sec, respectively. A copolymer of styrene containing two per cent 4-chlorostyrene emits phosphorescence having a τρ of 0.033 sec9.

These kinds of experiments suggest that organic copolymers behave in many ways like unidimensional analogues of mixed organic crystals. Unlike the organic crystals, in which the energy trap may be likened to a point in a tridimensional matrix, the copolymers may contain traps having rather long sequences of chromophores which may themselves be studied. Rise and decay times of the trap emission provide an insight into phenomena within the trap. We have found30 that phosphorescence is emitted from sites near

97 P.A.C—30/1—E

ROBERT B. FOX

the ends of the trap sequences and that both τρ and its energy level are affected by the length of the trap. Delayed fluorescence from within the naphthalene-containing trap sequences in 1 -vinylnaphthalene copolymers is the result of the annihilation of free triplets, while in poly(1-vinylnaphthalene) itself, the annihilation occurs between a free triplet and a triplet localized within the trap, perhaps at some chain defect.

Unanswered questions relate to the nature of the migrating species, the point at which intersystem crossing takes place, and to the absolute rate of migration. Some of these questions are partially answered in the styrene-1-vinylnaphthalene copolymer case by the finding that while piperylene will intermolecularly quench triplets (i.e. phosphorescence) in polystyrene at 77°K, it does not quench phosphorescence from short-sequence traps in the copolymers30. Either energy in polystyrene migrates as singlets (which are not quenched by piperylene) and triplet formation occurs at the point of trapping, or the rate of intramolecular migration of triplets greatly exceeds the rate of intermolecular transfer to piperylene.

The phenomena just discussed concern findings with polymers at quite low temperatures and in rigid dilute solutions. Such migration and transfer processes may very well also take place at room temperatures with fluid solutions or solid polymers, but there is little doubt that there will be other competing processes, not the least of which are the bimolecular interactions. Results such as those obtained with the photodegradation of the phenyl vinyl ketone-vinylnaphthalene copolymers29 strongly indicate that some form of migration and transfer does occur in this system in the solid at room temperature.

One photophysical process that stands out in solutions of vinyl-ring polymers at room temperatures is intramolecular excimer formation. Hirayama12 '33 observed intramolecular excimer emission from fluid solutions of polystyrene and two-unit model compounds for polystyrene, the 1,3-diphenylpropanes. Excimer formation in poly(l-vinylnaphthalene)34, poly(N-vinylcarbazole)13, and the corresponding two-unit model compounds have also been reported. In general, intramolecular excimer fluorescence should be observed from polymers derived from compounds that themselves form excimers, provided that certain conformational requirements are met. A fairly specific relationship must exist between adjacent chromophores at the excimer site. Excimer formation in the paracyclophanes34 and the diphenylalkanes12 indicates that the chromophores should be no more than 3.7 Â apart and should lie in a somewhat parallel conformation. Activation energies for intramolecular excimer formation in l,3-di(l- and 2-) naphthyl-propanes are approximately 4.0 kcal mol" \ a value similar to that for the rotation barrier in a méthylène chain35.

An example of the variation of fluorescence spectra with temperature where intramolecular excimer formation takes place is shown for poly(2-vinylnaphthalene) in Figure 836. The solutions are fluid throughout this temperature range, showing that a rigid matrix is not necessary to exclude the formation of excimers. As was done with the 2-unit model compounds35, activation energies for intramolecular excimer formation can be determined from plots of the logarithm of the ratio of excimer to molecular fluorescence intensities against 1/T°K. Such plots for poly(2-vinylnaphthalene) and

98


-

-161°C

L | -129° \

/ \ \ -71° \ \

20° ^^''

Excimer

-71°

y^I^v. r X s ^ N > ^ Ì 6 P ^ ^ ^ ^

300 400 Wavelength, nm

Figure 8. Variation of fluorescence of poly(2-vinylnaphthalene) with temperature

10 h

lil.O

0.1

Poly (2-vinvlnaphthalene)

Polystyrene

5 6 7 1000 IT

8

Figure 9. Excimer activation energy plots for poly(2-vinylnaphthalene) and polystyrene

99

ROBERT B. FOX

polystyrene are shown in Figure 936. Activation energies of 3.5 and 1.4 kcal m o l - 1 were obtained. For these polymers, excimer intensities reach a maximum not far below room temperature. The contrast between the 'high' and 'low' temperature regions has been discussed by Birks2.

In polymers, a second conformational requirement appears to involve alignment of the chain itself34'37. The observation of intramolecular excimer formation in polymers constitutes evidence for intramolecular electronic energy migration. Klöpffer13 has shown that in solid poly(iV-vinylcarbazole) excimer-forming sites and guest molecules compete in trapping migrating

300 400 500 Wavelength, nm

Figure 10. Fluorescence of poly(l-vinylnaphthalene): solid film (solid line); dilute solution in solid poly(methyl methacrylate) (dashed line) (Temperature in °K)

excitons. Energy transfer from an excimer can take place by the dipole-dipole mechanism13 '33. In Figure 10 is shown the fluorescence spectrum of poly(l-vinylnaphthalene) as a solid film and as a solid solution in poly(methyl methacrylate) at room temperature and at 77°K36. Unlike the fluid solution, the solid shows little decrease in excimer emission as the temperature is lowered. At 77°K in a rigid glass formed at low temperatures, excimer formation is absent34. It would appear that chain conformation at the time the rigid matrix is formed is the important criterion.

Clearly, excimer formation must influence the photochemistry of these polymers. If the excimer acts as a singlet energy trap, there arises the question of whether this trap is a source of observed photochemical reactions such as degradation. It is possible that some other excited state such as the triplet is the source of bond-breaking. We have found that the triplet population is to

100


a large extent controlled by the extent of excimer formation in a series of styrene-1-vinylnaphthalene copolymers9 '36. As the fraction of 1-vinyl-naphthalene-derived units in the copolymer increased, excimer fluorescence intensity also increased. At the same time, the triplet population, measured by the intensity of the 3M** <- 3M* absorption, decreased. It remains to be shown the extent to which each of these excited states is, in fact, responsible for photochemical degradation in vinyl-ring polymers.

SOME APPLICATIONS TO POLYMER PHOTOCHEMISTRY Armed with information on the fundamental photophysical processes

occurring in any or all constituents in a given polymer system, the polymer chemist not only can do much to explain the empirical observations made during light-induced polymer reactions, but he can also apply this information to predictably control the course of these reactions. One might consider as factors controlling the ultimate fate of a molecule that has absorbed light energy : (a) the thermal excitation in a given energy level ; (b) the lifetime of the excited state ; and (c) the orbital nature of the excited state. On excitation, normally to some high vibrational level within an excited state manifold, the molecule is said to be 4hot\ While the molecule usually returns to the ground vibrational level of the excited electronic state within picoseconds, during its brief 'hot' period the molecule can undergo dissociation to free radicals. On the other hand, the lifetime of an excited state will control its participation in bimolecular reactions. Thus, one commonly observes short-lived fluorescence at room temperature in fluid solution, but phosphorescence produced by long-lived triplets is only rarely seen under these conditions. The nature of the excited singlet state will certainly control the extent to which intersystem crossing to the triplet level will occur. Note may be made that the lowest excited singlet level is π* <- n in aromatic carbonyl compounds, characterized by high intersystem crossing rates and short-lived triplets. One might venture on this basis to predict that an aromatic carbonyl-containing energy trap in a polymer chain would be quite efficient as a stabilizer.

Attention in this report has been directed primarily toward the photo-physical processes in polymers ; the photochemistry of a polymer also may well be dependent on its interaction with some other excited species. A notable example is the attack of singlet oxygen on polymer chains containing unsaturation. Another is the apparent solvent effects seen in solution photodegradation studies of polymers such as poly(methyl methacrylate).

As illustrations of reports in which photophysical processes have been utilized in the interpretation of polymer phenomena, recent energy transfer studies in the polystyrene-pyrene38 and in the nylon-6,6-proflavine39

systems may be cited. In both papers, dipole-dipole energy transfer mechanisms were implicated. Briggs and McKellar40 have suggested that photostabilization of polypropylene by nickel oxime chelates proceeds by a triplet transfer mechanism to the chelate from carbonyl-containing oxidation sites in the polymer. The work of Lukâc and co-workers29 establishing an intramolecular transfer reaction in the photodegradation of poly(phenyl vinyl ketone) may again be mentioned.

101

ROBERT B. FOX

Polymers have been frequently used as photosensitizers and photo-reactions in polymers by way of external photosensitization have been studied intensively. For example, 'naphthoylatecT polystyrene has recently been used to photosensitize the photoisomerization of eis- and trans-stilbene41. Poly(vinylbenzophenone) and poly(phenyl vinyl ketone) have also been used as photosensitizers. Poly(vinyl cinnamate) is an example of a photosensitive polymer with which external photosensitizers have been used.

It is beyond the scope of this report to dwell exhaustively on all the areas of polymer science in which the underlying processes are photophysical ones. Suffice it to say that wherever a light-induced polymer interaction takes place, one or more photophysical processes will be involved. Once these processes can be identified in a given context, the polymer chemist is in a position to achieve a fair degree of control over the photochemistry of his polymer system.

ACKNOWLEDGEMENT The author acknowledges with thanks the many valuable discussions he

has had with his colleagues, Prof. Robert F. Cozzens of the George Mason College of Virginia, and Dr J. R. McDonald and Mr T. R. Price of the Naval Research Laboratory.

REFERENCES 1 J. G. Calvert and J. N. Pitts Jr, Photochemistry, Wiley: New York (1966). 2 J. B. Birks, Photophysics of Aromatic Molecules, Wiley-Interscience : London (1970). 3 C. A. Parker, Photoluminescence of Solutions, Elsevier: Amsterdam (1968). 4 S. P. McGlynn, T. Azumi and M. Kinoshita, Molecular Spectroscopy of the Triplet State,

Prentice-Hall : Englewood Cliffs, N.J. (1969). 5 J. N. Pitts Jr, G. S. Hammond and W. A. Noyes Jr, eds., Advances in Photochemistry, Inter-

science : New York ; a continuing series. 6 H. H. Jaffé and M. Orchin, Theory and Applications of Ultraviolet Spectroscopy, Wiley :

New York (1962). 7 S. K. Lower and M. A. El-Sayed, Chem. Rev. 66, 199 (1966). 8 M. A. El-Sayed, J. Chem. Phys. 38, 2834 (1963). 9 R. B. Fox, T. R. Price and R. F. Cozzens, unpublished results.

10 C. David, W. Demarteau and G. Geuskens, Europ. Polym. J. 6, 1397 (1970). 11 Th. Förster, Angexv. Chem., Intl. Ed. 8, 333 (1969). 12 F. Hirayama, J. Chem. Phys. 42, 3163 (1965). 13 W. Klöpffer, J. Chem. Phys. 50, 2337 (1969). 14 a W. Klöpffer, Chem. Phys. Letters, 4, 193 (1969).

6 W. Klöpffer and W. Liptay, Z. Naturforsch. 25a, 1091 (1970). 15 W. Klöpffer, J. Chem. Phys. 50, 1689 (1969). 16 C. V. Shank, A. Dienes, A. M. Trozzolo and J. A. Myer, Appi. Phys. Letters, 16, 405 (1970). 17 R. F. Cozzens and R. B. Fox, J. Chem. Phys. 50, 1532 (1969). 18 A. P. Pivovarov, Yu. A. Ershov and A. F. Lukovnikov, Plast. Massy, 1 (1966). 19 A. P. Pivovarov and A. F. Lukovnikov, Vysokomol. Soedin. A9, 2727 (1967); Polym. Sci.,

USSR, 9, 3087(1967). 20 M. Heskins and J. E. Guillet, Macromolecules, 1, 97 (1968). 21 R. B. Fox and T. R. Price, J. Appi. Polym. Sci. 11, 2373 (1967). 22 R. K. Swank and W. L. Buck, Phys. Rev. 91, 927 (1953). 23 J. B. Birks, The Theory and Practice of Scintillation Counting, Pergamon: Oxford (1964). 24 G. I. Lashkov and V. L. Ermolaev, Optics and Spectrosc. 22, 462 (1967). 25 S. Siegel and H. Judeikis, J. Chem. Phys. 43, 343 (1965). 26 C. David, W. Demarteau and G. Geuskens, Europ. Polym. J. 6, 537 (1970).

102


27 C. David, W. Demarteau and G. Geuskens, Europ. Polym. J. 6, 1405 (1970). 28 M. Inokuti and F. Hirayama, J. Chem. Phvs. 43, 1978 (1965). 29 I. Lukâc. P. Hrdlovic, Z. Manâsek and D. Bellus, J. Polym. Sci. ^ -1 ,9 . 69 (1971). 30 R. B. Fox, T. R. Price and R. F. Cozzens, J. Chem. Phys. 54, 79 (1971). 31 J. W. Longworth and M. del C. Battista, Photochem. Photobiol. 11, 207 (1970). 32 R. B. Fox and R. F. Cozzens, Macromolecules, 2, 181 (1969). 33 F. Hirayama, L. J. Basile and C. Kikuchi, Molec. Crystals, 4, 83 (1968). 34 M. T. Vaia Jr, J. Haebig and S. A. Rice, J. Chem. Phys. 43, 886 (1965). 35 E. A. Chandross and C. J. Dempster, J. Amer. Chem. Soc. 92, 3586 (1970). 36 R. B. Fox, T. R. Price, J. R. McDonald and R. F. Cozzens, presented at XXIIIrd Congress of

Pure and Applied Chemistry, Boston (1971). 37 I. H. Hillier and S. A. Rice, Proc. Nat. Acad. Sci. Wash. 53, 973 (1965). 38 J. Wilske and H. Heusinger. J. Polym. Sci. A-U 7. 995 (1969). 39 H. H. Dearman, F. T. Lang and W. C. Neely, J. Polym. Sci. A-2, 7, 497 (1969). 40 P. J. Briggs and J. F. McKellar, J. Appi Polym. Sci. 12, 1825 (1968). 4 1 H. A. Hammond, J. C. Doty, T. M. Laakso and J. L. R. Williams, Macromolecules. 3, 711

(1970).

103

PHOTOCHEMISTRY OF UNSATURATED POLYMERS

MORTON A. GOLUB

Ames Research Center, National Aeronautics and Space Administration, Moffett Field, Calif. 94035, USA

ABSTRACT The mechanisms of the direct photochemical reactions which occur in unsaturated polymers when irradiated as thin films in vacuo are reviewed. Important reactions in 1,4-polyisoprene and 1,4-polybutadiene are cis-trans isomerization, loss of 1,4-unsaturation, formation of new external double bonds (vinylidene and/or vinyl units) and cyclopropyl formation. In 1.2-polybutadiene and 3,4-polyisoprene, on the other hand, the main reaction is consumption of the external double bonds through cyclization. Recent work on the photoinduced microstructural changes in eis and trans polypentenamers

is also discussed.

INTRODUCTION During the past decade extensive work has been carried out on the photo

chemistry of polymers, especially their photodegradation and photo-oxidation. As recent reviews1-7 have given little attention to unsaturated polymers, and then mainly to photosensitized processes, this paper will comprise a review of the unsensitized photochemical transformations which occur in unsaturated polymers when irradiated as thin films in vacuo. The photoinduced microstructural changes to be discussed are those reported previously for polymers having principally internal C = C double bonds (eis- and trans- 1,4-polybutadiene and 1,4-polyisoprene) or external double bonds (1,2-polybutadiene and 3,4-polyisoprene). In addition, some new related work on the photochemistry of eis and trans polypentenamers will also be examined.

1,4-POLYBUTADIENE The first investigation of the direct photoinduced microstructural changes

in an unsaturated polymer was reported at the IUPAC Symposium on Macromolecular Chemistry in Prague8. That study, consisting of both near (2537 Â) and far (1236Â) ultra-violet irradiation of eis- and trans-1,4-polybutadiene films in vacuo, was an extension of earlier work in which these polymers were found to undergo an unsensitized radiation-induced isomerization in the solid state9. The latter process was attributed to excitation of the π-electrons of the double bonds to an antibonding state in which

105

MORTON A. GOLUB

free rotation and hence geometric interconversion can occur. This interpretation implied that an analogous unsensitized photoisomerization could also take place if the polymers were exposed to suitable radiation in the far or vacuum ultra-violet where unconjugated C—C bonds show significant absorption. Irradiation of very thin ds-1,4-polybutadiene films in vacuo with 1236 Â radiation8 did in fact lead to cis-trans isomerization, with a quantum yield of ~0.25. This was accompanied by a large decrease in unsaturation, with a quantum yield of ^1.53, suggesting a chain cyclization reaction. The vacuum photochemical results were rationalized on the basis that 1 236 Â photons have sufficient energy (MOeV) to cause ionization as well as excitation in polybutadiene and thus promote reactions similar to those obtained with gamma rays. The effective G-values (or 100-eV yields) for the isomerization and consumption of double bonds in the 1236 Â work (2.5 and 15.3) may be compared with the G-values obtained for the corresponding γ-ray-induced reactions in 1,4-polybutadiene (7.2 and 13.6-7.9, respectively)10.

It was further found8 that even near ultra-violet radiation at 2537 Â causes major microstructural changes in purified eis- and trans- 1,4-polybutadiene films in vacuo at room temperature. One important reaction is isomerization which proceeds towards a photostationary cis-trans ratio of ~ 60:40. In addition, the 1,4-polybutadienes exhibit photoinduced loss of unsaturation, although not as extensively as with 1236 Â radiation. The quantum yields for 2537-Â-induced cis-trans isomerization and consumption of double bonds were ~0.036 and ~0.061, respectively; these values represent a revision11 of earlier values8. In contrast to either 1236-Â-or γ-irradiation of 1,4-polybutadiene, 2537-Â-irradiation brings about the formation of vinyl double bonds and cyclopropyl groups distributed along the backbone. The vinyl groups have been depicted8 as arising through chain scission (reaction 1) while the cyclopropyl groups might form via process 211.

—CH2—CH=CH—CH2—CH2—CH=CH—CH2 ^ —CH 2—CH=CH—CH 2 · + -CH2—CH=CH—CH2— (1)

—CH 2—CH—CH=CH 2

—CH2—CH-=CH—CH2— ^ — C H 2 C H — C H — C H 2 — ^ütSi« —CH 2 CH—CH 2 tH ► —CH2CH—CH— (2)

CH72

eis or trans

Ultra-violet irradiation of eis-l,4-polybutadiene-2,3-rf2 [(—CH2CD= CDCH2—)„] showed that formation of trans —CH=CH— in ordinary eis- 1,4-polybutadiene was due to cis-trans isomerization and not to a double bond migration posing as an isomerization (reaction 3).

—CH2CH=CH—CH2— ^ — C H 2 C H = C H - t H — «-► —CH2tH—CH=CH— (3)

106


The occurrence of a direct photoisomerization of eis- 1,4-polybutadiene in vacuo at room temperature was reported also by Ho1 2 in connection with a study of the photo-oxidation of this polymer.

1,4-POLYISOPRENE Until recently the only fundamental study reported on the photolysis of

1,4-polyisoprene was the work of Bateman13 over twenty five years ago. He showed that irradiation of degassed, purified Hevea rubber films in the wavelength range 2300-3650 Â resulted in gas evolution, mostly hydrogen, with a quantum yield of ~4 x 10~4. The largest relative yield, at 2350-2850 Â, was about 2.5 times this value, indicating an upper limit of M O " 3

for efficiency of non-condensable gas formation. Hydrogen was assumed to originate through reaction 4, while the resulting polymeric radicals combine to form crosslinks. The attendant insolubilization of the irradiated 1,4-

CH3 CH3

—CH2—C=CH—CH2— * —CH2—C=CH—ÙH— + H· CH3 CH3 (4)

H· + —CH 2—C=CH—CH 2 ► —CH2—C=CH—OH— + H2

polyisoprene restricted its chemical analysis, but evidence was adduced for loss of unsaturation with a quantum yield13,14 approaching 0.1.

As a follow-up to Bateman's pioneering work, we carried out a detailed spectroscopic study of the photoinduced microstructural changes in 1,4-polyisoprene films11. These films, although somewhat crosslinked after irradiation, were nevertheless sufficiently soluble to permit n.m.r. analysis, while infra-red analysis was performed on the films before and after irradiation. As with the 1,4-polybutadienes, 2537-Â-irradiation of 1,4-polyisoprene films in vacuo resulted in cis-trans isomerization (with a photostationary cis-trans ratio in the range 70:30 to 50:50), loss of 1,4-double bonds, formation of cyclopropyl groups, and the appearance of new external double bonds (vinylidene and vinyl units). Quantum yields for the first three of these processes were estimated at 0.041, 0.083 and 0.018, respectively.

The photoisomerization of 1,4-polyisoprene, like that of 1,4-polybutadiene, has been considered11 to proceed through electronically excited 1,4-double bonds (probably triplet states with ^74 kcal/mole) which can rotate and assume either eis or trans configuration on return to the ground state. The smooth structureless absorption tail extending to around 3000 Â ( ^95 kcal/ mole) observed in the ultra-violet spectrum of the purified polyisoprene film has been associated with the trisubstituted ethylenic unit in the polymer, perhaps as a weak singlet-triplet transition15. On this basis, the polyisoprene double bonds could be excited by direct absorption of the 2537 Â photons without requiring inter- or intra-molecular energy transfer, a point of view adopted previously by Bateman13 and by Hart and Matheson14.

Some of the energy absorbed by the double bonds is undoubtedly diverted into rupturing C—C bonds connecting successive isoprene units. That particular bond is the 4weak link' in the chain with a strength ( ~~ 55 kcal/mole)

107

MORTON A. GOLUB

which is lowered from the normal C—C bond strength by the resonance energy of the two allyl radicals formed on chain scission (reaetion 5).

CH, CH,

CH,—C=CH—CH,- CH,—C=CH—CH h\>

I CH, CH3

-CH2—C=CH—CH2· + -CH2—C=CH—CH3 (I)

ÇH3

—CH2—Ç—CH=CH2

(HI)

(ID CH3

I CH2=C—CH—CH2-

(IV)

(5)

Radicals I-IV can recombine in any of four different ways or add to double bonds in the same or other macromolecular chain [reactions 6 and 7].

ÇH3 CH3

-CH2—C—CH -CH 2 + —C CH-

CH3 CH3

I I -CH2—CH—C^CH2 + —C=CH-

CH3

—CH2—C—CH^CH2

I —C—CH—

I ' CH3

ÇH3 I

—CH2—CH—C =CH2

—C—CH—

CH,

etc. (6)

etc. (7)

The overall effect is to create vinylidene and vinyl double bonds as well as some endlinks (or crosslinks) in the irradiated 1,4-polyisoprene. Since the resulting vinylidene content was three to five times the vinyl content, radical IV is that much more reactive than III, while I and II on subsequent reaction preserve the original 1,4-double bonds, with probable retention of cis-trans stereochemistry16. To the extent that radicals I-IV 'polymerize' on to nearby double bonds to form endlinks, there is an added loss of double bonds over and above those which are transformed into vinylidene, vinyl or cyclopropyl structures.

The mode of excitation of double bonds which leads to cis-trans isomeriza-tion was adapted1 ' to explain the appearance of cyclopropyl groups, namely, formation of a biradical followed by 1,2-hydrogen migration and then ring closure [reaction 8]. However, Carstensen17, in an e.s.r. study of free radicals produced in ultra-violet-irradiated 1,4-polyisoprene, has recently proposed

108

PHOTOCHEMISTRY OF UNSATURATED POLYMFRS

CH3 I CH 3

h I -CHz-C=CH-CH2- 4 -CH2-q-cH-CH2- +

CH, CH 3 I I -CH2-q-CH2-qH- + -CHzC-CH- (8)

\ / CH 2

cis or trans

(V) an alternative route to cyclopropyl groups: if radicals I and I1 formed on chain scission possess excess energy they might undergo ring closure [reaction 9a or 9b] : the resulting cyclopropyl radicals could then recombine with the

CH3 CH3

CH, I \ /

(9b)

CH3

.CH2-C=CH-CH2- + C-CH-CH2-

(11) CH2 counterpart radicals I1 and I (or their allylic resonance forms, IV and HI), respectively. For this mechanism to hold, the radicals in 9a or 9b must have enough energy to overcome the endothermicity of the cyclopropyl ring closure, i.e. -26 kcal/mole'*. That such 'hot' radicals might arise in the photolysis of 1,4-diene polymer films is indicated by the large energy difference between the 2537 A photons ( - 112 kcal/mole) and the strength ( N 55 kcal/mole) of the C-C bond leading to radicals I and 11.

It is interesting to note that the spectroscopic evidence advanced for cyclopropyl groups in ultra-violet-irradiated 1,4-polyisoprene and 1.4-

peaks in the former polymer and 9.4- and 10.3-7 peaks in the latter) was confirmed by Pinazzi' who reported practically the same infra-red and n.m.r. data for model compounds of 1,4-polyisoprene containing structure V.

That double bond migration (reaction 10) is not a significant reaction in the photolysis of 1,4-polyisoprene was shown by the absence of 10.3-p

-CHlG-CH=CH- (10) 109

MORTON A. GOLUB

absorption (trans —CH=CH—) in the infra-red spectrum of the irradiated polymer. Likewise, double bond migration in the opposite direction can be neglected, as seen in the fact that irradiated polyisoprene-3-d did not reveal 12-μ absorption [—C(CH3)=CH—units] indicative of reaction 11.

CH3 CH3 1 * I

—CH2C=CD—CH2— % —ÇH—C=CD—CH2— CH3

4-*—CH=C—CD—CH2— (11)

Moreover, the polyisoprenyl radicals in reactions 10 and 11 were not observed in Carstensen's e.s.r. work17. Besides radicals I and II, his e.s.r. spectra showed one additional, minor peak which was assigned to polyenyl radicals, —(CR|=CR2)M—CH2\ where Rl and R2 = H or CH3, and n is about three.

These considerations reinforce the conclusion, based on quantum yield data, that reaction 4 is unimportant relative to photoinduced cis-trans isomerization or loss of unsaturation ; as noted above, quantum yields for the latter processes are some 40-80 times as large as that for hydrogen production.

1,2-POLYBUTADIENE As an extension of work on the photoinduced microstructural changes in

the 1,4-diene polymers, where the double bonds are in the main chain, a study was made of the corresponding changes in two unsaturated polymers where the double bonds are external to the main chain, namely, 1,2-polybutadiene and 3,4-polyisoprene20. The only prior study of the photolysis in vacuo of a high 1,2-polybutadiene was that of a sodium-butadiene rubber (~68 per cent vinyl and ^32 per cent vinylene units) nearly twenty years ago21. In that work the polymer displayed a sharp drop in unsaturation (quantum yield of 0.73), the loss of double bonds occurring primarily in the main chain. In addition, there was evolution of gas ( ~ 64 per cent hydrogen and 32 per cent methane) with a quantum yield of ^ 2 x 10 - 3 . While the gas yields were not unlike those obtained by Bateman13, the results indicating only a small decrease in external double bonds do not carry over to 1,2-polybutadiene.

It was found more recently20 that a thin film of a polybutadiene containing 91.5 per cent vinyl and 8.5 per cent vinylene double bonds, upon exposure for 90 h in vacuo to 2 537 Â radiation, showed an 80 per cent loss of—CH CH2, as indicated by decreased intensities of characteristic infra-red bands. The absence of absorption at 9.8 μ implied that formation of cyclopropyl groups was not one of the photocyclization possibilities. Nor was there chain scission of the type shown in reaction 12 [the analogue of reactions 1 and 5 in the case of the 1,4-diene polymers] inasmuch as the characteristic 10.3-μ band for trans —CH=CH— was lacking in the spectrum of irradiated 1,2-poly butadiene.

110


-CH2CH

CH

CH2CH— fe —CH2CH· -CH2CH-Ί Ί Ί CH CH + CH

CH, CH, CH, 1

-CH,CH

CH

CH,

CH2

I CH2—CH-

\ / CH CH,

(12)

Although photocyclization of 1,2-polybutadiene could lead to fused cyclohexane rings, via reaction 13, this process was ruled out chiefly because all attempts to accomplish the free radical postpolymerization of 1,2-polybutadiene to a ladderlike polymer were unsuccessful22.

Iiv

(13)

Instead, the photocyclization of 1,2-polybutadiene was pictured as involving cycloaddition 14 of adjacent vinyl units to yield structures VI and

(14)

(VI) (VII) (VIII)

VII, along with rearrangement to VIII. The precedent cited for VI and VII was the mercury-photosensitized isomerization of 1,6-heptadiene (reaction 15) in the vapour phase to the analogous bicycloheptanes IX and X23'24. The concomitant formation of methylvinylcyclobutane XI on irradiation of

111

MORTON A. GOLUB

(IX)

(15)

(X)

(XX)

1,6-heptadiene at low pressures24, though apparently not at high23, might well find an analogue in a corresponding process in 1,2-polybutadiene, to form VIII. The latter structure could account for the appearance of a 7.3-μ (methyl) band in the spectrum of the irradiated polymer. The analogy with 1,6-heptadiene affords a further argument against fused cyclohexane rings in irradiated 1,2-polybutadiene : although certain substituted 1,6-heptadienes (e.g. 2,6-diphenyl-1,6-heptadiene) can undergo free radical cyclopolymeriza-tion, such a reaction is unknown for the unsubstituted 1,6-heptadiene22. To sum up the photolysis of 1,2-polybutadiene, it may be stated that the infra-red evidence, while not foreclosing other cyclized structures, is well represented by structures VI-VIII.

3,4-POLYISOPRENE In the ultra-violet irradiation of a high 3,4-polyisoprene (65 per cent

vinylidene and 35 per cent isoprenic or 1,4-double bonds)20, spectroscopic data disclosed partial consumption of the two kinds of double bonds present in the polymer. The major effect was photocycloaddition of the 3,4-units analogous to that discussed above for the vinyl units in 1,2-polybutadiene. At the same time, some of the 1,4-units were transformed into cyclopropyl groups in the manner described for 1,4-polyisoprene. In comparison to the photoinduced loss of ~80 per cent unsaturation in 1,2-polybutadiene, the high 3,4-polyisoprene showed, for the same radiation exposure, an overall decrease of ~40 per cent unsaturation [greater decrease in the 1,4- (~60 per cent) than in the 3,4- ( ~25 per cent) units].

The photocyclization of 3,4-polyisoprene was pictured as yielding mainly XII and XIII, rather than fused cyclohexane rings [reaction 16].

112


ÄV

(XII) (XIII) (XIV)

Here again, free radical postpolymerization of 3,4-polyisoprene has never been achieved22. The formation of structure XIV, the analogue of VIII in reaction 14, was not too important since there was no observable splitting of the 7.3-μ band (gem-dimethyls) in the spectrum of irradiated 3,4-polyisoprene. The 1,4-units in the high 3,4-polyisoprene, apart from forming

(17)

(XV)

cyclopropyl groups (and perhaps some vinyl and vinylidene units as well), would likely be involved in photocyclization with neighbouring 3,4-units to yield a structure such as XV. Reaction 17 is analogous to the mercury-sensitized photocyclization of 1,5-hexadiene to yield preferentially the bicyclohexane XVI23.

Η8(63Ρ,) (18)

(XVI)

POLYPENTENAMERS Having reviewed the published work on the direct photochemical trans

formations of diene polymers, we now consider the corresponding changes taking place in two additional unsaturated polymers, eis- and trans-polypentenamers25. These polymers (XVII, with head-to-tail monomer arrangement), obtained by stereospecific ring-opening polymerization of cyclopentene26, are of interest here because they possess at once a vinylene unit in common with the 1,4-polybutadienes and an extra méthylène group

(—CH2CH=CH—CH2CH2—)„

(XVII)

in each monomer unit, which leads to a stronger C—C bond at the allyl 113

MORTON A. GOLUB

position. Since rupture of that particular bond in stereoregular polypentenamer yields one allyl radical compared to two in 1,4-poly butadiene, the vulnerable C—C bond in polypentenamer has a bond strength of ~ 69.5 kcal/mole, i.e. midway between that of a normal C—C bond ( ~ 82 kcal/ mole, as in polyethylene) and that of the corresponding bond in 1,4-polybutadiene (82 minus twice the allyl resonance energy of 12.5, or 57 kcal/mole). Consequently, the polypentenamers should show photoinduced cis-trans isomerization similar to the 1,4-poly butadienes but with much less chain scission, and this has been observed25.

7 8 9 10 12 U Wavelength, microns

Figure 1. Typical infra-red spectra of eis- and fratts-polypentenamer films before (A, /) , respectively) and after (2?, C) ultra-violet irradiation in vacuo

Infra-red spectral changes produced in thin films of purified ds-poly-pentenamer (~95 per cent eis) and high iriws-polypentenamer (~80 per cent trans) on prolonged exposure to 2537 Â radiation in vacuo are shown in Figure 1. The sharp increase in intensity of the 10.35-μ band (trans

114


—CH—CH—) accompanied by a decrease in intensity of the 13.8-μ band (eis —CH=CH—) in going from spectrum A to £, in the case of the eis polymer; and the growth of the 13.8-μ band at the expense of the 10.35-μ band, in going from spectrum D to C, in the case of the trans polymer, signify direct photochemical cis-trans isomerization of polypentenamer. This reaction approaches an equilibrium structure with a photostationary cis-trans ratio of ~60:40, comparable to that of 1,4-polybutadienes.

As in the latter polymers, there is photoinduced vinyl formation in the polypentenamers, indicated by the minor ll.O-μ peaks in spectra Band C. However, this chemical transformation via reaction 19 is much less important

—CH 2CH=CHCH 2CH 2 - j -CH 2CH=CHCH 2CH 2—

^ — CH 2 CH=CHCH 2 CH 2 · ■ + -CH 2 CH=CHCH 2 CH 2 — (19) X

CH 2 =CHCHCH 2 CH 2 — than the corresponding process in the 1,4-polybutadienes [reaction 1]. While the vinyl content in an extensively irradiated 1,4-polybutadiene is estimated to be — 14 double bonds per 100 monomer units (as in spectrum C in Figure 5 of ref. 8), the vinyl content in spectrum B or C in Figure 1 here is estimated to be ~ 1.6 double bonds per 100 monomer units|.

The absence of distinct 9.8-μ absorption in the infra-red spectra of irradiated polypentenamers indicates that there is little or no photoinduced formation of cyclopropyl groups in these polymers, in contrast to the 1,4-diene polymers. To be sure, there is absorption at 9.75 μ in spectra Band C in Figure i, but this is related to the eis configurations in the polymer backbone since the initial eis polypentenamer (spectrum A) has a moderately strong band at that wavelength. As the eis content decreases from top to bottom in Figure 1 so also does the intensity of the 9.75-μ band. While this fortuitous absorption would obscure the development of a new weak peak at 9.8 μ, such a development clearly cannot be significant : the relative intensities at 9.75 μ in Band C not only reflect the eis contents in these spectra but they are also substantially lower than the relative intensity of the 9.8-μ cyclopropyl peak in a similarly irradiated 1,4-polybutadiene8. Unfortunately, n.m.r. analysis could not be applied to the cyclopropyl question here inasmuch as the polypentenamer films were completely insoluble after irradiation.

Since cyclopropyl structures are presumably not formed in the photolysis of polypentenamers J, the mechanism proposed originally11 for the formation

t The estimated vinyl contents are based on the relative values of the absorbance ratio, /4n.o//46.9, where the 6.9-μ band (CH2 bending vibration) is an approximate internal standard. The above estimate for 1,4-polybutadiene, which constitutes a revision of an earlier one (5 to 8 vinyl double bonds per 100 monomer units)8, is similar to the combined vinylidene-vinyl content of a comparably irradiated 1,4-polyisoprene (15.3 external double bonds per 100 monomer units)11.

X Further support lor this argument is provided by irradiations performed on films of a high frans-polyheptenamer [(—CH2CH CHCH2CH2CH2CH2—)J25. An irradiated film having an infra-red spectrum resembling a composite of Band C in Figure L so far as the 10.35-, 11.0-and 13.8-μ bands are concerned, showed scarcely any absorption at 9.8 μ and, therefore, negligible cyclopropyl formation, despite pronounced trans -> eis isomerization. The irûns-polyheptenamer film also showed the same small production of vinyl units that polypentenamer exhibited.

115

MORTON A. GOLUB

of such groups in the 1,4-diene polymers requires revision. For that mechanism to be valid, a process analogous to that given in reaction 2 or 8 would have to occur in the polypentenamers but this is evidently not so. The inference may therefore be drawn that whereas cis-trans isomerization in both 1,4-diene polymers and polypentenamers necessarily involves photoexcited vinylene units, cyclopropyl formation probably does not.

At this time, the only route to cyclopropyl structures which appears plausible is the one mentioned above in connection with Carstensen's e.s.r. work on irradiated 1,4-polyisoprene17. Reactions 9a and 9b imply that a pre-requisite for photoinduced formation of cyclopropyl groups in an unsaturated macromolecule is its ability to undergo chain scission with concomitant formation of 'hot' allyl radicals. Polypentenamer, which generates photochemically few vinyl double bonds and hence experiences little chain scission, consequently yields negligible cyclopropyls. This also offers a rationale for the failure of 1,2-polybutadiene to form cyclopropyl groups on ultra-violet irradiation: even though —CH=CH 2 units are excited photochemically, resulting in a different kind of cyclization [reaction 14], no main chain scission occurs and so no allyl radicals form which can undergo ring closure to cyclopropane structures. It is worth mentioning that this criterion also accounts for the absence of cyclopropyls in the radiolysis of the 1,4-diene polymers10 '27 (in contrast to their photolyses). Thus, despite the facile cis-trans isomerization and loss of unsaturation induced by ionizing radiation, there is negligible production of vinyl units in the 1,4-diene polymers and therefore no chain scission or transitory allyl radicals to give rise to cyclopropyl groups.

CONCLUSION From the foregoing discussion it is clear that the particular unsensitized

photochemical transformations occurring in unsaturated polymers depend on whether the C = C bonds are predominantly of the internal or external type. In polymers with internal double bonds the photochemical processes are further influenced by whether scission produces one or two allyl radicals per chain rupture. Thus, in 1,4-polybutadiene and 1,4-polyisoprene (which yield two allyls per rupture) the important reactions observed are cis-trans isomerization, loss of unsaturation, production of new double bonds and cyclopropyl formation. However, in eis- and irans-polypentenamers (which yield but one allyl per chain rupture) the principal microstructural change is cis-trans isomerization with very little vinyl production and negligible cyclopropyl formation. On the other hand, in 1,2-polybutadiene and 3,4-polyisoprene, polymers having external unsaturation, cycloaddition of adjacent double bonds is the major photochemical process. Although the above reactions are accompanied by gas evolution (mostly hydrogen) and crosslinking, the latter two processes are relatively unimportant on a quantum yield basis. In future work it will be of interest to see how these generalizations apply to the photochemistry of still other unsaturated polymers, such as those having vinylene units in the sidechains, or different distributions of double bonds along the main chain.

116


ACKNOWLEDGEMENT Support for the original research on the direct photolysis of 1,4-

polybutadiene films by the National Aeronautics and Space Administration (during 1964) while the author was at Stanford Research Institute, and the opportunity to carry out the polypentenamer work at Ames Research Centre in the course of a National Research Council Resident Research Associate-ship, 1968-1970, are gratefully acknowledged. The author wishes to thank Dr Gino Dall'Asta of Montecatini Edison, Milan, Italy, for kindly supplying the samples of polypentenamers and polyheptenamer used in this work.

REFERENCES 1 H. H. G. Jellinek. Pure Appi. Chem. 4, 419 (1962). 2 L. A. Wall and J. H. Flynn. Rubber Chem. Technol. 35, 1157 (1962). 3 M. B. Neiman (Ed.). Aging and Stabilization of Polymers, Consultants Bureau : New York

(1965). 4 N. Grassie. Encycl. Polymer Sci. Technol. 4, 647 (1966). 5 R. B. Fox. In Progress in Polymer Science (Ed. A. D. Jenkins) Vol. I, p 45, Pergamon : Oxford

(1967). 6 J. F. Rabek. Photochem. Photobiol. 7, 5 (1968). 7 R. B. Fox and R. F. Cozzens. Encycl. Polymer Sci. Technol. 11, 761 (1969). 8 M. A. Golub and C. L. Stephens. J. Polym. Sci. C16, 765 (1967); IUPAC Symposium on

Macromolecular Chemistry (Prague, 1965). 9 For reviews of the photo- and radiation-sensitized and other catalysed isomerizations, see :

(a) M. A. Golub in Polymer Chemistry of Synthetic Elastomers (Ed. J. P. Kennedy and E. G. M. Tornqvist), Part II, p 939, Interscience: New York (1969); (b) M. A. Golub in The Chemistry of Alkenes (Ed. J. Zabicky), Vol. II, p 411. Wiley-Inter-science: London (1970).

10 M. A. Golub. J. Phys. Chem. 69, 2639 (1965). 11 M. A. Golub and C. L. Stephens. J. Polym. Sci. A-l, 6, 763 (1968); Rev. Gen. Caout. Plast. 45,

749(1968). 12 T. C. Ho. K'o Hsueh CKu Pan She 365 (1963); Chem. Abstr. 64, 2253h (1966). 13 L. Bateman. Trans. Inst. Rubber Ind. 21, 118 (1945); J. Polym. Sei. 2, 1 (1947). 14 E. J. Hart and M. S. Matheson. J. Am. Chem. Soc. 70, 784 (1948). 15 C. P. Snow and C. B. Allsopp. Trans. Faraday Soc. 30, 93 (1934). 16 C. Walling and W. Thaler. J. Am. Chem. Soc. 83, 3877 (1961). 17 P. Carstensen. Makromol. Chem. 135, 219 (1970). 18 S. W. Benson. Personal communication. 19 C. P. Pinazzi. Personal communication; results presented at the 155th American Chemical

Society Meeting, San Francisco, Calif., 31 March-5 April (1968). 20 M. A. Golub. Macromolecules, 2, 550 (1969). 21 A. F. Postovskaya and A. S. Kuzminskii. Rubber Chem. Technol. 25, 872 (1952). 22 G. B. Butler, Encycl. Polymer Sci. Technol. 4, 568 (1966), and personal communication. 23 R. Srinivasan and K. H. Carlough. J. Am. Chem. Soc. 89, 4932 (1967). 2 4 R. Srinivasan and K. A. Hill. J. Am. Chem. Soc. 87, 4988 (1965). 25 M. A. Golub. Unpublished results; NRC-NASA Research Associateship (1968-70). 26 G. Natta, G. Dall'Asta and G. Mazzanti. Angew. Chem. 76, 765 (1964); Int. Ed. 3, 723 (1964). 27 M. A. Golub and J. Danon. Canad. J. Chem. 43, 2772 (1965).

117

RECENT WORK ON THE THERMAL DEGRADATION OF ACRYLATE AND METHACRYLATE HOMOPOLYMERS AND COPOLYMERS

N. GRASSIE

Department oj Chemistry, University of Glasgow, Glasgow W2, Scotland

ABSTRACT

The principal degradation reactions which occur in polymethacrylates are depolymerization to monomer and ester decomposition yielding methacrylic acid units in the polymer and liberating the corresponding olefin. The greater the number of ß hydrogen atoms in the ester group the greater the tendency towards ester decomposition. There is also a strong tendency to ester decomposition in polyacrylates incorporating large numbers of ß hydrogen atoms but the degradation processes which occur in primary esters are much more complex. The mechanisms of all these reactions are discussed. The behaviour of acrylate-methacrylate copolymers throws further light on these basic processes. In the light of an accumulation of experimental results an integrated mechanism for the thermal degradation of homopolymers and copolymers of

acrylates and methacrylates is presented.

It has for long been recognized that the principal products of the thermal degradation of polyacrylates (I) and polymethacrylates (II) are quite different from each other in spite of the close structural similarities between the two

H CH3 I I

—CH2—C— ~~ CH2— C~~ COOR COOR

(i) (in

families of polymers. On the one hand, large yields of monomer are obtained from many methacrylates, while, on the other, most acrylates give high yields of a mixture of chain fragments several monomer units in length.

However, analyses of minor products of reaction demonstrate that the degradation reactions in the two classes of polymers also have many similarities and the degradation mechanisms obviously coincide in the two tert-butyl esters1,2 (III) and (IV) from both of which almost quantitative yields of isobutene are evolved without either monomer or chain fragments.

119

N. GRASSIE

H CH3

I I ~~CH2— C ~ ~~CH 2—C—

COOC (CH 3 )3 COOC (CH 3 ) 3

(III) (IV)

With the examination of an increasing number of acrylate and metha-crylate polymers and especially with the availability of modern analytical techniques which has made accurate analysis of minor products possible, it has become clear that the whole range of behaviours of acrylates and methacrylates can be described in terms of a single overall mechanism and that the differences observed depend simply upon the relative importance of the various constituent parts of the overall mechanism.

Photodegradation studies have also contributed to our new understanding of these reactions but since this aspect is to be discussed by my colleagues elsewhere in this conference, the present paper is confined to thermal degradations.

Methacrylate homopolymers A great deal has been written about the thermal degradation of poly(methyl

methacrylate)3. Radicals are formed either by scission of the molecules at random or at vulnerable chain terminal structures. These radicals then unzip to give very high yields of monomer.

CH-j CH-» CH-i CH-i

I I I I I I

COOCH3 COOCH3 COOCH3 COOCH3

With higher methacrylate esters, ester decomposition reactions become possible resulting in methacrylic acid units in the polymer and evolution of the corresponding olefin. The most thoroughly studied system of this kind is poly(tert-butyl methacrylate)2 and it seems to be generally agreed that the reaction proceeds by a molecular mechanism involving interaction between the carbonyl group and hydrogen atoms on the ß carbon atom of the ester group

CH3 I C = C H ,

I I I C C CH3

O) ( p O OH

C H n ^ CH2

120

DEGRADATION OF ACRYLATE AND METHACRYLATE POLYMERS

Although yields of isobutene from poly(tert-butyl methacrylate) are almost quantitative, the rate of evolution of volatile products during the course of the reaction turns out to be quite complex as shown in Figure 1. Analysis

3

2

<b O l_ Φ > O * 1 ce _

A 200°C /

^^j ' ^^r

I

B

I I

\c

I I I

40 80 Time, min

120

Figure 1. Rates of evolution of volatile products from poly(tert-butyl methacrylate) heated to and maintained at 200°C.

of these volatile products during the course of the reaction reveals that the main peak (B) is due to isobutene and the shoulder (C) to water eliminated in the conversion of the methacrylic acid residues to methacrylic anhydride. But the initial peak (A) is due to pure monomer amounting to about one per cent of the original polymer although no monomer is detectable among the products thereafter. The most obvious explanation is that unzipping to monomer occurs initially but that the reaction is soon very effectively blocked by the ever increasing concentrations of ester decomposition residues, acid or anhydride, in the polymer chains.

Table 1. Influence of β-hydrogen atoms on the mechanism of degradation of polymethacrylates

Depolymerization

Methyl neo-Pentyl iso-Butyl Ethoxyethyl

0 0 1 2

Mainly depolymerization

Ethyl H-Propyl n-Butyl H-Hexyl n-Heptyl n-Octyl

3 2 2 2 2 2

Ester decomposition

(Ethyl) 2 iso-Propyl 6 sec-Butyl 5 tert-Butyl 9

121

N. GRASSIE

The importance of ß hydrogen atoms in facilitating ester decomposition at the expense of depolymerization is illustrated by the data in Table 1. Thus ester decomposition only becomes important when the nionomer unit incorporates five ß hydrogen atoms and depolymerization is quantitative when there are at most one or two ß hydrogen atoms. Inhibition of ester decomposition in the ethoxy ethyl ester is probably assisted by the electronegativity of the oxygen atom adjacent to the ß carbon atom which inhibits the movement of a ß hydrogen atom as a proton towards the carbonyl oxygen as required by the ester decomposition mechanism.

Acrylate homopolymers In a recent study of the thermal degradation of a series of poly(alkyl

acrylates)4 thermal volatilization analysis (TVA) provided a useful preliminary picture of the reactions involved.

Heated sample

1-200 mg

Trap temperatures 0,-45r75,-100°C

o—

0—■

I— Trap 1

—I Trap 2

— Trap 3

— Trap 4

μ^

l·3-^

LM,

Common trap

(liquid N2) Έ Έ

To vac. pumps

A,B,C,D,X are Pironi gauges, out put s fed to recorder

E.Fare receiver tubes or i.r. gas cells

Figure 2. TVA with differential condensation.

The principle of TVA is illustrated in Figure 2. A sample of polymer is heated from 20° to 500°C in vacuo and the volatile products are passed through four traps in parallel. These traps are normally maintained at 0°, — 45°, — 75° and — 100°C. Immediately after each trap a Pirani gauge measures continuously the pressure of volatile products passing through the trap. Thus the four gauges will record all products volatile at 0°, — 45°, — 75° and — 100°C, respectively. The result is finally displayed by a pen recorder. TVA thermograms were obtained for polymers of methyl, ethyl, n-propyl, iso-propyl, «-butyl and 2-ethyl hexyl acrylates.

It is immediately clear that poly(iso-propyl) methacrylate, which is a secondary ester, behaves differently from the other polymers, which are all primary esters. As shown in Figure 5, volatilization is discernible in poly(iso-propyl acrylate) at 260° building up to a maximum of 355°C. A secondary peak occurs at 442°C The behaviours of the five primary esters are very similar to one another and are typified by the n-butyl ester in Figure 4. They are more stable than the isopropyl ester, volatilization only becoming appreciable at 300°C and they exhibit only a single main peak. The coinci-

122


300 350 400 Temperature,°C

450

Figure 3. TVA thermogram of poly(iso-propyl acrylate): Trap temperatures, - 45°C and - 100°C; , - 75°C.

-, 0ÜC,

Φ (0 c o a S l_

<b

σ o> ._ c σ a.

/ / M / / » 1 /' '.1

/ ' I l / ' \ l / / M l ' A

1 i M / / Il / ' U Λ i

1 _L. . 1 1 350 400

Temperature, °C 450

Figure 4. TVA thermogram of poly(n-butyl acrylate): Trap temperatures, - 45°C; , - 75°C and - 100°C.

-, 0°C and

123

N. GRASSIE

dence of all the traces in the thermogram of the iso-propyl ester indicates that all the products volatile at ordinary temperature are also volatile at — 100°C. Quantitative analysis of the reaction products demonstrates that the only major products associated with the main TVA peak are carbon dioxide and propylene. Thus it is clear that the reaction associated with the main peak is ester decomposition combined with decarboxylation. This is quite different from the situation in poly(tert-butyl methacrylate) in which quantitative ester decomposition to form acid by a molecular mechanism is followed by quantitative elimination of water to form anhydride. It is suggested that at the higher temperature of the poly (iso-propyl methacrylate) decomposition reaction occurs principally by a radical mechanism as follows:

-CH 2 CH, \ / \

CU CU— I I

c c /\\ //\

0 o o o 1 I

C3H7 C3H7

RH

~CH2 CH2 \.V \ e cu-I I

c /\\

c //\

o o o o C3H7

CO, ~CH2 CH2 \ / \ CU CU-i ·

c / W o o

I C3H7

— C H 2 CH2 \V \2 e cu— II 0 c c

/ \ //\ 0 o-o to 1 <■ I

H7C3 H CU \Λ/ \ CH, CH,

This mechanism predicts a carbon dioxide/propylene ratio of unity. However, the observed ratio is 0.6 so that a certain proportion of ester decomposition probably occurs by way of the molecular mechanism.

c / / \

H CH

w \ CH, CH3

~CH2— CH— I c

/W o o I

H

CH3

+ C H 2 = C H

This is confirmed by the observation of anhydride structures in the residual polymer.

124


A very much more complex mixture of products was obtained from the primary esters studied, namely the polymers of ethyl, n-propyl, «-butyl and 2-ethyl hexyl acrylates. There are fairly wide variations in the relative yields of the various products, but, as previously suggested by the TVA thermo-grams, the range of products from each polymer is qualitatively similar so

Table 2. Products of thermal degradation of poly(n-butyl acrylate)

Main products Products in low yield Trace products

Carbon dioxide, Butyl acrylate, Methane, ethane, propane, butane, but-1-ene, butyl methacrylate ethylene, propylene, ds-2-butene, butanol, hydrogen, carbon monoxide chain fragments

Table 3. Mass balance table for degradation of poly(n-butyl acrylate) at 315°C

Degradation time, h 1 2 4 8 20 24 32

Products (wt % of initial polymer) Residue

Insoluble Soluble

Chain fragments Total liquids

Alcohol Monomer Methacrylate Remaining liquids

Condensable gases Carbon dioxide Butylène

Non-condensables Total volatiles Total products

90.3 5.3

85.0 7.7 0.4 0.2 0.00 0.00 0.02 1.8 0.5 1.3 0.1

10.0 100.3

84.6 6.0

78.6 4.1 2.2 1.6 0.04 0.00 0.6 4.0 1.9 2.1 0.0

10.3 94.9

74.6 5.6

69.0 11.1 10.6 2.6 0.08 0.03 7.9 5.1 2.0 3.1 0.2

27.0 101-6

67.2 13.1 54.1 11.5 9.8 9.0 0.10 0.04 0.7 9.4 4.2 5.2 0.5

31.2 98.4

55.8 29.6 26.2 16.3 19.8 17.6 0.16 0.05 2.0

14.8 8.3 6.5 1.1

52.0 107.8

35.2 33.6

1.6 17.1 28.7 19.6 0.20 0.03 8.9

15.4 9.1 6.3 1.1

62.3 97.5

33.9 33.9 0

21.7 29.6 20.7 0.20 0.05 8.6

15.5 9.6 5.9 1.2

68.0 101.9

it is convenient to discuss the mechanisms of their formation in terms of one of the polymers and poly(n-butyl acrylate) has been chosen for this purpose. Products are recorded in Table 2. Complete analyses were carried out at various times of reaction at 315°C and are recorded on a mass balance basis in Table 3.

The production of but-1-ene and carbon dioxide are illustrated in Figure 5. As in the methacrylate esters the yields of olefin vary with the number of ß hydrogen atoms. Thus most is produced from the ethyl ester, least from the 2-ethyl hexyl ester and similar intermediate amounts from the n-propyl and n-butyl esters. The ratio of olefin to carbon dioxide is of the order of unity over an appreciable part of the reaction although the relative carbon dioxide yield increases in the later stages.

Chromatograms of minor hydrocarbon products are illustrated in Figure 6 and are clearly produced by fragmentation of the butyl group. Butane is the

125

N. GRASSIE

Figure 5. Production of carbon dioxide. A; 1-butene, ■ ; and n-butanol,#, during degradation of poly(n-butyl acrylate).

60

40

201

170 Column temperature

120 70 50 50 32x10 Sensitivity

16 I

C2H6 C

AH<

S e n s i t i v i t y 3 2 x 1 0 2

A x 1 0

1-Butene p/s-2-butène y\ x*» 15 10 5

Retention time, min Figure 6. GLC of the gaseous products of degradation of poly(n-butyl acrylate) : (A) on a silica

gel column ; (B) on a silver nitrate-benzyl cyanide column.

most abundant of these products but the oletin/alkane ratio always lies in the range 100/1 to 1000/1.

Butanol is the most abundant of the products liquid at ordinary temperatures. The monomer and the corresponding methacrylate monomer are produced in very much lower, although significant yield. The reaction curve

126


for butanol is illustrated in Figure 5 from which an autocatalytic tendency is discernible.

The chain fragment fraction is an oily material, volatile at degradation temperatures but involatile at ordinary temperatures. It is usually yellow,

B Column temperature 230 230 210 160 110 60 40 40

M A Sensitivity

1x102 ·8χ102 1χ102

» I I I I L_3 I I I 35 30 25 20 15 10 5

Retention time,min Figure 7. GLC, (A),of the chain fragments produced during degradation of poly(n-butyl acrylate)

and (B), of a mixture of C18, C19, C20, C22 and C2 4 «-alkanes for comparison.

1800 1600 . 1400 1200 1000 Wavenumber

Figure 8. The 900-1 900 cm " l region of the infra-red spectra of undegraded, , and degraded, , poly(n-butyl acrylate).

127

N. GRASSIE

the intensity increasing with time of degradation. The infra-red spectrum is similar to that of the parent polymer although with some minor differences. The average molecular weight of the material obtained from the butyl ester was 470 which corresponds to three to four monomer units but it consists of a complex mixture as demonstrated by its chromatogram in Figure 7(A). The order of the molecular weight is confirmed by the chromatogram of a mixture of C18-C24 n-alkanes presented in Figure 7(B).

Insolubility develops progressively in the residual polymer although there is a rapid decrease in the molecular weight of the soluble part. The insoluble part is insoluble in all common organic solvents and even on prolonged treatment with sodium hydroxide solution. Since the principal volatile products, carbon dioxide, olefin and alcohol, must be products of decomposition of the ester group one would expect to observe quite significant changes in the infra-red spectrum of the residue during degradation. The

1800 1700 1800 1700 1800 Wavenumber

1700 1800 1700

Figure 9. The carbonyl region of the infra-red spectrum of poly(«-butyl acrylate) during progressive degradation. A, undegraded; B, 4 hours; C, 8 hours; D, 16 hours.

900-1900 cm"1 region of undegraded and degraded poly(n-butyl acrylate) are compared in Figure 8. The principal changes are the development of a shoulder on the carbonyl peak at 1760 cm" \ a general increase in absorption between 1650 and 1550 cm"1 and appearance of a new peak at 1563 cm -1. There is a shift in the carbonyl maximum from 1730 to 1720 cm"1 and in the C—O stretch region from 1165 to 1175 cm" l. Outside the 90O-1900 cm"l

region there are no significant changes. Changes in the carbonyl region are illustrated in more detail in Figure 9.

The fraction described as 'remaining liquids' is probably mainly water. 128


TVA and the data in Table 3 have demonstrated that the principal characteristics of the production of all the degradation products are similar. Alcohol production, which exhibits slight autocatalytic properties, is a possible exception. It therefore seems probable that all the products must be accounted for in terms of a complex series of radical processes with a common initiation step. It has been suggested4'5 that the initial step is exactly the same as that suggested for the poly(iso-propyl acrylate) degradation, resulting in the

R,

I II / / \ / \

O OR -O OR

relatively stable radical, A, and that all the major decomposition products are initiated by this radical.

Carbon dioxide and olefin are produced from these polymers in a molar ratio close to unity, at least in the early stages of reaction. It therefore seems

CH2 ,H CH2

~ C C ~ ~~CH CH~~ II P I c c c

/ \ / / > / / \ R, 0 0 0 Xp o o / 1 I ~ I + co2 + C H 2 = C

CH2 / CH2 CH2 \ 1 I ^ i

CH * H — C CH / \ / \ / \

R| R2 Rj R2 Rj R2

probable that they are produced in a reaction exactly analogous to that proposed to account for the poly(iso-propyl acrylate) reaction. Unlike the poly(iso-propyl acrylate) reaction, however, carbon dioxide production exceeds that of olefin at longer reaction times. It has been suggested5-8 that this can be accounted for in terms of the following reaction :

COOR H COOR C H 2 ^ | ^ C H 2 ^ | ~ ~ C H 2 ^ | ^ C H 2 ^ |

c c— + co2 " ' I

CH,—R

Two possible mechanisms for the production of alcohol are represented in the following equations. In each case they are represented to occur intramolecularly although an intermolecular version is also possible. The first results in a γ, δ unsaturated δ lactone ring system while the second

129 P.A.C—30/1—F

N. GRASSIE

/ C H 2 ^ ^ C H 2 N

~CH C~~ —CH I II I

o=c «. c c c + -OR CU · °" OR ° ^ ο ^ "OR

COOR COOR I I

-C ~ C H 2 - C RO—C—O CH2 O—C CH2 + -OR

w l I I I — CH HC—COOR —CH HC—COOR

^CHf "CHf

produces a β-ketoester. The former receives strong support from infra-red data. As shown in Figure 9 a shoulder develops at 1760 cm"1 which is at higher frequency than most carbonyl absorptions and suggests the presence of a lactone. Normal absorption for δ lactones is in the region 1750—1735 cm -1 but with γ, δ unsaturation in the ring absorption could easily be moved to 1760 cm" l.

Carbon-carbon double bond absorption is normally much weaker than carbonyl absorption. However, in this case the polarity of the bond will be greater because of the presence of two carbon-oxygen bonds at one end and two carbon-carbon bonds at the other end. This may explain the enhanced absorption in the carbon-carbon double bond region between 1680 cm"1 and 1620 cm"1 as degradation proceeds although no clear peak can be distinguished. However, the carbonyl absorption overlaps into this zone at later stages of degradation so no firm conclusions about the structure can be drawn from absorption in this region. The second route suggested for alcohol formation yields a saturated ketone which would absorb in the region 1720-1700 cm -1. This is slightly lower than the absorption of ester carbonyl so that the presence of ketonic structures in the degradation residue may explain the shift in the carbonyl peak to lower frequency.

Thus there is strong infra-red spectral evidence for lactone formation while changes in infra-red spectra are in accordance with, but do not prove, the formation of ketonic groups.

One striking feature of the production of alcohol is its apparent auto-catalytic characteristics. The mechanisms proposed for alcohol production are all initiated by the same polymer radical as are the other component parts of the overall reaction and since no other part of the reaction is auto-catalytic, there is no obvious direct chemical route by which the products of the alcohol elimination reaction might facilitate further production of alcohol. Alternatively, the answer may be associated with the changes in the physical nature of the polymer molecule brought about by elimination of alcohol. Both reactions proposed above involve either the formation of a ring (when they occur intramolecularly) or of a crosslink (when they occur intermolecularly). These new structures would decrease the flexibility of the polymer molecules and this should be expected to encourage intramolecular reactions involving six-membered rings such as those proposed for alcohol

130

DEGRADATION OF ACRYLATE AND METHACRYLATE POLYMERS production. This explanation of autocatalysis must be regarded as highly speculative, however.

The rapid decrease in the molecular weight of the soluble residue indicates considerable chain scission of polymer radicals which may be represented :

COOR COOR COOR COOR

II I I — CH2-C~CH2-CH— —CH2-C=CH2 + -CH—

or -CH CH2 ÇOOR _ C H ^ CH2 COOR ; \ i 2^ // \ , CH - ^ Ç— ^Cl & CO, + -C~

*< iCH2~R' CH2R' O O

The chain terminal radicals may then undergo intramolecular transfer thereby accounting for the high yields of chain fragments obtained. Trace yields of monomer are a measure of the very small tendency for acrylate terminated radicals to unzip.

The residual polymer from all the poly(primary acrylates) becomes progressively more intensely coloured as degradation proceeds. Unfortunately, all the good solvents for these polymers absorb in the ultra-violet region so that no significant spectra could be obtained. It is presumed, however, that the colour is the result of conjugation involving principally carbon-carbon double bonds, but possibly also carbonyl groups. Carbon-carbon double bonds may be formed in the polymer in a reaction analogous to the loss of hydrogen chloride from poly(vinyl chloride). Thus hydrogen will be liberated from sequences of adjacent acrylate units resulting in carbon-carbon conjugation in the polymer backbone. This kind of reaction has

— CH2-C-CH2-CH-CH2-CH~~ ~~CH2-C=CH-C=CH-C~~ + H2 I I I I I I COOR COOR COOR COOR COOR COOR

previously been discussed in some detail8. It certainly occurs in polyethylene9, polystyrene10 and methyl acrylate-methyl methacrylate copolymers8 and it has been suggested that it may be a general reaction of ethylenic type addition polymers. Very little hydrogen is observed among the degradation products so that the reaction is comparatively unimportant quantitatively.

It has been suggested that the trace of the corresponding methacrylate obtained from each polymer is the result of reactions at unsaturated chain ends formed in the reaction :

~CH,—CH—CH2—C=CH2 - ™CH2—CH + -CH2—C=CH?

I I I I COOR COOR COOR COOR

the bond ß to the double bond being particularly vulnerable to attack.

131

N. GRASSIE

Carbon monoxide is another minor product from all the polyacrylates studied. It must obviously be derived from the ester group and the most likely source is through homolytic scission of the acyl-oxygen bond,

— C H — I

Λ O OR

RO.

~CH~

// O

-CH-

+ CO

The carbonyl radical is very unstable11 and will decompose immediately to give carbon monoxide.

In view of the overlap in the behaviour of the polyacrylates and poly-methacrylates it has been of interest to examine the reactions which occur in their copolymers and in particular in a series of methyl methacrylate-methyl acrylate and methyl methacrylate-n-butyl acrylate copolymers where a wide range of mole ratios have been examined8 '12.

The products of decomposition of these copolymers are qualitatively what one would expect from the behaviours of the homopolymers but there are some quite significant features which throw further light on the nature of certain of the component reactions. Data for the methyl methacrylate-methyl acrylate system are presented in Table 4.

Table 4. Composition of volatile products of degradation of MMA-MA copolymers

Composition, % by weight of total volatiles

Polymer

PMMA 112/1 26/1 7.7/1 2/1 PMA"

co2

Trace 1 1 3 7.5

Permanent gases

— Trace 0.1 0.4 1

Methanol

— — — — 15

MMA

96 96 93 87 64 —

MA

__ —

0.8 2.5 7.0 0.76

Chain fragments

Trace 5

10 25 75

The high yields of alcohol, typical of the primary acrylates, are drastically reduced by incorporation of methyl methacrylate. The effect of methyl methacrylate units is to break up the methyl acrylate units into short sequences and calculations demonstrate that at least three acrylate units are required in sequence in order that alcohol be formed. This conclusion is in accordance with the thermal degradation behaviour of ethylene-methyl acrylate copolymers13. Thus block copolymers produce methanol in the quantities expected from the methyl acrylate content while random copolymers of the same overall composition produce very much less.

It may also be surprising that the yield of methyl acrylate increases with decreasing concentrations of acrylate in the copolymer. It seems that single acrylate units can readily participate in an unzipping process but that unzipping cannot pass through groups of methyl acrylate units. In this

132


case transfer reactions are preferred which is reflected in the strong tendency for the yield of chain fragments to increase with acrylate content.

Thus studies to date allow the following integrated mechanism for the thermal degradation of polyacrylates, polymethacrylates and acrylate-methacrylate copolymers to be drawn up (volatile products are underlined).

Ester decomposition (Molecular)

Olefin 4- acid residues

Anhydride residues + water

Polymer or copolymer

Chain scission

I Terminal chain radicals

Deprópagation

Methacrvlate monomer

Alcohol +

carbon monoxide

Deprópagation

Acrylate monomer

Acrylate terminated radical

I \ intramolecular transfer

I Chain fragments

Ester decomposition (Radical)

Oletin + carbon dioxide Methane + hydrogen + unsaturation + coloration

Intermolecular transfer

Chain radicals

Chain scission +

carbon dioxide Alcohol

The pattern of products obtained from any given polymer or copolymer will depend upon the interplay of all these processes which, in turn, is determined by structural factors as outlined in this paper. Further studies may, of course, necessitate additions or modifications to the reaction scheme.

One may summarize the application of this reaction scheme to the thermal degradation of polyacrylates, polymethacrylates and acrylate-methacrylate copolymers in the following way.

1. If the number of ß hydrogen atoms is large as in the tert-butyl esters then ester decomposition by the molecular mechanism occurs preferentially giving the corresponding olefin and acrylic or methacrylic acid units in the polymer.

2. When fewer ß hydrogen atoms are present, a higher temperature is required for the molecular mechanism and radical processes occur preferentially.

3. Among the methacrylates chain scission to form radicals is followed by unzipping to monomer although the yields of monomer may be limited by small amounts of ester decomposition, the extent of which depends upon the ß hydrogen atom content.

4. Among the polyacrylates radical induced ester decomposition predominates if the monomer has a large number of ß hydrogen atoms as in poly(iso-propyl acrylate).

133

N. GRASSIE

5. In polyacrylates with fewer ß hydrogen atoms a much more complex series of reactions occurs in competition with ester decomposition yielding the alcohol corresponding to the ester group and chain fragments as major products and the monomer, the corresponding methacrylate, hydrogen and carbon monoxide as the principal minor products.

6. Products of degradation of acrylate-methacrylate copolymers are usually qualitatively what one would expect from the behaviours of the individual homopolymers although alcohol, monomer and chain fragment yields are strongly dependent upon copolymer composition.

REFERENCES 1 J. R. Schaefgen and I. M. Sarasohn, J. Polym. Sci. 58, 1049 (1962). 2 N. Grassie and D. H. Grant, Polymer, Lond. 1, 445 (1960). 3 N. Grassie in Chemical Reactions of Polymers, p 565. Ed. E. M. Fettes, Interscience : New

York (1964). 4 N. Grassie and J. G. Speakman, J. Polym. Sci. Al, 9, 919, 931, 949 (1971). 5 G. G. Cameron and D. R. Kane, Makromol. Chem. 109, 194 (1967); 113, 75 (1968). 6 G. G. Cameron and D. R. Kane, Polym. Letters, 2, 693 (1964). 7 R. B. Fox, L. G. Isaacs, S. Stokes and R. E. Kagarise, J. Polym. Sci. A, 2, 2085 (1964). 8 N. Grassie and B. J. D. Torrance, J. Polym. Sci. A-Ì, 6, 3303 and 3315 (1968). 9 S. Ohnishi, S. Sugimoto and I. Nitta, J. Polym. Sci. A, 1, 605 (1963).

10 N. Grassie and N. A. Weir, J. Appi. Polym. Sci. 9, 999 (1965). 11 H. W. Anderson and G. K. Rollefson, J. Amer. Chem. Soc. 63, 816 (1941). 12 N. Grassie and J. D. Fortune. To be published. 13 K. J. Bombaugh, C. E. Cook and B. M. Clampitt, Analyt. Chem. 35, 1834 (1963).

134

FUNDAMENTAL PROCESSES IN THE UV DEGRADATION AND STABILIZATION OF

POLYMERS

J. E. GUILLET*

Department ofChemistry, University of Toronto, Canada

ABSTRACT

The phenomenon of 'weathering' of polymeric materials is usually caused by a complex series of chemical reactions initiated by the absorption of ultra-violet light which ultimately result in the deterioration of the physical properties of the polymer. One of the most likely initiating sites for ultraviolet absorption in polymers is the ketone carbonyl group which is often introduced into the polymer by oxidation during its processing or preparation.

Studies have been made of the photochemistry of polymers containing various types of ketone groups to determine the effects of polymer structure, external and internal viscosity, and temperature on the quantum yields of the chemical and physical processes which dissipate the energy of the absorbed light. The application of this information to the stabilization of polymers against

ultra-violet degradation is discussed in this paper.

The term 'weathering' refers tó a wide variety of chemical reactions and physical processes which occur when macromolecules, either natural or synthetic, are exposed for extended periods of time to outdoor conditions. Extensive research over the past thirty or forty years has shown that for the great majority of synthetic macromolecules, used as plastics and synthetic fibres, the most important degradative mechanisms are associated with the absorption of ultra-violet (u.v.) light. Natural polymers, on the other hand, although they also degrade by photochemical mechanisms usually are more rapidly degraded biologically, that is by the attack of micro-organisms. The protection of polymers against the effect of u.v. radiation, thus becomes of particular importance to the plastics and synthetic fibre industries.

The problem of stabilizing synthetic polymers against u.v. light is not a simple one because polymers break down by a variety of mechanisms. Furthermore the reactions frequently involve the presence of moisture, oxygen or pollutants such as sulphur dioxide, hydrochloric acid, ozone, etc. This makes it very difficult to predict the effective lifetime of plastics in

* Visiting Professor, Centre de Recherches sur les Macromolecules, Strasbourg, France.

135

J. E. GUILLET

various locations of the earth's surface because the concentration of these ingredients, with the exception of oxygen, varies considerably from place to place over the surface of the earth. Furthermore, both the intensity of u.v. radiation and the ambient temperature change markedly with geographic location and with the seasons of the year.

Although their chemical reactions differ widely from one polymer to another, synthetic polymers can be classified into two groups with respect to their changes under the influence of u.v. light. In the first category are polymers such as polyvinyl chloride and polyacrylonitrile which tend to retain their physical properties for extended periods of irradiation but discolour rapidly when exposed to u.v. light. In these polymers the main result of radiation is a change in the chemical structure of the polymer inducing chromophoric groups, but not causing a scission of the backbone of the polymer chain. In the second category are polymers such as polyethylene, polypropylene and polystyrene which tend to embrittle under the action of u.v. light. This effect can be caused by one or a combination of three effects: (1) scission of the main chain, (2) photoinduced crystallization, (3) crosslinking. Although the latter two processes can, initially, improve the properties of the polymer, if the action is allowed to go on for an extended period of time the ultimate result is usually deleterious.

Because of the complexity of the reactions which occur after the absorption of light by synthetic polymers, the simplest route to the stabilization of polymers would seem to be either to prevent the polymer from absorbing the u.v. radiation or, if this is not possible, to prevent the chemical reactions of the excited states induced in the polymer by the absorption of light. To do this it is necessary to obtain a fundamental understanding of organic photochemistry in macromolecular systems.

In the first place it is important to establish the nature of the light to which the polymer will be exposed. The sun has an approximate Boltzmann distribution of energy with a peak maximum at a wavelength of approximately 5000 Â. However, the shorter wavelengths are not available at the earth's surface because they are absorbed by the ozone layer in the upper atmosphere. As a general rule, only light having a wavelength exceeding 3000 A reaches the earth's surface. This restricts the number of reactions which may occur, since the energy of a quantum of a particular wavelength λ is given by E = hc/λ where c is the velocity of light, λ is the wavelength and h is Planck's constant. One Einstein, which is defined as one mole (6.02 χ IO23) of quanta will have an energy which is inversely proportional to the wavelength of light and may be expressed in kilocalories per mole (kcal/mole).

Figure 1 shows a plot of the distribution of energy from the sun in terms of the intensity at the earth's surface as a function of the wavelength or energy per Einstein. Both scales are shown together with the bond strengths of a number of typical co valent bonds. Since the major reaction in the degradation of physical properties is usually the breaking of chemical bonds in the backbone of the polymer, we are interested in what proportion of the total radiation of the sun is sufficiently energetic to break a chemical bond. It is apparent from this figure that, although a large portion of the sun's radiation is sufficiently energetic to break weak bonds such as the O—O bond in a peroxide or an N—N bond, very little of the total radiation is sufficiently

136

THE UV DEGRADATION AND STABILIZATION OF POLYMERS

energetic to break strong bonds such as the carbon-carbon bond and none is available to break bonds such as C = 0 or C = C , which have energies greater than 100 kcal/mole. The chemical bonds involved in the formation of the backbone of most stable polymers usually have strengths comparable to those of the carbon-carbon bond and since biphotonic processes are rather rare in organic photochemistry, we can, therefore, expect from the data shown in this figure that radiation with wavelengths longer than about 4000 Â will be ineffective in bond-breaking processes. Since the radiation having wavelengths shorter than 3000Â is filtered out by the earth's atmosphere, for many practical purposes we can restrict our photochemistry to the wavelength range of 3000 to 4000 Â.

CM

Έ

o

D in ω

σ

o

c <D

50000 10 000 6 000 4 000 3 000 20 000 8 000 5 000

Figure 1. Distribution of solar energy

Polymers are known to undergo nearly all of the typical photochemical reactions exhibited by small molecules having the same chromophoric groups ; however, the quantum yields are modified and usually reduced by the fact that these reactions must now take place in the solid state. For example, butadiene polymers have been shown by Golub1 to undergo cis-trans isomerization under the action of u.v. light. Polymers containing cinnamic acid groups crosslink by a mechanism similar to the dimerization of cinnamic acid under u.v. radiation, and polycarbonates and polyesters containing phenyl groups undergo reactions analogous to the photo-Fries reaction. Although the occurrence of such reactions does affect the physical properties of the polymer to a certain extent, they do not lead to the ultimate breakdown of the polymer into smaller and smaller fragments.

Many authors refer to any reaction which alters the properties of a polymer as a photodegradation but in our opinion only those reactions which lead to an ultimate destruction of the polymer's most important physical

137

50 100. Energy per Einstein(kcal)

J. E. GUILLET

properties can be considered to fall under the latter classification. Extensive studies have been made on the weathering of a very wide variety of synthetic polymers and in almost every case the reactions involved in the true photo-degradative processes have been shown to involve photooxidation. For example, in the photodegradation of polyethylene Tamblyn and co-workers2

showed that there was an excellent correlation between the decrease in the physical properties of the polyethylene and the increase in the oxygen content caused by photodegradation. They also showed that the rate of oxidation was closely correlated with the intensity of the u.v. light absorbed and suggested that polyethylene can be used as an actinometer to monitor the amount of near u.v. radiation absorbed by a specimen.

Oxidation processes are notoriously complex and, although in principle it is possible to retard such processes by the addition of additives which interfere with the oxidation chain, a more promising approach to the stabilization of polymers would be to inhibit the photoreactions leading to initiation of the reaction. Consequently, in our work at the University of Toronto, we have concentrated on the fundamental photochemistry of reactions which might be involved in the initiation of photodegradative processes.

One of the most likely chromophores absorbing in the near u.v. is the ketone carbonyl group which is formed in the thermal oxidation of poly-olefins and other polymers such as polystyrene containing hydrocarbon backbones. This has been shown to be the group responsible for major damage in the weathering of polyethylene and possibly also in polypropylene. If the carbonyl is located in the main chain or on the adjacent carbon in the sidechain, degradation may occur by one of two processes :

O O

~ ~ C H 2 — C H 2 — C H 2 — C — C H 2 ~ ^ ! ~ C H 2 — C H 2 — C H 2 + C—CH2— hv

and O O

— CH 2—CH 2—CH 2—C—CH 2™T^i ] C H = C H 2 + CH3—C—CH2—

If the carbonyl is in the main chain, both of these processes contribute to the scission of the main chain of the polymer, and hence the degradation of molecular weight and subsequent deterioration of the physical properties of the polymer. If the ketone group is in the sidechain but adjacent to the main chain, only the type II reaction results in scission and degradation of the molecular weight. In both cases the radicals formed by the type I process may initiate further oxidation to produce more chromophoric groups. Additional scission usually occurs as a result of this photooxidation, which can also be initiated by additives which give radical intermediates under the action of u.v. light.

138


STABILIZATION

The stabilization of polymers against weathering damage involves the retardation or elimination of primary photochemical processes similar to those we have discussed above. There are, in principle, three general ways in which this can be done : (1) by preventing the light from reaching the polymer by use of a coating or an u.v. screen; (2) by preferential absorption of the light by some compound which can dissipate the energy harmlessly (such compounds are known as u.v. absorbers) ; and (3) by addition of a compound which can remove the excited state energy from the polymer before harmful reaction can occur. This process is designated as quenching.

(1) Ultra-violet screens These compounds function by rendering the polymer opaque to both

visible and u.v. light and thereby prevent the penetration of u.v. radiation beyond the surface thus restricting the total amount of degradation to a thin surface layer. The most important of these substances is carbon black which is the most effective stabilizer for most polymers. The effectiveness of carbon black is dependent on first the type, second, the size of the particles, and third, the degree of dispersion of the particles within the polymer. Channel black, about 250 Â in diameter and highly dispersed throughout a polymer in a concentration of from one to two per cent is probably the best weathering stabilizer known. The stabilization of polyethylene increases with concentration of carbon black up to some practical limit but concentrations greater than five per cent may result in the loss of mechanical properties such as elongation and impact strength. Carbon black is considerably more efficient as a weathering stabilizer than would be predicted on the basis of just its ability to screen the polymer from u.v. light. The increased efficiency of carbon black is usually ascribed to its ability to trap radicals produced during the photooxidative processes which lead to chain degradation. However, Heskins and Guillet3 have suggested that carbon black may also stabilize polymers by its ability to quench the excited states induced in the polymer by the absorption of u.v. radiation.

(2) Ultra-violet absorbers Let us consider some of the factors which must be taken into account in

choosing an u.v. absorber to stabilize a polymer system. The first of these is high absorptivity in the region of wavelengths most harmful to the polymer. Since this region of maximum sensitivity varies from one polymer system to another, a universal stabilizer should absorb throughout the entire solar u.v. region. The most widely used stabilizers consist of derivatives of salicylic esters, benzotriazoles and orthohydroxybenzophenones. The structures of typical compounds are shown in Table 1. They may be listed in order of increasing absorptivity in the 3000 to 4000 A range as follows: phenyl salicylate < 2-hydroxy-4-methoxybenzophenone < substituted benzotri-azole < 2,2'-dihydroxy-4-methoxybenzophenone. A typical absorption spectrum for 2-hydroxy-4-dodecyloxybenzophenone is shown in Figure 2 along with the absorption of a ketone group in a typical hydrocarbon polymer. On a molar basis the absorptivity of the stabilizer is about 1 000

139

J. E. GUILLET

times that of the polymer, so that relatively small concentrations will absorb most of the incident light. The substance should also impart no colour to the medium, especially when clear films or coating are required. Ideally, then, a stabilizer should absorb strongly in the u.v. region of the solar spectrum incident on the earth's surface, but should have a sharp cutoff at the visible region.

Ultra-violet absorbers, to be effective stabilizers, must also be stable to u.v. radiation or the stabilizer will eventually be used up if it undergoes an irreversible reaction. Compounds which do not chemically react on exposure may still be unsuitable as stabilizers. Fluorescent compounds, unless they re-emit their absorbed energy at sufficiently low energies and sufficiently high rates relative to collisional processes may sensitize degradation by the process of reabsorption by the polymer. Compounds with long-lived excited

Table I. Typical structures of u.v. absorbers

Name Structure Wavelength of 0

maximum kS P e c , , l c

absorbance,Â absorpt.vHy

2-hydroxy-4-methoxy benzophenone

C H , 0

2,2'-dihydroxy-4-methoxy benzophenone

C H , 0

2,2'-dihydroxy- 4,4'-dimethoxy benzophenone

CH 3 0

phenyl salicylate

3 260 42.4

3270

3430

41.2

50.2

OCH,

3100 23.6

2-(2'-hydroxy-5'-methyl phenyl)-benzolylazole

HO N \ v OX> 3400 73.0

H,C

140


states may also transfer energy through collision to the polymer. Therefore it is essential that the electronic energy of excitation of a stabilizer must be dissipated quickly and in a harmless manner.

220 250 300 350 £00 Wavelength,m/>i

Figure 2. Absorption spectra of polymer and stabilizers in hydrocarbon solution

The orthohydroxybenzophenones represent an important class of light stabilizers. Their effectiveness appears to depend on the reversible formation of a six-membered hydrogen-bonded ring.

The two tautomerie forms in equilibrium apparently provide a facile pathway for deactivation of the excited state induced by absorption of a quantum of light by a mechanism which does not cause reaction in the polymer and leaves the stabilizer molecule unchanged. The stabilizing efficiency of these compounds increases with the strength of the intramolecular hydrogen bond as measured by the chemical shift in the proton magnetic resonance.

A further requirement of any stabilizer is compatibility. Since protection over long periods of exposure is expected from a stabilizer it must remain soluble in the polymer for an appreciable time.

For similar reasons it is imperative that good light stabilizers have low volatility.

In addition to all of the above requirements, u.v. absorbers should also be heat-stable since the stabilizers may be subjected to heating when they are incorporated into the polymer or during subsequent processing or fabrication.

(3) Quenchers Recently it has been demonstrated that compounds which do not absorb

light can also stabilize polymers by abstraction of the excited state energy from the polymer molecule. This process is known as energy transfer.

141

J. E. GUILLET

There are two basic mechanisms by which such transfer may occur. These are known as resonance energy transfer, which is caused by dipole-dipole interactions over relatively long distances, from 50-100 Â and requires a large overlap of donor emission and acceptor absorption spectrum. The second is exchange energy transfer which requires actual overlap charge clouds of the donor and acceptor molecule and total electronic spin conservation. The latter process usually occurs at diffusion-controlled rates.

Table 2. Quantum yield of chain breaking in ethylene-CO copolymers

Concentration of COD, mol/1.

0.000 0.038 0.079 0.092

mol/Einstein

0.0505 0.0430 0.0355 0.0350

Concentration of COD, mol/1.

0.122 0.207 0.387 0.410

mol/Einstein

0.0326 0.0281 0.0283 0.0270

Exchange energy transfer was shown by Heskins and Guillet3 to stabilize the photodegradation of ethylene-carbon monoxide copolymers when relatively large amounts of 1,3-cyclooctadiene (COD) were added to the polymer. These results are shown in Table 2. The data are plotted as in Figure 3 in the form of a Stern-Volmer plot, which shows that not all of the degradation can be quenched by the addition of COD. The reason for this is that COD quenches only that part of the reaction which comes out of the triplet state and in aliphatic ketones only about 65 per cent of the reaction results from a triplet excitation state, the remainder coming from an excited singlet.

In the case of aromatic ketones such as in poly(phenyl vinyl ketone) nearly all of the scission reaction results from the triplet excited state and in this case COD can quench nearly all of the degradation reaction.

2.0

0.1 0.2 0-3 0M 0.5 Concentration COD, moles /I.

Figure 3. Stern-Volmer plot for quenching by COD

142


Experimental results on the photodegradation of poly(phenyl vinyl ketone) and the copolymer of phenyl vinyl ketone and styrene are shown in Tables 3 and 4. In this case it is interesting to note that the slopes of the two curves are not the same and this is attributed to the fact that there are two types of

Table 3. Photolysis of poly(phenyl vinyl ketone) (PPVK) in the presence of cyclo-octadiene (COD)

Concentration COD, mole/litre

0.000 0.079 0.185 0.250 0.314

Concentration PPVK g/litre

7 7 7

15 7

Φη (avg)

0.245 0.053 0.030 0.0194 0.0175

Number of determinations

8 2 3 2 3

Φο/Φ (avg)

1 4.6 8.2

12.6 14.2

kqT (avg)

45.6 38.9 46.4 42.0

ketone groups in the copolymer, namely an isolated ketone group in which a monomer unit of PVK is surrounded by two styrene groups, and PVK groups in sequence containing two or more PVK monomer units. It appears from this that the isolated ketone groups are quenched more rapidly than ketone groups in sequence. In this case the photodegradation is quenched entirely by a compound which does not absorb light of the wavelength used in the experiment (in this case 3 130 Â). Unfortunately, compounds such as COD are not particularly effective as stabilizers because of the relatively short lifetime of the excited states of these ketone groups. From the Stern-

Table 4. Photolysis of copolymer styrene-PVK (14 mole % PVK) in the presence of COD

Concentration COD, moles/litre

0.000 0.021 0.040 0.067 0.067 0.145 0.285 0.285

0rs

0.176 0.051 0.031 0.021 0.022 0.0143 0.0089 0.0098

Φο/Φ

1.0 3.4 5.6 8.4 8.0

12.3 19.8 18.0

* f t

114 115 110 104 78 66 60

Table 5. Comparison of copolymer and homopolymer

Type φ0 kqT τχ sec x 108

Homopolymer 0.245 43 1.7 Copolymer 0.176 120 (initial slope) 4.8

143

J. E. GUILLET

Volmer equation the slope of the quenching curve is proportional to the lifetime of the excited state and thus, to have complete quenching with a small amount of additive it is necessary that the excited state have a long half-life. Alternatively, one must use larger quantities of the stabilizing additive. The lifetimes estimated for various types of ketone groups are summarized in the tables. Ketone groups located in glassy or crystalline regions of the polymer might be expected to have longer excited state lifetimes than those listed in Table 5. One can estimate, however, that the lifetimes would have to be several orders of magnitude longer (i.e. of the order of 10"5 sec) in order to have 95 per cent quenching of excited states with an additive concentration less than one per cent by this mechanism. Such long lifetimes may well occur in ketone groups contained in certain aromatic polymers particularly in the glassy state.

Recently Chien and Connor4 have reported that certain organometallic complexes may stabilize polypropylene by a resonance energy transfer mechanism as well as by light absorption. The compounds studied were the nickel chelates of 2,2'-thio-bis[4-(l,l,3,3,-tetramethyl-butyl)phenol] which quenches the phosphorescence of diethyl ketone in a low temperature glass and also the photooxidation of eumene in solution. In this case the stabilizing molecule also absorbs at wavelengths of 3130 À and so it is necessary to separate the two effects. However, their evidence appears to be quite strong in this case and these compounds are known to be exceptionally good stabilizers against the photooxidation of polypropylene. These authors suggested that the most probable process for resonance exchange in this case involves quenching of the excited singlet of the ketone group by the nickel chelate which is postulated to have a triplet ground state.

This can be written schematically in the following way ;

^ketone)* + 3(Ni - R)° - Hketone)0 + 3(Ni - R)*

By quenching of excited carbonyl singlets in this way, it is postulated that the photooxidation process is drastically reduced. The chromium chelate is also effective and both the nickel and chromium compounds are characterized by high absorptivity and excellent stability to u.v. light.

Ershov and his collaborators5 have also postulated that hydroxybenzo-phenones stabilize the photooxidation of polymers by a resonance energy transfer mechanism but in this case the evidence is much less strong. However, it is quite clear that the use of stabilizers which are capable of quenching photoexcited states in polymers could well lead to great advances in the stabilization of polymers. In order to be used effectively, however, it will be necessary to develop very much more fundamental information about the nature of the excited states involved in the photodegradation of polymers and the general photochemistry of polymeric materials.

REFERENCES 1 M. A. Golub, J. Polym. Sci. 25, 373 (1957). 2 G. C. Newland and J. W. Tamblyn, Appi. Polym. Symposia, 4, 119 (1967). 3 M. Heskins and J. E. Guillet, Macromolecules, 3, 224 (1970). 4 J. C. W. Chien and W. P. Connor, J. Amer. Chem. Soc. 90, 1001 (1968). 5 Y. A. Ershov, S. I. Kuzina and M. B. Neiman, Usp. Khim. [Russ. Chem. Revs], 2, 149 (1969).

144

SOME ASPECTS OF THE LIGHT PROTECTION OF POLYMERS

H. J. HELLER and H. R. BLATTMANN

CIBA-GEIGY Limited, CH-4002 Basel Switzerland

ABSTRACT Some aspects of the protection of polymers against light degradation are discussed. (1) It is shown that the actually measured efficiency of light stabilizers depends very markedly on the brands of polymers used, on the preparation of the test sample, and on supplementary additives. (2) Two mechanistic conceptions are discussed on how u.v.-absorbers of the o-hydroxyphenyl type can dissipate their excitation energy. (3) As experiments show, the efficiency of u.v.-absorbers in thin samples and/or at low concentration cannot be explained by their filtering effect alone. (4) An empirical model is developed for the classification of the spectroscopic properties of u.v.-absorbers. (5) It is shown that the efficiency of esters of 4-hydroxy-3,5-di-tert-butylbenzoic acid as light stabilizers is not related to rate or yield of the photo-Fries-rearrange-

ment of these compounds.

1. INTRODUCTION The breakdown of the properties of polymeric substrates upon light expo

sure is a very complex phenomenon. Except for polymers especially designed to be highly photosensitive, direct photolytic cleavage of the backbones seems to be a minor cause of the degradation observed in air. This is evidenced by the well known fact that irradiation is much less harmful in an inert atmosphere than in the presence of oxygen1. Quite often the light exposure acts merely as a trigger of the oxygen induced ageing, observed even in the dark.

Oxygen is known to have at least two major ways of enhancing photo-degradation. One widely observed mechanism is the autoxidation, i.e. the formation of hydroperoxides, the other mechanism is the quenching, i.e. the formation of singlet oxygen. The latter process is somewhat ambiguous in its consequences insofar as the quenching of excited species by oxygen reduces the chances for start-ups of harmful autoxidative chain reactions. Singlet oxygen so produced is on the other hand very reactive and liable to form hydroperoxides also. However, one excited state produces at most one hydroperoxide by the singlet oxygen route ; in the autoxidative reaction mode it is by definition the source of many hydroperoxides. From this several possibilities for the light protection of polymers become obvious :

The filtering of light by u.v.-absorbers which can dissipate the absorbed energy in a perfectly innocuous way. This is, however, only practicable for u.v.-light, as coloured filters substantially alter the aspect of the material to be protected. Other limitations will be discussed later.

145

H. J. HELLER AND H. R. BLATTMANN

The quenching of harmful excited states by quenchers which themselves form harmless, i.e. non-reactive excited states. Since such deactivation mechanisms are very specific and demand highly elaborate and laborious methods of investigation, quenchers have only lately entered the arsenal of commercially useful light protective agents.

The breaking of autoxidative chains by antioxidants, which themselves form radicals unable to start up new chains. Antioxidants are the topic of Prof. Scott's and several other papers and will be dealt with only in respect of the interdependence with light protective agents.

The suppression of the harmful effects of hydroperoxides by : (a) metal deactivators or chelators which lessen the catalytic effect of traces of certain metals, and/or (b) peroxide decomposers or synergists which promote a harmless decomposition of peroxides, either by themselves or through their degradation products such as sulphur dioxide.

2. PROBLEMS IN TESTING LIGHT STABILIZERS More and more it has become clear that a single member of the cited classes

of protective compounds alone provides inadequate protection, be it that too high concentrations are needed, which causes other problems, be it that the desired level of protection cannot be reached at all. Today it is the practice in industry to use complex systems to ensure the kind of protection needed in each major application of a particular polymer. For the producer of stabilizers this poses a very difficult problem. The truly scientific investigation of a compound necessitates work in solution and/or supercooled solvent matrices. However, the thus obtainable thermodynamic and kinetic data have not necessarily any bearing on the actual behaviour of this compound in the complex system comprising the polymer, several stabilizers, processing aids and possibly fillers and pigments. What complicates matters even more is the fact that today many, if not most, polymers themselves consist of at least two distinct phases, be it a so-called crystalline and amorphous phase in the chemically homogeneous crystalline polymers like the nylons or polypropylene, be it two amorphous phases in the chemically inhomogeneous polymers of the high impact strength type like ABS resin.

For the above reasons the testing of stabilizers in our laboratories has been dealt with in a very pragmatic way. Most tests bear some or even a close resemblance to the actual end use exposure of the polymer in question. Obviously some kind of acceleration is called for in the early screening. The results of highly accelerated tests, however, have always to be regarded with great caution.

The primary problem is the preparation of a suitable test specimen. There, the first hurdle to take, is the reproducibility of the mechanical and/or electrical properties of these test specimens. Highly crystalline polymers are particularly trying. This is illustrated by the following example. Fibre grade polypropylene was pressed for six minutes at a temperature well above the crystalline melting point to 0.1 mm thick sheets. These were immediately annealed for 60 minutes at 150°C and then air-quenched to room temperature. Table 1 gives the properties of the resulting test specimens. Thus, not even

146

LIGHT PROTECTION OF POLYMERS

quite extensive annealing can always be counted upon to correct differences in the previous history of a specimen.

Table I. Properties of pressed polypropylene sheets

Temperature in press Property 200°C 260°C

Aspect of specimen translucent large crystals almost perfectly clear very tine crystals

Tensile strength (kg/cm2) 299 ± 7 290 ± 19 Elongation at break ( %) 25 ± 21 997 ± 60

The next example illustrates (Table 2) the known and sometimes very marked effect of the molecular weight of one and the same polymer—albeit from two different producers—upon its light stability.

Table 2. Light stability of 0.1 mm polypropylene specimen [Stabilizer system: 0.5% 2-i2-hydroxy-3,5-di-tert-pentylphenyl)-benzotriazole; 0.2% octa-

decyl ß-(4-hydroxv-3,5-di-tert-butylphenyl)-propionate] Indicated is the exposure time in a Xenotest-150 to produce 50 per cent loss of the ultimate

elongation.

Melt index 19-22 3.2-3.5 1.3-1.4

hours 840 1400 2100

Another complication is the different composition, particularly with respect to trace impurities, of otherwise identical polymers as produced by different producers. So caution has to be exercised in comparing two different brands of polymers.

A typical example of what can happen is given in Table 3. Otherwise comparable polypropylene samples from two different producers were used in the screening of experimental u.v.-absorbers. Using otherwise identical stabilization, light stabilities in the two polymers were normally identical within the limits of error (Nos. 1-4). In some instances, however, one polymer responded much better to specific light stabilizers (Nos. 5, 6) than the other.

Table 3. Light stability of polypropylene (the time to 50 per cent loss of elongation at break is recorded)

light stabilizer

1 2 3 4 5 6

polymer h _

930 1270 1450 830

1400 1550

A polymer B h

855 1290 1340 765

2100 2500

147


But not only the polymer and the conditions during the preparation of the test specimens are critical. The following example illustrates how important the exact details of the whole testing sequence can be. The light stability of polypropylene containing 0.2 per cent octadecyl ß-(4-hydroxy-3,5-di-tert-butylphenyl)-propionate as antioxidant and optionally 0.5 per cent light stabilizer was determined as the time to 50 per cent loss of the ultimate elongation of small probes after light exposure in the Xenotest-150. The tensile-probes were obtained in one of three ways. 0.1 mm thick sheets were pressed for six minutes at 260°C. These sheets were then either quenched to room temperature and the probes were punched out and irradiated (Method A) or else the sheets were annealed for one hour at 150°C. The probes in this case were either punched out directly and irradiated (Method B) or else punched out only after irradiation of the intact sheet (Method C). The results in Table 4 indicate that only method C is useful for the testing of light stabil-

Table 4. Light stability of polypropylene (Conditions see text)

Sequence of operations

Light stabilizer

A B C quenched annealed annealed punched punched irradiated

irradiated irradiated punched

None f-Am

odK} HO i-Am

t-Bu

H O - ( ( ) ) -CH 2 P0 7 OEt

i'-Bu

l-DU

)-<P^CH2f Ni

200 h

260 h

295 h

225 h

255 h

225 h

310h

840 h

830 h

izers. The punching out of the tensile-bars apparently results in surface lesions which lead to failure in the subsequent light exposure independent of the presence of light stabilizers.

3. THE LIGHT STABILITY OF UV-ABSORBERS Of the known u.v.-filters or u.v.-absorbers the o-hydroxybenzophenones

and o-hydroxyphenylbenzotriazoles are most frequently used in industrial practice. This popularity is based mainly on their extreme light stability in polymeric substrates. Two kinds of explanations have been put forward to explain this striking superiority of the o-compounds over their p-isomers.

Based upon the observation of an extremely large Stokes-shift of the fluorescence of o-hydroxyphenylpyrimidines2, Otterstedt attributes the unique

148


behaviour of such o-hydroxyphenyl systems to the interplay of 4enoF- and 'keto'-forms. In generalized notation they can be formulated as follows:

<Lf V \-Fy 'enol'-form 'keto'-form

It is assumed that in the ground state the fcenoF form is energetically preferred, whereas the reverse is true for the first excited singlet. Upon absorption of a photon the following cycle ensues :

Arguments for this viewpoint are the facts that in the excited state phenols become much more acidic, whereas heteroatoms of sp2-hybridization become more basic than in the ground state. However, the question remains whether these differences in acidity can account fully for the magnitude of the effect. Also one would expect that the 'keto'-form would be the more preferred, the more basic the heteroatom Y is in the ground state and hence the more light stable is the compound. This extrapolation of the postulated mechanism is, however, at variance with some experimental evidence3.

Furthermore the 'keto'-form—even in the ground state—should be quite vulnerable to attack by a variety of chemicals and since the combined lifetimes of the keto-forms should be appreciable—the establishing of two keto enol equilibria being involved—the extreme light stability is not self-evident.

In view of the much poorer protective capacity of o-hydroxyphenyl-pyrimidines in unsaturated polyester resins as compared to the industrial u.v.-absorbers one critical experiment is missing: nothing is known about the fluorescence of the o-hydroxybenzophenones and o-hydroxyphenyl-benzotriazoles in the i.r.-region. The observation of such fluorescence would lend much weight to the postulated 'keto-enol'-mechanism.

In another attempt to explain the singularity of the two o-hydroxyphenyl classes of u.v.-absorbers it is postulated that the rate of internal conversion from the excited singlet is particularly high4. It is speculated that these molecules exist in the ground state in 'perpendicular' and 'coplanar' forms in comparable concentrations. The 'perpendicular' form, i.e. a configuration in which the o-hydroxyphenyl group is substantially twisted out of the plane of the benzotriazole nucleus, is favoured sterically and possibly stabilized by the interaction of the lone pair electrons on the 2-nitrogen with the π-system of the hydroxyphenyl group. The 'coplanar' form on the other hand is favoured by resonance between the triazole nucleus and the hydroxyphenyl

149


group as well as by hydrogen-bonding. Apparently the various effects practically cancel out, so that the o-hydroxyphenyl ring can rotate with respect to the triazole ring. Hence a great number of vibrational and rotational levels exist in both the ground and the first excited singlet state which fact accounts for a large Franck-Condon factor. If this picture is right and the further assumption is made that the first absorption band is mostly charge-transfer in nature, the intensity of this first band provides an estimate of the fraction of the 'coplanar' form.

In a preliminary investigation it was found that the integral absorption intensity of the first band of 2-(2-hydroxy-4-methylphenyl)-benzotriazole increases with increasing temperature, whereas the intensities of the second and third band (mainly locally excited transitions of the benzotriazole system) are temperature independent. (All the data were corrected for changes in volume of the solvent.) These changes in intensity of the first band are reversible and are established within the time needed for heating and cooling the samples (for example, i\\e minutes from 80° to 20°C).

The absolute concentration of the 'coplanar' form cannot be determined from the absorption spectrum. But the change in intensity of the first band (six per cent increase from 20° to 80° in isooctane) indicates that the 'perpendicular' form is more stable than the 'coplanar' form by less than 2 kcal/mol. (0.3 kcal/mol is found if the first band is assumed to be a pure charge-transfer transition.)

These results prove that two forms are present in about the same concentration and that the activation energy between them is rather small.

4. SYNERGISM BETWEEN UV-ABSORBERS AND ANTIOXID ANTS

As pointed out earlier, autoxidation is quite often the main contributor to photoinduced degradation of polymers. Interestingly, there exists little correlation between the activity of an antioxidant against pure thermal autoxidation and light injury.

An illustration of this is given in Figure 7, the data of which were collected on cold stretched (1:3.5), 5 mil thick polypropylene fibres stabilized with 0.2 per cent antioxidant and 0.5 per cent u.v.-absorber. The irradiated light energy needed to reduce the tensile strength to 75 per cent and 50 per cent of the original value was determined.

While indeed no correlation between the effect of the antioxidants in the oven-ageing and the light stability is found, the ranking of the various antioxidants with two different light stabilizers is, within the limits of error, exactly the same. As the two tests are conducted at widely different temperatures—in polypropylene 147°C is customary for oven tests, while approximately 30°C is common for light stability tests—one contributing factor is the difference in the activation energies of the diffusion coefficients of the various antioxidants. Also different antioxidants might respond to the photogenerated primary radicals differently than to the chain-propagating hydroperoxy radicals.

150


100

ιΛ

£50

0

S v-^-757.

1 1 1 1 A B C D 55 550 25 165 Ovenlifeat 1A7°C, h

Figure 1. Influence of antioxidants on the light stability of polypropylene monofilaments. Light stabilizers: 1: 2-hydroxy-4-octoxybenzophenone ; 2: 2-(2-hydroxy-3,5-di-tert-butyl-

phenyl)-5-chlorobenzotriazole.

Antioxidants

A: HO- / (^VcH-CH 2 CH CH, )— f e

CH

OH

Oven-life (147°C)

B:

C:

D:

H O

H O -

HO-

* >

* >

b

-CH2CH2COOCH2

-CH2CH2COOC18H3

OC,8H37

-CH—P—O OC18H37

r Oven-life C (147°C)

4

Oven-life i {UTC)

Oven-life (147°C)

55 h

550 h

25 h

165 h

( + dénotes a tert-butyl group in these formulae) For experimental conditions, see text.

151


5. QUENCHING PROPERTIES OF UV-ABSORBERS Commercially used u.v.-absorbers can quench the luminescence of car-

bony 1 compounds. This has been demonstrated by Ershov et al.5 for atactic polypropylene and by Shlyapintokh et al.6 for chemiluminescent benzo-phenone in solution.

We have tried to determine the approximate importance of the quenching versus the filtering effect of benzophenones and benzotriazoles under practical conditions. The underlying model is based upon the following assumptions.

(i) The amount of light absorbed by the substrate to produce a certain effect is independent of the amount of u.v.-absorber present, other variables being kept constant, and

(ii) The quantum yield of degradation is—within the frequency range of the irradiation source and the absorption of the substrate—independent of the wavelength and the absolute intensity of the incident light.

This leads to the following formula for the protection by a pure filter effect of a light stabilizer :

E D ^ A 1 E0 D t o t A0 Rs

where fis the protective factor. E is the total light energy (klys) with which the sample is irradiated in order to produce a given mechanical, electrical or optical effect; the subscript F (0) denotes a sample with (without) u.v.-filtering stabilizer. Dtoi is the optical density of the sample with stabilizer and DF is the optical density due solely to the stabilizer in the sample. A stands for absorption, Aiot = l-10"Dtot and A0 = 1 -10~ ( D U »- D F ) . This notation has been used to account for possible light absorption of the degradation products.

Obviously E can also stand for a time, if the light source is of constant light intensity or an effect, if the latter is linear with time.

If the optical density of the substrate (or the sensitizer in the substrate) and the one of the degradation products is very small, as compared to DF, formula 1 can be reduced to7 :

/ = I - = ^ , o 8 e = a 4 3 4 3 : l z ^=f ,2) L0 Utot DF Ls

The model according to equations 1 and 2 would be absolute if monochromatic light were used for degradation experiments. In order to produce measurable degradations in technically interesting samples, however, high light intensities are needed, a fact which precludes the use of monochromatic light.

Average DF and Z)tot (and the corresponding A-values) have therefore to be determined. They depend upon the light source and the polymer and are calculated from one experimental point, the protective factor (/) of which can be assumed to arise solely or overwhelmingly from a pure filtering effect of the stabilizer. Such experimental points correspond to very high stabilizer concentrations and/or very thick samples.

152


If no such point is available, the one with the highest share of filtering effect is chosen, but then only a lower limit of the quenching effect can be estimated.

Two such experiments, both of preliminary character, have been carried out to test the one parametric model. In the first of these experiments the light stability of polypropylene test specimens of varying thickness (fibres, films and plaques) but otherwise identical stabilization and heat history was studie<i_ Three different light stabilizers were used and the average Devalues (DF) were calculated from the thickest plaques. For stabilizer 1 (2-hydroxy-4-octoxybenzophenone) a DF of 1.5 was calculated, which corresponds to an average absorptivity of 10 [whereas the peak absorptivities are 33.3 (327 nm) and 48.2 (289 nm)]. For stabilizer 2 [2-(2-hydroxy-3:5-di-tert-butylphenyl)-5-chlorobenzotriazole] an average optical density DF of 1.9 corresponding to an average absorptivity of 13 was determined. The absorptivities at the maxima are 44.4 (351 nm) and 40.6 (312 nm). For stabilizer 3 (nickel salt of monobutyl-4-hydroxy-3,5-di-tert-butylbenzyl-phosphonate) no DF was calculated, as this stabilizer absorbs very little in the longwave u.v. and is considered to be a typical quencher. From the above DF values the protective factors expected on the basis of a pure filter effect of stabilizers 1 and 2 were calculated as a function of the sample thickness. These values correspond to the heavy lines in Figure 2.

5

4

3

2

1 10 15 30 60 100 150 300

Figure 2. Protective factors in polypropylene as function of the sample thickness (in μ). 1 : 2-hydroxy-4-octoxybenzophenone (0.5%); 2: 2-(2-hydroxy-3,5-di-tert-butylphenyl)-5-chloro-benzotriazole (0.5%); 3; Nickel salt of monobutyl 4-hydroxy-3,5-di-tert-butylbenzylphos-

phonate(0.5%).

The protective factors actually observed are indicated by the numbered circles, which are connected by the thin dotted lines. It is obvious that both stabilizers offer a higher protection in thinner samples than would be expected on the basis of a pure filter effect and it is logical to attribute this effect to the quenching capacity of the benzophenone and benzotriazole u.v.-absorbers.

153


As it is very difficult to obtain polypropylene samples of different thickness but exactly identical degrees of crystallinity and orientation, the observed effects are only qualitative in nature and cannot be used for quantitative purposes.

The second of these experiments to check for quenching capabilities of u.v.-absorbers in polymeric substrates is based upon the well known yellowing of chlorinated polyester resins. The yellow colour which develops as a consequence of polymer degradation has itself a protective power and retards further degradation. This renders the computations somewhat cumbersome, so the average optical ^density of the unstabilized polymer (Dp) and of the degradation product (Dg) ha!d to be determined from the_ action spectrum (absorption x source intensity). It was_then assumed that Dp was the same in all the samples of the same thickness. DF was calculated from the protective factor of the thickest plaque (6.3 mm) at the highest concentration of u.v.-absorber (0.5 per cent). The self-protecting effect of the degradation products is much smaller in stabilized than in unstabilized samples. So the protective factors were determined and calculated for an irradiation time of 200 hours. In Table 5 the measured Ks-values (reciprocal of the protective factor) are compared with those calculated by equation 1. This example also indicates the quenching properties of u.v.-absorbers.

Table 5. The measured #s-values after 200 h of irradiation and those calculated by equation 1. (Measurements on 1.25 mm thick plaques of Polylite X-133-Z). Stabilizers used:

A = 2-(2-Hydroxy-5-methylphenyl)-benzotriazole; B = 2-Hydroxy-4-octoxybenzophenone

Stabilizer

A

B

Concentration (%)

0.1 0.25 0-5 0.1 0.25 0.5

Rs (meas.)

12 + 1 19 + 1.5 31 + 2.5

5 + 1 8.5 + 1 17 ± 1.5

Rs (cale.)

6.5 15 30 3.5 8

16

6. THE LONG WAVELENGTH ABSORPTION OF UV-ABSORBERS Theoretically no simple mathematical description of absorption curves

is to be expected. Purely empirical curve-fitting, however, has often been tried. For nearly symmetrical absorption curves the Gaussian normal distribution curve has attained some popularity in the u.v.-field much more so than the Lorentz function. On closer inspection, however, both these two parametric curves fail to describe actual absorption curves. A typical example is the absorption curve of dimethyl p-methoxybenzalmalonate as depicted in Figure 3.

On account of the asymmetry of the absorption curve the Gaussian curve marked in can only describe a small sector. Such behaviour is quite common and in general the precise mathematical description of whole peaks is very difficult.

154


15

'ο

5

0 38 36 34 32 30 28

νχΊΟ"3

Figure 3. Spectrum of dimethyl p-methoxybenzalmalonate in chloroform. For explanation, see text.

The long wavelength tail-end of an absorption band, on the other hand, is frequently easy to describe. As illustrated in Figures 4 and 5 log ε of the long wavelength branch of the curve approaches a straight line asymptotically. It is further seen that the fit is somewhat better in a plot versus wavelength than versus frequency. That this is not an isolated phenomenon is shown in Figure 6, in which the long wavelength part of two commercial u.v.-absorbers is shown in a plot of log a versus λ. The accuracy of the linear relationship is approximately ± 1.5 nm in the range of square densities of 0.25 g/m2 to 50 g/m2 (square density = concentration x thickness of layer).

In order to test the applicability of this rule of linearity, a large number of compounds have been investigated. In most cases linearity of log a versus the wavelength is found between log a values of about — 2 up to close to + 1. So far we have no measurements taken in the log a region below — 2 as such measurements necessitate high concentrations of u.v.-absorbers and demand a very high degree of purity of compounds. Conversely the deviation from linearity is often a very sensitive method of detecting impurities.

In some cases, however, broken lines are obtained, indicating the presence of a rest-absorption, possibly some weak n -► 7i*-transition normally hidden under a dominating π -► 7i*-absorption.

An illustration of the variety is given in Figure 7. The difference between the short wavelength absorbing u.v.-filters for cosmetic purposes (Nos. 1,2,3) and the long wavelength absorbing industrial u.v.-filters (Nos. 4 to 10 inclusive) is seen clearly.

In practical usage only this very tail-end of the absorption curve, which is linear in its logarithmic form, determines the optical quality and the filtering capacity of an u.v.-absorber. This is illustrated by the transmission curves

. of Figure #, of which the logarithmic plot appears in Figure 6 below. Absorber 155

> ^ \ /COOCH3


1 which is characterized by a straight line of a lower gradient in Figure 6 produces distinctly yellow solutions at a square density of 100 g/m2. In Figure 8 this is visualized with the help of the circle (86 per cent transmission at 420 nm). If a steep monotopic rise of the transmission curve is presumed, an object is the more yellow the further below this circle its transmission curve passes. Absorber 2, with a steeper gradient in Figure 6 and whose transmission curve passes through the circle in Figure 8 yields practically colourless solutions at a square density of 100 g/m2.

k" <0

o

\

X V

\

J I I l I \ 280 300 320 340 360

A.nm Figure 4. log ε of dimethyl p-methoxybenzalmalonate (Solvent: chloroform).

3

2

<o

O

1

0

- Vy \ \

\ \

\ \

\ \

\ Vv

\ \ . . ♦ >

350 370 30000 28000

390 λ 26 000 v

Figure 5. log ε of dimethyl p-methoxybenzalmalonate versus wavelength and frequency. (Solvent : chloroform).

156


o2-(2-Hydroxy-5-methylphenyl)-benzotriazole(l)

•2-Hydroxy-A-methoxybenzophenone(2)

distance of parallel lines 3nm

Figure 6. Long wavelength absorption in chloroform of 2-hydroxy-4-methoxybenzophenone (2) and 2-(2-hydroxy-5-methylphenyl)-benzotriazole (1).

1h

-2

* 3 2 &

\

I I K \

X \ ^

2 3

5%\V

\ \ \ \ \

\ V Vx\ % "5 6 7 8 9 10

340 360 380 400 420 440 λ,ηηη

Figure 7. Log a values of various classes of u.v.-filters. Benzoates (1). Salicylates (2.3). oc-Cyano-cinnamates (4, 6). 2-(2-hydroxyphenyl)-benzotriazoles (5, 7, 9). 2,4-bis-(2-hydroxyphenyl)-s-

triazines (8). Azocompounds (10). All values were measured in chloroform.

157


100

80

60

^40

20

1.'· / 2 / /

- / / / / / /

h / ' / /

φ'>" /

/ * 2/ h

1 1

\ . 1 /0-1g/m2/ 100g/m2/ / - \ / ' ·' / ;/

1 I 1 I I

II

1! 1 1

1 /

/

280 300 320 340 360 380 400 420 440 λ,ηπη

FigureS. The transmission-curves of 2-hydroxy-4-methoxy-ben7ophenone (1) and 2-(2-hydrox_\-5-methylphenyl)-benzotriazole (2). (Solvent: chloroform).

At low square densities, however, absorber 1 is a distinctly poorer u.v.-tilter than absorber 2, its transmission curve being shifted to much shorter wavelengths.

From this the general conclusion can be drawn that an u.v.-absorber is the better a filter the steeper is the gradient of the straight line of the logarithmic plot of the long wavelength tail-end of its absorption curve. This can be quantified easily :

log a = (g- X)/s - 1 (3) a denoting the absorptivity, s the 'steepness'* and g the 'applicational limit of absorption'. The latter is defined arbitrarily as the wavelength at which a is 0.1 which corresponds in practice to the utmost limit at which an absorber is useful.

Combination of equation 3 with the law of Beer-Lambert yields j _ 2Q- io - 1 < ( ; [ + *;-i/)-iogi . . v

wherein δ stands for the square density. Differentiation yields άΤ/άλ= - (In 10/s)Tln T (5)

This expression depends explicitly only on the 'steepness' and the transmission itself. Accordingly, for one compound, i.e. for constant s, all transmission curves have to be parallel at high square densities for which the linearity of log a holds.

As illustrated in Figure 9, this is really the case. At long wavelengths the transmission curves are equidistant and a change in the square density by a factor often produces a shift of the curve by the exact amount of the 'steepness' 5.

* The steepness in this definition is reciprocal to the gradient discussed above. Hence the smaller s is, the more favourable is the u.v.-absorber for filtering purposes.

158


100

300 320 340 360 380 400 420 440 A.nm

Figure 9. The transmission-curves of 2-hydroxy-4-octoxy-benzophenone at different concentrations (Solvent : chloroform).

Thus the 'steepness's and the 'applicational absorption limit' g are essential characteristics of an u.v.-absorber coming close in practical importance to the value of the maxima.

In order to be able to determine these characteristics in a simple manner from transmission curves, a normalization is indicated.

With 1 = (λ + s - g)/s - log δ (6) the normalized transmission curve is obtained

T = \0-10~x (7) which is graphically represented in Figure 10. A table of I as a function of

Figure 10. The normalized transmission curve.

159


T (in per cent) is given in the appendix. The practical use of this concept is simple. For two convenient transmission values the wavelengths on two transmission curves are determined.

Using the 1 values according to Figure 10 or the table in the appendix, s and g are then calculated from the expressions :

λί - λ2 + log δι — log δ2

0 = ^ + 5(1 -λ, - I o g a , ) (8b)

An example is found in the appendix. The 'steepness' s and the 'appHcational limit of absorption' g of some

u.v.-absorbers are given in Figure 11. It is seen that the two spectral characteristics of one class of compounds are markedly influenced by substituents ; however, the differences between the various classes of compounds clearly surpass these variations.

The substituted cinnamates offer appHcational limits of absorption at the shortest wavelengths. Correspondingly, these compounds are only used in special applications in which absolute insensitivity towards base and metal ions is required.

The two most popular u.v.-absorber systems—the o-hydroxybenzo-phenones and o-hydroxyphenylbenzotriazoles—exhibit appHcational limits of absorption in the middle range, i.e. approximately from 400 to 430 nm. On average, the benzotriazoles show a much smaller, i.e. more favourable, "steepness's than the benzophenones. The absorption of the latter compounds is much more sensitive to the square density, i.e. to changes in concentration and/or thickness of layer.

The s-triazines and particularly the hydroxy-xanthones and azo compounds derived from activated méthylène compounds have their appHcational limit of absorption at very long wavelengths. Generally they are suitable in substrates which are very sensitive in the long wavelength region of the u.v.-spectrum. Accordingly they provoke in thicker layers a distinct yellow discoloration.

In the benzotriazole series, for which most data are available, we find quite reproducible substituent effects. The replacement of a hydrogen by a chlorine in the 5-position results in a shift of the appHcational limit of absorption to longer wavelengths by about 7 to 8 nm while the steepness is hardly affected. The kind of alkyl group in the 5'-position is relatively unimportant; changing from a methyl- to a tert-butyl-group lowers the appHcational limit of absorption by 1 to 2 nm, and the steepness is again unchanged within the limit of error. On the other hand the two spectral characteristics are very sensitive to changes in substitution in the 3'-position, e.g. the replacement of hydrogen by a tert-butyl group results in an increase of the appHcational limit of absorption by 11-12 nm and of the steepness by 1 to 2 nm. It is quite interesting to note that the effects—chlorine in 5-position and i-butyl group in 3'-position—are additive within the limits of error.

In view of these regularities and particularly the possibilities to classify u.v.-absorbers with respect to their filtering properties in the long wavelength

160

LIG

HT PR

OT

EC

TIO

N O

F POL

YM

ER

S

<

X

SS

s a

CQ

X m ro

U

- —·

X *

<Ν

ο ^

rt (N

n

S3 ~H rn

OO

SS

^

*r< rsi r-

üü

SS

^:

(U O

^

S§

2

04 0έ

0έ <3>

# Ή.

Z ON

^

ft oo

m

U

ro

*-H

COOMe

394

18.4

X <»

«

CU

CU

D i a S

m

o

04

C3> co

O-

^V

oi

o=\ J>

®

«u oo

^

S 9

22

X es

oo

û4

<3> C/5

04 \

κοΓ g-fe>

04

SO

©

^> c

j

UJ

z o

I 1

I d UJ 1

a a

161

P.A.C

—30/1—

G


region the notations of 'steepness' and 'applicational limit of absorption' are considered to be of value as spectral characteristics of a compound.

7. /j-HYDROXYBENZOATES AS LIGHT STABILIZERS

Esters of 4-hydroxy-3,5-di-tert-butylbenzoic acid are effective light stabilizers for polyolefins. The ester with 2,4-di-tert-butylphenol has received the widest attention of this class in the literature8. On the basis of its filtering capacity this compound should make a poor stabilizer. In practice, however, it exhibits a good efficiency. It has been speculated, therefore, that the observed photo-Fries-rearrangement leading to an o-hydroxybenzophenone

HO coo H O ^ Q > - C O

HO

could be the source of the efficiency. However, it is hard to see how the protection of a polymer by an u. v.-filter slowly produced during the irradiation could be so effective. Dr H. Lind in our laboratories has recently found convincing evidence that the photo-Fries-rearrangement is not the basis of the effectiveness of the 2,4-di-tert-butylphenyl 4-hydroxy-3,5-di-tert-butylbenzo-ate. First, the rearrangement product has been isolated in sufficient amounts to allow a separate test of its efficiency as a light stabilizer. The 2,4'-dihy-droxy-3,3',5,5'-tetra-tert-butylbenzophenone is distinctly less effective in polypropylene than its precursor when tested in exactly the same way. Secondly, esters of 2,4,6-trialkylphenols with 4-hydroxy-3,5-di-tert-butyl-benzoic acid are light stabilizers with an effectiveness comparable to 2,4-di-tert-butylphenyl 4-hydroxy-3,5-di-tert-butylbenzoate, in spite of the fact that they undergo rearrangement at a much slower rate (Figure 12). Hence the

20 000 A and B before irradiation A after 60 min irradiation B after 60 min irradiation

^ _ _ , -—I

250 300 350 400 A,nm

Figure 12. The u.v.-spectra of 2,4-di-tert-butylphenyl 4-hydroxy-3,5-di-tert-butylbenzoate (A) and of 2-methyl-4,6-di-tert-butylphenyl 4-hydroxy-3,5-di-tert-butyl-benzoate (B) before and after 60 min irradiation with a mercury medium pressure arc filtered by a 1 mm Pyrex glass.

Solvent: cvclohexane.

162


Fries-rearrangement is not the cause of the efficiency of these compounds ; at best it is a coincidental consequence of a protective mechanism not known as yet. As the photo-rearrangement of the 2,4-di-tert-butylphenyl 4-hydroxy-3,5-di-tert-butylbenzoate does not constitute the basis of its efficiency as light stabilizer the simultaneous use of u.v.-absorbers is not precluded. In fact mixtures of 2,4-di-tert-butylphenyl 4-hydroxy-3,5-di-tert-butylbenzoate and o-hydroxyphenylbenzotriazoles give excellent results in the light stabilization of polyolefins. This is yet another step in the direction of more complex stabilizer systems as outlined earlier.

ACKNOWLEDGEMENT We wish to acknowledge with gratitude the supply of applicational data

by our colleagues from product development under the supervision of Dr H. Gysling.

REFERENCES 1 Results for different polymers are summarized in M. B. Neiman, Ageing and Stabilization of

Polymers, Consultants Bureau: New York (1965). 2 J.-E. A. Otterstedt, unpublished results cited in R. Pater, J. Heterocyclic Chem. 7, 1113 (1970). 3 H. J. Heller, Europ. Polytn. J. Supplement, 122 (1969). 4 Ref. 3, p 126. 5 A. P. Pivovarov, Yu. A. Ershov and A. F. Lukovnikov, Plast. Massy, 10, 7 (1966). 6 V. I. Goldenberg, V. Ya. Shlyapintokh and L. M. Postnikov, Preprint of lecture to the con

ference on 'Chemical Transformation of Polymers', Bratislava (1968). 7 Ref. 3,p 114. 8 G. M. Coppinger and E. R. Bell, J. Phys. Chem. 70, 3479 (1966).

NOTE: Appendix follows on pages 164, 165.

163


APPENDIX. The 'normalized transmission' curve (equation 7)

0 1 2 3 4

0 10 20 30 40 50 60 70 80 90

— 0.00000 0.15554 0.28160 0.40018 0.52139 0.65394 0.80994 1.01363 1.33954

-0.30103 0.01836 0.16891 0.29359 0.41205 0.53398 0.66823 0.82757 1.03851 1.38765

-0.23019 0.03583 0.18205 0.30553 0.42395 0.54669 0.68276 0.84567 1.06456 1.44115

-0.18267 0.05254 0.19499 0.31742 0.43589 0.55953 0.69754 0.86430 1.09193 1.50145

-0.14549 0.06861 0.20776 0.32927 0.44788 0.57250 0.71260 0.88350 1.12079 1.57070

Example of use of the table Given the two points

(1) δ = 5 mg/cm2; 90 per cent transmission at 424 nm (2) ô = 2.5 mg/cm2 ; 20 per cent transmission at 405 nm

λ{ = 1.340 and λ2 = 0.156 are read from the table and hence

424-405 19 s = = = 12.8

1.340-0.156 + 0.699-0.398 1.485 g = 414.5 + 12.8 (1-0.748-0.548) = 414.5-3.8 = 410.7

164


5 6 7 8 9

-0.11429 0.08412 0.22036 0.34110 0.45994 0.58563 0.72796 0.90330 1.15131 1.65216

-0.08702 0.09915 0.23282 0.35291 0.47206 0.59892 0.74363 0.92377 1.18375 1.75132

-0.06255 0.11376 0.24517 0.36472 0.48426 0.61239 0.75964 0.94497 1.21838 1.87850

-0.04017 0.12800 0.25740 0.37653 0.49654 0.62604 0.77601 0.96696 1.25557 2.05681

-0.01943 0.14192 0.26954 0.38835 0.50891 0.63988 0.79277 0.98981 1.29576 2.36004

165

THE POLARITY OF POLYMER RADICALS

A. D. JENKINS

The School of Molecular Sciences, The University of Sussex, Brighton BN1 9QJ, Sussex, UK

ABSTRACT The main theories of radical reactivity are reviewed and compared. It is shown that the 'Patterns' treatment remains the only approach to avoid assignment

of arbitrary standard parameters of reactivity.

The polarity of polymer radicals has engaged the attention of chemists for à full twenty five years so that any review of the subject must, if it is to be complete, include some material which many will regard as ancient history. In fact, in various ways modern approaches to the subject are always subject to comparison with the Alfrey-Price Q-e scheme1,2 of 1946-47 so it will be as well to begin with a brief statement of the basic problem and the first solution toit.

Why is it necessary to take polarity into consideration in discussing the reactivity of radicals in polymerization reactions, frequently carried out in solvents of low dielectric constant? If polar effects were negligible radicals would be reactive or unreactive according to the level of delocalization of their unpaired spin so that it would be possible to draw up an unambiguous list of radicals in order of reactivity, one which would apply to all substrates. The failure of this proposition in practice is borne out by reference to only two radicals, polystyrene (PS) and polyacrylonitrile (PAN), in their reactions with a few substrates. Inspection of Table 1 shows that not only is a statement that 'the PS radical is n times more reactive than the PAN radical' invalid but that one cannot say, in general, that one is more reactive than the other: towards ferric chloride the PS radical is the more reactive by a factor of 100 but towards triethylamine the PAN radical is more reactive by a factor of 50003.

Table 1. Relative reactivities of radicals derived from styrène (S) and acrylonitrile (AN)

Reactivity ratio for S*/AN·

Ferric chloride

100

Substrate

Acrylonitrile

2

Vinyl chloride

0.05

Styrene

0.002

Triethylamine

0.0002

This single fact belies the fable, still found, that radicals are indiscriminate in their tendency to attack any molecule which they encounter: while

167

A. D. JENKINS

reactivity is, on the whole, at a high level, it is plain that the degree of selectivity which is deployed is high, and the only reasonable explanation is that the transition states for radical reactions can be strongly influenced by contributions in which charge separation takes place and which are therefore dependent upon the polar character of both substrate and radical. It remains to formulate an expression for a rate constant based on parameters representing the 'general reactivity' (i.e. lack of delocalization) of the radical and its polarity.

The Q-e scheme of Alfrey and Price makes the assumptions that : (i) general reactivity of reactants can be denoted by Q factors ;

(ii) polar properties of reactants can be denoted by e factors ; (iii) for a given monomer and its derived radical the e values are identical ; (iv) a rate constant will be related to these factors by the equation

% M = Ô R Ô M C

or log fcRM = log QR + log QM - eReM

A reactivity ratio will then be given by

rl = (Q1/Q2)^eiiei-e2)

The model underlying this treatment assumes electrostatic interactions between permanent charges on radical and monomer. No-one would support that idea these days and it is therefore easy to dismiss the Q-e scheme as without serious foundation, but the fact remains that, regarded as a purely empirical exercise, it achieves a remarkable degree of success, so much so that it may be too late to expect it to be superseded by a better treatment with firmer foundations.

It is necessary, in view of what comes later, to dwell for a moment on the fact that the allocation of individual Q and e values depends upon an arbitrary assignment of two such parameters, and that the reference points chosen were Q = 1.0, e = — 0.8 for styrene. Kawabata, Tsuruta and Furukawa4 recalculated Q values after changing the styrene e parameter to 0.00 and discussed the effect on Q values in general.

Very early on Wall5 had suggested modifying Q-e to Q-e-e* by allowing different polar parameters for conjugate monomer-radical pairs but, of course, there are further problems in assignment of a reference e* value and this scheme, although superior in terms of accuracy, did not catch on. (Wall's arguments were based on an analysis of reactivity ratio data for dienes which did not fit the Q-e scheme at all well.) A completely fresh attempt at a treatment of radical reactivity which avoids any arbitrary assignment of reference values was advanced in 1958 by Bamford, Jenkins and Johnston3 '6-8 , according to whom an individual rate constant for reaction between a radical and a substrate is given by

log k = log k3T + ασ + β

where log fc3T is the (measured) chain transfer constant for reaction of the radical with toluene, σ is the Hammett para sigma constant (determined in classical fashion) for the substituent on the carbon atom bearing the unpaired spin and α,β are substrate parameters, experimentally determined by reaction

168


of the substrate with a series of calibrated radicals, that is radicals of known k3T and σ. This scheme was derived from a collection of graphs which displayed patterns of reactivity points and is conveniently known as the 'Patterns' treatment.

'Patterns' successfully eliminates the need for any arbitrary element and it also deals with propagation and transfer reactions alike. A formal comparison of the Q-e and Patterns treatments3 reveals a basic similarity but a most important distinction, corresponding to allocation of separate polar parameters to radical-monomer pairs. It therefore represents an advance on Q-e in the same sense as Q-e-e* but with the invaluable advantage that its basis rests upon experimentally determined reference data devoid of arbitrary assignment. To summarize the use of the Patterns treatment, Figure 1 shows how the order of radical reactivity is expected to depend upon the polarity of the substrate, and comparison with Table 1 demonstrates that the data therein are in excellent accord with expectation.

Log k - β

2 Γ

Figure 1. Relative radical reactivity as a function of substrate polarity

An alternative formulation which includes both resonance and polar terms has been put forward by Yamamoto9 and developed by Yamamoto and Otsu10. Essentially, this treatment is concerned with chain transfer processes with aromatic substrates in which a comparison is made between the rates of reaction of a substituted and the unsubstituted transfer agent with a standard radical (styrene or methyl methacrylate)

169

A. D. JENKINS

R· + CH3CHCH3 RH + CH3CCH3

X X

R· + CH3CHCH3 RH + CH3CCH3

O ~ The equation proposed is

log k = log fc0 -f ρσ + yER

where the σ is the Hammett parameter for the substituent group in the eumene (in this case) and ER is a resonance parameter for the same moiety. p and y are essentially coefficients which denote the relative weights to be attachai to the polar and resonance contributions to the value of the rate constant. The problem arises of determining the ER values, and Yamamoto's solution is to regard reaction with a styrene radical as a standard for which p = 0 and y = 1.0. This seems to be quite arbitrary so that one is nearer to the Q-e than to the Patterns situation.

Once a list of ER values has been obtained to complement the as, data on the rates of other reactions can be analysed by a suitable plot to obtain the p and y values. In practice this appears to mean adjusting y to obtain the best linear plot of log (k/k0) versus σ and deducing p from the slope of this plot. The treatment of several reactions in this way seems to demonstrate that y is usually close to unity: the corresponding p values can be surprising, for example, for reaction with cumenes the p value is 0.7 for the polystyrene radical and 0.03 for the polymethylmethacrylate radical, indicating that polar contributions are much stronger in the former case. This is certainly in clear contrast to the Patterns interpretation of the characters of these two radicals.

The arbitrary character of the assignment of basic values has been mentioned above : a further shortcoming of this treatment is that it only appears to lend itself to reactions involving aromatic substrates so that it cannot be used for more than a very small portion of the available polymerization data.

The most recent work in the field is that of Hoyland11 who has tackled the problem in two ways, of which the first attempts to relate polarity to the electronegativity (in the Pauling sense) of the radicals and monomers. The general reactivity is similarly related to the relative localization energy for the monomer-radical pair and Hoyland then postulates that the reactivity ratio r, in a copolymerization is given by the equation

log ri = L(2) - L(l) + |XR(1) - Xu(\)\ -\XR{\) - XM(2)\ 170

Ô


Here L(l) denotes the localization energy for monomer-radical 1 and XM(i) and XR(\) are the electronegativities of monomer 1 and radical 1 respectively. This, of course, corresponds to the competition between the two following reactions :

1 XT 1 1 where rx = fcn/fc12 Ri- + M 2 4 R 2 . J

Hoyland's equation implies that one would write for an individual specific velocity constant

!og kRM = L(R) - L(M) + XR - XM + constant To put figures to the individual L and X values, Hoyland took known reactivity ratio data for five monomers (styrene, methyl acrylate, methyl methacrylate, 2-vinylpyridine and 4-vinylpyridine) and computed the best values of the various [L(2)-L(l)] and [XR-Xu[] terms. Attribution of numerical values of L and X parameters then depends on the arbitrary assignment of Land XR for one of the monomers, L = 0 and XR = 0 being selected for styrene. When the L and X values for the five primary monomers had been determined, data for another twelve were processed by computer to obtain the best fit with experimental reactivity ratios.

It is concluded that the results accord well with experiment, except for systems in which acrylonitrile is a component.

It seems that the following criticisms can be levelled against this method of tackling reactivity : (1) An arbitrary assignment of parameters for a standard monomer is

required. (2) The five primary monomers are not ideally chosen since three of them,

2-vinylpyridine, styrene and 4-vinylpyridine are too similar in character. (3) It would be better to select acrylonitrile as one of the primary monomers

as it has the most polar single substituent ; any other choice involves an implicit extrapolation to account for the behaviour of acrylonitrile and may therefore (as found) fail satisfactorily to account for it.

Hoyland's second treatment alternatively employs the concept of charge transfer so that the charge transfer energy AECT (R.M.) for reaction of radical R with monomer M is a factor which contributes to the specific velocity constant along with the localization energy as before. We then have

log r1 = L(2) - L(l) + A£CT(1,2) - A£CT(1,1) This procedure requires a rather more elaborate analysis than his electro

negativity approach but an arbitrary reference value for one of the E terms is required and the 'best values' of the parameters are derived as before, and again the systems containing acrylonitrile are exceptions to the general good agreement with experiment.

As might be expected, Hoyland finds a close correlation between XR,XM and some of his E values. In short, the two approaches are equivalent for practical purposes.

Dr Hoyland has provided a very useful comparison of the accuracy of all the schemes discussed here, except Patterns. The basic Q-e scheme and the

171

A. D. JENKINS

Hoyland charge transfer model without the localization parameters are rather poor but the three parameter Q-e-e* model is significantly better, and both the Hoyland schemes (electronegativity and charge transfer) are very much better still. By including a fourth parameter the value of the models becomes very high but, of course, this is only natural in view of the decreasing gap between the number of equations and the number of unknowns. It should be clear from the foregoing that all these methods require arbitrary attribution of standard parameters and really only differ in whether one chooses to work in terms of two, three or four parameters.

It remains to assess the accuracy of the Patterns scheme by the same standard, and this work is currently in hand. However, one may observe in conclusion that the Patterns scheme alone has the advantages of using experimental measures of radical general reactivity and polarity, and of treating propagation and transfer reactions with equal facility.

REFERENCES 1 C. G Price, J. Polym. Sci. 1. 83 (1940). 2 T. Alfrey and C. C. Price, J. Polym. Sci. 2, 101 (1947). 3 A. D. Jenkins, Advanc. Free Radical Chem. 2, 139 (1967). 4 N. Kawabata, T. Tsuruta and J. Furukawa, Makromol Chem. 51, 70 and 80 (1962). 5 L. A. Wall, J. Polym. Sci. 2, 542 (1947). 6 C. H. Bamford, A. D. Jenkins and R. Johnston, Trans. Faraday Soc. 55, 418 (1959). 7 C. H. Bamford and A. D. Jenkins, J. Polym. Sci. 53, 149 (1961). 8 C. H. Bamford and A. D. Jenkins, Trans. Faraday Soc. 59, 530 (1963). 9 T. Yamamoto, Bull. Chem. Soc. Japan, 40, 642 (1967).

10 T. Yamamoto and T. Otsu, Chem. & Ind. 787 (1967). 11 J. R. Hoyland, J. Polym. Sci. (A-1), 8, 885, 901, 1863 (1970).

172

PEROXIDE CROSSLINKING REACTIONS OF POLYMERS

L. D. LOAN

Bell Laboratories, Murray Hill, New Jersey 07974, USA

ABSTRACT The current status of our understanding of the chemical mechanism of peroxide

vulcanization is reviewed.

INTRODUCTION The subject of peroxide curing has always been of interest to chemists and

physicists working in the elastomer field. The reasons for this interest stem mainly from the relatively simple chemistry involved and the simple resulting network structure. The more widely used vulcanization systems based upon sulphur or sulphur compounds are very much more complex in mechanism and produce more varied crosslinks together with other chain modifications1. Indeed until recently both this mechanism and the resultant network structure were unknown.

The introduction of newer saturated rubbers gave the study of peroxide curing reactions some impetus and a large number of rubbers were studied. Much of this work was reviewed by the author in 19672.

Up to that time substantially all of the work had been performed using one of two or three peroxides in simple combination with the rubber of interest, the influence of co-vulcanizing agents having received little attention. Since then further work has been performed with perhaps a somewhat different emphasis. The importance of coagents such as polyfunctional monomers has been recognized and investigated and a substantial increase in the number of suitable peroxides has been exploited.

There is now, perhaps, one further factor which makes the study of peroxide vulcanization systems very timely. The continuous production of vulcanized extruded articles is of practical importance and the limitations on such processes often result from difficulties in vulcanization. The current development of radiation technology suggests that electron irradiation crosslinking may soon be economically desirable and indeed some suggest that it is so already3. Crosslinking during electron irradiation occurs by a free radical process and utilizes the same types of coagents as are, at present, used for peroxide curing. Its mechanism is probably very similar to that of peroxide crosslinking and thus information obtained on this latter process may be useful in what may be the technology of the future.

The aim of this presentation is to summarize briefly the material previously reviewed and devote more attention to the more recent developments.

173

L. D. LOAN

UNSATURATED RUBBERS It is now well established that two types of reaction occur with unsaturated

rubbers. The first of these is typified by natural rubber in which the simple reaction listed in textbooks really occurs4. This consists of hydrogen abstraction by the alkoxy radical derived from the peroxide followed by pairwise combination of the resultant polymer radicals to give crosslinks. The process is shown in general terms in Figure 1. Such a mechanism results in a unit crosslinking efficiency as measured by the ratio crosslink density : peroxide and has been shown to be operative with two peroxides. cumy15 and t-buty14. There are complications of a minor degree iri so far as some alkoxy radicals break down to give a ketone and a methyl radical before abstracting hydrogen. This does not. however, interfere with the general picture of one crosslink per molecule of peroxide decomposed.

ROOR -2RO'

- CH,C(CH,)=CHCH, - + RO' + ROH + - CH,C(CH,)=CHtH - 2 - C H , C ( C H ~ ) = C H ~ H - -+ - CH,C(CH,)=CHCH -

I - CH,C(CH,)=CHCH - Figure I . Peroxide crosslinking in natural rubber

ROOR -+ 2RO'

RO I

RO' + - CH2CH=CHCH2 - + - CH2CH-CHCH2 - RO RO

I - CH2LH-eHcH, - + - CH,CH=CHCH, - -+ - CH,CH-cHCH, - I - CHZCH-CHCH,

Figure 2. Peroxide crosslinking in cis-polybutadiene

The second type of reaction which can occur between peroxides and unsaturated rubbers is what we might call a polymerization reaction. Here the alkoxy radicals derived from the peroxide initiate a 'polymerization' of the double bonds in the polymer. The polymerization chain is quite short but is sufficient to give crosslinking efficiencies well above unity and values of around ten have been reported for cis-polybutadiene and styrene- butadiene rubber (SBR). This type of reaction is shown again in an idealized way in Figure 2. Of course. this simple reaction sequence is incomplete. hydrogen abstraction does occur as shown by the presence of alcohol in the reaction products, but some cumyloxy residues (in the case where cumyl peroxide was used) are unrecoverable. presumably due to their permanent attachment to the network'. This highly efficient reaction is. as might be

174


expected, inhibited by the presence of radical traps in the shape of anti-oxidants; similarly any centres easily attacked by free radicals may interfere. As an example of this latter possibility the acrylonitrile groups present in nitrile rubber limit the crossHnking efficiency to unity6.

What factors in the structure of the rubber determine which of these reactions occurs? Leaving aside the more obvious factors such as the presence of new reactive groups in nitrile rubber, the detailed structure around the double bond and the peroxide structure may conceivably be important. We also shall leave aside the latter factor to be dealt with in a later section and consider here only the polymer structure.

The presence of adventitious antioxidants in the natural rubber does not explain the magnitude of the difference between the two types of behaviour. Experiments using synthetic ds-1,4-polyisoprene show a crossHnking efficiency close to unity where natural antioxidants cannot be present. Likewise the presence of vinyl groups in the polybutadiene (resulting from some 1,2 addition) do not explain the high crossHnking efficiency with this polymer7. The choice between hydrogen abstraction or polymerization appears to be determined by the degree of substitution of the double bond. This seems quite reasonable when one recalls that 1,2 substituted olefins do not polymerize in the normally accepted sense of the word and trisubstituted olefins (with substituents on both double bonded carbons) would obviously be less likely to react in this way.

One exception to the above two alternative crossHnking reactions should also be mentioned: butyl rubber. This has long been known to degrade when heated with peroxide8-10 and the mechanism of its degradation and of the closely related polyisobutylene by cumyl peroxide have been studied by Thomas9 and by Loan10. The latter author showed that degradation depended very sensitively on the amount of unsaturation present, the reactivities of isobutylene and isoprene units differing by some one hundredfold. More recently a peroxide curable butyl has been made which in addition to isoprene as a comonomer also includes divinyl benzene; this leads to some gel during the polymerization reaction.

SATURATED RUBBERS The situation portrayed above for unsaturated polymers is paralleled here

in so far as two main reactions occur. The first of these results in scission and the second in crossHnking. Mechanisms for these reactions are shown in Figure 3 using polypropylene and polyethylene as examples.

The type of reaction observed in a given polymer obviously depends upon two major factors, the nature of the hydrogen atom most easily abstracted and the polymer structure around this radical. In the case of linear polyethylene and poly(ethylene oxide) only one type of hydrogen atom exists and the polymer radicals formed can most easily react by combination to give crosslinks. With nonlinear polyethylene, polypropylene and polypropylene oxide the possibility for scission arises and, indeed, in the latter two polymers proves to be the predominant reaction.

A much more interesting situation exists where both types of unit are present 175

L. D. LOAN

CH, CH, CH, CH, I I I I - CH2CHCH2CH,CH, - + RO’ + * CH,CHCH,CCH, - + ROH

CH, CH, CH , CH , I 1 i I - CH,CHCH,CCH, - -+ - CH,CH + CH,=CCH, -

- CH2CH,CHz - + RO’ + CH2CHCH, - + ROH

2 - CH,CHCH, - -+ - CH~CHCH, - I - CHzCHCH, -

Figure 3. Scission and crosslinking in saturated polymers

in a single polymer chain. This. of course. occurs in ethylene-propylene rubbers and as might be expected a mixture of crosslinking and scission reactions occur’ I , 12. The most labile hydrogen atom, towards radical attack, is the tertiary hydrogenI3 of the propylene unit. Abstraction of this hydrogen gives rise to a tertiary radical which. unlike the similar radical from polypropylene. cannot easily lead to scission. As shown in Figure 4 the simple reaction leading to scission produces a primary radical from the original tertiary one and thus seems unfavourable. Abstraction of secondary hydrogen p to the methyl group. however. leads to an easy scission reaction as shown in Figure 4. This specific abstraction occurs quite simply where two propylene units occur together when the reaction shown for polypropylene occurs.

CH , CH I I - CH,CHZCCH,CH, - -+ - CH; + CH,=CCH,CH, -

CH , CH , I I - CHCH,CHCH,CH, - -+ - CH=CH, + ‘CHCH~CH, -

CH, CH, CH , CH, i l I I - CCH,CHCH,CH, - + - C=CH2 + ‘CHCHZCH,

Figure 4 . Possible scission reactions in ethylene-propylene rubbers

Thus although the propylene groups in ethylene-propylene rubbers have been shown to be the really important groups in providing scission sites. their exact part in the reaction is not yet clear. A study of the dependence of scission upon propylene content and sequence distribution would do much to eliminate this uncertainty but with the narrow band of compositions available until recently this has not been possible.

The introduction of unsaturation into ethylene-propylene rubbers has a marked effect on the crosslinking efficiency. Whereas the saturated copolymers have been shown to have crosslinking efficiencies in the range 0.4 to 0.7 typical unsaturated terpolymers may show efficiencies greater than unity”, 14.

It seems reasonable to suppose that the lability of hydrogen atoms in the 176


different structures will be different and thus give rise to different rates of reaction and this effect has been recently studied12.

Using a variety of unsaturated monomers in an ethylene-propylene copolymer the effect of the unsaturated unit on crosslinking efficiency was measured. The summarized results are shown in Table 1. The results quoted

Table 1. Crosslinking efficiencies in EPDMs

5-Methylene-2-norbornene 5-Ethylidene-2-norbornene 5-Vinyl-2-norbornene 5-Propenyl-2-norbornene 5-Isopropenyl-2-norbornene 5-Crotyl-2-norbornene 5-{2-buten-2-yl)-norbornene 5-Methallyl norbornene 5-Methyl-5-vinyl norbornene None*

Crosslinking efficiency

1.78 0.61 1.55 0.61 1.06 0.51 0.62 0.79 0.61 0.28

* Cumyl peroxide concentration 7.54 pphr.

in this table are those observed at the highest cumyl peroxide concentrations (1.9-2.0 pphr) as measured by compression modulus. The monomers giving the highest efficiencies are those with unsubstituted terminal double bonds and since the efficiencies observed are greater than one some 'polymerization' must be occurring. A somewhat lower efficiency is found where the residual unsaturation is terminal but substituted as in the case of the polymer made with the isopropenyl substituted norbornene. The monomers with internal double bonds show the lowest efficiencies. One surprising exception to these conclusions is the polymer containing 5-methyl-5-vinyl norbornene. A somewhat higher efficiency might have been expected in this case and indeed is observed at lower peroxide concentrations. All of the terminal unsaturation polymers show a similar dependence of crosslinking efficiency on peroxide concentration. The remaining polymers have efficiencies showing no dependence upon peroxide concentration. This variation in efficiency may be explained by analogy with polymerization reactions where the chain length increases as the initiation rate is decreased.

The variation of crosslinking efficiency in EPDMs has also been studied as a function of temperature14. In the temperature range 150° to 180°C the efficiency decreases with increasing temperature for a polymer having a dicyclopentadiene unsaturated unit.

A much more complex dependence was observed in the SBR cumyl peroxide system. As the curing temperature is increased from 140° to 180°C the efficiency first» increases and then decreases14. This type of dependence may be connected with a change from a polymerization to a hydrogen abstraction mechanism but an insufficient number of systems have so far been investigated. As in so many aspects of peroxide curing more work is necessary.

177

L. D. LOAN

PEROXIDES AND THEIR EFFICIENCIES Perhaps the first matter to be raised under this heading should concern

the decomposition of the peroxide. It would be reasonable to say that until recently everyone took shelter behind the concept of a unimolecular decomposition unaffected, at least to any substantial extent by the environment. It is strange that it took so long to reveal this gross over-simplification for what it is; recent data from Hercules Inc.15 show that the decomposition rates of both cumyl peroxide and p-bis[2-(2-i-butylperoxy)propyl]benzene may vary by as much as a factor of three in different media. The slowest rates were observed in polyethylene and the fastest in SBR and natural rubber.

The majority of the work referred to above had been done prior to the last review. Since that time some effort has been directed towards the study of a variety of different peroxides. Until 1960 the only peroxides widely considered in rubber vulcanization were cumyl and i-butyl peroxides. Since that time a variety of peresters, alkyl and aryl peroxides have been used. The primary incentive has been perhaps the elimination of the unpleasant odour of dicumyl peroxide cures, the volatility of i-butyl peroxide making it undesirable for industrial use. Work with these different peroxides has, however, revealed some interesting effects which will be discussed.

Some of the published data are summarized in Table 2. This shows

Table 2. Crosslinking efficiencies for various peroxide-polymer combinations

Cumyl peroxide p-Bis[2-(2-i-butylperoxy)propyl]benzene 2.5-Dimethyl-2.5-di-i-butylperoxyhexane 2.5-Dimethyl-2.5-di-i-butylperoxyhexyne-3 ί-Butyl peroxide i-Butylperbenzoate i-Butylperoxyisopropylate Benzoyl peroxide

1 (a)

1.5 1.5 0.44 0.62

EPDM (b)

1.5

0.87

SBR (b)

13

2.5 0.4 0.37

PE

1.0C

1.0d

a Hercules data16. The EPDM value of 1.5 was assigned for easy comparison with b. Values obtained from peroxide concentration necessary to produce given modulus in a black vulcani/ate.

b Data of ref.14. c Data of ref.17. d Data of ref.18.

that all peroxides do not behave similarly and such a conclusion was foreshadowed by some earlier work. Viohl, Little and Stiteler19 found that, while polyisobutylene could be crosslinked by a combination of sulphur and i-butyl peroxide, sulphur and cumyl peroxide were ineffective. The present author has found that the scission efficiencies of these two peroxides in polyisobutylene differ by a factor of two2 and Lenas also has noted a variation of efficiency with peroxide20.

The reasons for the difference in effectiveness are not at this stage clear. A number of the peroxides used contain more than one peroxy link and hence

178


questions about the effectiveness and indeed the formation of the potential diradicai have been raised. However, the simple possibility of diradicai formation need not diminish the efficiency of crosslinking as evidenced by the high efficiency of /?-bis[2-(2-r-butylperoxy)propyl] benzene. In this material the radical sites are separated by a rigid structure and it has been suggested that where the structure is more flexible (e.g. in 2,5-dimethyl-2,5-di-r-butylperoxyhexane) cyclization may occur with a resulting loss of radicals. No definite evidence seems to be available on this point and further investigations would appear to be desirable.

As is now well known the tertiary alkoxy radicals formed by decomposition of some peroxides decompose easily to a methyl ketone and a methyl radical. This decomposition provides a useful way of measuring the ease of hydrogen abstraction as this is quite simply obtained from the alcohol/ ketone ratio

+ R H ^ R ' + (CH3)3COH (CH3)3CO· /c,

^ (CH3)2CO + CH3

kx _ [(CH3)3COH] k2 [(CH3)2CO][RH]

This type of measurement thus allows the measurement of the rate of one of the basic processes in peroxide vulcanization. Two recent papers have used this measurement13 '18. In studies with five different elastomers13 it was found that the presence of sulphur in the reaction mixture hardly affected the alcohol : ketone ratio for cumyl peroxide thus suggesting only a minimum influence at the hydrogen abstraction stage. Other measurements showed qualitative agreement between abstraction rate ratios for different types of hydrogen for cumyloxy and ί-butoxy21 radicals. A similar agreement was also found with the rates for methyl radicals. This agreement was not exact however, and further comparisons between the relative reactivities of different alkoxy radicals would be interesting.

COAGENTS The use of coagents in peroxide cures started to receive serious attention

upon the introduction of ethylene-propylene rubbers. It was quickly discovered that the efficiency of crosslinking using peroxide alone was low and that the introduction of sulphur improved some vulcanizate properties. Sulphur did not, however, increase the state of cure in a gum vulcanizate and also resulted in a very objectionable odour. Other additives were therefore sought.

Most of the additives now used are multi-functional monomers such as ethylene glycol dimethacrylate and triallyl cyanurate. Such monomers are now used in the peroxide curing of a variety of rubbers not only to increase the efficiency of curing but to give high hardness vulcanizates.

179

L. D. LOAN

Until recently little work had been done on the mechanism of the peroxide coagent cure and a recent paper by Cornell, Winters and Halterman22

makes the first detailed investigation of this mechanism. Using a model system it was found that in a simulated crosslinking reaction using n-decane, cumyl peroxide and methyl methacrylate that the main high boiling point product was an addition compound of decane and methyl methacrylate. A lesser but significant product was probably a similar compound in which a short chain of methyl methacrylate units were present. When the methyl methacrylate was replaced in the reactants by ethylene glycol dimethacrylate a much greater tendency towards polymerization was observed leading to insoluble products.

In some related experiments involving radiation crosslinking Salmon and Loan23 have found that the multifunctional coagent does indeed polymerize and appears to form a network into which the base polymer is subsequently bound.

REFERENCES 1 L. Bateman. C. G. Moore. M. Porter and B. Saville in The Chemistry and Physics oj Rubber-

like Substances. L. Bateman. ed.: Maclaren and Wiley: New York (1963). 2 L. D. Loan. Rubber Chem. Technol. 40. 149 (1967). 3 K. H. Morganstern. Rubber Age (NY). 49 (1971). 4 C. G. Moore and W. F. Watson. J. Polym. Sci. 19. 237 (1956). 5 D. K. Thomas. J. Appi Polym. Sci. 6. 613 (1962). e L. D. Loan. J. Appi Polym. Sci. 7. 2259 (1963). 7 B. M. E. van der Hoff. Industr. Engng Chem. Prod. Res. Develop. 2. 273 (1963). 8 R. Rado and D. Simunkova. Vysokomoi Soedin. 3. 1277 (1961). 9 D. K. Thomas. Trans. Faraday Soc. 57. 511 (1961).

10 L. D. Loan. J. Polym. Sci. A. 2. 2127 (1964). 11 L. D. Loan. J. Polym. Sci. A. 2. 3053 (1964). 12 F. P. Baldwin. P. Borzel. C. A. Cohen. H. S. Makowskii and J. F. Van de Castle. Rubber Chem.

Technol. 43. 522(1970). 13 J. Lai. J. E. McGrath and R. D. Board. J. Polym. Sci. A. 1. 6 821 (1968). 14 N. Ashikari. I. Kawashima and T. Kawashima. Bull. Chem. Soc. Japan. 40. 2597 (1967);

Rubber Chem. Technol. 42. 1245 (1969). 15 Hercules Inc. Peroxide Bulletin PRC-101 (November 1969). 16 Hercules Inc. Peroxide Bulletin PRC-103 (November 1969). 17 A. A. Miller. J. Polym. Sci. 42. 441 (1960). 18 D. Simunkova. R. Rado and O. Mlejnek. J. Appi Polym. Sci. 14. 1825 (1970). 19 P. Viohl. J. R. Little and C. H. Stiteler. Rubber Age (NY). 94. 594 (1964). 20 L. P. Lenas, Industr. Engng Chem. Prod. Res. Develop. 5. 138 (1966). 21 C. Walling and W. Thaler. J. Amer. Chem. Soc. 83. 3877 (1961). 22 J. A. Cornell. A. J. Winters and L. Halterman. Rubber Chem. Technol. 43. 613 (1970). 23 W. A. Salmon and L. D. Loan, to be published.

180

SOME ASPECTS OF SOLID STATE POLYMERIZATION

SEIZO OKAMURA

Department of Polymer Chemistry, Kyoto University, Japan

ABSTRACT Previous investigations on solid state polymerization are reviewed, and a few examples of new experimental results are added. Radiation-induced polymerizations of methyl methacrylate in the supercooled liquid phase, and of cyclohexene oxide in the plastic crystalline phase, have been examined. Polymerizations of acrylamide and 7V-vinylcarbazole were investigated on a solid surface with initiation induced by chlorine atoms. Solid suspension polymerization of trioxane was studied with cationic initiation. Differential thermal analysis, and electron-microscope investigations of radiation-induced

polymerizations of tetraoxane and acrylamide are discussed.

1. INTRODUCTION Chemical reactions have usually been investigated in homogeneous gas

or liquid phase, mainly because solid state reactions involve heterogeneous characteristics which are difficult to treat reproducibly.

In the last fifteen years, however, extensive studies have been concentrated on radiation-induced solid state polymerization in the homogeneous phase. High energy radiation can penetrate into the solid system and initiate homogeneous polymerization. Crystalline acrylamide or methacrylamide have been found to polymerize in the solid state by gamma- or x-ray irradiation into amorphous polyacrylic or polymethacrylic amide. These facts have been reported by Henglein and Schulz1, Morawetz and Rubin2. Ballantine et al.3. Bamford et al\ and by others. There seemed to be possibilities for both fusion by heat and disordering by displacement during polymerization. To avoid these possibilities high melting acrylic acid salts of alkaline earths were examined in radiation-induced solid phase polymerization by Morawetz et ai5, O'Donnell et al.6, and others.

Low temperature polymerization has been studied more easily by using high energy radiation at low temperatures ; at temperatures as low as that of liquid nitrogen almost all organic liquids are frozen solid. The x- or y-ray-induced polymerization of formaldehyde and acetaldehyde. also in the solid state, has been reported by Chachaty and Magat7, and by others. Some vinyl monomers such as acrylonitrile, methylmethacrylate. vinyl acetate and styrene have been examined in γ-initiated solid state polymerization by Pshejetskii and Kargin8. Amagi and Chapiro9, Goldanskii et ai10. Hardy

181

SEIZO OKAMURA

et al.11, Chen and Grabar12, and by others. All these cases were recognized as polymerizations of crystalline monomer into amorphous polymer. Finally, crystal-into-crystal solid state polymerizations have been reported with trioxane, 33-bischloromethylcyclo-oxetane, ß-propiolactone and diketene by Hayashi et al.13 in our laboratory.

So-called solid phase polymerization has been the subject of continuing interest and has been examined in three cases: (1) crystal-to-crystal type polymerization has been examined on tetra-oxane, pentoxane and recently hexa-oxane by Tadokoro et a/.14; (2) Crystal-to-amorphous polymer type polymerization has been examined on the surface-initiated catalytic polymerizations of acrylamide, methacrylamide or N-vinyl carbazole, and this work is summarized here ; and (3) glass-forming systems have been examined and reported on from radiation-induced polymerization in the supercooled liquid states of methylmethacrylate or acrylonitrile. The main results of this work are also summarized here. In order to clarify further the intricacies of solid state polymerization behaviour, I would like to report here (4) some of the preliminary work on the plastic crystal phase polymerization of cyclohexene oxide compared with its liquid and crystalline phases.

2. RADIATION-INDUCED POLYMERIZATION OF VINYL COMPOUNDS IN THE SUPERCOOLED LIQUID PHASE15

(a) Thermal measurement A differential thermal analyser (DTA) (AGNE Research Centre Co. Ltd)

was used for estimating the glass transition temperature. Tg, and other critical temperatures of the systems. For example, DTA traces of some acid-amide systems are represented in Figure 1. First (A in bottom figure)

-120 -100 -80 -60 -40 -20 0 20 Polymerization tempérâture,°C

40

ω-120 -100 -80 -60 -40 -20 Temperature,0 C

Figure 1. Polymerization and DTA of monomer-acid-amide systems indicating glass transition15c. MMA-propionic acid-acetamide (13:38:49).

182


in a propionic acid-acetamide mixture, Tg is seen at -92°C and with the crystallization temperature, Tc< at -42°C. After adding methylmethacrylate (MMA) to the system, (B in the figure), Tg decreases to - 104°C, and Tc becomes uncertain, roughly about -25°C. Between these two (Tg - T^) critical temperatures, the system is said to be in the supercooled liquid state, being glassy to outward appearances.

(b) In-source polymerization For polymerization experiments, each sample is degassed three times in

glass ampoules and sealed off in a vacuum. The sample is then irradiated by γ-rays from a cobalt-60 source (90000 Ci) at the desired temperature. The polymer obtained (PMMA) is precipitated in cooled methanol and the polymer yield determined gravimetrically. The molecular weight is determined by viscometry in benzene solution at 25°C.

The rates of polymerization in a glass-forming matrix are shown in the upper part of Figure 1. It is obvious that polymerization does not occur below Tg, but proceeds rapidly in the temperature range above Tg. The rate of polymerization shows an apparent maximum value between Tg and Tcn being due to changes in the viscosity of the system. Diffusion of the molecules is restricted near Tg. With raising of the temperature above Tg< the restriction for a propagation reaction is lessened but not yet sufficiently for a termination reaction to commence. This causes acceleration of the reaction rate, as well as increase of the molecular weight. Polymerization in a supercooled liquid behaves in much the same way as solid state polymerization.

(c) Post polymerization The samples are irradiated by γ-rays at liquid nitrogen temperature

(—196°C) and then carefully heated at a controlled rate. As shown in the DTA curves of Figure 2 B, the post-polymerization reaction of the MMA-propionic acid-acetamide system does not proceed below Ύφ but starts at about -92°C

E f 0) H

° X

LU I <l

o I E L-Q>

-C ■ * ■* ° "° 1 LU

-KO -120 -100 -80 -60 -40 -20 0 20 40 Temperature, °C

Figure 2. DTA traces of methylmethacrylate-propionic acid-acetamide system. A: Before irradiation; B: After irradiation (4 x 105 (rad/h) x 1 (h) at -196°C)

-104°C

Y t

-1CK°C Tp

Tg

I 1 1

/ \

92°C

I i

V~>-N/"-""\ SA VH A / \

1 1 1 1 i

183

SEIZO OKAMURA

-KO -120 -100 -80 -60 --40 Temperature ,°C

Figure 3. DTA traces of methylmethacrylate-propionic acid-acetamide irradiated system. Heating rate (mV/min) A: 0.04; B: 0.1 ;C: 0.2; D: 0.4 (4 x 105 (rad/h) x 1 (h) at -196°C)

By increasing the heating rate, the temperature for initiating polymerization (Tp) increases, but Tg remains constant. These facts are shown in Figure 3.

As shown in Figure 4, the temperature difference (ΔΤ= Tg — Tp) tends to zero when the heating rate is extrapolated to zero. This indicates that post polymerization starts at the glass temperature in the supercooled liquid state. Similar results are also obtained in the systems of acrylamide-itaconic

0.08 0.16 0.24 0.32 0.40 Heating rate^V/min

0.48

Figure 4. Temperature difference (ΔΓ = Tg - Tp) between Tg and Tp as a function of heating rate. O Methylmethacrylate-propionic acid-acetamide. · Methylmethacrylate-succinic acid-

acetamide. Δ Acrylonitrile-succinic acid-acetamide

184


Table 1. Molecular weight of polymer obtained in the polymerization of glass-forming systemsI5a (acrylamide-propionie acid-

formamide. 1:1:0.5, irradiation dose. 0.003 Mr)

Irradiation temperature (°C)

- 7 8 - 6 3

- 4 8 - 3 0 - 2 0

DP x 10"4

2.1 2.4

1.6 1.1 1.3

Huggins constant (V)

3.40 3.00

0.52 0.51 0.48

acid (or -malonic acid) and acrylic acid-acetamide (or -formamide). Unusually large values of Huggins's constant k in the viscosity equation are found between Tg and Tc in the acrylamide-propionie acid-formamide system, as shown in Table 1. The reason for the large /c'-value in supercooled liquids arises from increased reactivity in branch formation which chain transfer may increase by increasing the polymerization rate between Tg and Tc.

The glass-forming systems are obtained by carefully selecting the cooling rate. Acrylamide is found to form its glassy state by mixing with acids, like itaconic, malonic, or acrylic acid. Similarly acrylic acid can form glass with formamide, acetamide or propionamide. Here Tg values dilatometrically measured are set out in Table 2.

Table 2. Transition temperatures of glass-forming systems1

Systems Composition Glass transition temperature. (in volume) (°C)

Acrylamide-itaconic acid 1 Acrylamide-malonic acid 1 Acrylamide-succinic acid-acetamide 1 Acrylamide-acrylic acid 1 Acrylamide-propionie acid-acetamide 1 Acrylamide-propionie acid-formamide 1 Acrylamide-propionie acid-formamide 1 Acrylic acid-acetamide 1 Acrylic acid-formamide 1

0.8 1 0.5:1 1 1:0.5 1:0.5 1:1 0.5 1

- 3 9 - 6 0 - 7 4

-100 -100 --110 -115 ~ - 1 1 0 --135 -

-105

- 1 2 0 -115 -145

3. CATALYTIC POLYMERIZATION OF VINYL AND CYCLIC COMPOUNDS IN SOLID PHASE16

(a) Solid state polymerization of acrylamide initiated by a Cl atom at the solid surface

The crystalline monomer of acrylamide or methacrylamide has long been recognized to polymerize in the solid state by u.V.-, x- or γ-rays. During polymerization the solid shape of the monomer does not change in outer

185

SEIZO OKAMURA

0.5

«A

0.3

0 225 30 35 40 45 50 Temperature, °C

Figure 5. ^/sp/c-values of the polymer obtained in the solid state polymerization of acrylamide and polymerization temperatures168. [M] 0 : 1 g in 33 ml, [Cl2] : 21 m moles per litre

appearance but the polymer obtained is always found to be amorphous. The essential point lies in whether the propagation occurs before or after disordering of the monomer lattice. The behaviour depends critically on the rate-relationship between the solid state reaction and the crystallization of polymer molecules. The polymerization might proceed in a highly viscous state. It depends also upon the order of disturbance by the polymerization reaction.

Now, crystalline monomers of acrylamide and methacrylamide are found to polymerize in the solid state in the presence of chlorine gas under u.v.

Figure 6

186


radiation. As the shape of the monomer remains unchanged, the polymerization proceeds from just below the surface of the crystals. It is obvious here that the chlorine atoms produced by the decomposition of chlorine molecules initiate the polymerization at the surface of a monomer crystal. As one apparent characteristic of solid state polymerization the molecular weights of the polymers obtained are shown to increase by raising of the polymerization temperature. One of these results is shown in Figure 5.

After unreacted monomer was sublimed from partially polymerized crystals of acrylamide, the polyacrylamide obtained was found to be crystalline by the Debye-Scherrer ring diagram in x-ray diffraction analysis and to be somewhat oriented by polarized microscopic analysis, as shown in the following x-ray diagrams (Figure 6) and photographs (Figure 7).

Ordinary microscopic

Polar ized microscopic

Figure 7

187

SEIZO OKAMURA

Polyacrylamide or methacrylamide as reported in radiation-induced solid-state polymerization has already been recognized to be amorphous, but here the polymer obtained in the surface polymerization was found to be crystalline.

Once Professor I. Nitta et al17 examined the molecular alignment of monomer acrylamide in a single crystal and assumed that monomer molecules existing in the 'a- plane' should be easily polymerized into a polymer layer in random directions within the plane. If the sites for initiation are distributed at random in the crystal, then propagation in the direction vertical to the 0-plane may also be random which will make the polymer amorphous. In the case of the surface initiation mentioned here, however, the initiation sites are concentrated, at least in the early stage of polymerization, in the surface of the crystal ; in the a-plane, then, propagation proceeds step by step along each a-plane. Regularity will be obtained by slow surface-initiated solid-state polymerization at the interaction between polymer molecules probably by hydrogen-bonding at ^)NH and ^ C = 0 groups. Gradual sublimation of unreacted monomer was the only way of obtaining a crystalline state of the polymer, but solvent extraction by methanol or acetone destroys the polymer structure in which no birefringence shows.

(b) Solid state polymerization of N-vinyl carbazole initiated by redox-catalyst in the suspension

An /v-vinylcarbazole crystal was polymerized in suspension in water using ammonium persulphate and sodium bisulphate as redox catalysts. As one characteristic of solid state polymerization, the molecular weights of the polymer obtained18 were shown to increase with rising polymerization temperature (Figure 8). Birefringence was clearly observed in the polymer layer of a partially polymerized crystal. The crystalline Debye-Scherrer rings were also observed in x-ray diffraction diagrams of the polymer.

0.101 1

0 0 2 l 1 1 1 1 1 0 0Ό5 0.10 015 0 2 0

[CLmole/l.

Figure 8. (rçsp/c) (100 ml/g)-values of the polymer and the catalyst concentration in the medium [C]. (mole/1). Polymerization temperature : O 50°C; Δ 40°C: at [M]0:1 gin 20 ml medium18

188


These are also cases of surface-initiated solid-state polymerization. Sublimation of the monomer in this case, however, destroys the regularity of the polymer structure in which no strong bonding exists, in contrast to acrylamide.

(c) Solid state polymerization of trioxane initiated by cationic catalyst in suspension

Trioxane was polymerized in a suspension of n-hexane, having stannic chloride as a cationic catalyst. The effect of polymerizing temperature on the molecular weight of polymer obtained is represented in Figure 9. The general

8.00

6.00

iu)o

2.00

0 10 20 30 Conversion, %

Figure 9. Effect of polymerization temperature on the molecular weight (nsp/c) of polymer16c. /i-Hexane, (MJ; 0.74 mole per litre; [SnClJ: 11.25 mmole per litre, SnCl4; TAA = 1; 0.57;

polym. temperature. O 50°C, € 40°C. 3 31.5°C. · 20°C

tendency for solid state polymerization as a function of molecular weight is also confirmed in these cases.

The polyoxymethylene obtained here was highly crystalline and well oriented in three dimensions as shown in Figure 10.

In these cases no large differences were observed between radiation-induced and surface-initiated solid state polymerization.

(d) Differential thermal analysis of radiation-induced post-polymerization of tetra-oxane and acrylamide in the solid state16e

Thermal measurements were done using the Shimazu-DT-10 analyser for tetra-oxane and acrylamide. Tetra-oxane has the benefit of a smaller tendency to sublime than that of trioxane. Figures 11(a) and 77(b) show DTA curves for non-irradiated (each a) and irradiated tetra-oxane and acrylamide, respectively.

189

SEIZO OKAMURA

Figure 10. X-ray diffraction diagrams of POM obtained in solid state. Polymerization in n-hexane by stannic chloride-TCA at 50°C.

Conv. Polym. time (%) (min)

A.-j Containing a.-) After 23 9 B. > unreacted b. I sublimation 39 60 C.J monomer c. J of monomer 56 360

The melting point of tetra-oxane was seen at 112°C and this endothermic peak decreased by irradiation (shown in b, c and d) due to decrease of the amount of monomer remaining. In irradiated samples, exothermic broad peaks were observed to start at about 62°C —àHp was found to be 0.8 ±0.1 kcal/mole from calculation of the exothermic heat content divided by the amount of polymer obtained. AHf was similarly calculated to be 5.8 ± 0.3 kcal/mole from the endothermic heat content divided by the amount of monomer remaining. AHf for pure monomer was found to be 5.2 ± 0.1 kcal/ mole with which the calculated value mentioned above could well coincide.

190


0) o c 0) l _ 0)

Endothermic

a (A) A

' b 1/ K ' . c I — ' i ' ' y f [' ■ - d - —i . 1 y\ - '

Exothermic 1 I I I M 1 . 1 1 I I

50 60 70 80 90 100 110 120 130 Temperature, °C

60 70 80 90 Temperature, °C

Figure 11. DTA of tetra-oxane (A) and acrylamide (B)16e.

(AW

Pre-irradiation dose Polymer yield (r) (%)

a 0 0 b 1.0 x 106 24.0 c 3.4 x 106 35.2 d 6.4 x 106 50.2

(B)

Pre-irradiation dose Polymer yield (r) (%)

la 0 0 )b 2.2 x 105 39.6 c 4.1 x 105 61.2 d 1.2 x 105 76.7

The melting point of acrylamide was found to be 84.5°C and this endothermic peak also decreased by irradiation due to decrease in the amount of monomer remaining. In irradiated samples, exothermic broad peaks were also observed between about 35°C and 70°C, due to post-polymerization. The post-polymerization seemed to be fast as represented by the steep exothermicity. In this case, the heat of polymerization, AHp could not be obtained due to the fusion of some part of the monomer by a rise of temperature. For the calculation of AH^ the extrapolation to zero heating rate was done by which AHP was found to be 8.1 ±1.1 kcal/mole. However, the AHp for solution polymerization has been known to be 13.8 ± 0.3 kcal/mole. In a similar way to that for tetra-oxane, AHf in solid state polymerization for the monomer was found to be 0.7 ± 0.1 kcal/mole which was very much smaller than AHf for the pure monomer : 4.1 ± 0.4 kcal/mole.

191

SEIZO OKAMURA

Table 3. Thermal data for post-polymerization

- Δ Η Ρ AHj

- Δ / / ρ Mf

Observed 8.1 + 1.1 0.7 + 0.1 0.8 + 0.1 5.8 + 0.3

Calculated 13.8 + 0.3 4.1 + 0.4 0.7 + 0.1 5.2 +0.1

These experimental facts (Table 3) seem to be well explained by the assumption of propagation in the molten state in acrylamide and propagation in the crystalline state in tetra-oxane both in the solid-state post-polymerization.

(e) Electron-microscopic observation by low-temperature replica for post-polymerizations of trioxane and acrylamide in the solid state16f

To demonstrate some differences of chain propagation between cyclic and vinyl polymerization, the replica method at low temperature was adopted with the electron microscope for observation.

(a)

Specimen Figure 12. Low-temperature replica apparatus, a: Photograph, b : Schematic diagram of a

single-stage carbon replica using a shield plate16f

192


5 ^

Figure 13. Electron-micrograph of trioxane. 1. Trioxane: 8000 x ; 2. Polymerized trioxane (P: 4%) 9300 x ; 3. Polymerized trioxane (P: 10%) 9300 x ; 4. Polymerized trioxane (P: 10%)

67000 x

In the radiation-induced solid-state polymerization of trioxane, crystalline polymer fibrils of 200 to 500 Â diameter and above 1 μ in length were observed in electron micrographs taken using a low temperature replica, as shown in Figures 13 and 14.

Table 4. Relations between intermolecular contacts in the monomer crystals and conformations of polymers produced

Compound Intermolecular contacts (A)*

Monomer unit length (B)t B/A

Trioxane Tetra-oxane Pentoxane Hexa-oxane

Trithiane BCMO

Acrylamide Acrylic acid

4.175 À 4.160 6.74 7.913

5.17 3.3

3.9a

3.52a

5.80 Λ 7.73 9.66

11.59

6.45 4.8

2.52b

2.52b

1.39 1.86 1.43 1.46

1.24 1.46

0.65 0.72

* The distance between the monomer molecules along the direction of greatest chain growth. t Occupied length of the monomer unit in the polymer chain. a. The shortest distance between the vinyl bonds. b. Fully-extended zigzag chain was assumed.

P.A.C.—30/1-H 193

SEIZO OKAMURA

Figure 14. Electron -micrograph of acrylamide. 5. Acrylamide : 6 700 x ; 6. Polymerized acrylamide (P: 1.1.%ì 83Q0 x ; 7. Polymerized acrylamide (P: 4.1 %) 20000 x ; 8. Polymerized acrylamide

(P: 20%) 10000 x

In those of acrylamide, amorphous globular polymer particles of 300 to 400 Â diameter were found. It has been concluded that chain propagation of trioxane was controlled by the regularities of ordered monomer molecules in the crystal lattice, but with acrylamide the reaction was not controlled.

Tables 4 and 5 summarize the displacements of monomer assumed during solid-state polymerization in ring-opening and vinyl polymerizations14.

Table 5. Cross section of monomer molecule (E) and that of polymer (Q)14

Cyclic monomer E Q Trioxane 25.48 17.30 Tetra-oxane 22.7 17.30 Pentoxane 25.6 17.30

4. RADIATION-INDUCED POLYMERIZATION OF CYCLOHEXENE OXIDE IN A PLASTIC CRYSTAL19

By thermal measurements on various kinds of monomers, cyclohexene oxide has been found to have two kinds of solid phases, Solid-II (below the

194


Gas

B.pt 132°C

M.pt

Liquid

Su

-36°C

1-81 °C

Sjand S n : Solid phases

Figure 15. Phases for 1,2-cyclohexene oxide 1 9

melting point, Tm = -36°C, to the transition point, Tt = -81°C) and Solid-I (below the transition point, Tt = -81°C) (see Figure 15). Thermal data are shown in Table 6, with some compounds already known as plastic crystals.

Table 6. Transition temperatures of several cyclic compounds1

Compounds

Cyclohexene oxide Cyclohexane Cyclohexanol

Transition Tt AS,

(°C) (E.U.)

- 81 - 8 7 . 1 - 1 0

7.5 8.66 7.45

Melting AS,

(°C) (E.U.) Δ 0 «

- 3 6 6.3

25.5

1.0 7.5 2.0 3.94 1.37 5.45

Figure 16 shows the x-ray diffraction diagrams. At the phase S-II the reflections at high orders have relatively weak intensities. The vapour snake phenomenon was also observed in the S-II-phase. These facts indicate that the S-II-phase may be a plastic crystal in which the molecules in the solid are fixed at the heavy points of a crystal lattice, but can rotate freely around the lattice points.

The radiation-induced polymerization behaviour has been presented in Figure 17 as the rates and degrees of polymerization. We can see that the rates in a plastic crystal are similar to those in liquids but the degrees of polymerization are nearer those of the crystal S-I.

195

SEIZO OKAMURA

A Cyclohexeneoxide

P las t i c c r y s t a l ST a t - 7 5 + 5°C

C r y s t a l S j a l - 1 2 8 ± 5 ° C

L i q u i d ( -36°C) S n (-81 °C) S j

B Cyclohexane

P l a s t i c c r y s t a l S j Ta t - 8 1 ± 5 ° C

C r y s t a l S na t - 1 3 0 ± 5°C

L iqu id ( ± 6 . 6 ° C ) SIT (-87.1°C ) ST

Figure 16. X-ray diffraction diagrams of cyclohexene oxide and cyclohexane19

196


Rp oc /"; n : 0-79 0.98 0-95 Gi = 60.7 £6.1 1.37

Figure 17. Polymerization of cyclohexene oxide, initial rates and degrees of polymerization at various temperatures in three phases

5. SUMMARY Here solid state polymerizations of various kinds are reconsidered.

Between the liquid and crystalline states, there seem to be supercooled liquid and plastic crystal phases, in which polymerization behaviour should also be interesting and worth more thorough investigation.

The author wishes to express his gratitude to the colleagues engaged here in this work.

REFERENCES 1 A. Henglein and R. Schulz. Z. Naturforsch. B9, 617 (1954). 2 H. Morawetz and I. D. Rubin, J. Polym. Sci. 57. 669 (1962). 3 G. Adler, D. Ballantine and B. Baysal, J. Polym. Sci. 48. 195 (1960). 4 C. H. Bamford, G. C. Eastmond and J. C. Ward, Nature. Lond. 192. 1036 (1961). 5 T. A. Fauder. I. D. Rubin and H. Morawetz, J. Polym. Sci. 37. 549 (1959). 6 J. H. O'Donnell. B. McGarvey and H. Morawetz. J. Amer. Chem. Soc. 56. 2322 (1964). 7 C. Chachaty and M. Magat. J. Polym. Sci. 48, 139 (1960). 8 B. S. Pshejetskii and V. A. Kargin. Vysokomol. Soedin. 3. 925 (1962). 9 Y. Amagi and A. Chapiro. J. Chim. Phys. 59, 357 (1962).

10 I. M. Barkalov. V. I. Goldanskii, N. S. Enikolopyan. S. F. Terekhova and G. M. Trofimova. J. Polym. Sci. C4. 909(1961).

11 G. Hardy. J. Varga and G. Nagy, Makromol. Chem. 85, 58 (1965). 12 C.S. H. Chen and D. G. Grabar. J. Polym. Sci. C4, 869 (1964). 13 K. Hayashi. Y. Kitanishi and S. Okamura. J. Polym. Sci. 58. 925 (1962);

K. Hayashi. H. Ochi. M. Nishii. Y. Miyaké and S. Okamura. Polymer Letters. 1. 427 (1963); K. Hayashi. H. Ochi and S. Okamura, J. Polym. Sci. 2, 2929 (1964).

14 Y. Chatani. T. Uchida. H. Tadokoro. K. Hayashi. M. Nishii and S. Okamura. J. Macromol. Sci.—Phys. B2(4). 567 (1968); B4. 61 (1970).

15 (a) I. Kaetsu. K. Tsuji. K. Hayashi and S. Okamura. J. Polym. Sci. A-i. 5. 1899 (1967); (b) I. Kaetsu. H. Kamiyama. K. Hayashi and S. Okamura. J. Macromol. Sci.—Chem. A3(8). 1509(1969); (c) I. Kaetsu. Y. Nakase and K. Hayashi, J. Macromol. Sci.—Chem. A3(8). 1525 (1969).

197

SEIZO OKAMURA 16 (a) T. Matsuda. T. Higashimura and S. Okamura. J. Macromol. Sci.—Chem. A4(l). 1 (1970);

(b) T. Matsuda. T. Higashimura and S. Okamura, J. Polym. Sci. Al. 8. 483 (1970); (c) S. Okamura. E. Kobayashi and T. Higashimura. Makromol. Chem. 88. 1 (1966); (d) S. Okamura. E. Kobayashi and T. Higashimura. Makromol. Chem. 94. 20 (1966); (e) M. Nishii. K. Hayashi and S. Okamura, JAERI-report 5018. 31 (1968); (0 M. Nishii. K. Hayashi and S. Okamura, JAERI-Osaka Annual Report. 151 (1968).

17 M. Shiogi. S. Ohnishi and I. Nitta, J. Polym. Sci. A. 3373 (1963). 18 T. Matsuda. T. Higashimura and S. Okamura. J. Macromol. Sci—Chem. A2(l). 43 (1968). 19 T. Hiramoto and M. Nishii. Annual Meeting of the Polymer Society of Japan (May 1970).

198

MODIFICATION OF POLYMERS FOR THE PREPARATION OF SEMIPERMEABLE MEMBRANES

MARIO PEGORARO

Istituto di Chimica Industriale del Politecnico, Piazza Leonardo da Vinci 32. 20133 Milan, Italy

ABSTRACT A short survey is made of the principles to be followed in order to prepare membranes fit for the fractionation of a solute from a solvent. First it is essential to use substances with convenient chemical properties and then to prepare different membrane structures depending on the fractionation to be done.

The reaction of grafting is recognized as a very important means for the preparation of convenient materials. As an example, we report the results obtained in the desalination process using membranes consisting of polyacrylic acid grafted on polypropylene. Membranes may be obtained from solutions or by grafting in the heterogeneous phase on films bi- or mono-oriented through a radical mechanism by peroxidation or y-irradiation.

Membrane structures observed with the electron microscope are described and the various results obtained in the desalination process are interpreted on the basis of the membrane structure. An interpretation is also given of the several transmission rates of water vapour, detected in the different types of

membranes.

As is well known, a membrane is an interface between two fluid regions. By the action of a convenient driving force, chemical substances may pass through the membrane from one region to the other.

The search for artificial membranes through which various chemical species may pass at quite different rates has interested scientists, chemists and physicochemical researchers for more than a century. Research work led on the one hand to a better knowledge of the transfer principles, to the definition of osmotic pressure, to the theory of thermodynamic properties of solutions and to a considerable development of the thermodynamics of irreversible processes1,2 ; on the other hand, it led to the study and production of permselective membranes that find very important practical applications : for example desalting of sea and brackish water, artificial kidney, water pollution control, and the separation of isotopes 235 and 238 of uranium3 '4.

The knowledge of the mechanism of selective transfer of substances on the molecular scale is fundamental to producing a membrane convenient for a given fractionation ; up to now, however, it is not contained within an adequate theoretical system. Different mechanisms of transfer are often involved and a reliable interpretation can hardly be found : however, some useful qualitative interpretations have been refined.

199

MARIO PEGORARO

The simplest possible mechanism for the selective transfer of a substance through a membrane is filtering. For example, using cellophane membranes, and since only the solvent may pass through them, it is possible to measure the osmotic pressure of high polymer solutions. By contrast, using only the simple 'sieve' mechanism, it is impossible to explain the selectivity of membranes hindering the transfer of ions dissolved in aqueous solutions; in fact, their size does not differ markedly from that of water that passes through the membrane (Table 1).

Table 1. Sizes of some typical ions21

Ion

Radius ofhydrated ion, Â

H +

2.82

O H -

2.46

Na +

3.58

cr 3.32

Mg2 +

4.28

The permselectivity mechanism can then be associated both with the solubility of the solvent (in the above case, water) in the membrane and with its diffusion (solution diffusion membranes). Although the membrane is macroscopically compact and homogeneous, it is never homogeneous on a molecular scale: actually, it consists of pores and channels whose size fluctuates with time; on the whole they represent the free volume of the polymer. If only the molecules of the solvent—and not those of the solute— can be adsorbed at the pore walls, the solvent fills up the pores and may easily pass from one adsorption centre to a neighbouring one, practically without requiring energy5. With cellulose acetate membranes, the adsorption forces of water to the polymer are essentially due to the hydrogen bond6.

In solutions in which the size of the solute molecules or ions does not differ much from those of the solvent, the selective transfer of the solvent may be also explained7 on the basis of simple considerations of surface tension and use of a membrane model with holes of very small size. It is sufficient that, at the membrane/solution interface, a negative adsorption of the solute takes place : i.e. the solute concentration at the interface is lower than that in the solution. In this way, a layer of pure solvent with thickness t is formed on the membrane. If holes exist with maximum diameter 2i, there may be a perfect fractionation of the solvent.

By considering the value of It (a few Â : if the interface is between air and a salt solution, 2i = 8 Â) this mechanism does not require a physical model of a membrane differing markedly from the previous one.

SOLUTE REJECTION In order that a membrane may exert a non-transfer action for the solutes,

it is useful that the equilibrium partition constant of the solute between the material of the membrane and that of the solution is as low as possible.

However, equilibrium considerations are not enough, as usual, to explain the mechanism of mass transfer and rejection.

200

PREPARATION OF SEMIPERMEABLE MEMBRANES

A possible explanation that is very convenient for desalination is the one based on the fact that membranes consist of materials having a dielectric constant lower than that of the solvent8. It is possible to demonstrate that the energy of an ion contained, for example, in an aqueous hole, is much higher than for an ion in solution : consequently the probability of the ion remaining in solution is high. This mechanism is especially apt for non-polar membranes.

The selective action of repulsion of the salts dissolved in water may be improved by the use of polymers containing ionic active centres (charged membranes). Since electroneutrality must be maintained, the passage of cations will be hindered if the salt anions are blocked by active centres in the membrane and vice versa (Donnan effect9).

MEMBRANE CLASSIFICATION Actually, membranes with sufficiently large pores (porous membranes),

several sizes larger than the average free run of passing molecules (100 Â -10μ), are interesting owing to the possibility of high flowrates, obeying the Poiseuille rule (viscous flow)10 but less interesting if one has to separate small dissolved ions with dimensions comparable to those of the solvent. Separation of the solvent from the solute is possible only if the concentration of the solute in the liquid pore differs from the concentration in the external solution. This happens only if a membrane effect exists. In general viscous flow (bulk flow) transfer is less selective than that occurring with a solution-diffusion mechanism in compact and non-porous membranes constituting the second limit class of membranes (solution diffusion membranes). Obviously intermediate types of membranes are possible (finely porous membranes), in which transfer occurs partly through a bulk mechanism, although the interactions between membrane and penetrating substance are very important11. Pore diameters in this type have a mean value of 7 to 50 Â.

As clearly shown from the above, in order to obtain a membrane that will allow the selective passage transfer of the solvent and not of the substances dissolved in it, it is necessary: (1) to choose a solid substance insoluble in the solvent, but with a good chemical affinity toward the solvent and not toward the dissolved substances ; moreover, it must possess good mechanical properties and good ageing properties ; (2) the substance chosen must have an adequate structure, i.e. containing pores with a small equivalent diameter.

Composite structures consisting of a compact thin skin (porous on the molecular scale) and of a thick porous (on a macroscale) layer acting as a skin support, were largely successful12.

Owing to their easy processability and convenient properties, polymers are widely used for the production of membranes. A number of homo-polymers and copolymers13 have been studied and some of them are now also used in considerable quantities on a pilot plant or semi-industrial scale. Grafting is certainly one of the most versatile tools for regulating the properties defined at the outset. For example, water-soluble polymers, e.g. polyacrylic acid, may become insoluble by grafting them to an insoluble polymer14; hence the system may be swollen in water and has good properties of water transfer; polymers that are convenient to desalting (cellulose

201

MARIO PEGORARO

acetate but showing irreversible compressibility phenomena become stiffer after polystyrene grafting15. The reaction of grafting is an additional operator : it does not substantially change the main properties of the backbone polymer (A), but it adds the properties of the grafted polymer (B). Although the two polymers A and B have different cohesive energies, they can become compatible on the visible and macroscopic scale. However, this does not occur on a molecular scale ; in fact in the solid state a biphasic structure is formed16 with finely interdispersed domains of phases. Therefore grafting also markedly influences the structure of the membranes. In particular, grafting may disturb packing of the macromolecules : depending on circumstances, structures may assume a higher or lower free volume (and more or less low densities) than those corresponding to the simple additivity rules and consequently may become more or less permeable.

For example, in the case of polypropylene, grafting of a polyhydrocarbon, such as polystyrene, improves the permeability to water vapour17.

Moreover, when the membranes are obtained from a solution, the use of a solvent that is good for A and bad for B, makes the molecules of B contract in solution: the structures obtained are completely different from those obtainable with a solvent good for B and bad for A. In this respect, we mention, by way of example, poly-2-vinylpyridine grafted on polystyrene18.

REACTIONS OF GRAFTING Materials convenient for membrane production may be prepared by

reactions both in the heterogeneous and in the homogeneous phase. In the first case, the starting product is generally a pre-formed film (e.g. nylon, teflon, polypropylene), and grafting is usually done with the chosen monomer in the presence of convenient initiators, often by irradiation19 or sometimes by using peroxidic functions20 fixed by per oxidation on the film to be grafted or other convenient chemical functions.

These methods are quite handy and relatively simple since they start from a pre-formed matrix. By contrast, when operating in the homogeneous phase, all the reagents must be in solution ; moreover, the membrane must be filmed—which is a further difficulty. The latter method, although less easy, is very interesting : by it, in fact, it is possible to obtain different structures depending on the operating conditions.

Among the methods quoted, the technologically simplest one uses radiation grafting starting from a polymeric film. This method is very flexible since it allows grafting of the desired monomers on films of practically any polymer. Since 1957, i.e. since Chen published his research on the preparation of membranes consisting of sulphonated polystyrene grafted on polyethylene by radiation21, many articles have been published on this topic19.

We were the first who in 1968 obtained membranes by chemical grafting in solution of high-melting crystalline polymers followed by high-temperature film-making20.

We report here the main results obtained from the research we carried out concerning the preparation and study of the properties of some membranes from 1967 to the present time.

202

PREPARATION OF SEM1PERMEABLE MEMBRANES

(a) Preparation of membranes from preformed films In our laboratory we used polypropylene and Saran films. Grafting was

done with different monomers : in the first case the backbone polymer was peroxidized and in the second copper complexes were used22. The structure of the starting film is very important, since phenomena of monomer diffusion and the grafting sites depend on the structure. Looking at the surface and cross section of a bioriented polypropylene film (density 0.90) by scanning microscope, we observed23 that its structure consists of compact lamellae not easily distinguishable. In order to perform grafting, it is necessary to introduce by oxidation20, after elimination of the stabilizers by carbon tetrachloride extraction, hydroperoxide groups (range 0.1 to 0.6 per cent of active oxygen) which disturb the structure causing reliefs in some regions (Figure la) and detach the lamellae (Figure lb). After grafting with acrylic acid, the lamellae become much more distinct than before (Figure 2), and the shape of superficial irregularities is changed23.

When starting from a commercially so-called un-oriented film (density 0.89, thickness 20 μ) we observe (Figure 3) the section structure prevailingly with one-direction orientation (thick laminae) and several irregularities in the direction perpendicular to it23. Un-oriented films also show a higher permeability to water vapour than bioriented ones24. After extraction of stabilizers at room temperature by benzene, peroxidation causes some structural modification (laminae detachment23 and grafting may occur in such a way that the original structure is modified in depth (Figure 4))\ laminae disappear and a microporous structure is generated.

Radiation of moplefan by y-rays (60Co source) causes some changes that appear23, however, not very important in the scanning microscope especially for bioriented moplefan.

Grafting by irradiation of polyacrylic acid produces modifications of structure with both bioriented and un-oriented (Figure 5) moplefan: the original lamellar structure is not maintained.

fb) Preparation of membranes from solutions of grafted polymers Table 2 shows some of the extraction membrane types obtained after

starting with two different backbone polymers, one crystalline and one amorphous ; the main monomers used and the conditions adopted for the extraction. Since polypropylene is soluble only at high temperatures, high temperatures must be used for grafted membrane extraction (120° to 140°C). It is convenient to carry out grafting directly at these temperatures, immediately before the extraction. A thorough study of the reaction under such conditions for the grafting of acrylic acid was recently published14. After completing the reaction, in order to avoid degradation, stabilizers should be added and film-making of the membrane should be carried out quickly.

Film-making may be carried out by various techniques : we found it very convenient to extract a glass cylinder from a hollow cylinder containing the solution operating in a thermostat at a high temperature (120° to 140°C). In this way, depending on the extraction rate and on the composition of the reaction mixture, it is possible to obtain by evaporation membranes of different thicknesses25 between 2 and 30 μ. After forming, membranes are

203

(a)

Figure 1. Surface (a) and cross section (b) of bioriented peroxidized moplefan (scanning).

204

Figure 2. Cross section of bioriented grafted (by oxidation) moplefan (scanning) (PPA content : 55 per cent).

Figure 3. Cross section of un-oriented moplefan. extrusion direction (scanning).

205

5jfe

Figure 4. Cross section of grafted (after oxidation) un-oriented moplefan (PAA (scanning).

81 per cent)

FigureS. Cross section of radiation grafted un-oriented moplefan (scanning) (PAA = 43 per cent. dose 380000 rad).

206

PR

EP

AR

AT

ION

O

F

SEM

IPE

RM

EA

BL

E

ME

MB

RA

NE

S

~ 2 b

ci ^ 'J

ω S

a (U

._

(U

H

o

1) > o ζ/}

ÜC

JC

JC

JC

J4

J4

JC

JC

JO

CJ

ce

cc

cc

cc

cc

e C

JC

J<

JC

JD

CJ

CJ

CJ

1)C

JC

J Ν

ΝΝ

ΝΝ

ΝΝ

ΝΝ

ΝΝ

υ C

CC

CC

CC

CC

CC

c c

jc

jc

jü

cj

cj

wc

jc

jc

je

jr

)

οο

οο

οο

οο

οο

ο^

oo

oo

oo

oo

oo

og

XJ

XlX

lXlx

:x1

XlX

!X:x

:xi.fi

ου

οο

ου

υο

υυ

υθ

S S S S S S S S S S S £ Ò

ÓÓ

ÓÓ

OÓ

ÓÓ

ÒÓ

U

o 2 o eu

43

32

32

33

32

38

*8*8

*

U

u υ

υ

XX

X o

o o U

, U

, i-i

<υ <υ <υ

G. a

a

>> >* >* o

o o N

N

N

fi

G

G

<υ ω

<υ OQ

CQ O

ù O

N

C cj

X)

UH <υ Π

,

>* 3

OQ

*L

<υ cj ω

<υ <υ <υ <υ . rR

!Η 3 Γ° 3 .Ή

Η2 !Η '

^ Χ

ΧΧ

ΧΧ

ΧΧ

ΓΛ

/^

ΓΛ

οο

οο

οο

ο·

2°

·£

*>, ^ ^

^ ">. ">Ϊ *>> £ 'S

3 O

OO

OO

OO

cd

JS

cd

CC

CC

GC

CU

g-

W

DÜ

D(

UD

Da

>'

G'

jC

CQ

ÎQC

QC

QC

QO

QP

QU

U

1 υ-

; £

'-> Γ3 «

O

Cd x <υ fi .>»

x:

a> ■

>Ï

<υ cd

>> 0 cd

X!

<υ

ε >> X

I <υ

>> (U

cd

>> CJ cd

X <υ

ε >» X <]>

>%

TJ ^ T

3 Ό

"Ό *0 Ό

<u cj

'υ 'ο 'Ü 'Ü

'Ü "o 'o 32 .32 ^ ., .>

cd

cd

cd

cd

cd

cd

cd

prjO

OO

O

CJ

OO

OO

CJ

55

X:

±:

±:

.M .^ .— .»^ ._ .« .—

cd cd »o

-Q 73

UU

I-

IU

UU

IU

II

-I

U. >

T<

ITH

IT'

>* -5

<υ <υ fi fi

<υ <υ

<l> fi <υ

Q> fi (U

« fi <υ

<υ <υ

fi fi cu

<υ

CJ fi <υ

<u <υ

fi fi <u

cj CJ

O

c "fi

cj X

αα

.α.α

,αο

.αα

.α-ο

- D-^L

OO

OO

OO

OO

OO

Oc

xT* " XT' XT' XT' XT'

XT' XT' XT' XT' XT' 0

00

00

00

00

00

0 P

HC

MP

-ICU

&ÌC

UC

MU

SC

HC

UP

MC

U o

OH

207

MARIO PEGORARO

Figure 6. Section of an extracted membrane (PAA = 57 per cent) (transmission).

Figure 7. Section of an extracted membrane (PAA = 12 per cent) (transmission).

208


stored at room temperature in water which may (at least partially) extract the homopolymer when it is soluble in it (e.g. polyacrylic acid, etc.).

If the backbone polymer (for example PVC, Table 2, and its grafted derivative) is soluble in the reaction solvent at room temperature, filmmaking of the reaction products is identical, but much easier.

The structures of the membranes obtained may be well shown up in the electron microscope (transmission technique). Figures 6 and 7 show the cross sections of two polypropylene membranes one with 12 and the other with 57 per cent of polyacrylic acid. After treatment with osmium, it is possible to distinguish darker domains (polyacrylic acid) intermingled with lighter regions (polypropylene) arranged in a fairly regular way. The domain area is related to the composition. Also, in the case of moulded products completely free from homopolymer obtained from the same material as that used for membranes the existence of two phases is proved by mechanical dynamic measurements, which show the existence of different relaxation processes characteristic of polypropylene and of polyacrylic acid16.

PROPERTIES OF THE MEMBRANES On the basis of the foregoing investigations, both mechanical, dynamic

and optical, we conclude that the membranes illustrated above (no matter how obtained) are biphasic after grafting and consist of physically separated regions of polyacrylic acid and of polypropylene. Only the shape and distribution of the two components differ depending on how the membranes were obtained : if made from solution, the components will be distributed in distinct regions : PAA is immersed in the PP matrix ; if made from per-oxidized bioriented films, the PAA regions are distributed among the PP lamellae in such a way that lamellae persist but become more spaced out ; if made from un-oriented peroxidized films the initially stratified structure appears to be severely disturbed. In irradiated films, the distribution of PAA regions in the PP matrix is probably more homogeneous than that in the previous case.

Water vapour transfer through membranes from a saturated vapour

Table 3. Permeability ASTM-E 96-66 cup vapour test : T = 23°C. Drying agent : phosphorus pentoxide

Type of material

Extraction membrane

Bioriented irradiated moplefan

Not oriented irradiated moplefan Peroxidized bioriented moplefan Peroxidized not oriented moplefan

PAA grafted

°/o

r41 Î43 152 L57 f 30

23 l 4 5

26 81 81

Thickness, μ

25-40 3-6

10-15 2-6

6 initial 10 initial 10 initial 20 initial 10 initial 20 initial

mg H20/h cm2

0.7 1 1.4 1.5 1.2 0.2 0.8 0.8 0.3 1.4

209

MARIO PEGORARO

space to a dry one at 23°C was measured by the ASTM E 96-66 cup method. Transfer occurs according to data summarized in Table 3. Transfer rates appear to be generally an increasing function of the percentage of PAA.

Transfer rates decrease in the order : extraction membranes, un-oriented irradiated moplefan, bioriented irradiated moplefan, un-oriented per-oxidized moplefan, bioriented peroxidized moplefan. This order is connected with the structure, which is more compact for moplefan than for extraction membranes and is more compact for bioriented films than for those not oriented.

With regard to extraction membranes, the rate is independent of thickness : this suggests that the rate of the process is not limited by diffusion inside the membrane, but by evaporation or condensation of water at the interfaces. As to membranes obtained from irradiated films, their rates depend, however, on thickness.

Extraction membranes show higher compressibility than peroxidized or grafted films. Figure #, shows, for example, the effects of compaction on pure water flux through an extraction membrane (PAA = 33 per cent) with particularly high permeability (low thickness) versus time.

The higher compressibility of extraction membranes may act in such a way that the above order of permeability, determined by the cup method, is changed in reverse osmosis plants under high pressures: permeability of membranes may be considerably reduced.

PERMSELECTIVE PROPERTIES OF MEMBRANES Membranes of PAA grafted on PP may be used for sea or brackish water

desalination. Several runs were performed by us in a reverse osmosis plant with water containing sodium chloride (10000 p.p.m.), under pressures of 50 (or 100) atm.

Figure 9 shows the experimental flux/salt rejection curve obtained with all the extraction membranes tested by us, containing 40 to 50 per cent of grafted PAA: rejection increases with decreasing flux. Figure 10 shows flux at 50 atm versus thickness. The highest rejections obtained (80 per cent) were found by operating with a thickness of 17 μ (flux was 1.1 l./hm2 at 50 atm, grafted PAA = 43 per cent).

When using membranes obtained from not oriented peroxidized and grafted moplefan (thickness = 20 μ), fluxes are comparable with those of extraction membranes, and rejections are even better (maximum obtained, 97 per cent) (see Table 4).

If membranes are obtained from bioriented moplefan, grafted after per-oxidation (thickness = 20μ), fluxes are zero for a long time; with high percentages of grafting, the flux only begins to rise after more than twenty hours (0.04 l./h m2) with fairly high rejections (66 per cent).

Finally, when using radiation membranes higher fluxes are observed and rejections are zero. After operating at 50 kg/cm2 and 10000 p.p.m. of sodium chloride and after an induction period of more than ten hours, we found fluxes of 3.5 l./hm2, using initially bioriented moplefan (thickness 10μ, PAA = 38 per cent) and we found fluxes of 8.1 l./hm2 using un-oriented moplefan (thickness 20 μ, PAA = 40 per cent). Radiation-grafted moplefans

210


I*

' · ' · · · 25

J I I L _l I I L 0 1 2 3 4 5 6 7 8 9 10

/,h

Figure 8. Compaction effects on an extraction membrane 5 μ thick. PAA = 33 per cent. Flux versus time at different constant pressures.

Table 4. Desalination properties of grafted moplefans (10 000 p.p.m. sodium chloride in water, pressure 100 atm)

Type

Not oriented Not oriented Bioriented

Bioriented

Type of grafting

Peroxidation Peroxidation Peroxidation

Peroxidation

Initial thickness

μ

20 20 12.5

20

PAA grafted

% 81 69 55

32

Flux l./h m2

0.8 0.5 0.04

Mean rejection

/ o

92 87 66

Notes

Flux starts after 23 h

No flux for 48 h

211

MARIO PEGORARO

uu

80

60

40

20

1-

" ^ :

""·." " ^ ^ — _ - . _ . - ] ι 1 ι 1

0 10 20 30 F, t /hm2

Figure 9. Extraction membranes: flux versus rejection (Sodium chloride 10000 p.p.m.; 50 atm).

30

20

£ -C

10

P=50 kg/cm2

10000 p.p.m. NaCl

\

10 20

Figure 10. Extraction membranes: flux versus thickness (Sodium chloride 10000 p.p.m.; pressure 50 kg/cm2).

212


Extraction membrane

Grafted film

Figure 11. Model of PP-PAA grafted membranes.

30

x 20

10 l·

Exchangeable g.ion H+=620 x 10"

0 1 2 3 4 5 6 7 8 9 10 11 I.

Figure 12. Hydrogen ions exchanged versus volume (litres) of 3.5 per cent sodium chloride solution eluted through grafted powder (PAA = 45 per cent).

213

MARIO PEGORARO

using other monomers (say, 2-vinyl pyridine) after iodomethylation appear to give satisfactory salt rejections26.

Desalination may be interpreted by assuming a membrane model of the type shown in Figure 11 (a) (extraction membrane) or in Figure 11 (b) (membrane from film). PAA and PP regions must necessarily show low adhesion : furthermore, water swells PAA remarkably, causing internal stresses and possible microfractures, which must necessarily have an extremely small equivalent diameter ; in fact, determinations of surface area made by us have shown that extraction membranes are compact. Microfractures may constitute continuous channels : their possibility of passing through the sample is the greater the lower is the extraction membrane thickness. This explains both high fluxes and low rejections of extraction membranes with low thickness : in this case, the liquid flow should be of the bulk type. As to membranes with fairly large thickness, salt water reaches the PAA regions either directly or through microfractures ; however, it cannot pass through the membrane by continuous channels. PAA is very hydrophilic and can repel salt owing to the Donnan effect. In fact, it reacts with sodium chloride. Figure 12 shows the hydrogen ion exchange of a pure grafted powder (PAA grafted on PP equal to 45 per cent, weight 1 g, put in a continuous flux of 3.5 per cent sodium chloride solution). The membrane may then act with a diffusion solution mechanism, which is most effective for desalination. However, too great a thickness reduces the flux, by increasing the number of exchanges and the overall resistance.

Table 5. Ultrafiltration properties of extraction membranes27.

% Rejection operating with

%of T h . . ° p e r a t i n g Flux 5 % °·5 ,%

Type /0... Thickness pressure , ., 2 saccarose protein grafting , , _ 2 l./h nr , . , +. 0 & μ kg/cm z solution solution

Polyacrylamide grafted on polypropylene 30 10 100 5.4 50 —

Poly-2-hydroxyethyl methacrylate grafted on polypropylene

Polyacrylic acid grafted on polypropylene

Poly-2-hydroxyethyl methacrylate grafted on polypropylene

In addition to desalination, the membranes prepared by us may also be useful for many other applications. Table 5 shows rejections of saccarose and of proteins (a and β lactoalbumins and β lactoglobulins : MW 70 to 250000) from aqueous solutions27.

CONCLUSIONS Grafting is a very important means for obtaining materials fit for the

production of permselective membranes. It is not only important to use polymers with adequate chemical properties, but also to prepare a membrane

53 33 33 78 78

5 5 2 7 2

50 5 20 4 3.5

5.4 8.3 — 0.9 4

50 — — — —

— 41 60 60 80

214


structure suitable for fractionation. When using entirely incompatible polymers, such as polyacrylic acid (hydrophilic) and polypropylene (hydro-phobic) it is possible to obtain low-thickness very permeable membranes, which may be used for filtering, although they are not selective for low-molecular weight substances, probably owing to the very frequent micro-fractures generated by grafting.

On the other hand, when using thicker membranes, the transfer process changes and becomes of the solution-diffusion type; hence it allows high rejections, but not very high fluxes.

ACKNOWLEDGEMENT The financial support of the Italian Institute for Research on Water

(IRSA) of the National Council of Research (CNR) is gratefully acknowledged. I thank Dr A. Penati and Dr G. Alessandrini for their valuable cooperation.

REFERENCES 1 S. B. Tuwiner. Diffusion and Membrane Technology, Reinhold: New York (1962). 2 A. Katchalsky and P. F. Curran, Non-equilibrium Thermodynamics in Biophysics, Harvard

University Press: Cambridge, Mass. (1967). 3 S. Sourirajan, Reverse Osmosis. Logos Press: London (1970). 4 New Scientist, London, p 225 (29 October 1970). 5 C. E. Reid in Desalination by Reverse Osmosis, p 11 Edited by V. Merten. MIT Press: Cam

bridge, Mass. (1966). 6 C. E. Reid and E. J. Breton, J. Appi. Polym. Sci. 1, 133 (1959). 7 S. Sourirajan, Industr. Engng Chem. Fundamentals, 2, 51 (1963). 8 E. Glueckauf, Proceedings of the First International Desalination Symposium, Paper SWD/1.

Washington, D.C. (3-9 October 1965). 9 F. G. Donnan, Chem. Rev. 7, 373 (1937).

10 V. Merten, Desalination by Reverse Osmosis, p. 22. MIT Press: Cambridge, Mass. (1966). 11 H. F. Mark and N. G. Gaylord, Encyclopedia of Polymer Science and Technology. Vol. VIII,

p 620. Interscience-Wiley : New York (1968). 12 S. Loeb and S. Sourirajan, Advanc. Chem. Ser. 38, 117 (1962). 13 See for example H. F. Mark and N. G. Gaylord, Encyclopedia of Polymer Science and Techno

logy, Vol. VIII, p 628. Interscience-Wiley : New York (1968). 14 M. Pegoraro and A. Penati, Chim. e Industr. 53, 235 (1971). 15 H. B. Hopfen berg, V. Stannett, F. Kimura and P. T. Rigney, Membranes from Cellulose and

Cellulose Derivatives, p 139. Interscience: New York (1970). 16 M. Pegoraro, L. Szilagyi, A. Penati and G. Alessandrini, Europ. Polym. J., 7, 1709 (1971). 17 G. Albanesi, E. V. Zaitseva and E. Beati, Chim. e Industr. 49, 1300 (1967). 18 V. Stannett, J. Macromol. Sci. Chem. A, 4 (5), 1177 (1970). 19 See for example S. Munari, Fourth International Congress of Radiation Research. Evian

(28 June-4 July 1970). 20 Quaderni de La Ricerca Scientifica No. 58, 91 CNR : Roma (1969). 21 W. K. Chen and co-workers, J. Polym. Sci. 23, 903 (1957). 22 Italian Pat Appi. No. 20217 A/71. 23 G. Alessandrini, M. Pegoraro, A. Penati and G. Mossa, Chim. e. Industr., 54, 105 (1972). 24 Montecatini Edison Bulletin Moplefan (1966). 25 M. Pegoraro and A. Penati, unpublished work. 26 S. Munari, private communication 27 C. Peri and C. Cantarelli, private communication. 28 Y. Marcus and A. S. Kertes, Ion Exchange and Solvent Extraction of Metal Complexes, p 28.

Wiley: New York (1969).

215

SOME PROBLEMS OF CHEMICAL AND PHYSICAL MICROMODIFICATION OF POLYMER SYSTEMS

N. A. PLATE

Institute of Petrochemical Synthesis, Academy of Sciences of the USSR, Moscow, USSR

ABSTRACT The experimental results from, and the approaches to, the study of the role of microadditives in polymer systems are reviewed. The influence of low-molecular weight plasticizer at levels of 0.1 to 0.5 per cent added to polymer which results in the abnormal decrease of the glass transition temperature, in the sharp decrease of specific viscosity of polypropylene melts, and other non-classical phenomena, is discussed. Some features of gel formation in the case of a system of comb-like polymer (polyhexadecyl acrylate) with aliphatic hydrocarbon or aliphatic alcohol are discussed. Such gels are structurally ordered systems, they are formed at very low concentrations of polymer (0.1 to 0.3 per cent), they behave as typical lyotropic liquid crystals and the role of the structure-forming agent of the polymer toward the low-molecular medium is clearly demonstrated.

The role of chemically attached microadditives is demonstrated on the example of copolymers of isopropyl acrylate and hexadecyl acrylate which retain the hexagonal crystalline lattice parameters typical of the homopolymer of the hexadecyl acrylate even at low content of the latter.

The other examples are self-stabilized polyvinyl chloride containing chemically attached organotin stabilizing groups in the chain, and macromolecular models of proteolytic enzymes with partially alkylated polyvinyl pyridine. In these systems a slight change in the content of chemical modifier in the chain

provides inadequate change in the macroproperties of the total system.

The marked tendency of polymers toward structure formation not only within the crystalline state but in the amorphous one as well, and the ability of macromolecules to form associations even in dilute solutions oblige us to reconsider now the role of microadditives in polymer systems. Such additives being chemically bound to the macromolecular chain or being introduced as admixtures can change decisively all the physical and mechanical properties as well as the reactivity of the system. This direction of research in the field of modification of properties of polymer substances which has received much attention in the last few years has already resulted in a series of interesting and useful results. The aim of the present review is to discuss some recent results in this field, when the introduction of microquantities of 'foreign' groups or units into chains or microadmixtures in the amount of one molecule on average over dozens and hundreds of macromolecules brings about desired effects in the change of macroproperties of polymer systems.

217

N. A. PLATE

As one of the examples of the influence of microadditives on physical properties of polymers, I would like to cite results on the structural plasti-cization of amorphous and crystalline polymers1. In the early publications of P. V. Kozlov and V. A. Kargin2 it was shown that besides the well-known mechanism of molecular plasticization of polymers there is another mechanism of structural plasticization when a low molecular substance scarcely compatible with the polymer is being introduced into the polymer and is situated on the boundaries between supermolecular structures where it plays a role of lubricant and favours the easy displacement of structural elements. This results in a sharp decrease of the glass transition temperature when the content of plasticizer is between 0.05 and 0.1 per cent. The typical situation is represented in Figure 7, where Tg of triacetylcellulose is plotted against the plasticizer content, the latter being a molecular or structural one depending on its chemical nature.

160

120 o o

80

40

0 10 20 30 c,%

Figure J. Glass transition temperature of cellulose triacetate as a function of plasticizer content: 1—l-nitro-2-methyl-2-propanol; 2—trimonochloroethylphosphate; 3—butyl stéarate.

Much more effective are the results obtained by G. P. Andrianova and V. A. Kargin1 '3 on the micromodification of polymeric melts. In Figure 2 the effective viscosity of two polypropylene melts at fixed shear modulus is plotted against the content of oligomeric polydiethylsiloxane additive. The introduction of an amount of additive as small as 0.05 per cent results in a tenfold viscosity decrease without any change of mechanical properties of the crystalline polypropylene in the block. These data show the possibility of displacing structural elements on the supermolecular level in polymeric melts.

Depending on the nature of an additive the concentration dependence of the viscosity can be either an extreme or a permanent low viscosity value is reached which is not changed with further increase of the low molecular

218

PROBLEMS OF CHEMICAL AND PHYSICAL MICROMODIFICATION

Admixture, %

Figure 2. Effective viscosity of polypropylene melts at constant shear stress as a function of the ethylsiloxane oligomer content: 1—Moplen (=1 x 105 dyne/cm2); 2—ICI sample ( = 5 x

104 dyne/cm2).

component content. Both types of dependence are strictly different from the known mechanisms of plasticization of melts which lead to the exponential viscosity function in relation to plasticizer content.

Microadditives which act on the supermolecular level are capable1 of increasing fatigue resistance sharply during the periodical loading of rubbers (12 to 15 times at the additive content of 0.1 per cent and without changing the stress/strain characteristics or other mechanical properties of the polymer). These additives can also produce a marked increase of adhesion of polymer material (the maximum of adhesional content of plasticized Polyvinylchloride is at the additive content of 0.02 to 0.03 per cent).

The above results show that the efficiency of action of microadmixtures overlaps the usual additive scheme of interaction on the molecular level. The reason for this is that here one takes advantage of a chance to use and to develop the strong tendency of polymer substances to be organized and structured. Due to this fact the role of boundaries between structures in such microheterogeneous systems greatly increases and, consequently, very small amounts of additive are sufficient to ensure the desired effect. When one goes to higher concentrations of additives, the result is to change the mechanism of action and transfer all the phenomena into another more usual and in this sense less effective category.

A further problem concerns the role of microadditives if they are chemically bound to the polymer. The strong struclurability of the macromolecules provided by the presence of a small number of interacting units and groups can be demonstrated by taking as an example a curious class of comb-like polymers of the acrylic series, which were studied in detail on account of their liquid-crystalline state4. We have recently found that polyacrylates, polymethacrylates and polyvinyl esters which have as ester group an aliphatic side group containing 10, 14, 16 or perhaps 18 carbon atoms, i.e. long side branches in each monomeric unit, are crystallized in a hexagonal

219

N. A. PLATE

lattice with participation of side aliphatic chains. This type of packing is none other than one of the forms of smectic liquid crystals stable over a wide temperature interval4"6. As has been proved by x-ray analysis, such packing is achieved by close placement of the cylinders of side aliphatic branches with some mobility in azimuth around the C—O bond at the place of attachment to the backbone chain. Such structures give the appearance of long periods within the liquid crystalline state derived from the parallel packing of layers formed by the side branches. Small angle x-ray diagrams of polyhexadecyl acrylate and of polyhexadecyl methacrylate show a long period which is normally the sum of the lengths of two sidegroups plus the width of the main chain in the case of polyacrylates, and the length of sidegroups in the case of polymethacrylates (in the latter case the more rigid methacrylate chain does not enter the crystalline lattice)4-6. One of the peculiarities of the structure of such comb-like polymers is the abnormal stability of layer packing in these systems and a strong tendency to inter-branch interaction of the aliphatic chains. The interplanar distances for some polymers and copolymers of this type derived from x-ray large angle and small angle analysis as well as data on temperature and heat of fusion are given in Table 1.

The interaction between side chains and sufficient mobility of fragments of macromolecules within side branches permits introduction of the 'foreign' units into comb-like polymers without essential disturbance of their crystalline structure. The r1 and r2 values in the radical copolymerization of hexadecylacrylate and isopropylacrylate are rx = 0.58 ± 0.13; r2 = 0.70 ± 0.15 and this allows us to consider the copolymers of these two components as statistical with the corresponding unit distribution along the chain. The data given in Table 1 show that the introduction of up to 80 Mol.% of isopropylacrylate does not result in the disappearance of the hexagonal type structure or of layer packing. The long period is preserved, although its value changes for small contents of isopropylacrylate. This period tends to correspond not to two-layer but to one-layer packing, i.e. the introduction of isopropylacrylate units into the chain is equal, from this standpoint, to the appearance of an α-methyl group in the backbone chain because one-layer packing is typical for polymethacrylates.

The most interesting phenomenon is the existence of layer packing at the mole ratio one unit of hexadecylacrylate to four units of isopropylacrylate. There is some reason to suppose that the minimum critical concentration of long side monomer in the chain can be even less and still conserve layer packing, in other words, the long branches can still find each other in the polymer mass. On the other hand, the introduction of 4 to 6 mole per cent of hexadecylacrylate totally prevents the crystallization of polyisopropyl-acrylate—the polymer becomes amorphous due to the combined effect of two different tendencies toward crystallization—the one of polyisopropyl-acrylate which 'wants7 to produce helical conformation of the main chains, and the other, of polyhexadecylacrylate, for which the hexagonal packing of side chains is typical. All this clearly demonstrates the influence of long side groups on crystallizability. This opens the way for regulation of such processes in polymers by means of the introduction of comb-like monomers into linear polymers by copolymerization.

220

Table 1. Interplanar distances, Tm and AHm for comb-like polymers and copolymers.

O BJ r m

O *n n m g n > r > Ö

n > r

o o Ö s n > H δ

Interplanar distances, Â. Numbers of diffractional maxima

Polymer

1

Ά2 44.5 47.0 29.0 30.0 26.5 26.5 27.0 26.5

Small angles

2

14.7 14.9 15.8 14.7 14.9 13.6 13.9 13.6 —

3

8.34 8.84 9.50 8.67 9.90 8.69 8.70 8.81 —

4

6.06 6.30 6.60 — —

6.34 — — —

5 6

— 4.19 — 4.17 — 4.15 — 4.19 — 4.17 — 4.19 — 4.19 — 4.19 — 4.19

8.49

Large angles

7

2.43 2.41 2.40 2.41 2.42 2.43 2.43 — —

5.15

8

2.10 2.08 2.10 2.08 — — — — —

4.2

T„ °C (±0.5)

38.0 46.0 49.0 22.0 40.0 32(26.1)* 26.5(21.3)* 18.0(11.4)* 14(-5.7)*

—

àHm cal/g

19.9 24.3 24.6 9.4 — — 9.2 7.0 — —

Polyhexadecyl acrylate, PA-16 Polyheptadecyl acrylate, PA-17 Polyoctadecyl acrylate, PA-18 Polyhexadecyl methacrylate, PMA-16 Polyoctadecyl methacrylate, PMA-18 Copolymer A-16-IPA (67: 33) Copolymer A-16-IPA (56: 44) Copolymer A-16-IPA (42 ;58) Copolymer A-16-IPA (22:78) Copolymer A-16-IPA (2: 98)

* Melting point predicted by Flory's equation for copolymers.

N. A. PLATE

The tendency described above of comb-like polymers to structurize also results in the development of the role of these polymers themselves as effective modifiers of the system.

The intensive intramolecular interaction of the sidegroups leads to the fact that with increase of side chain length the optical anisotropy of the macromolecules strongly increases as will be seen from the data on dynamic birefringence (Table 2)4. In comb-like polymers the rigidity of the main

Table 2. Number of monomer units in thermodynamic Kuhn segment v, the segment anisotropy (αι — «2) a n d the anisotropy of monomer units (a^ — a±) for some esters of polymethacrylic acid.

Polymer

Polymethyl methacrylate PMA-1

Polybutyl methacrylate PMA-4

Polyhexyl methacrylate PMA-6

Polyoctyl methacrylate PMA-8

Polyhexadecyl methacrylate PMA-16

Cetyl ester of poly-p-methacrylyl

v (<xl - a2) x 1025 cm3 (,

7.0

6.7

8.6

7.9

19.0

25.0

+ 2

- 1 4

- 4 0

- 4 7

- 1 7 0

-2500

«II - fl±) x 1

+0.2

-2 .1

-4 .6

-5 .9

-8 .9

-100.0 oxybenzoic acid

PMOB-16

chain increases and the orientational order in the side chains goes up due to the interaction of alkyl radicals. The most perfect intramolecular structures arise in the solutions of esters of polymethacryloxybenzoic acid, which contains sidegroups capable of forming liquid crystals. The negative segment anisotropy of the PMOB-16 macromolecules is already comparable with the anisotropy of crystal-like molecules. Evidently, in these systems, specific physical polymer network formation can occur even in dilute solutions. This phenomenon was indeed observed4'7 in a series of our researches with polyhexadecylacrylate, which was found to be a very effective structure inducer and medium modifier, particularly a modifier of low-molecular solvent. The unexpected role of small amounts of polymer additives as structure organizers of the liquid medium becomes very evident.

Such an effect can be observed clearly when one uses as solvent the substances according to their chemical nature analogous to the side branchings of comb-like polymers. Thus, solutions of PA-16 in n-aliphatic hydrocarbons C10H22 to C16H34 and n-aliphatic alcohols (C7H15OH to C12H25OH) are capable of forming gels. The latter are stable within a wide temperature interval at extremely low (from the viewpoint of gel formation) polymer concentrations (0.30 to 0.35 per cent by weight)4,7. In practice, the melting temperature of gels in alcohols does not depend on the length of the solvent molecule, being 28° for heptyl and dodecyl alcohol. In hydrocarbons there is a strong relationship of the Tm of gels with the number of carbon atoms in the solvent (13° for decane, 18° for dodecane and 26° for cetane). Crystallization of the solvent over a wide concentration interval does not

222


destroy the gel structure and the gel exists in the range between the melting point of the solvent and that of the gel. DTA analysis shows two characteristic endothermic peaks of melting which correspond to melting of the solvent and independently to that of the gel.

The gels of PA-16 in aliphatic alcohols are characterized by the high degree of order in the form of layer structures which can be detected by x-ray analysis. They have a sharp diffraction maximum d1 at small angles of diffraction and its value is a linear function of the length of the solvent molecule (Figure 3). Simultaneously, with the transition from heptyl to dodecyl alcohol the decrease of intensity and the increase of diffuseness of this reflex is observed showing a growth in defects in the gel structure of higher alcohols.

19 ^

17

13 7 9 11 13

n

Figure 3. Interplanar distance άγ (Â) in polyhexadecyl aery late gels as a function of the aliphatic alcohol molecule length.

X-ray analysis of gels in these hydrocarbons has drawn attention to the absence of the d± period. In both cases the x-ray diagram has a diffractional maximum corresponding to the interplanar distance of 4.5 to 4.6 Â. This maximum arises in gels independently from the number of carbon atoms in solvent molecules and it corresponds to the van der Waals interaction of méthylène side chains. The melting of an alcohol gel is followed by disappearance of the dl reflex while preserving the d2 reflex. The essential role in creating and stabilizing gel structure in aliphatic alcohols should be played by the hydrogen bonds between both solvent and polymer molecules because the addition of substances such as dimethylformamide results in the destruction of the gel.

The study of the temperature dependence of dielectric loss and dielectric permeability (Figure 4) for PA-16 in cetane has shown that in the interval from 20° to 30° there is a phase transition from gel to solution. The change of tan δ and ε' in the region of their maximum values at various frequencies is typical for the relaxation process of dielectric polarization and testifies to the high mobility of structural elements in the gel.

The ordered state of gels, small value of heat elimination during the gel-solution transition, sufficient mobility of structural units in the gel—all this is characteristic of the liquid-crystalline (mesomorphous) state of the system of comb-like polymer with solvent. This state is close to that of the lyotropic liquid crystals which are formed on dissolution of several biopolymers (e.g. poly-y-benzyl-L-glutamate) and low molecular substances in suitable solvents. Gels in alcohols being characterized by the layer structure can be

223

N. A. PLATE

0.05

0.04

0.03 to

-20.02

0.01

0 10 20 30 40 50 60 r,°c

2.3

2.2

2.1

10 20 30 40 50 60 r,°c

Figure 4. Temperature dependence of dielectric losses for PA-16 gels in cetane (1-3) and pure cetane (4) at the frequencies 800 (1); 5000 (2) and 104 (3) Hz.

considered as smectic liquid crystalline systems and gels in hydrocarbons without layer structure are close to the mesomorphic nematic type. The above-described ability of comb-like polymers to make association on intra- and inter-molecular levels shows that these polymers can be used as active structure-inducing agents. The specific structure of these branched polymers which combine the mobility of sidegroups with order in their arrangement facilitates consideration of this class of polymers as self-organizing systems towards their proper macromolecules as well as towards molecules of low-molecular solvents.

Another problem in the aspect under consideration is the relationship between physical and chemical methods of modifying polymer materials. One of the interesting questions is the introduction into the macromolecule of a particular chemical group which, being analogous to a known stabilizer, can behave as a stabilizer inside the macromolecule. In fact, it is the problem of macromolecular stabilizers or that of self-stabilized polymers. At the previous Bratislava Conference I reported on some possibilities of modification of halide-containing polymers with organotin compounds8. As a result

224

\

L T

I 2 · ^ ·ι

43,SdC2Z£ii3fc*^*os^ ^

J I I I 1 L


of such modifications one can successfully introduce the SnR3 sidegroups into the polymer :

~ C H 2 — C H 2 — C H - +R 3 SnLi = ^ ^ ~CH 2 —CH 2 —CH~

C\ SnR3 R = Ph, Bu

Such tin-containing Polyvinylchloride possesses increased stability toward dehydrochlorination, having stabilizing groups in its macromolecules. This was one of the first examples of self-stabilized polymers obtained by reactions in the chain without further mechanical admixture of stabilizers.

But the special features and complications of chemical transformations with macromolecules (configurational, conformational and supermolecular effects)9 make the synthesis of derivatives of given structure difficult. An easier way to introduce analogous organotin groups into PVC is the radical copolymerization of vinyl derivatives of trialkyltin with vinyl chloride10:

C H 2 = C H + C H 2 = C H - -CH 2 —CH—CH 2 —CH~ I I I I -Cl Sn(C2H5)3 Cl Sn(C2H5)3

The copolymer composition curve looks like that shown in Figure 5 and rx = 1.2 ±0.3 (vinyl chloride); r2 -> 0 (triethylvinyltin). The introduction of organotin monomer into the chain results in the decrease of dehydrochlorination to a constant value as well as causing modification of Polyvinylchloride with trialkyltinlithium8 (Figure 6).

-5 20h E ε io

& ^

2/ / 1

— o

20 40 M2,mole %

Figure 5. Copolymer composition curve of vinyl chloride and triethylvinyltin (w2-organotin monomer content in feed mixture; M2—organotin monomer unit content in the copolymer),

measured as Sn (1) and Cl (2) contents in the reaction product.

The study of the mechanism of hydrogen chloride uptake by PVC organotin groups brought out certain features by comparison with the earlier known mechanism of dehydrochlorination of triphenyl- and tributyl-tin derivatives11. The application of Mössbauer spectroscopy was very successful in elucidating the chemical scheme of the process. If for triphenyltin derivatives the two-step mechanism of phenyl group substitution by chlorine was established with subsequent rupture of the polymer-tin bond to form SnCl4, for tributyl derivatives the immediate formation of tributyltinchloride was observed, then with triethyl derivatives the first step

225 P.A.C.—30/1 - I

N. A. PLATE

5

I C E 3

if) -O

1h

90 100 m^mole %

Figure 6. Dehydrochlorination rate constant k of the vinyl chloride-triethylvinyltin co-polymers (mi—vinyl chloride content in the copolymer).

is substitution of one ethyl group on to chlorine followed by rupture of the polymer-tin bond with formation of diethyltinchloride. The ethyl group has an intermediate stability toward hydrogen chloride between those of the phenyl and butyl groups. All the reactions can be represented by Scheme I11 :

~ C H 2 — C H ~ ~CH 2 — C H ~ ~CH 2 — C H 2 ~ I -?aa. I ;ci ™L - 2- + snci4

Sn(C6H5)3 -2C6H6 C6H5—SnX - Q H 6

ei

~ C H 2 — C H ~ _ " £ ! CH 2 —CH 2 ~ +(C4H9)3SnCl

Sn(C4H9)3

~ C H 2 - Ç H ^ J i ç U j ^ C H 2 - Ç H ~ H C I - C H 2 - C H 2 ^

SniC.Hs)^2"6 C2H5—Sn—Cl 2 + (C2H5)2SnCl2

C2H5

Scheme 1.

The main role in thermostabilization of an organotin derivative of PVC at low contents of the Sn(Et)3 group is played by the reaction of the ethyl group substitution with formation of organic tin monochloride as a side group in the polymer.

The role of chemically attached microadditives becomes especially important if one is dealing with biologically active polymers. We know very well how dramatic the consequences of chemical distortion in the case of enzymatic action can be. V. A. Kabanov and Yu. E. Kirsch have recently shown that synthetic models of proteolytic enzymes reveal exclusive sensitivity to chemical micromodification of the chain.

It has been found that poly-4-vinyl pyridine partially alkylated with benzyl chloride provides high catalytic activity in the hydrolysis of ester bonds (in p-nitrophenyl acetate, for instance) which is three to five orders higher than the activity of its monomeric analogue 4-ethyl pyridine. It is known that the catalytically active part is a non-alkylated pyridine nucleus playing the role of an active nucleophilic agent :

226


-CH2-ÇH—CHj—ÇH-CH2-CH

ci-/7 Hfi^P a-'f

°>"-^y°~ Such highly catalytically active polymers are only those polyvinyl pyridines

which have a very specific and particular degree of substitution òr a particular content of non-alkylated pyridine rings (a per cent). It has been discovered that the catalytic efficiency values v (the fraction of active pyridine rings over the whole amount of non-alkylated pyridine nuclei) have a sharp extreme character. For each molecular weight there is a maximum of catalytic activity at a definite value of acrit (Figure 7). The interval of the sharp change

4 000 34 000

; «>5-5Qp250ooo

* 5 0

Figure 7. Catalytic efficiency v and solution viscosity of polyvinylpyridine catalyst as a function of the fraction of free pyridine nuclei (a).

227

N. A. PLATE

of activity around acrit is very small and corresponds to between two and three per cent of alkylation. It seems that at the critical value of a a sharp conformational change takes place which results in the formation of 'active holes' consisting of free pyridine nuclei and side alkylated groups. The substrate nitrophenyl acetate is extracted into these active holes by the forces of hydrophobic interaction and reacting with pyridine as with a nucleophile to form nitrophenol and iV-acetyl pyridinium. It has been shown that kinetic behaviour of these polymers is totally analogous to the behaviour of natural hydrolytic enzymes, e.g. a-chymotrypsin.

This example demonstrates that a small quantitative change in a macro-molecule (one to two per cent in degree of alkylation) leads to a sharp qualitative shift in the properties following the principle 'Yes-No').

The effects and reactions described above do not definitely fit all the possible pictures of the important role of micromodification of polymers. I am sure that such an approach will lead us to many other interesting results in both theory and practical application.

REFERENCES 1 G. P. Andrianova, N. F. Bakejev and P. V. Kozlov, Vysokomoi Soedin. 13A, 266 (1971). 2 P. V. Kozlov, V. G. Timofejeva and V. A. Kargin, Dokl. Akad. Nauk SSSR, 148, 886 (1963). 3 G. P. Andrianova and V. A. Kargin, Dokl. Akad. Nauk SSSR, 183, 587 (1968). 4 N. A. Plate and V. P. Shibaev, Vysokomoi. Soedin, 13A, 410 (1971). 5 N. A. Plate, V. P. Shibaev, B. S. Petrukhin and V. A. Kargin, J. Polym. Sci. C23, 37 (1968);

V. P. Shibaev, B. S. Petrukhin, Yu. A. Zubov, N. A. Plate and V. A. Kargin. Vysokomoi. Soedin. AIO, 216(1968).

6 V. P. Shibaev, B. S. Petrukhin, N. A. Plate and V. A. Kargin, Proceedings of the International Symposium on Macromolecular Chemistry, Toronto (1968).

7 V. P. Shibaev, R. V. Talrose, B. S. Petrukhin and N. A. Plate, Vysokomoi. Soedin. 13B, 4 (1971). 8 N. A. Plate, Europ. Polym. J. Suppl., pp 173-188 (1969). 9 N. A. Plate, Plenary lecture at the IUPAC International Symposium on Macromolecular

Chemistry, Budapest (1969), pp 651-675. Publishing House of the Hungarian Academy of Sciences (1971).

10 N. A. Plate, T. B. Zavarova, V. V. Maltzev, K. S. Minsker, G. T. Fedosejeva and V. A. Kargin, Vysokomoi. Soedin. 11A, 803 (1969).

I I A. Yu. Alexandrov, V. I. Goldansky, T. B. Zavarova, A. A. Korytko and N. A. Plate, Vysokomoi Soedin. 13B, 78 (1971).

12 Yu. E. Kirsch, S. K. Pluzhnov, T. S. Shomina, V. A. Kabanov and V. A. Kargin, Vysokomoi. Soedin.UA, 186(1970).

13 Yu. E. Kirsch, L. Ya. Bessmertnaya, V. P. Torchilin, I. M. Papissov and V. A. Kabanov, Dokl. Akad. Nauk SSSR, 191, 603 (1970).

228

GRAFTING AND BRANCHING OF POLYMERS

PAUL REMPP and EMILE FRANTA

Centre de Recherches sur les Macromolécules, CNRS, 6, rue Boussingault, 67—Strasbourg, France

ABSTRACT

This review summarizes the present state of investigations concerning synthesis of various types of model macromolecules. Emphasis was given to those methods which enable preparation of well characterized macromolecules, suited well enough for morphological and thermodynamic investigations in dilute solution, and to get information on the influence of heterocontact interactions and/or of enhanced segment density on the properties of the polymers considered. It can be seen that for such an aim, the anionic polymerization techniques which are versatile and well adapted to such problems, have yielded decisive progress. The main disadvantage of these methods is that the number of adequate monomers is rather limited. Other polymerization techniques do not involve active sites which retain their activity long enough

to be of great interest for such synthesis.

The properties of linear homopolymers have been investigated intensively for the past two years in many laboratories, in dilute solution as well as in bulk. The behaviour of such macromolecules is now more or less established. It is known, for example, that two parameters are sufficient to account for the dimensions of these molecules in dilute solution, one of these parameters characterizing the unperturbed gaussian distribution of segments around the centre of mass, the other the expansion which is due to the so-called polymer-solvent interactions.

But it has been shown that the theories which account satisfactorily for the behaviour of linear homopolymers fail to explain the experimental results on several other types of macromolecules, as block or graft copolymers and as branched homopolymers1. This is one major reason for the recent development of methods of synthesis of model macromolecules of various types :

(1) Block-copolymers which are linear in structure and where incompatible homopolymeric sequences are connected by chemical bonds. Three different interaction parameters have to be considered here : between two A segments, between two B segments, and between an A and a B segment ; the latter type is much less frequent in block-copolymers, owing to intramolecular phase separations which have been shown to occur2.

(2) Graft-copolymers. Homopolymeric grafts A are linked, at random, on to a homopolymeric backbone B of a different chemical nature. In this case both the effect of heterocontact interactions and the effect of the branched structure of the molecule have to be considered3.

229

PAUL REMPP AND EMILE FRANTA

(3) Branched-homopolymers. The high density of segments which characterizes branched structures influences the solution properties of these macro-molecules4. The well-known 'simple contact approximation' is therefore not valid here. Amongst branched homopolymers two types of macromolecules have been synthesized, characterized and studied : (a) Comb-polymers : these are graft-copolymers in which grafts and backbone are of the same chemical nature. The grafts are distributed at random along the backbone; (b) Star-shaped polymers ; in these p individual homopolymeric chains are connected by one of their chain ends on to a small central nucleus.

The purpose of this paper is to review briefly the chief methods of preparation of these various types of macromolecules ; emphasis will be given to those methods which lead to real 'model macromolecules'. i.e. to macro-molecular samples which are little polydisperse in mass, of satisfactory homogeneity in composition, and in which the molecular structure is defined unambiguously by the conditions of preparation.

Many attempts were made in the past fifteen years to synthesize block-copolymers, graft-copolymers, and branched-homopolymers ; several methods, however, yielded products which could not be characterized adequately. Let us first consider the preparation of block-copolymers and of graft-copolymers and, in a second part, the methods of synthesis of branched homopolymer molecules.

I. BLOCK- AND GRAFT-COPOLYMERS (A) Methods involving radical polymerization of the monomers

Obviously, polymerization of a mixture of two monomers cannot yield block- or graft-copolymers, whatever the reactivity ratios of the system may be.

It should be recalled that the lifetime of a radical site in a liquid is very short, and attempts to initiate polymerization of monomer A in a tube while flowing into a vessel filled with another monomer B yielded only very small amounts of block-copolymer5.

Many attempts to synthesize block- or graft-copolymers proceed in two steps. The first step involves preparation of homopolymer sequences fitted with active sites. The second step involves polymerization of the second monomer, using the active sites as initiators. If the sites are located at chain ends a block-copolymer will be obtained. If they are distributed at random along the chain, the process will yield graft-copolymers. In both cases. homopolymers will be present in the reaction medium, and have to be separated by careful fractionation.

Numerous methods of this type have been developed and they have to be mentioned, even if none of them has ever yielded what we call model macromolecules, as defined above.

Many different ways have been used to create active radical sites on a polymer backbone, either randomly distributed or selectively located at chain ends. Let us briefly recall some of them :

(1) Some authors6 have used bifunctional initiators, such as di-isopropyl-230


benzene dihydroperoxide, to initiate the polymerization of a monomer (styrene). The polymer is thus fitted with hydroperoxide end groups which may then participate in redox initiation for the polymerization of a second monomer (methylmethacrylate).

(2) Peroxidation of polystyrene results in chain scissions and in formation of terminal hydroperoxide groups, able to initiate the polymerization of methylmethacrylate and of acrylonitrile7.

(3) Various chemical methods have been described8. One can for instance start with a random copolymer of methylacrylate and acrylylchloride. and react the latter units of the copolymer with r-butylhydroperoxide. The perester functions obtained may initiate various polymerizations, such as that of styrene.

(4) Much work has been performed on grafting processes involving irradiation of a polymer, to create active sites, and using the latter to initiate the polymerization of a second monomer.

Different types of experimental procedures have been u sed 9 1 3 : (a) One can irradiate a polymeric film swollen in a monomer. Grafting

takes place, but much homopolymer is formed, too; furthermore gel effect is involved in the process, owing to an extended lifetime of the radical sites. Therefore this grafting process, though very efficient, does not yield very homogeneous samples.

(b) The pre-irradiation method is also used, especially with crystalline polymers. Irradiation of such polymers yields trapped radicals, and these sites act as initiators on contact with another monomer. The grafting process lasts only for a short time, until all radicals have disappeared. The same technique can also be carried out in the presence of oxygen, yielding thus peroxy radicals which are more stable but can also act as initiators for the polymerization of a second monomer.

(5) To be exhaustive, it should be mentioned that in some cases grafting may result from a transfer reaction. Several monomers, when polymerized using a radical initiator, in the presence of a polymer, yield transfer reactions between the growing chains and labile C—H bonds of the polymer. This process yields radical sites on the polymer, which can either contribute in turn to the initiation of the second monomer, or react with the growing radicals by recombination14-17.

All these methods involving radical processes have been used to produce graft copolymers, but in none of these cases has it been possible to characterize precisely the obtained species: from analysis and spectrographic methods the content of each of the constituents can be established, but neither the number of branch points, nor the length of the backbone, nor the size of the individual grafts can be determined; besides, the samples are quite polydisperse in mass and in composition.

(B) Methods involving ionic polymerization of the monomer Decisive progress was made when the anionic polymerization techniques

in aprotic media, involving no termination reactions, had been developed, especially by Szwarc and his co-workers18. It is well known now that metal-organic derivatives such as butyl-lithium, cumyl-potassium, fluorenyl-sodium do initiate the polymerization of various monomers having an electro-

231


attractive substituent. The polymerization goes to completion and the polymers thus obtained are called "living" because they retain reactive carbon-metal bonds at chain ends. Further addition of monomer results in an increase of the length of the chains. It has been established that the obtained polymers exhibit a narrow molecular weight distribution, if they have been synthesized under adequate conditions.

We shall distinguish here between methods meant to prepare block copolymers and ways of preparation of graft copolymers. Obviously these methods are restricted to anionically polymerizable monomers.

(1) Block-copolymers—To synthesize block-copolymers a monomer B is added to a solution of living polymer A. The carbanionic sites initiate the polymerization of monomer B and, since initiation is an addition process, a two-block-copolymer is formed, provided the electroaffinity of monomer B is at least that of monomer A1 9 , 2 0 . Many systems were tested in recent years, and a whole series of different block-copolymers were obtained by this method, as can be seen from a recent review20. Vinyl, diene and acrylic monomers are involved, but also heterocyclic ones, such as ethylene oxide, butyrolac-tone, thiocyclanes21, or even cyclic dimethylsiloxane-trimer or tetramer.

Similarly it is possible to synthesize BAB triblock-copolymers by using a bifunctional initiator (such as naphthalene sodium, in which case the initiation is an electron transfer process to the monomer, followed by dimerization18) to initiate polymerization of monomer A. After completion of the polymerization, the obtained bifunctional living polymer A is reacted with monomer B, thus yielding a triblock-copolymer.

In both cases, adequate characterization of the block-copolymer is possible, and it was confirmed that if the experimental conditions are well chosen, the homogeneity in composition of the sample is satisfactory and the corresponding molecular weight distribution is narrow22.

Several attempts were made to carry out similar syntheses using cationic initiators and adequate monomers. But since transfer reactions often occur in cationic polymerization processes, most of these attempts were unsuccessful: chain ends do not retain active sites23. However, a very nice method of preparation of styrene-tetrahydrofuran block-copolymers should be mentioned here : cross-termination between living' carbanionic polystyrene and a living' cationic polytetrahydrofuran results in block-copolymer formation with rather good yields24.

Another method was recently described: living anionic polystyrene is reacted with a small amount of p-divinylbenzene and quickly deactivated with a proton donor : the molecular weight is unchanged but some double bonds remain pendant at the end of the polystyrene chain. The species obtained can then be copolymerized through these double bonds by any method applicable to styrene : free radical, cationic or anionic polymerization. The best results so far have been obtained anionically45.

Another possible way to synthesize block-copolymers is to start with homopolymers A and B, both fitted with reactive end groups, and to have them reacted selectively on each other. A recent example for this type of process is the reaction of polyoxyethylene glycol with large excess of toluene diisocyanate, yielding no increase of the molecular weight. In a second step, the remaining isocyanate functions, at chain end, are reacted with the terminal

232


OH functions of a polyester whose preparation involved a slight excess of the glycol25.

A similar method was used to prepare BAB triblock-copolymers of styrene and ethylene oxide, the PEO block being the central A block. Living polystyrene is reacted with an excess of phosgene to yield acid chloride end groups. The latter are then reacted on to the terminal OH functions of a polyoxyethylene glycol chain26. One can also start from a living AB block-copolymer and react it with a difunctional deactivator to yield ABA triblock-copolymers27.

(2) Graft-copolymers—To synthesize graft-copolymers using anionic polymerization techniques two different paths can be used, namely grafting via carbanionic initiation, and grafting via carbanionic deactivation20.

(a) Grafting via carbanionic initiation A polymer chain is fitted with organometallic sites distributed at random,

and these sites are used to initiate the polymerization of an adequate monomer. Several examples of such reactions have been described, leading to graft-copolymers28,29. However, the number of grafts is usually lower than the number of sites ; their length is not experimentally accessible and may fluctuate quite a bit within the sample (cf. réf. 20, review article).

The main difficulty of grafting reactions carried out via anionic initiation is of course to find ways to get organometallic sites on a polymer chain. Several methods have been used to perform this :

(1) Direct metallation of polymers containing vinyl-aromatic monomer units: vinyl-fluorene, vinyl-biphenyl, vinyl-naphthalene. In the two latter cases, however, the initiation process for vinyl monomers involves electron transfer, instead of addition30. Therefore only ethylene oxide can be grafted here ; with vinyl monomers homopolymers are formed, solely.

(ii) Polyvinylfluorene and poly-3,3-diphenylpropene can be metallated by naphthalene-sodium28. The organometallic sites formed may initiate the polymerization of various monomers, yielding graft-copolymers.

(iii) Exchange reactions between halogen and metal can take place on to poly-p-bromostyrene, yielding p-lithiophenyl sites, which are rather active as initiators. Butyl-lithium31 or naphthalene-lithium32 have been used as metallating agents.

(iv) Though polystyrene itself is rather hard to metallate this was achieved recently by using a butyl-lithium complex with iV-tetramethylethylene diamine33. The same reagent was also used successfully to yield anionic sites on a poly-(2,6-dimethyl-l,4-phenylene oxide) chain34. Metallation takes place both on the benzene rings and on the methyl side groups, but the latter isomerize yielding the former species.

(v) Another way of getting metallated sites on a polymeric chain is addition of butyl-lithium on to some pyridine rings of a polyvinylpyridine chain35, or by reaction of butyl-lithium on to vinyl-naphthalene units36. In both cases graft-copolymers could be obtained.

(b) Grafting via carbanionic deactivation The second path for anionic grafting proceeds by carbanionic deactivation.

It is well known that carbanionic sites do react with various electrophilic 233

PAUL REMP AND EMILE FRANTA

functions such as halogenides, acid chlorides, esters, nitriles. and anhydrides18,20. If a living polymer A is reacted with electrophilic functions linked to a polymer chain B, graft-copolymers are obtained: the grafts which are attached to the main chain are randomly distributed, and they are, to a first approximation, of the same length. Though on reacting living polystyrene on to polymethylmethacrylate37, only a small proportion of the ester functions do react—because of problems of accessibility—the graft-copolymers obtained are entirely characterizable. Side reactions sometimes occur, leading to ungrafted polymer A, but separation of this polymer is usually rather easy. Polymethylmethacrylate37, Polyvinylchloride37, permethylated xylanes38 and even polyvinylpyridine39 served as backbone, i.e. as polyfunctional deactivators. Polystyrene, polyisoprene, polyvinylpyridine, were used as grafts, but not polymethylmethacrylate, because of insufficient reactivity.

This method of anionic grafting is restricted to a small number of cases, but it is one of the preferred methods to synthesize graft-copolymers for morphological investigations, since characterization of the structure of the molecule is easy : the length of the backbone, the number and the average length of the grafts are experimentally accessible; furthermore the random distribution of the grafts and the low level of fluctuations in composition within a sample have been established.

It should be mentioned that one example of cationic deactivation grafting process was published recently40: grafting of cationic polystyrene on to poly-2,6-dimethoxystyrene, the cationic polymerization of styrene being carried out in the presence of the latter polymer. Yields are moderate, but characterization of the obtained graft-copolymers is not very easy.

II. BRANCHED HOMOPOLYMERS

It is well known that some industrial polymers of major importance, such as polyethylene, polyvinylacetate and acrylic polymers exhibit a certain number of branch points in their macromolecules. Branching affects the solution properties of the polymers as well as their mechanical behaviour. But to investigate systematically and efficiently the influence of branching on behaviour, model macromolecules must be used. Various methods have been developed to build up macromolecules of precisely known structure.

The first branched macromolecules to be investigated were polycondens-ates. If in a polyesterification of an acid-alcohol a small proportion of an acid-dialcohol (or of an alcohol-diacid) is added, the copolycondensation of the latter monomer results in the formation of branch points, which are more or less randomly distributed. These polycondensations can never yield networks (in contrast with reaction of a glycol with a diacid containing some triacid). But the polydispersity of the samples is high, and the average number of effective branch points per molecule is not known precisely. These randomly branched macromolecules, though of great interest, cannot be considered as model polymers.

More recently major interest has been devoted to branched molecules of rather simple molecular structure. Two types of branched model macro-

234


molecules have been investigated in great detail in several laboratories around the world : comb-like polymers and star-shaped polymers. The methods of preparation of these two model structures will be described below :

(A) Comb-like polymers These are in fact graft-polymers in which grafts and backbone are of the

same chemical nature. It is quite understandable therefore that the methods of anionic grafting should have been used to synthesize such model molecules. Since the methods using anionic initiation from a metallated backbone do not lead to well characterizable polymers, we shall not go into the details of such methods, which have been applied nevertheless to polystyrene and to polyvinylpyridine comb-polymers, but we shall describe two methods of grafting via carbanionic deactivation.

(1) The first method41 uses as a backbone a random copolymer of sty rene and methylmethacrylate, the content of the latter monomer being of the order of ten per cent. Such copolymers are obtained using radical initiators, and care is taken to stop the reaction at rather low yields, in order to avoid fluctuations in composition within the samples obtained. Their poly-dispersity is, however, rather high. The ester functions of the backbone can be used in a second step as electrophilic deactivators for an anionically prepared 'living' polystyrene. This second step leads thus to a comb-like polystyrene, the polystyrene grafts being randomly distributed along the polystyrene backbone. A small proportion of ester functions remains untouched, but this is not important. A small proportion of ungrafted polystyrene has to be removed by fractionation, which is relatively easy, since grafts are usually not too long. But this method has two disadvantages : (i) the grafts are attached to the main chain by means of a carbonyl group which is quite sensitive to photochemical oxidation. As a matter of fact, prolonged exposure to bright light will yield chain scissions and free grafts ; (ii) The molecular weight distribution of the samples thus obtained is as broad as that of the backbone itself, because the number of grafts is proportional to the length of the main chain. Therefore fractionation of the sample has to be undertaken with great care, prior to investigation of any physical property of the comb-polymers.

(2) The second method42,43 is more rigorous. An anionically prepared polystyrene sample—of sharp molecular weight distribution—is chloro-methylated cationically under experimental conditions chosen so as to avoid any coupling or degradation process44. The degree of chloromethylation is limited to values of the order of five to eight per cent. The chloromethylated groups are selectively located in para position to the phenyl groups. These functions are used in a second step as electrophilic functions to deactivate 'living' monocarbanionic polystyrene. This deactivation is unambiguous, because no elimination reaction is possible when benzylic chlorides are used. Some authors43 have claimed that exchange reactions between metal and halogen may take place, but later results have shown that these side reactions can be neglected if adequate experimental conditions have been chosen42. The grafting reaction is not far from being quantitative, and the comb-polymers thus obtained can be considered to be real model macromolecules. The polydispersity is very low, the grafts are identical, to a first approxima-

235


tion, and randomly distributed on a backbone which can be characterized by itself.

(B) Star-shaped polymers These are characterized by one single p-functional branch-point connecting

p chains of identical length; they have been synthesized in various laboratories, two main experimental techniques being used : (1) The first method uses polyfunctional electrophilic derivatives to react with a monocarbanionic polymer. The deactivators used are various poly-halogen derivatives such as SiCl4

27, tri- or tetrachloromethylbenzenes46,47. the trimer of phosphonitrilic chloride46; more recently triallyloxytriazine48

gave good results. A slight excess of living polymer, with respect to stoichio-metry is introduced ; it has to be separated carefully afterwards by fractiona-tion. The main disadvantage of this method is that it may only lead to star-molecules with three, four or at most six branches. Furthermore, the reaction is not always quantitative; with tetrachloromethylbenzene49 it yields approximately equal amounts of stars with three and four branches, mixed with some of the initial polystyrene. Quantitative separation of these constituents is a hard job. (2) The other method which was developed in recent years proceeds by anionic block-copolymerization of a vinyl monomer and a di vinyl monomer. the latter being used in small amounts50. If the polymerization of styrene is carried out with an effective monofunctional initiator, it yields a living polymer, with active sites located at the chain end. Addition of a small amount of divinylbenzene (say three moles per living end) results in formation of small polydivinylbenzene nodules, each of them being connected, surrounded and protected by the p-polystyrene chains which participated in initiation of the DVB of the central nodule. As a matter of fact, much greater amounts of DVB can be used without gelation of the reaction medium, so effective is the protection of the nodule by the surrounding chains. But for model molecules it is better to use very small proportions of DVB, to keep the central nodule as small as possible. It was found that the value of p is determined by the concentration of the reaction medium. and, to a lesser extent, by the amount of DVB added. Star-molecules with six to twenty branches could be obtained thus.

The polydispersity of the samples obtained is merely the problem of the fluctuations of p within a sample, the branches being to a first approximation identical. To study the distribution of numbers of branches within a sample, careful fractionation had to be undertaken. The standard precipitation fractionation method is of no interest for that purpose, since differences in solubilities between molecules differing only by the number of branches are very small. Elution chromatography using a column fitted with cyclic variation of temperature51 was sensitive enough to yield satisfactory fractionation. It was found that the fluctuations of p around its average value are rather small, provided the experimental conditions are adequate. It is thus established that this method of preparation of star-shaped macro-molecules yields well characterizable model macromolecules, and that the homogeneity of the samples is satisfactory. Because of their simple structure and of their high degree of symmetry, these star-shaped molecules are of

236


particular interest for investigations on the effect of enhanced segment density on the morphological and thermodynamic properties of branched macromolecules52.

It should be mentioned briefly here that the same preparation technique, carried out with an efficient bifunctional initiator, yields model networks53

in which the average distance between two successive branch points is more or less constant within a sample and can be known in advance, by a proper choice of the styrene-to-initiator ratio, i.e. the molecular weight of the di-carbanionic polystyrene precursor. If the experimental conditions are satisfactory, the gels obtained are homogeneous. From their swelling behaviour54

it was established that these model gels are very close to 'ideal' gels, in which all chain ends are linked to a branch point.

REFERENCES 1 H. Benoit. D. Decker, A. Dondos and P. Rempp, J. Polym. Sci. C. 30. 27 (1970). 2 E. Franta, A. Skoulios, P. Rempp and H. Benoit, Makromol. Chem. 87. 271 (1965) 3 A. Dondos. P, Rempp and H. Benoit, J. Polym. Sci. B-4. 293 (1966). 4 D. Decker, Makromol. Chem. 125, 136 (1969). 5 J. A. Hicks and H. W. Melville, J. Polym. Sci. 12. 461 (1954). 6 G. M. Burnett. P. Meares and C. Paton, Nature. Lond. 58, 723 (1962). 7 G. Riess and A. Banderet. Bull. Soc. Chim. Fr. 733 (1959). 8 G. Smets and M. Van Beylen, Makromol. Chem. 69, 140 (1963). 9 A. Chapiro and A. Matsumoto. J. Polym. Sci. 57. 743 (1962).

10 A. Chapiro, J. Polym. Sci. 48. 109 (1960). 11 A. Chapiro and V. Stannett, Internat. J. Appi. Rad. Isotopes. 8. 164 (1960). 12 A. Chapiro et al. Europ Polym. J. 4, 627 (1968). 13 A. Chapiro. Radiation Chemistry of Polymeric Systems. Interscience: New York (1962). 14 K. H. Kahrs and J. W. Zimmermann. Makromol. Chem. 58, 75 Π962). 15 G. Smets and Claesen, J. Polym. Sci. 8. 289 (1952). 16 W. Kobryner and A. Banderet. J. Polym. Sci. 34. 381 (1959). 17 P. W. Allen, G. Ayrey and C. G. Moore, J. Polym. Sci. 36, 55 (1959). 18 M. Szwarc, Carbanions. living polymers and electron transfer processes. Interscience: New

York (1968) 19 E. Franta and P. Rempp. C.R. Acad. Sci. Paris. 254, 674 (1962). 20 M. Morton and L. J. Fetters. Macromol. Rev. 2. 71 (1967).

L. J. Fetters, J. Polym. Sci. C. 26, 1 (1969). 21 R. Kammereck, Akron Summit Polymer Conference (1970).

S. Boileau and P. Sigwalt. C.R. Acad. Sci. Paris. 252. 882 (1961). 22 D. Freyss. P. Rempp and H. Benoit, Polymer Letters. 2. 217 (1964). 23 P. H. Plesch, The Chemistry of Cat ionic Polymerization. Pergamon : London (1963). 24 G. Berger. M. Levy and D. Vofsi, J. Polym. Sci. B-4. 183 (1966). 25 M. Galin and J. C. Galin. Makromol. Chem. to be published. 26 G. Finaz, P. Rempp and J. Parrod, Bull. Soc. Chim. Fr. 262 (1962). 27 M. Morton, T. E. Helminiak, S. D. Gadkary and F. Bueche. J. Polym. Sci. 57. 471 (1962).

J. Heller et al. J. Polym. Sci. A-\. 7. 73 (1969). 28 G. Greber. Makromol. Chem. 101, 104(1967). 29 A. Dondos. Bull. Soc. Chim. Fr. 2762 (1963).

G. Greber. J. Tolle and W. Burchard. Makromol. Chem. 71. 47 (1964). 30 J. E. Herz. D. J. Worsfold and P. Rempp, Europ. Polym. J.. Supplì. 453 (1969).

J. Gole. G. Gouttière and P. Rempp, C.R. Acad. Sci. Paris. 254, 3867 (1962). E. Cuddily, J. Moacanin and A. Rembaum. J. Appi. Polym. Sci. 9, 1385 (1965).

31 D. Braun. Angew. Chem. 73, 197 (1961). 32 A. Dondos and P. Rempp. C.R. Acad. Sci. Paris. 258. 4045 (1964). 33 Y. Minoura and H. Harada. J. Polym. Sci. A-l, 7. 3 (1969). 34 A. J. Chalk and A. S. Hay. J. Polym. Sci. A-l, 7, 691 (1969): 7. 1359 (1969).

237

PAUL REMPP AND EMILE FRANTA 35 A. Dondos and P. Rempp. C.R. Acad. Sci. Paris, 254, 869 (1967). 36 J. Heller, Akron Summit Polymer Conference (1970). 37 Y. Gallot, P. Rempp and J. Parrod. Polymer Letters, 1, 329 (1963). 38 J. J. O'Malley and R. H. Marchessault. J. Phys. Chem. 70. 32.55 (1966). 39 A. Dondos and P. Rempp, C.R. Acad. Sci. Paris, C-264, 869 (1967). 40 C. G. Overberger and C. M. Burus. J. Polym. Sci. A l . 7. 333 (1969). 41 P. Rempp and D. Decker-Freyss. J. Polym. Sci. C-16. 4027 (1968). 42 F. Candau and E. Franta, Europ. Polym. J. Makromol. Chem. 149, 41 (1971). 4 3 T. A. Altares, D. P. Wyman and V. R. Allen, J. Polym. Sci. A-2. 4533 (1964). 4 4 F. Candau and P. Rempp, Makromol. Chem. Ill, 15 (1969). 4 5 P. E. Black and D. J. Worsfold, J. Appi. Polym. Sci. 14, 1671 (1970). 46 J. A. Gervasi, D. P. Wyman, V. R. Allen and K. Meyersen, J. Polym. Sci. A-3. 4131 (1965). 47 J. Herz and C. Strazielle. C.R. Acad. ScL Paris, C-272, 747 (1971). 4 8 J. C. Meunier and R. Van Leemput, Makromol. Chem. 142, 1 (1971). 4 9 D. J. Worsfold, J. G. Zilliox and P. Rempp, Canad. J. Chem. 47. 3379 (1969). 50 A. Kohler, J. Polacek. I. Koessler, J. G. Zilliox and P. Rempp, Europ. Polym. J. to be published. 51 J. G. Zilliox, Makromol. Chem. to be published. 52 P. Weiss. G. Hild, J. Herz and P. Rempp, Makromol. Chem. 135. 249 (1970). 53 P. Weiss, J. Herz and P. Rempp, Makromol. Chem. 141. 145 (1971). 54 M. Morton, R. F. Kammereck and L. J. Fetters, Macromolecules, 4, 11 (1971).

238

THE COMPARISON OF ANALOGOUS REACTIONS OF MACROMOLECULES WITH LOW-MOLECULAR

MODELS

ROLF C. SCHULZ

Institute of Macromolecular Chemistry, Technische Hochschule Darmstadt, West Germany

ABSTRACT

Most experiments for the chemical transformation of polymers start from the assumption that the reactivity of a functional group in a macromolecule has to be the same as one in a low molecular compound. But it turns out that usually the reaction on the polymer is faster or slower than the reaction of the model compound, or the reaction with the polymer proceeds in a different way and the reaction products differ from those of the model compound. These phenomena are called 'polymeric effects'. It depends on the property to be compared as to which model reflects the polymer in the best way. These general considerations are exemplified by three reactions.

(a) The optical and chiroptical properties of polymers with atropisomeric groups in the pendant sidegroups and in the backbone are the same as for three low-molecular model compounds. But the racemization rates of the polymers are much slower than those of the models.

(b) Electron-donor-acceptor complexes between polyesters with fluorene rings and TCNQ or TCNE were prepared and compared with the corresponding models. The value of the equilibrium constant {K) is only reached by models with at least two donor groups per molecule. In the oligomer range, the constant K depends upon the degree of polymerization. The presence of neighbouring donor groups seems to be the cause of the greater stability of the polymer complex.

(c) Preparation and properties of JV-chlorinated nylon 66 is described. The i.r. spectra are the same as those of the model compounds but the reactivity of the chlorine is usually different. JV-chloro-nylon oxidizes secondary alcohols faster and often to a higher yield than the models. Primary alcohols are transformed to the esters. The advantages of a polymeric oxidation reagent are discussed.

At higher temperature or upon irradiation, a rearrangement takes place and the chlorine moves from the N-atom to a C-atom in the model as well as in the N-chloro-nylon. This reaction can be followed by i.r. spectra or by the decrease

in oxidation ability.

GENERAL CONSIDERATIONS

Chemical reactions with synthetic or biological polymers play a very important role in technology as well as in science and medicine. For more than fifty years they have therefore been the subject of a great deal of research

239

ROLF C. SCHULZ

and there is an enormous amount of patents and publications on them1. The aims and methods of these investigations differ widely. Usually the macro-molecules are chemically changed either to modify their properties or to analyse the primary and secondary structure. Sometimes the polymer is used as a support, matrix or template for other reactions2, i.e. the polymer becomes part of a cycle of reactions but returns to its original chemical state after the reactions. Furthermore there are many experiments, in which macromolecular substances with specific functional groups behave as catalysts3, inhibitors, photosensitizers4 etc. It should be mentioned that some of these experiments can help to elucidate the mechanism of enzyme reactions.

In all these experiments we have to start from the assumption that the reactivity of a functional group in a macromolecule has to be the same as one in a low-molecular compound. But we often learn that this assumption is not right, i.e. the reaction on the polymer turns out to be different from an analogous reaction with a low-molecular model compound. As this differing behaviour is a consequence of the polymeric structure of the reaction partner this phenomenon is called 'polymeric effects'. We talk of polymeric effects if the reaction of the polymer is faster or slower than the reaction of the model compound, or if the reaction with the polymer proceeds by a different route and the reaction products differ from those of the model compound. Sometimes the polymeric effects depend on the molecular weight of the polymers.

There are different reasons for polymeric effects, for instance the inaccessibility of the functional groups, secondary structure, influence of neighbouring groups, charge distribution, solvation etc. It depends on the nature of the macromolecule and the type of reaction as to which influence predominates and whether there are polymeric effects. Often a definite explanation for polymeric effects has been impossible up to now. Certainly it is very important which model compounds for the polymers are chosen for a comparison and which characteristics are examined. For instance, we can compare physical (u.V., i.r., n.m.r.) or chemical properties (dissociation constants, rate constants, yields etc.) reaction mechanisms or stereochemistry. But sometimes it is difficult to find a low-molecular compound which is comparable with the polymer and could therefore serve as a model. It should be emphasized that model compounds mostly imitate only one property. That is why we should have to use various model compounds if we want to compare several characteristics of the polymer. If we, however, compare a model compound with several features of the polymer we often find a correspondence with only a few qualities ; we see great differences, however, as far as other qualities are concerned. So we find here, too, that every comparison would be defective.

In order to compare the chemical reactivity of a polymer one usually takes a model compound which corresponds with a base unit of the macromolecule. If, however, we want to study chemical or physical interactions of adjacent functional groups it is necessary that the model compounds shall have at least the structure of dimers or trimers. But for stereochemical comparisons or as models of secondary structures we generally need linear or cyclic oligomers.

Many examples in the literature demonstrate these general considerations and conclusions. I prefer, however, to cite our own research comparisons

240

THE COMPARISON OF ANALOGOUS REACTIONS OF MACROMOLECULES

between model compounds and polymer reactions. I shall discuss the rate constants of an isomerization reaction, the equilibrium constants of EDA-complexes and the chemical reactivity of N-chlorocarboxamides.

1. THE RACEMIZATION OF POLYMERS WITH ATROPISOMERIC DIPHENYL GROUPS5

(Together with R. H. JUNG)

There are some organic compounds which are chiral because of hindered rotation around a C—C single bond and can therefore be separated into antipodes6. These stereo isomers are called 'atropisomers'. Some of them are very stable; in other cases the enantiomers racemize when heated up. This reaction proceeds by first-order kinetics and can be clearly observed in the change of optical activity. The following equation applies :

and therefore : dc/dt = — da/di = kTac x a

- In ao/a, = fcrac x t

(1)

(2) The rate of the racemization depends on the chemical structure, and the substituents as well as the solvents.

In our experiments on optically active polymers we also produced some which contain atropisomeric diphenyl derivatives either as pendant groups or in their backbone5. We examined the absorption spectra, the circular dichroism and the racemization of these polymers7'8. We then compared these properties with those of the corresponding model compounds. The preparation of a polymer with pendant atropisomeric groups is shown in Scheme 1 (equation 3). 2-Methyl-6-nitrobiphenyl-2'-carboxylic acid (I) was separated with the aid of quinine (according to Bell9 into the antipodes, and the dextrorotatory form ([a]D = + 67.6 in dioxane) was transferred into the acid chloride (II) ([<x]D = + 183).

HOOC CH3 A H I L A J

NO,

(ID

L: 0 = C

NO,

CH3—CH2 I

o I R

(IV)

o I R

(V)

o I R

O I R

(VI)

Scheme 1

241

ROLF C. SCHULZ

By the reaction of II with polyvinyl alcohol we obtained the polymer HI. The esters of 2-methyl-6-nitrobiphenyl-2'-carboxylic acid with ethanol (IV), isopropanol (V) and butanediol (VI)10 were prepared as low-molecular model compounds (see Schemel). The absorption spectra of these compounds have no characteristic maxima, but are very similar to one another. The molar rotations of the model compounds and the polymer are shown in Table 1.

Table 1. Molar rotation [M]* of the polyvinylester of 2-methyl-6-nitrobiphenyl-2'-carboxylic acid (III) and the model compounds (IV). (V) and (VI).

λ(ητη)

III IV V

VI

589

139.7 144.1 112.5 347.4

578

149.6 154.9 120.6 372.4

546

185.0 192.0 149.7 462.8

436

644.5 671.3 509.4

1631.8

405

1346 1406 1062 3485

A0 x 10~7

0.85 0.95 0.77 1.2

Λ0 (nm)

385 + 5 385 + 5 382 + 5 285 ± 5

* In dioxine at 22°C: concentration c = lg/100 ml : for III calculated on base units.

Between 589 and 436 πιμ a single-term Drude equation is valid. The constants determined by Yang-Doty plots can be found in the last column of Table 1.

[α] = Α0/(λ2 - λ20) (4)

The similarities between the polymers and the model compounds are particularly significant in the ORD- and CD-curves (Figure i, Figure 2 and Table 2\

With regard to these properties the different low-molecular model compounds coincide very well with the polymer. Experiments on the kinetics of the racemization, however, lead to a completely different result. If we plot the relative change of the rotation on a logarithmic scale as a function of time we get straight lines for all the compounds examined here. This means that the polymers as well as the model compounds racemize according to a first-order reaction. The rate constant fcrac and the half-life time for the model compounds are the same within the limit of the error; in contrast to this racemization of the polymer takes place with only half the rate (Figure 3 and Table 3). A chain degradation or a split-off of sidegroups did not take place during the thermal treatment which was tested by viscosity and N-analyses. A possible influence of the viscosity of the polymer solution could also be excluded by comparative experiments11. We therefore conclude that this low rate of racemization is to be explained by the fact that the whole coiled macromolecule and the adjacent substituents cause an additional hindrance to internal rotation around the biphenyl single bond.

This polymeric effect should become even more evident if the rotation-hindered biphenyl bond does not belong to the sidegroup but is part of the main chain. We therefore prepared some polyamides, polyurethanes and model compounds based on ( + )-2,2,-diaminobinaphthyl-(l,l·) (VII) 2,2',3,3'-tetramethylbenzidin (VIII) and 2,2-'dimethoxydiphenic acid (IX)10

(equations 5, 6 and 7). The dextro-rotatory form of IX was obtained with 70 per cent optical

purity and the laevo-rotatory with 98 per cent optical purity. For the racemization experiments we only used the polyamide (X) and the corresponding

242


(5)

(VII)

CH,

Η , Ν ^

H3C CH3

NH2 (VIII)

< C H 2 ) 8 -

(VIII a) (Vlllb)

H3CO OCH3

(IX)

HOOC COOH

/ \ H.CO OCH, H3CO OCH 3

IX) (X!)

243

(6)

P)

ROLF C. SCHULZ

^9000

► 7000

^5000

+ 3000

H 1000

-1000

-3000

Figure 1. ORD of esters of 2-methyl-6-nitrobiphenyl-2'-carboxylic acid with polyvinylalcohol (III) and low molecular model compounds (in dioxane).

244


+ Δε

1: CH3-CH-CH3 V

2: CH2-(CH2)2-CH2 VI OR OR

- C H 2 - C H - III 2 I OR

240| \260 280 A 3 0 0 _ / r 3 2 0 340 360 380 400 λ,ηπη

15.0

Figure 2. CD of esters of 2-methyl-6-nitro-biphenyl-2'-carboxylic acid with polyvinylalcohol (III) and low molecular model compounds (in dioxane).

245

ROLF C. SCH

ULZ

o +· o m

o

NO «r> O

<N

ϊ^ ^

dr

nd

n I

I I

I

<N <N

<N <N

m r-

^ q

oo

oq

T-Î T-Î ö en I

I I

I

o NO r- NO

^

o ©

©

w o

f*i w

Ή

Ή

rH

ff) + + + +

m m

«o NO

Ci

CO C"> C*")

a£

>>

O X

"e1

o X

<N CN

OO Γ>-

+ + + +

O O

r- oo M

M (N

M

O O

^ « I

I I

ON

O

N CN O

N

r- r- oo r-(N

(N (S (N

O O

O O

+ + + I

NO NO

ON

Tt" O

N

ON t—

O

N

CN

ÎN

<N

rsi

O

O

—« <N

I

I I

I

m o

»-« ^t CN

<N (N

<N

m m

rn m

ON ON

^ OO

T}· »Λ» NO

OO

Ö Ö

Ö Ö

+ + + +

O

O

OO

O

OO 00 f-

OO

ro t*">

ΓΛ

co

246


1: CH3-ÇH-CH3

0.9 h

Figure 3. First-order plots of racemization of a polymer with atropisomeric biphenyl groups (III) and corresponding model compounds.

Table 3. Racemization of polyvinylester of 2-methyl-6-nitrobiphenyl-2-carboxylic acid (III) and the model compounds (IV), (V) and (VI); temp.: 120°C; solvent: dioxane; R : -C—O CH3 concentration: 0,1 % , 1 J v

NO,

-CH2—CH— CH3—CH2 CH3—CH—CH3 CH2—CH2—CH2—CH2

OR OR OR (HI)

OR (IV)

OR (V) (VI)

kTac x 106

(sec-1) Mh)

4.0 48.1

8.3 23.2

7.9 24.5

7.7 25.0

247

ROLF C. SCHULZ

250 \/270 290 310 330 350 370

2.0 co

4.0

46.0 Figure 4. Some u.V.- and CD-spectra of a polyamide based on 2,2'-dimethoxy-diphenic acid

(X) ( ) and a corresponding model compound (XI) ( ).

Table 4. Specific rotation [a] of dextro-rotatory and laevo-rotatory polyamide (X) and model compound (XI)*. Solvent: DMF; temp. 21°C.

( + ) X (+) x i

( - ) X ( - ) XI

Concentration g/100ml

0.527 0.284

0.5 0.5

578

+ 62.4 + 58.1

-95.9 -73.1

546

+ 73.1 + 68.7

-114.2 -86.0

436

+ 162.5 +142.6

-250.0 -183.2

405

+ 223.0 + 193.7

-342.9 -246.3

* The optical purity of the dimethoxydiphenic acid (IX) was ~ 70 per cent for ( + ) (IX) and 98 per cent for (-MIX).

248


Table 5. Cotton effects and extrema in the ORD curves of polyamide (X) and model compound (XI)*. Solvent : chloroform

( + ) X ( + ) XI

( + ) X ( + ) x i

λ (nm)

286 284

λ (nm)

298 298

Δε

+ 6.32 + 5.74

[M]

+ 12250 + 11900

CD λ (nm)

250 247

ORD Λ(ηηι)

273 273

Δε

+ 7.57 + 9.55

[M]

-16600 - 7200

* Optical purity ~70 per cent ; [M] and Δε calculated on the basis of molecular weight M = 352.38.

model compound (XI) derived from the dextrorotatory form of the dimethyl -diphenic acid because the polymers Villa, b were not soluble enough and the polymers of VII were not available in sufficient quantities. Again we notice that the polymer X and the model compound XI correspond in their u.v. spectra and their chiroptical properties (see Tables 4 and 5, and Figure 4). These properties arise from the interaction of electrons in the chromophore groups and are therefore not influenced by the molecular weight. But the shape of the molecules and the flexibility of the groups depend to a high degree on the size of the molecule. That is why we notice also in this case where the atropisomer groups are parts of a macromolecule a strong retardation of the racemization (see Table 6). The half-life time for the racemization

Table 6. Racemization of dextro-rotatory polyamide (X) and the model compounds (+) (XI) and ( —) (XI). Solvent: DMF; temp.

110°C.

( + ) X ( + ) X I ( - ) XI

/crac x 106(sec_1) 2.4 6.12 6.09 τ^(η) 81.6 31.4 31.6

of the polymer at 110° C in DMF is almost three times as long as that for the model compound. We regard this case as a sterically caused polymeric effect. The next example shows the influence of the molecular weight on the equilibrium constant of EDA complexes.

2. EDA COMPLEXES OF SOME OLIGOESTERS AND MODEL COMPOUNDS12

(Together with H. TANAKA)

During recent years many authors have examined polymeric EDA complexes13. We have to distinguish four types of complexes (see Scheme 2).

249

ROLF C. SCHULZ

CT complexes with polymers

D

A

1 D

A 1

1 D

A

( a )

1 D

A 1

( c )

— Γ D

A

1 D

A 1

Scheme 2

A

D

1 • D

T A

D

( b )

1 ·■ A ··

( d )

Γ A

D

1 D

This means : in (a) a great number of donor groups D are linked to the macro-molecular backbone and the acceptor A is a low-molecular compound. In (b), however, the acceptor groups A are linked to the macromolecule. Furthermore, it is possible [as in (c)] that both components of the EDA complex are attached to macromolecules14. Some experiments have been reported in which the polymer contains donor groups as well as acceptor groups (d) and therefore an intra-molecular complex-formation is possible15.

Most of the polymer-EDA-complexes which have been examined, however, belong to the first-named type, i.e. the donor groups are part of the macro-molecules and become complexed with low-molecular acceptors. The equilibrium constant K can be determined by the absorption spectra. Here the Benesi-Hildebrand-Scott equation is usually taken; it applies for low-molecular EDA complexes16.

ΑΟ x C, DO

D K x ε ε (8)

(XII)

— O O C ^ ^ - COO—CH — CH2-J „

(XIII) (XIV)

COO— CH2— CH2—OOC

250


In equation 8, CAO and CDO are respectively the concentration of the acceptor and donor groups, D is the optical density, K is the equilibrium constant, and ε is the molar coefficient of extinction.

In plotting (CAO x CDO)/D as a function of CDO the constants K and ε are determined by graphical methods.

For our experiments we prepared a polyester (XII) from ethylene glycol and £rans-2,3-dicarboxyspirocyclopropane-1,9'-fluorene ; furthermore we

CAOx103 =1.026, mol/l.

λ,ΓΠμ

Figure 5. Absorption spectra of ED A complexes between TCNQ and a polyester of trans-2,3-dicarboxyspirocyclopropane-1,9'-fluorene.

synthesized the model compounds (XIII) to (XV) which correspond respectively to one and two base units of the polymer12. The synthesis is shown in Scheme 3. The fluorene-groups are π-donors which together with the acceptors tetracyanoquinodimethane (TCNQ) or tetracyanoethylene (TCNE) form EDA-complexes. In u.v. spectra we have two CT-bands. The following considerations, however, are restricted to TCNQ-complexes and the long wave CT-band. Experimental details and a detailed discussion have recently been published12. In Figure 5 the absorption spectra of the polymer donor-acceptor complex with constant acceptor and variable donor concentration are shown.

251

RO

LF C. SC

HU

LZ

v\ ^ //

252

COOH H3COOC COOCH3

ClOC COC1 COC1

H X m o o 2: > 2 o z o > > r o 0 o G

m > o H O 2!

(XII) Scheme 3 (XV)

2 > o O

o r o c r m

Table 7. EDA complexes of polyester (XII) and model compounds (XIII) to (XV) with tetracyanoquinodimethane (TCNQ). Solvent: CH2C12; temp. 19°C.

Λ:τ(πιμ) ^ χ ε ( m o l 2 x l 2 c m !) K (mol l 1.) ε (mol 1 1. x cm l)

[—H2C—OOC—D—COO—CH2—]* XII H3C—OOC—D—COO—CH3 XIII

H—D—COO—CH3] XIV [H—D—COO—CH2—]2 XV

547 557 567 557

20000 2884 4386

20833

5.0 ±0.7 0.3+0.1 1.0 ±0.1 6.4 ±0.9

4032 ±356 10300±2850 4350±380 3 260 ±250

O

Λ7-4400: pn ~ 15 D: n x c r N


In all cases there is a great excess of donor groups, (D :A = 1:0.78 x 10"2

to 1:36 x 10"2) that means 1 mol acceptor per 28 to 127 mol donor groups. But it can be shown that there is only one kind of complex whose composition is probably 1:1. The absorption maximum of the CT-bands is 547 πιμ. Calculation according to Benesi-Hildebrand-Scott yields for the product Κε = 20000 (see Table 7).

The dimethyl ester of the irans-dicarboxylic acid (XIII), whose structure corresponds almost exactly to one base unit of the polyester, forms an EDA-complex with a CT-band at 557 ιημ under the same conditions; the product Κε, however, is only 2900, that is ~ 15% of that of the polyester. Neither does the complex formation of model compound XIV correspond to the polymer. But the compound XV which corresponds to two base units of the polyester, forms an EDA complex with a similar equilibrium constant to that of the polymer (see Table 7). We can be almost certain that in this case, too, we have a 1:1 complex, despite the excess of donor, i.e. only one fluorene ring of the dimer is complexed with the TCNQ-molecule. Although it has not been proved it seems obvious that here an acceptor-molecule is embedded sandwich-like between two donor-groups of the dimer.

In any case these experiments show that a simple model compound does not really reflect the properties of the polymer. The presence of neighbouring donor groups seems to be the cause of the greater stability of the polymer complex although the donor groups are not directly involved in the formation of the complex. This reaction can therefore only be imitated with models which also contain several donor groups.

Furthermore we prepared a homologous series of oligoesters of the ds-2,3,-dicarboxyspirocyclopropane-l,9'-fluorene (XVI) in order to determine the

- H X - O O C n = 3 to 10

influence of the number of donor groups per molecule on complex formation17. Together with TCNQ or TCNE they form EDA complexes, whose absorption spectra were recorded with different acceptor/donor ratios. The CT-bands always lay between 548 and 550 ιημ. In all cases the data could be described by the equation of Benesi-Hildebrand-Scott. It is surprising that the intercept decreases with increasing degree of polymerization, i.e. the product K x ε increases (see Figure 6 and Table 8). According to Liptay's method18 we can show that here, too, we find only one type of complex; furthermore we may assume that the degree of polymerization has no influence on the ionization potential of the donor group (i.e. of the fluorene ring) and on the Coulomb interaction energy. The molar extinction coefficient should therefore be the same for all oligomer complexes. This means, how-

255

ROLF C. SCHULZ

8, = 3.0

0.6 1.0 CDOx10 (mol/l.)

Figure 6. Benesi Hildebrand-Scott plots of EDA-complexes between ds-oligoesters (XVI) and TCNQ.

ever, that the equilibrium constant increases with an increasing degree of polymerization, at least in this oligomer range.

By statistical calculations and the introduction of an 'effective donor concentration' it is possible to derive a modified Benesi-Hildebrand-Scott equation17. The degree of polymerization Pn is part of this equation as well as the quantity m which gives the number of all donor groups being involved in complex formation.

CDO x CA 1 (m — 1)] ε D K' x ε' [Ρ,

K x εχ Pn = Κ'ε' [Pn-(m- 1)]

(9)

(10)

According to these equations the plotting of K χ ε χ Pn against Pn should result in a straight line with the abscissal section (m — 1) and the slope K' x ε'. In fact the observed values for complex formation with TCNQ as well as with TCNE fit this equation very well (Figure 7). For the TCNQ complexes we find for the product K' x e' = 23200; and for m we get a value of 2.8 ; this means that we have an EDA complex, in which on the average two or three donors are complexed with one acceptor. For the TCNE complexes we get m = 2.4. These values do not tell us anything about the geometrical arrangement and the spatial structure of the complex, but it is likely that the structure here is again sandwich-like, because of the close neighbourhood and the rather restricted mobility of the donor groups attached very tightly to the backbone.

256


This assumption is supported by space-filling models and also explains the greater stability of the oligomer complexes compared with monomer complexes.

Although in the oligomers three donors per acceptor are involved in the complex because of these spatial arrangements, the electronic interaction

18

o X

lOflO

2

2 6 10

Figure 7. Plot according to equation 10 for EDA-complexes between cis-oligoesters (XVI) and TCNQ ; values of Figure 6.

only takes place between one donor and one acceptor group (as may be concluded from the u.v. spectra).

Other authors19 too, assume that polymeric EDA-complexes may have sandwich-like structures; but they did not see any relationship between complex stability and degree of polymerization. Probably this effect becomes sufficiently evident only in the oligomer range.

These examples show how important it is even with apparently simple equilibrium reactions with polymers, to study low-molecular model compounds. In the following section some irreversible reactions between a highly reactive polymer and low-molecular substrates will be described.

Table 8. EDA complexes of oligomeric esters of cis-2,3-dicarboxyspirocyclopropane-l,9'-fluorene (XVI) with tetracyano-

quinodimethane (TCNQ). Solvent : CH2C12 .

Fn λ€Ύ (τημ) Κ χ ε ( m o l - 2 x l 2 x c m 1 )

- 3 . 0 550 9470 - 3 . 9 550 10180 - 4 . 9 550 14840 - 7 . 2 550 16540 - 7 . 8 546 18510 - 9 . 9 548 19020

257 P.A.C.—30/1—K

A: TCNQ

(m-1

D:[-CH2 -OOC-D-COOCH2-L eis

ROLF C. SCHULZ

3. PREPARATION AND PROPERTIES OF W-CHLORO-POLYAMIDES20 2 *

(Together with H. SCHUTTENBERG, M. MÜLLER and K. HAHN)

Carboxamides or imides, in which an H-atom is replaced by a halogen atom are extremely reactive compounds and are therefore often used in preparative organic chemistry. The best known compounds of this kind are JV-bromo-succinimide, A/-bromo-hydantoin, A/-bromo-acetamide etc. Although there are many polymers with amide groups or imide groups in the main chain or in pendant groups, polymer iV-halogenamides have as yet only been described in two examples22. In continuation of earlier experiments23 on polymers with reactive groups we prepared the N-chloro

—<CH2)2—C—N— ; —<CH2)5—C—N-II I II I O H O H

-N- (CH 2 ) 6 -N-C-<CH 2 ) 4 -Ç- ; -CM \ ç - N - ( C H 2 ) 6 - N -I .. .. ..

H H O O O O H H

(/ Vs02-Ij^(CH2)6~N-

Ç - N - C H 2 - ^ y - C H 2 - C H — C H — C H -

o<^NAo O H

Scheme 4

derivatives of the seven following polyamides : nylon 3, nylon 6, nylon 66, polyhexamethylene terephthalamide, polyhexamethylene-w-benzene-sulph-onamide, polystyrene-co-maleimide and poly-N-benzylacrylamide. But the ensuing considerations are restricted to the N-chlorination of nylon 66 and to the reactions of ΛΓ-chloro-nylon (equation 11).

- C - { C H 2 ) 4 - C - N H - ( C H 2 ) 6 - N H - -

O O (XVII) (Π)

—C—(CH2)4—C—N—(CH2)6—N— II II I I O O CI Cl (XVIII)

We used acetyl and propionyl derivatives of monamines and diamines (e.g. XIX to XXI), as model compounds. The chlorinations were done with i-butylhypochlorite (dissolved in tetrachloroethane) or chloromonoxide (dis-

258


Nylon 66 and model compounds

—C—HN—CH2—(CH2)4—CH2—NH—C—CH2—(CH2)2—CH2—

O O

CH3—(CH2)4—CH2—NH—C—CH3

O

CH3—CH2—CH2—NH—C—CH2—CH3

O

H 3 C - C — HN—CH2—(CH2)4—CH2—NH—Ó—CH3

(XVII)

(XIX)

(XX)

(XXI)

O Scheme 5 O

solved in tetrachloroethane); gaseous chlorine and aqueous solutions of potassium hypochlorite20 can also be used. Finely divided nylon 66 was suspended in C2H2C14 and the chlorination agent was added between — 20° C and + 15° C. After 3 to 12 hours we got a clear solution which was dropped into ether. The resulting iV-chloro-nylon (XVIII) is a white powder. As it does not contain any NH-groups and therefore no H-bonds can be formed, it is easily soluble in tetrachloroethane, chloroform and benzene in contrast to nylon 66. In the i.r. spectrum the NH-bands are missing at 3 300 and 3080 cm"1 and the amide bands II at 1530 cm - 1; the carbonyl band is shifted by 35 cm"1 towards longer wavenumbers (Figure 8). In the model compound (XIX) and its N-chloroderivatives we find the same bands and band shifts (Figure 9). The n.m.r. spectra confirm the structure mentioned above.

2 5 λ 3 3.5 U lOoF

80F

6θ[

40l·

2θί

°L 4000 v 3000

0 0

8 9 10 μ 15 20 25 I 1 1 1 1

» /> · ' 11 '

I ; ; (\! 1 ; \i)

Γ 1 1

1 1 1 1 1 1 1

\. h\ \\ \ ! v :\ \/\ 1 v 'V

1 1 . 1

2000 1500 1000

0

500 400

-C- (CH 2 ) 4 -C-N- (CH 2 ) 6 —N-

Cl Cl

-C-(CH2 )A -C-N-(CH2 )6—N-

H H

Figure 8. The i.r. spectra of nylon 66 (XVII) and chlorinated nylon 66 (XVIII).

259 P.A.C—30/l-K·

ROLF C. SCHULZ

2.5 λ 3 35 4 5 6 7 8 9 10 μ 15 20 25 00

80

60

40

20

0

1

\ / ^

ϊ '\ ί 1 ' ''

-

1 I

c~ Ì 1

ν,;

ι ι l i l t ! ι ;

Λ Α ~ - - ΐ « . - ^ — ν I

Λ A .r "'"''Λ

\ ' 1 / "*' 1/ *'

il \\ ι ' ι r

4000 3000 2000 1500 1000 500 400

CH3-(CH2)A— CH2— N - C - C H O I II

H 0

CH.— iCHol-CH.-N-C-CHo I II ό

Cl 0

Figure 9. The i.r. spectra of N-acetyl-fî-hexylamine (XIX) ( ) and the iV-chlorinated compounds ( —).

261 °C

Nylon 66

T C -1?-(ΟΗ2)4-?-Ν-(ΟΗ^Ν]n / \ / V v ^ Cl Cl*—*

9 o CH3-C-N-(CH2)6-N-(i-CH3

Cl Cl

157°C

140 °C

Figure 10. Thermograms of nylon 66 (XVII), N-chloro-nylon (XVIII) and chlorinated model compound (XX).

260


As we have no H-bonds in chlorinated products, the melting points are much lower than those for the amides. In the sample of nylon 66 we have been using here, we get an endothermal melting peak at 261 °C (Figure W) in the thermogram; after the chlorination the melting peak is at 71° C. The chlorinated model compound melts at 48° C At 140° C and at 157° C there are great exothermal peaks. This transformation will be discussed further on page 263.

In order to determine iV-chloro-amide groups quantitatively the polymer is treated with a potassium iodide solution and then the resulting iodine is titrated. The chlorine content of 30 per cent to nearly 100 per cent of theory depends on the chlorination agent and the experimental conditions. Some examples are to be seen in Table 9. During chlorination a chain degradation

Table 9. Chlorination of nylon 66 (P ^ 180).

Chlorination reagent

tert-BuOCl

KOC1 in H 2 0 Cl2 + KHC0 3 C120 in CCU

Temp. C

- 2 0 + 15

0 + 15 - 2 0

Conversion / o

69-95-70-80-98-

70 96 73 89

100

P

137 118 132 29 34

also takes place. The strongest degradation can be observed when chlorine or chloromonoxide are used. Tert-butylhypochlorite causes only a small reduction of the degree of polymerization and is therefore very suitable indeed. Naturally the polymers with the pendant amide- and imide-groups mentioned above (Scheme 4) are not decomposed during chlorination.

After having elucidated the structure of the chlorinated nylon we examined the chemical reactivity of the polymer as compared with the model compounds. We knew from former publications24 that N-chloroderivatives of cyclic and linear amides and imides react with diazomethane by inserting a méthylène group; one gets N-chloromethyl derivatives (equation 12). This

R' R

,N—Cl + CH2N2 -+ N—CH2C1 + N2 (12)

R R'

R : CH3—C— : C6H5—C—; p-tosyl II II

o o R': CH3—: C6H5—CH2—; cyclohexyl-; i-propyl-

reaction goes on under very mild conditions and with yields of 50 to 80 per cent, depending on the kind of substituent. All attempts to get this reaction also with N-chloro-nylon failed ; we always got insoluble polymers, whose structures could not be defined.

261

ROLF C. SCHULZ

The reaction with potassium iodide shows that the N-chloro-polyamides are very good oxidizing agents. Therefore we tested whether they were also suitable for the oxidation of alcohol in analogy to N-bromosuccinimide.

If one dissolves N-chloro-nylon in benzene and adds the equivalent amount of pyridine and, say, cyclohexanol one observes that the content of the active chlorine decreases after a certain induction period. Finally a powder-like polymer precipitates which does not contain any chlorine and is nearly pure nylon. The solution contains more than 80 per cent of the theoretical amount of cyclohexanone (proved and determined by GPC). The reaction depends on the solvent. In DMSO no ketone is formed; in CC14 or CHCI3 the reduced polymer precipitates as a gel and the separation of the oxidation products is difficult. We get an optimal reaction with benzene or cyclohexane, although the polymer is not soluble in cyclohexane. The oxidation then takes place in a heterogeneous medium. The possibility of separating the polymeric oxidation agent by filtration after the reaction is a great advantage. Thus many other secondary alcohols can be oxidized into ketones with nearly quantitative yields. The reaction can in general be described by equation 13.

0

1 Cl

0 II

- N - C - -1 H

R

- - + CH—OH + Pyr.

R

R \

- - + C = 0 + Pyr.HCl

R'

(13)

We chose some of the experiments described in the literature25,26 in order to make comparisons with the oxidation reaction of low-molecular N-halogen amides. The results are listed in Table 10.

The yields determined by gas chromatography exceed 80 per cent when oxidized with N-chloro-nylon. The yields are also determined by the weights of semicarbazones and dinitrophenylhydrazones in order to compare the results with published data. We find that we get the same yields or in some cases even better ones with N-chloro-nylon compared with N-bromo-hydantoin or N-bromoacetamide.

Kinetic experiments on oxidation proved difficult because the reactions with N-chloro-nylon beyond a certain stage proceed in a heterogeneous medium27. It was clearly shown, however, that the active chlorine of the polymer is consumed faster than in model compounds and N-chloro-caprolactam.

Primary alcohols can also be oxidized with N-chloro-nylon. At 35°C benzylalcohol is transformed into benzaldehyde after 24 hours with a yield of 95 per cent. The aldehydes formed are not oxidized even with a great excess of N-chloro-nylon. The oxidation of aliphatic primary alcohols yields the esters of the corresponding acids. We can assume that in this case semi-acetals as intermediates are formed in a similar manner to the oxidation with

262

THE C O M P A R I S O N OF A N A L O G O U S REACTIONS OF M A C R O M O L E C U L E S

Table 10. Oxidat ion of secondary alcohols with ΛΓ-chloro-nylon (JVCnyl). N-bromo-hydan to in (NBH) and N-bromo-ace tanude (NBA); temp. 35°C;

solvent : benzene ; t ime : 24 h.

Yields of ketones with ^nyl JVBH

(a) (b°) (b) (c)

A1 , - NCnyl ΛΓΒΗ NBA Alcohol 0 / ^ 0 / 0 / 0 /

/o /o /o /o

Cyclohexanol 82 — — 45 P h — C H 2 — C H — C H 3 90 62 51 35

I OH

Ph—(CH2)2—CH—CH3 95 68 74 —

OH (Ph—CH2)2—CH 62 — — < 1

OH Ph—CH2— CH—i-prop. 92 60 53 <1

OH

(a) with GPC. (b) semicarbazone. (c) 2,4-dinitrophenylhydrazone.

chromic acid28. Actually, an equimolar mixture of octanol and octanal with N-chloro-nylon forms 70 per cent n-octyl-caprylate after a few hours. The oxidation of pure octanol proceeds much more slowly.

As mentioned above a strongly exothermal transformation takes place at 140°C. Without any solvent highly chlorinated nylon can decompose in an explosive way. We suppose that this transformation is an analogous reaction to the Hofmann-Löffler rearrangement of N-halogen amines29. It proceeds by a radical mechanism and leads to chain-halogenated N-alkyl amines. These rearrangements have also been described with iV-bromo- and N-chloro-carboxamides (equation 14). They lead partly to the y-chloro-

R — C H 2 — C H 2 — C H 2 — C — N — R

O Cl

R — C H — C H 2 — C H 2 — C — N — R (14) i n R':alkyl;R:f-C4H9

carboxamides30 with satisfactory yields; usually a cyclization reaction follows in the case of N-bromo compounds. If we proceed with the rearrangement of the JV-chloro-nylon under comparatively mild conditions we actually get a product which contains approximately the same amount of total chlorine but no active chlorine linked to nitrogen. The position of the chlorine and the structure of the polymer are still under investigation31.

Further, it is known that this rearrangement is also induced by light, 263

ROLF C. SCHULZ

although JV-chloro-carboxamide above 250 ιημ has no absorption maximum32. The mechanism of this reaction is still obscure. Nevertheless we have also tried to make this photo-rearrangement with N-chloro-nylon. It we cast a thin film of this polymer on plates of rocksalt and record the i.r. spectrum we get the spectrum already mentioned above (Figure 8, solid line). Even a longer i.r. irradiation does not cause any changes. Upon irradiation of the film with a mercury vapour lamp (365 nm) for some seconds the NH-bands appear in the spectrum at 3300 and 3030 cm"* (Figure 11). This

25.

100

3.5

80

60

ω

20

Photo rearrangement of A/-chloro- nylon

£000 3000

Figure lì. Change of i.r. spectra of chlorinated nylon during irradiation with a mercury lamp.

means that the rearrangement proceeds also in a solid state. After about five minutes of radiation we get an i.r. spectrum similar to that of the original nylon. With increase of the NH-bands the ability to oxidize disappears ; in fact irradiated N-chloro-nylon does not actually form any iodine from potassium iodide solution.

264


We gratefully acknowledge the financial support of 'Deutsche Forschungsgemeinschaft'. H. Schuttenberg expresses his sincere gratitude for an 'Alexander von Humboldt-Stipendium' and R. H. Jung is grateful for a 'Liebig-Stipendium' from the 'Fonds der Chemischen Industrie'.

REFERENCES 1 E. M. Fettes (Ed.) Chemical Reactions of Polymers High Polymer Series, Vol. XIX, Inter-

science: New York (1964). Houben-Weyl, Methoden der Organischen Chemie E. Müller (Ed.) 4th ed., Vol. XIV/2, p 637. Thieme: Stuttgart (1963). IUPAC Symposium, Brussels-Louvain, Belgium (June 1967); Pure Appi Chem. 16, Nos. 2-3 (1968). International Conference on Chemical Transformations of Polymers, Bratislava (June 1968), Europ. Polym. J. Suppl. (August 1969).

2 R. B. Merrifield, J. Amer. Chem. Soc. 85, 2149 (1963); 86, 304 (1964); R. L. Letsinger and V. Mahadevan, J. Amer. Chem. Soc. 87, 3526 (1965); F. Cramer, R. Helbig, H. Hettler, H. Scheit and H. Seliger, Angew. Chem. 78, 640 (1966).

3 C. G. Overberger, J. C. Salamone, J. Cho and H. Maki, Ann. N.Y. Acad. Sci. 155, 431 (1969). G. Manecke, Naturwissenschaften, 51, 25 (1964). G. Manecke and G. Günzel, Naturwissenschaften 54, 531 (1967). A. S. Lindsey, J. Macromol. Sci., Rev. Macromol. Chem. C3, 1 (1969).

4 R. E. Moser and H. G. Cassidy, J. Polym. Sci. BZ 545 (1964). R. Searle, J. L. R. Williams, J. C. Doty. D. E. DeMeyer. S. R. Merrill and T. M. Laakso, Makromol. Chem. 107, 246 (1967). C. David, W. Demarteau and G. Geuskens, Europ. Polym. J. 6, 537 (1970). D. Bellus et al. J. Polymer. Sci. A-l, 9, 69 (1971).

5 R. C. Schulz and R. H. Jung, Makromol. Chem. 96, 295 (1966); 116, 190 (1968); R. C. Schulz and R. H. Jung, Angew. Chem. 79, 422 (1967); Angew. Chem. Internat. Ed. 6 461 (1967).

6 E. L. Eliel, Stereochemistry of Carbon Compounds, McGraw-Hill: New York (1962). K. Mislow, Introduction to Stereochemistry, Benjamin: New York (1965).

7 R. C. Schulz, IUPAC Symposium, Budapest (1969). Main Lectures, pp. 185-212. 8 Some other polymers with atropisomer base units were described by C. G. Overberger et al.

IUPAC Symposium, Tokyo (1966), Preprint 2.3.02 J. Polym. Sci. A-l. 8, 2275 (1970). 9 F. Bell, J. Chem. Soc. 835 (1934).

10 R. H. Jung, unpublished results. 11 R. C. Schulz and R. H. Jung, Tetrahedron Letters. 4333 (1967). 12 A. Braun, H. G. Cassidy, R. C. Schulz and H. Tanaka, Makromol. Chem. 146, 195 (1971). 13 W. Slough, Trans. Faraday Soc. 58, 2360 (1962);

G Smets, V. Balogh and Y. Castille, J. Polym. Sci. C4, 1467 (1964); R. Knoesel, Bull. Soc. Chim. France, 4299 (1967); W. Klöpffer and W. Willicks, Makromol. Chem. 115, 156(1968).

14 T. Sulzberg and R. J. Cotter, Macromolecules, 1, 554 (1968). 15 N. C. Yang and Y. Gaoni, J. Amer. Chem. Soc. 86, 5022 (1964). 16 G. Briegleb, Elektronen-Donator-Acceptor-Komplexe, Springer: Berlin (1961);

R. Foster, Organic Charge-Transfer Complexes, Academic Press: London (1969). 17 H. Tanaka, Mainz and Darmstadt (1970), unpublished results. 18 W. Liptay, Z. Elektrochem. 65, 375 (1961). 19 J. Parrod, P. Rempp and R. Knoesel, J. Polym. Sci. C16, 4049 (1965). 20 H. Schuttenberg and R. C. Schulz, Makromol. Chem. 143, 153 (1971) 21 H. Schuttenberg and R. C. Schulz, Angew. Makromol. Chem. 18, 175 (1971). 22 M. Okawara and H. Shinohara, J. Chem. Soc. Japan, Industr. Chem. Sect. 60. 75 (1957);

Chem. Abstr. 53, 5730h US Pat. No. 2853 475; W. A. Murphey, Chem. Abstr. 53, 3725h

23 R. G Schulz, Pure Appi. Chem. 16, 433 (1968). Angew. Makromol. Chem. 4/5. 1 (1968). 2 4 R. A. Corral and O. O. Orazi, Tetrahedron Letters 1693 (1964);

O. O. Orazi, R. A. Corral and H. Schuttenberg, Tetrahedron Letters, 2639 (1969).

265

ROLF C. SCHULZ 25 R. A. Corral and O. O. Orazi, Anal. Asoc. Quim. Argentina, 55, 205 (1967); Chem. Abstr. 69,

105 572r 26 J. Lecomte and H. Gault, C.R. Acad. Sci., Paris 238,2538 (1954);

J. Lecomte and C. Dufour, C.R. Acad. Sci., Paris 234,1887 (1952). 27 K. Hahn, Darmstadt (1971), unpublished results.

28 W. A. Mosher and D. M. Preis, J. Amer. Chem. Soc. 75, 5605 (1953). 29 R. S. Neale, Synthesis, 1, 1 (1971). 30 R. S. Neale, N. L. Marcus and R. G. Schepers, J. Amer. Chem. Soc. 88, 3051 (1966). 31 M. Müller, Darmstadt (1971), unpublished results. 32 R. S. Neale and M. R. Walsh, J. Amer. Chem. Soc. 87, 1255 (1965); A. L. J. Beckwith and

J. E. Goodrich, Austral. J. Chem. 18, 747 (1965).

266

MECHANISMS OF POLYMER STABILIZATION

G. SCOTT

Department of Chemistry University ofAston in Birmingham, Warwickshire, UK

ABSTRACT

The behaviour of antioxidant in polymers differs markedly depending on the type of test used. The performance of antioxidants can be optimized on the basis of intrinsic activity, compatibility with the polymer and rate of loss from the polymer by volatilization. Molecular interactions in antioxidants and stabilizers have a profound effect on all these factors which may be influenced in opposite directions. Present trends in antioxidant and stabilizer technology are toward agents designated to meet specific requirements. The progress of recent studies on the mechanism of sulphide and phosphite antioxidants is

reviewed.

INTRODUCTION

Industrial companies spend many hundreds of thousands of pounds a year on the evaluation of new stabilizing systems for polymers. Much of this testing is carried out under conditions which bear little relation to practical performance and considerable doubt exists about the ability of accelerated tests to predict correctly the useful service life of the fabricated article. Three types of prediction test can be distinguished.

(a) Melt stability tests The evaluation of polymer melt stability by the measurement of changes

in melt viscosity has direct relevance to the performance of the polymer under processing conditions. The effect of the stabilizer in this case is to extend the induction period to rapid viscosity change and this is normally carried out by monitoring changes in melt-flow index at intervals during the processing operation1 or by continual measurement of polymer viscosity in a torque rheometer under oxidative conditions2. Assuming that this type of test is carried out under the same conditions of temperature, shear and environment then results are directly relevant to engineering practice. Apart from noting at this stage that what happens to the polymer under these conditions determines the subsequent performance of polymer articles under service conditions3, this type of test is the most satisfactory predictor of performance of those normally used and will not be discussed further.

267

G. SCOTT

(b) Thermal stability of fabricated test pieces The second type of test which is widely used as a predictor of long term

performance involves high temperature evaluation of fabricated component performance. This is normally carried out by subjecting sheet or film to temperatures between 100° and 150°C in an oxygen or air stream4-8. This type of evaluation, which is used almost universally by polymer and stabilizer manufacturers is open to the severe criticism that these conditions bear little relationship to those encountered in the normal service life of polymers. The exceptions to this are polymers Used in hot water systems or in engineering components where high temperature is one of the conditions of service. For ambient use, the amounts of antioxidant incorporated to stabilize the polymer in the melt are normally more than adequate to protect the polymer during service. The amount of antioxidant used to give polypropylene adequate processing stability at 200° can be calculated (assuming the Arrhenius relationship holds) to give the fabricated article a service life of between ten and twenty years if environmental degradation occurs simply as a result of thermal reactions. In practice service behaviour is much more complex than this and factors other than thermal oxidation become dominating. In plastics by far the most important of these is u.v. light.

(c)The u.v. and simulated weathering stability of fabricated test pieces The u.v. exposure cabinet test is much more relevant to the service life of

the polymer than is the high temperature thermal stability test, assuming an u.v. source has been selected to provide a similar spectral distribution to that of sunlight. Accelerated u.v. testing also has the advantage that it can be correlated directly with outdoor exposure. Moreover, lifetime extrapolation can be made directly to service conditions based on previous experience of the u.v. source. Since other speakers at this conference will be discussing u.v. stabilization, this type of technique will not be discussed further in this lecture.

ANTIOXIDANT ENVIRONMENT IN THE POLYMER In spite of the practical lack of relevance of the thermal stability test (b) as

a direct predictor of service life for most applications, it is still widely used in industry. Recent work has thrown a good deal of light on the factors which determine stabilizer activity under practical conditions. It has been known for some time as a result of experience in rubber and more recently polypropylene that a factor of major importance in determining antioxidant activity is the molecular weight of the antioxidant9.

The work of Gordon10 is particularly illuminating in this connection, since he has shown that the way in which an antioxidant is tested determines its relative effectiveness in relation to other antioxidants. Table 1 shows that antioxidants rate in effectiveness in different orders depending on the sample thickness in an oven test at 125°C

First of all, it should be noted that all antioxidants become apparently more effective as the sample thickness is increased but that some (e.g. antioxidants A and D) show this effect much more markedly than others (e.g. antioxidant I).

268


Table 1. Days to failure in oven ageing at 125°C.

2.5

| _ j U 2 j B29 C34 E34 H34

LD50J F50 G75

! 195 !

Sample 25

B117 I A130 |

C140 E187 F226 D256 G259

• 1288 ; H297

thickness, cm 2

60

B151 C185

| A240 | E287 [ F375

i 1418 ; j G460

1 D479 | H552

120

B166 C219 E368 A390I F453 1501 i G > 600 D > 600 H > 6 0 0

A similar pattern emerges if the sample thickness is held constant but the temperature of testing is increased (Table 2).

Table 2. Days to failure in oven ageing tests for 25 cm samples.

100

B167 C183

[A286J E300 G387 H466

: 1507 ; F > 600 D > 600

Temperature, 125

B117 1 A130I

C140 E187 F226

1 D256 1 G259

: 1288 : H297

°C 140

1 A22 1 B34 C35

1 D53 1 E57 F64 G66 H72 ;I7_4;

150

EU B15

1 PIS 1 C17 H19 F27 G30 E31

;_I33;

Here again all antioxidants decrease in effectiveness with increasing temperature but some (e.g. antioxidants A and D) are much more susceptible to temperature increase than are others (e.g. antioxidant I) and their relative ratings as the temperature is increased are quite different. It is clear then that factors other than inherent antioxidant efficiency are playing a part and the obvious inference to be drawn from Tables 1 and 2 is that antioxidants A and D are relatively volatile antioxidants whereas antioxidant I is relatively involatile.

We have carried out a more systematic study of the effect of antioxidant molecular weight and volatility in two series of antioxidants. each of which contained the same antioxidant function (Table 5)11.

269 P.A.C.—30/1—L

G. SCOTT

Table 3. Induction periods of two series of isofunctional antioxidants in decalin.

Antioxidant function

(ROCOCH2CH2)2S

i-Bu

HO—ft V - C H 2 C H 2 C O O R

/ f-Bu

Side chain

R

Methyl Hexyl Lauryl Stearyl

Methyl Hexyl Lauryl Stearyl

M.pt °C

—

41.2 57.8

67.8 10 15 51.2

Mol. wt

206 346 514 702

292 362 446 530

Induction period h

28 27 26 26

>2 x 10 - 5 mole/100 g

27 24 22 18

>2 x 10~3 mole/100 g

Members of both these series give a slightly decreasing antioxidant activity on a molar basis when measured in a sealed system by oxygen absorption. Behaviour in polypropylene in a sealed system is rather different in that there is maximum effectiveness with the lauryl ester in the case of the phenols (see Figure 1). There is an interesting inverse correlation with melting point suggesting that under these conditions compatibility with the polymer is playing a role in antioxidant effectiveness.

f -Bu 5oor y-\

HO-C XV-CH2CH2COOR

t-Bu 400

| 3θομ CD Q .

c Î 2001 D

100

M 5 ° C )

/—In polypropylene V (52°C)

P(68°C)

In decalin A -

1 6 12 18 Number of C atoms in pendant chain

Figure 1. Effect of sidechain length on the effectiveness of the isofunctional phenolic antioxidants at 140°C by oxygen absorption.

270


It can be seen (see Table 3) that the sulphur series is very much more effective than the phenol series. It has been suggested earlier1 2 1 3 that sulphur dioxide, which is known to be formed from the thiodipropionate esters by reaction with hydroperoxides and which is a very efficient catalyst for hydroperoxide decomposition14, is one of the compounds which contribute to the powerful antioxidant activity of this class of antioxidant. Some weight is given to this by the finding that in a closed system and at the same molar concentration, sulphur dioxide is about as effective as the thiodi-propionates. However, other lower molecular weight species derived from the thiodipropionate esters by oxidation are also formed which effect the same acid-catalysed process15.

Oxygen absorption tests give valuable information about inherent antioxidant activity but predict little of practical value about behaviour under service conditions. In addition results from the plaque oven ageing test referred to earlier are notoriously irreproducible. The reasons for this are associated with the difficulty in assuring that each sample is identical with respect to dimensions, prior heat treatment, antioxidant dispersion, etc. and an accuracy better than +50 per cent is not normally possible.

Torsional spindle

Spindle clamp mounted on race bearing

Spring

Glass/Polymer braid

Glass tubing draught shield

Cable release Release arm and pointer

Protractor

Fixed spindle bearing

Pin vice

Pin vice

Torsional disc pendulum

Figure 2. Apparatus for determination of torsional period of braids.

271

G. SCOTT

In our own work considerable attention was paid to these factors. The technique used was a modification of that used by Lewis and Gillham16,17

for following crosslinking reactions in polymers. The test piece was a glass-polymer braid, whose torsional period was measured (see Figure 2) at intervals on ageing in a single-cell oven at temperatures between 100°C and 140°C with a constant flow of oxygen past the sample (28.3 l./h). Particular care was taken that the samples were reproducible. Unstabilized polypropylene was used in all experiments and stabilizers were incorporated at concentrations between 2 x 10~4 and 2 x 10~2 mol/100 g by solvent application and evaporation. This technique was found to give more reproducible results than ball-milling of the antioxidant and polypropylene powder. The polypropylene was applied to the glass (which was carefully heat treated at 400°C for two hours to remove the size) by wetting the glass with distilled water and drawing through the powder. Fusion was carried out in the complete absence of oxygen by heating to 200°C in nitrogen after first displacing air from the polypropylene powder11. After one fusion cycle, the melting point of the polymer was found to be constant indicating satisfactory antioxidant dispersion.

35 r

)l 1 1 LxJ i i I ■ ' i 0 25 50 75 100 125 150

Ageing time, h

Figure 3. Typical polypropylene torsion braid decay curves at 100°C: — v — dilauryl thio-dipropionate; —Δ—lauryl-ß-(3,5-di-tert-butyl-4-hydroxy phenyl)-propionate. Antioxidant

concentration 2 x 10 ~4 moles/100 g polymer.

The polymer-glass braids produced by the above procedure were found to be remarkably consistent. The polypropylene content of the braids was 5 2 + 2 per cent and the radius 0.3 + 0.03 mm. It was found that the end of

272


the induction period, as indicated by a rapid increase in the torsional period of the braid, was consistent to ± 10 per cent (see Figure 3) and although the initial modulus of the braids varied, this did not seem to affect the accuracy of the result.

As anticipated, under the conditions outlined involving high surface area to volume ratio and continual change in atmosphere, the effectiveness of antioxidants in the same homologous series varied markedly with molecular weight. The results obtained at 100°C at two molar concentrations are shown in Table 4.

Table 4. Effect of molecular weight on the induction period of two series of antioxidants.

Antioxidant Mol. wt

Induction period, h 2 x IO"4 2 x IO"3

mol/100 g mol/100 g

(ROCOCH2CH2)2S

J^w. i-Bu

Methyl 206 25 42 Hexyl 346 30 42 Lauryl 514 65 80 Stearyl 702 130 430

Methyl 292 25 45 Hexyl 362 40 80 Lauryl 446 75 400 Stearyl 530 9800 J>L04

A simple volatility cell was used to measure the relative volatility of antioxidants at various temperatures in an atmosphere of nitrogen to eliminate the possibility of antioxidant destruction by oxygen. A plot of log IP against the reciprocal of the volatility (Ì/V) showed (see Figure 4) that the two series of antioxidants differed very considerably in their efficiency/ volatility relationship; the phenol series is very much more effective than the sulphides as volatility is reduced. The reason for this is almost certainly associated with the fact that in performing an antioxidant function, thio-dipropionate esters break down to lower molecular weight fragments which will be more readily lost from the polymer (see later). The results also explain why the higher molecular weight antioxidants discussed above are so much more effective than dilauryl thiodipropionate and do not show very marked synergism with it under high temperature test conditions18.

There appears to be a superimposed factor which introduces a limit to the effectiveness/inverse volatility relationship. Figure 4 shows that the pentaerythritol-based antioxidant containing the same functional group is less effective on a molar basis than might have been expected on the basis of its 1/V value. This appears to be associated with compatibility/mobility factors in the polymer. The melting point of this antioxidant (120°C) is

273

G. SCOTT

10V

10'

10'

7/V'-Bu H O - ^ VcH2CH2COOCH2 l C

V - B u

/-Bu

/ ^ H O - ( ^ V-CH2CH2 COOR

2 3 U 5 1/Volatility g h"1 x 105 (at 100°C)

Figure 4. Effect of sidechain length on the effectiveness of isofunctional antioxidants at 100°C by torsion braid.

considerably higher than those of the others studied implying limited compatibility in the polymer. DTA studies have shown that unlike the other propionic ester-based phenols, this antioxidant has almost no effect on the melting point of the polymer, confirming its incompatibility.

The importance of factors other than molecular weight has been observed in a similar study of increasing molecular weight in the series19

CH, where n = 0 to 3

These antioxidants were compared on a molar basis (i.e. mole/100 g polymer of functional group) in the three tests discussed, namely decalin in oxygen absorption at 140°C, polypropylene in oxygen absorption at 140°C and polypropylene by torsional braid at 140°C. Figure 5 shows a pronounced alternation with the value of n but the alternation is in opposite directions depending on the type of test used. In oxygen absorption in decalin there is a general trend toward decreased activity with increasing molecular weight,

274


Induction periods;h

260r

90

O C 0) Ό)

c ">* Q. O

o

1 100

c o

.a σ

il· 55

OH OH OH Μ θ γ Λ ^ Ο Η 2 ν Λ ^ Ο Η 2 ν Λ ^ ^ Μ θ

V V V

50

>* X o c O u O 1 1

>. CL o Q. >N O

O.

1 1 1

1

Decalin, w\ oxygen absorption

O·0 Polypropylene, * w torsion braid ^ /

Me Me in polypropylene

Polypropylene, oxygen ,χ absorption / \

' / \ / \ /

v / \

P

Me

n=0 n=1 n=2 M. pt =144-6 M. pt = 109-12 M.pt= 191-3

200 300 2246 400 500 Molecular weight

π=3 M.pt=152-4 600

Figure 5. Antioxidant activity at 140°C of polyphenols.

but antioxidants having n = 0 or 2 are more effective than those having n = 1 or 3. In the torsion braid test there is a general increase in effectiveness but n = 0 or 2 are less effective than n = 1 or 3. In the oxygen absorption test in the polymer there is a more marked decrease in activity with n = 0 or 2 being less effective than n = 1 or 3. It seems likely that the alternating effects are all associated with intra- and inter-molecular associations. Intramolecular hydrogen bonds seem to increase the inherent antioxidant activity but this is associated with a decrease in intermolecular hydrogen bonding and hence increase in volatility. The overall effect is to reduce the effectiveness of the antioxidant in the torsion braid test. Internal hydrogen bonding also increases the melting point and hence decreases compatibility so that n = 0 or 2 have higher melting points (see Table 5) and hence lower compatibilities than n = 1 or 3.

It is constructive to compare the position of a commercial bisphenol, Antioxidant 2246 in these three tests. Its intrinsic antioxidant activity and compatibility are good and it performs well in both the closed system tests. It is traditionally noted for its good performance in the Oxygen bomb' test which is of this type. In the torsion braid test it is singularly ineffective, underlining the fact that volatility can dominate other factors in practical conditions.

275

G. SCOTT

Table 5. Physical characteristics of polyphenolic antioxidants.

M.pt °C

Solubility in hexane, 25"C g/100 cm3

0 1 2 3

2246

144-6 109-12 191-3 152-4 119-20

0.25 1.93 0.08 0.22 9.95

10'

10

01 101

10l

O)

10"

10" 100 120 U0 160 180 200

Temperature of volatilization, °C 220

Figure 6. Relationship between temperature and volatility of antioxidants of various molecular weights.

276


Summarizing then, three factors appear to be important in determining antioxidant activity and any of these might have a dominating influence depending on conditions of test. These are : (a) Intrinsic antioxidant activity, which is influenced primarily by the

structure of the molecule, including factors such as intramolecular interactions.

(b) Compatibility/mobility of the antioxidant which will again be determined by intra- and inter-molecular interactions in the molecule but generally in the opposite direction to the above.

(c) Volatility of the antioxidant which will be determined by molecular weight and molecular interaction in the polymer.

Although testing of the last factor has been largely carried out at 100 C and above, the results appear relevant to service performance since volatility/ temperature plots appear to be linear and generally parallel (see Figure 6). It seems likely therefore that in fabricated articles with a large surface area to volume ratio, persistence of antioxidants and stabilizers will be considerably more relevant to service performance than intrinsic antioxidant activity. Careful consideration must therefore be given to selecting the type of test which is most relevant to service performance.

DESIGN OF ANTIOXIDANTS FOR HIGH TEMPERATURE, HIGH SURFACE AREA PERFORMANCE

The future of antioxidants for this type of application lies therefore in the direction of increasing the retention of the antioxidant in the polymer. Interesting results have been reported by Phillips, Thomas and Wright20 for high temperature rubbers. By building a conventional antioxidant structure into a polymer chain, for example by reacting hydroquinone with /?-phenyl-enediamine. an involatile antioxidant (I) was obtained which was much

Γ\ Γ\ N H - e X)-NH- Γ\

(I)

more effective in EPDM and Viton rubbers at 150° and 250°C respectively than the conventional arylamine antioxidants normally used for high temperature applications.

The effect of molecular weight on rubber antioxidant performance has also been reported by Monsanto21 who have found that increasing the size of the iV-alkyl group (R) in the p-phenylenediamine antioxidants (II) considerably increases their antioxidant activity under practical conditions where extraction by water may be of importance. Figure 7 shows that Santofex IP (II, R = isopropyl) although initially more effective than

277

G. SCOTT

Fatigue life

1 5 θ Κ

125

25h

^oSantof lex 13

Santoflex 77

0 3 6 Days immersion

Figure 7. Effect of water extraction on the fatigue life of rubbers containing antioxidants.

2.Or

1.5

n, '

addi

tivi

-C CL

0.5

< ^ ^ ^ » ^ ° \

\

\ Δ

\ X

I

■ - ^ ^ S a n t o f l e x 13

Λ__ Santoflex 77

^ ^ Santoflex IP ^ — - χ

A

._ 1 . . ._ .-._.. i 3 6 Days immersion

Figure 8. Effect of water extraction on the concentration of antioxidants in rubber.

278


P V - N H - ^ V-NHR

(ID

Santofìex 13 (IL R = 1,3-dimethylbutyl) is much less effective after immersion in water. Figure 8 shows that this is due to more rapid extraction of the lower molecular weight antioxidant from the polymer.

An interesting approach to this same problem has been the development of rubbers with antioxidants chemically attached to the polymer chain. By reacting nitroso-containing antioxidants (e.g. diethylaminonitrosoaniline, DENA) with rubbers (a modification of the reaction involved in the cross-linking of rubber by dinitrosobenzene), Cain and his co-workers22 have found that the antioxidant can no longer be removed from the rubber by extraction with water at 100°C.

CH

NEU NEt2 NEU

Table 6. Oxygen absorption at 100°C of water. Extracted vulcanizates (Cain et al.)

Vulcanizate Additive Time to 1 % w/w absorption, h Unextracted Extracted

CBS (white) CBS (black) CBS (white) CBS (black)

DENA DENA IPPD IPPD

39 30 47 40

33 32

Here again, although the isopropyl compound (IL R = isopropyl) is more effective initially, the polymer-bound antioxidant persists in the rubber and is much more effective under conditions of extraction.

279

G. SCOTT

NEW DEVELOPMENTS IN ANTIOXIDANT MECHANISMS It seems likely that apart from the type of development discussed above,

the future of antioxidants acting by the kinetic chain-breaking mechanism is somewhat limited23. This is primarily because the 2,6-di-tert-butyl phenol structure probably combines the best balance of properties for radical capture. Few improved commercial phenolic antioxidants have emerged in recent years and it seems likely that the future lies in the direction of further developments of synergistic systems involving phenols particularly for environmental stabilization. This will be increasingly important as bulk polymers are increasingly accepted in the building and automotive industries. In this connection the peroxide decomposers play a dual role since not only do they stabilize polymers during fabrication operations but they also act as powerful u.v. stabilizers due to their ability to remove the hydroperoxides on which u.v. light acts to produce a powerful photo-initiation of autoxi-dation3.

Two main types of peroxide decomposer have achieved industrial importance in synergistic systems. The first is the sulphur-containing stabilizers of which the thiodipropionate esters (DLTP) (III), the zinc dialkyldithio-carbamates (IV) and a variety of thiols (V and VI) and related disulphides (VII) have become important particularly in thermal stabilization of polypropylene13.

Î (ROCOCH2CH2)2S (R2NCS)2 Zn

(IV)

(VI)

(VII)

The second is the phosphite esters (VIII)

(RO)3P where R = alkyl, aryl and cycloaryl (VIII)

which are again effective as melt stabilizers particularly in combination with phenolic antioxidants with which they show pronounced concentration ratio optima24 as do the sulphur compounds9. It is of some interest that the peroxide decomposers which act as u.v. stabilizers react with hydroperoxides

280


even at ambient temperatures. This is particularly true of the metal dialkyl dithiocarbamates and dithiophosphates which rapidly evolve sulphur dioxide and catalytically destroy hydroperoxides at room temperature14, and the trialkyl phosphites which are equally effective at 25°C25 but act stoichiometrically not catalytically.

10ϋο

Q_ X CJ

Q-X CJ

Time, min

Figure 9. First order plot for the decomposition of 0.2 M eumene hydroperoxide by 0.2 M

°x P —OR

O in chlorobenzene at

The action of the catechol phosphite esters is remarkably similar in many ways to the behaviour of the sulphur-containing peroxide decomposers26. Both appear to act by producing powerful Lewis acid catalysts for the decomposition of hydroperoxides. This is shown for the catechol phosphites in Figure 9. And both appear to involve, in the early stages of the autoxidation

281

G. SCOTT

process, an initial radical generating reaction in the presence of hydro-peroxide which gives rise to pro-oxidant effects (see Figure 10). Pobedimskii and his co-workers27 have postulated on the basis of e.s.r. evidence the

Control

2.5 x 1Cf M

.— x — x x *· x -χ-1.25χ10'2Μ

Figure 10 -inhibited oxidation of eumene initiated by ).2 M eumene hydroperoxide in oxygen at 75°C.

f-Bu

formation of a radical cage reaction by interaction of the phosphorus compound with hydroperoxide to account for this.

(RO)3P + R'OOH -> [(RO)3P~0—R'] I

OH

(RO)3P=:0 + ROH

[(RO)3f>—OH ÒR]

and a similar reaction with sulphide leading to sulphoxide and alcohol. Pobedimskii27 suggests that a small proportion of the radicals escape from the cage. Humphris and Scott26 have found considerable support for this mechanism with the catechol phosphite ester (IX) since changing from a

282


non-polar to a polar solvent, such as nitrobenzene, changes the nature of the products of eumene hydroperoxide decomposition from those which would be expected on the basis of a radical reaction (α-methyl sty rene and acetophenone) to those expected on the basis of a Lewis acid catalysed reaction (phenol and eumene). The Lewis acid is formed in a step subsequent to the above reaction.

i-Bu

f-Bu

(IX)

evidence has been obtained for the participation of tree radicals in the reaction between catechol phosphite esters and hydroperoxides not only in observation of pro-oxidant effects but also in enhanced rate of initiation of styrene by this system (Figure 11).

6r

02 03 0-4 [Cumene hydroperoxide]72,mole/l.

Figure 11. Polymerization of styrene at 75°C in air: O uninhibited; Δ with 0.01 mole/1. f-Bu

A> \ v

> - ° A O > - C H > i-Bu

283

G. SCOTT

The evidence for Pobedimskii's mechanic is much less unequivocal for the sulphides since in this case, the derived sulphoxide has also been found to be a powerful pro-oxidant with hydroperoxide12.

This system has been studied in some detail and it has been found that dimethylsulphinyldipropionate breaks down in a first order reaction to give methyl àcrylate. This reaction is reversible particularly in polar solvents (see Table 7) and the products are the expected thiolsulphonates with the formation of the theoretical amount of water and disproportionation products.

Table 7. Yield of methylacrylate from dimethylsulphinyl dipropionate (0.04 M) in the presence and absence of added methylacrylate (0.04 M) at 75°C in various solvent.

Solvent

Yield (mole litre

Carbon tetrachloride Dioxane Chlorobenzene

With methylacrylate 0.0375

Without methylacrylate 0.045 0.038 0.038

(MeOCOCH2CH2)2SO

O

^ MeOCOCH=CH 2 +

MeOCOCH2CH2SOH fcGÌG'

H 2 0 + MeOCOCH2CH2SSCH2CH2COOMe i

Disproportionation products

MeOCOCH2CH2SO + GH I o

II MeOCOCH2CH2SSCH2CH2COOMe

II o

i-Bu i-Bu

G" = 0 = / V c H - ^ V-O·

f-Bu f-Bu

In the presence of the stable free radical galvinoxyl no water is formed and the product is the thiolsulphonate which is more stable than the thiolsulphin-ate. The first order rate constant for the disappearance of galvinoxyl in carbon tetrachloride measured in an e.s.r. spectrometer is in good accord with that for methylacrylate formation (Table 8) considering the reversible

284


Table 8. First order rate constants for methylacrylate formation (/q) and galvinoxyl disappearance (kG) in the decomposition of dimethylsulphinyl dipropionate.

Temperature, °C

l O 5 ^ 105/cG

75

5.8

70

3.58 5.40

65

2.54 3.37

60

1.59

55

0.73 0.99

50

0.54 0.65

nature of this reaction. The activation energies of the two processes are 93 and 101 kJmol"1 respectively.

[Sulphoxidejx 10

Figure 12. Rate of polymerization of styrene in the presence of dimethylsulphinyl dipropionate at 75°C; A 0.01 mole 1"1 AZBN; B without initiator.

The sulphoxide appears to be an effective retarder for styrene polymerization when initiated by AZBN or purely thermally (see Figure 12). This is almost certainly due to the ability of the sulphenic acid to transfer a hydrogen to a growing polymer radical in competition with the chain propagation step.

It acts both as a retarder and as an activator in the hydroperoxide-initiated polymerization of styrene (see Figure 13) and which process super-

285

G. SCOTT

0 0.02 0.04 0.06 0.08 0.1 Sulphoxide or eumene hydroperoxide molarity

Figure 13. Polymerization of styrene in the presence of both sulphoxide and hydroperoxide in vacuo. A 0.055 mole Γ 1 dimethylsulphinyl dipropionate; B 0.038 mole Γ 1 a-cumene

hydroperoxide.

venes depends mainly on the ratio of hydroperoxide to sulphur compound. The competing reactions are :

P.

MeOCOCH2CH2SOH ROOH

I Monomer

Adduci

Dimer | Retardation

MeOCOCH2CH2SO' + PH

MeOCOCH2CH2SO + H 2 0 + O R IP. 1

o MeOCOCH2CH2SP

Retardation

Initiation

Confirmation for the reaction of sulphenic acid, formed by 5-centre elimination from sulphoxide, has been found in that in a solvent such as dioxane in which the dissociation is reversible, good first order kinetics for methylacrylate formation are obtained in the presence of galvinoxyl. styrene

286


and hydroperoxide, all of which react rapidly with the sulphenic acid. As might be expected, the first order rate constants are substantially the same (see Figure 14).

Time, h

Figure 14. Decomposition of dimethylsulphinyl dipropionate (0.04 M) in dioxane at 75°C. A No galvinoxyl : B 0.0815 mole 1 "l galvinoxyl ; C 0.09 mole 1 ~l eumene hydroperoxide.

Similar behaviour has been observed in autoxidation. In contrast to the findings of previous workers on the behaviour of sulphoxides in autoxidation. dimethylsulphinyl dipropionate is an effective inhibitor for the AZBN-initiated autoxidation of eumene (Figure 15). In the presence of hydroperoxide, however, a much less unequivocal result is found. At high hydroperoxide to sulphoxide ratios, a strong inhibition is found (see Figure 16). This must be due to a combination of chain-interruptive and peroxide-decomposing mechanisms working together since it has also been shown that the hydroperoxide concentration is rapidly reduced to zero under these conditions14. As the amount of sulphoxide is increased in the system, however, radicals are produced as described above and pro-oxidant effects are observed. This gave a satisfactory explanation for the fact that there is a concentration optimum for a thiodipropionate ester acting as a stabilizer in the processing of polypropylene1. It also explains why phenolic antioxidants appear to reduce an initial melt degradation which occurs with most sulphur compounds in technological media1 '9-13. It also seems likely that the reason for the well known auto-retarding curves observed in the autoxidation of

287

G. SCOTT

0 20 40 60 80 100 120 W Time,min

Figure 15 Effect of dimethylsulphinyl dipropionate (DMSD) on the oxidation of eumene initiated with AZBN (0.00042 M) at 75°C.

most sulphur-vulcanized rubbers is due to the formation and breakdown of sulphoxides to give redox active sulphenic acids. Further work is in progress to establish this.

Time, min

Figure 16. Eii'ect of dimethylsulphinyl dipropionate on the oxidation of eumene initiated by a-cumene hydroperoxide (0.05 mole Γ *). Temperature 75°C. Numbers on curves are the ratio

[eumene hydroperoxide]/[DMSD].

288


ACKNOWLEDGEMENTS I wish to acknowledge my indebtedness to my colleagues who have

participated in the studies described in this paper. In particular to Messrs M. A. Plant, C. Armstrong and K. J. Humphris and to Dr W. W. Wright and Mrs G. Knight for providing samples of the polyphenolic antioxidants.

REFERENCES 1 G. Scott and P. A. Shearn, J. Appi. Polym. Sci. 13, 1329 (1969). 2 J. E. Goodrich, Polym. Engng Sci. 10 (4), 215 (1970). 3 G. Scott, J. Plastics Inst. In press. 4 J. P. Forsman, S.P.E. Journal 20 (8), 729 (1964). 5 J. W. Tamblyn and G. C. Newland, J. Appi Polym. Sci. 9, 2251 (1965). 6 D. A. Jorden and E. C. Rothstein, Polym. Engng Sci. 6 (3), 231 (1966). 7 B. Wright, Plastics, 28, 111 (1963). 8 N. P. Neureiter and D. E. Bown, Industr. Engng Chem. Prod. Res. Develop. 1, 236 (1962). 9 G. Scott, Europ. Polym. J. (Suppl.), 189 (1969).

10 D. A. Gordon, Advanc. Chem. Ser. 85, 224 (1968). 11 M. A. Plant and G. Scott, Europ. Polym. J. In press. 12 G. Scott, Chem. Commun. 1572 (1968). 13 G. Scott, Mechanisms of Reactions of Sulfur Compounds, 4, 99 (1969). 14 J. D. Holdsworth, G. Scott and D. Williams, J. Chem. Soc. 4692 (1964). 15 C. Armstrong, M. A. Plant and G. Scott, unpublished work. 16 A. F. Lewis and J. K. Gillham, J. Appi. Polym. Sci. 6, 422 (1962); 7, 685 (1963); 7, 2293 (1963). 17 J. K. Gillham, American Chemical Society, Division of Polymer Chemistry, Preprint, 7, 513

(1966); Polym. Engng Sci. 7, 225 (1967). 18 Geigy ; Technical Pamphlet, Irganox 1076 (1966). 19 M. A. Plant and G. Scott, unpublished work. 20 L. N. Phillip, D. K. Thomas and W. W. Wright, Brit. Pat. Appn No. 900/65. 21 Monsanto Lab. Rep. LA25/2 (March 1966). 22 M. E. Cain, G. T. Knight, P. M. Lewis and B. Saville, Rubb. J. 150 (2), 10 (1968). 23 G. Scott, Brit. Polym. J. 3, 24 (1971). 24 P. I. Levin and T. A. Bulgakova, Polym. Sci. USSR, 6, 769 (1964). 25 J. E. Bonkowski, Text. Res. J. 243 (1969). 26 K. J. Humphris and G. Scott, unpublished work. 27 D. E. Pobedemskii and A. L. Buchachenko, Izvest. Akad. Nauk SSSR, 6, 1181 (1968). 28 C. Armstrong and G. Scott, J. Chem. Soc. In press. 29 D. G. Pobedemskii and A. L. Buchachenko, Izvest. Akad. Nauk SSSR, Chem. Ser., No. 12,

2750 (1968).

289

chemical transformations of polymers

Documents