400 JSDC NOVEMBER 1972; GRIFFITHS, MANNING AND RHODES

References I Peters and Stevens, BP 826,479; 856,381 (1956). 2 Karrholm and Lindberg, Text. Research J., 26 (1956) 528. 3 Peters and Stevens, Dyer, 115 (1956) 327; J.S.D.C., 72 (1956) 100. 4 Sidi and Zviak, ‘Probltmes Capillaires’ (Paris: Cauthier-Villars,

5 Steinhartt, Fugitt and Harris,J. Res. Nut.Bur. Stand.,26 (1941) 293. 6 Human and Speakman, J. Textile Inst., 45 (1954) T162. 7 Crank, ‘Mathematics of Diffusion’ (Oxford: Clarendon Press,

8 Medley and Andrew, Text. Research J., 30 (1960) 855. 9 Barrer, in ‘Diffusion in Polymers’, Ed. by Crank and Park (New

I0 Frisch, Wang and Kwei, J. Polymer Sci., 7 (1969) 879.

I 1 Medley, Trans. Furuduy SOC., 40 (1964) 1010. I2 Bell and Breuer, J. CoIloid und Interface Sci., in press. 13 Cockett, Rattee and Stevens, 4th Znternat. Wool Text. Research

14 Medley, Proc. 3rd Znternat. Wool Text. Research Conf., Paris, 3

15 Rosenbaum, J. Appl. Polymer Sci., 7 (1963) 1225. 16 Smith and Whitney, Text. Research J., 39 (1969) 392. 17 Taylor, private communication. 18 Lightfoot and Cussler, Chern. Engng Prog., (Symposium Series), 61

19 Cussler and Breuer, Nature, 235 (1972) 74; Amer. Inst. Chetn. Bzg.

20 Beal and Corbishley, J.S.D.C., 87 (Oct 1971) 329.

Conf., Berkeley, California (1970).

1966), p. 158. (1965) 117.

1956), p. 72 et seq.

York: Academic Press, 1968).

(1965) 66.

J., 18 (1972) 812.

The Prediction of Colour Change in Dye Equilibria. of Ortho-Hydroxyazobenzenes

11-Metal Complexes

J. GRIFFITHS, A. N. MANNING AND D. RHODES

Department of Colour Chemistry and Dyeing, The University, Leeds LS2 9JT

The visible speara of two series of ortho-hydroxyazobenzenes and their metal complexes have been studied and the effect of ring substituents has been evaluated. A linear relation exists between the values of the complexes and those of the parent dyes, enabling the colour change accompanying metallisation to be predicted with reasonable precision. The linear relations can be accounted for

qualitatively.

Introduction Prediction of the effect of substituents on the colour of dyes is of

considerable theoretical and practical importance. The most usual approach attempts to establish linear relations between A,,, ,values and the Hammett rs constants of substituents. although in most cases the correlations are poor because of the large changes in electron distribution in the excited state. This then leads to large uncertainties in the suitability of ground state 5 values for such correlations. In certain cases this difficulty can be avoided if the coloiir change (i.e. frequency shift) between the two equilibrium forms of a dye are plotted against 0, since for theoretical reasons ground-state 5 values are then appropriate (I, 2). For example, the frequency shifts between the azo and hydrazone tautomers of substituted 4-arylazo-1 -naphthols give a good linear correlation with 0 (2).

A particularly important equilibrium process in this field which might be amenable to such treatment is that between o-hydroxyazo dyes and their metal complexes. The colour change accompanying metallisation is made use of practically, and the prediction of this change could be of great value. Various studies of the absorption spectra of metal complexes of azo compounds have been made, although substituent effects have generally received only qualitative consideration. In a series of papers, Yagi (3) has examined the spectra of tervalent metal-ion complexes of various dihydroxyazo compounds, and has shown that substituent effects were not simply related to Hammett G

values, although general trends could be predicted. The systems studied were too complex for more than a general treatment to be made. To evaluate substituent effects and to attempt to predict these effects in simpler systems, we have considered the complexes of mono-substituted o-hydroxyazobenzenes. Two systems were selected, the first based on dyes (I) in which substituents are in the phenyl ring remote from the hydroxyl auxochrome. The second series is based on dyes (11) in which the substituents are in the same ring as the hydroxyl group. The copper, nickel and cobalt complexes of series I and the copper and nickel complexes of I1 have been examined.

Experimental Dyes I were prepared by coupling various diazotised m- and

p-substituted anilines to p-cresol. Dyes 11 were prepared similarly from diazotised aniline and p-substituted phenols. The dyes were purified by column chromatography and recrystallisation. The purity was confirmed by thin-layer chromatography.

(1 ) (11)

Copper and nickel complexes were prepared in the pure form by reaction of the dye with an excess of the metal acetate in hot aqueous ethanol, neutrality being maintained by the cautious addition of ammonium hydroxide. The precipitated complexes were purified by recrystallisation from pyridine.

Measurement of the spectra of the complexes prepared thus was unsatisfactory, since many of the complexes decomposed in the very dilute solutions required for spectral analysis. Decomposition occurred in all the solvents studied (ethanol, chloroform, methyl- ene chloride, dimethylformamide and pyridine) and was irreversible. Thus, for example, the spectrum of a 6 x 10 l-hi solution of the copper complex of I (R=H) recorded in 1 .O-mm cells in methylene chloride showed hmax =498 nm. In 5 .O-mm cells, a 1 a 2 x l O P 4 - ~ solution showed Amax=494 nm, and in IO.O-mm cells a 6 x 10-5-~ solution showed only an inflexion at approx. 490 nm. When the solution was diluted further, the spectrum was identical to that of the parent dye and did not show the typical long-wavelength band of the complex. The dilute solution was concentrated by evaporation at 30°C to 6 x 10-4~ , when the spectrum still indicated the presence of uncomplexed dye only. Thus the complexes are decomposed irreversibly on dilution, and this leads to unreliable hmax values. The various Am:L, values of such complexes quoted in the literature (4) must therefore be treated with reserve, unless this particular point has received attention.

PREDICTION OF COLOUR CHANGE IN DYE EQUILIBRIA-II 401

Decomposition can be prevented, even in very dilute solutions by the presence of a large excess of the complexing metal ion. Thus for accurate spectroscopic measurements the complexes were best prepared in situ by addition of a solution in pyridine of the metal acetate (copper, nickel or cobalt) to a ~ O + M solution of the dye in pyridine, until a final concentration of metal ion of approximately 5 x lo-, M was reached. The solutions were then measured in 1.O-mm cells. Under these conditions the long-wavelength maxima of the complexes showed no further bathochromic shift on increasing the concentration of the metal ion, and it was presumed that complete conversion to the com- plexes had occurred.

The spectra of the ions were obtained from solutions of the dyes (< IO-,-M) in pyridine containing potassium hydroxide (lO-l-~). The spectra of the dyes themselves were determined in approximately ~O-,-M solutions in pyridine; in both cases cells of 1 .O-mm path length were employed. All spectra were measured on a Unicam SP800 spectrophotometer.

Results and Discussion

DYES I AND COMPLEXES

Nine derivatives of I were prepared, with R-p-OMe, p-Me H, p-CI, p-COMe, m-CF,, m-CN, rn-NO, or p-NOa. This re- presents a wide range of Hammett a constants. The dyes are yellow to orange in pyridine solution, with a long-wavelength band in the region 390-420 nm. The copper, nickel and cobalt complexes show a bathochromic displacement of this band of 90-130 nm, the displacement depending on both the metal and the substituent in the aryl ring. The complexes are deep orange to magenta in solution. For comparison, the spectra of the anions of dyes I were recorded, and showed much larger bathochromic shifts of the visible band, viz. 130-215 nm.

The dependence of the hmax values of the dyes, complexes and anions on substituent a constants was examined, and, although there was a general trend for wavelength to increase with the electron-withdrawing power of the substituent, no useful cor- relation with a could be found. In Part I of this series (2), it was noted that, in several dye equilibria (e.g. azo-hydrazone tauto- merism or the protonation of aminoazobenzenes), the frequency shift Av between the absorption maxima of the equilibrating forms showed a better linear correlation with a than did the individual vmax values, and the theoretical reasons for this were discussed. The metal complexes of the o-hydroxyazobenzenes can also exist in equilibrium with the free dyes in solution, and thus a similar linear relation between a and Av (or the wavelength shift Ah) might be expected. However, this proved not to be the case, and it was found that the correlations between +s and Ah were not significantly better than those between Amax values and a. The failure of this approach, which has succeeded in several other instances, may be accounted for as follows.

The effects of substituents on the Amax of the dyes are very similar to those on the Amax of the corresponding metal complexes, i.e. the displacement of the visible band of the unsubstituted dye by the presence of a particular substituent is similar in magnitude to that observed for the metal complex. For example, with dyes I and their copper complexes, the magnitude of the wavelength shift [Amax (complex)-hrn,, (dye)] for nine derivatives falls within the narrow range 92-95 nm. Similarly, for seven dyes based on structure I1 the corresponding range is 88-95 nm. Thus the effect of substituents on A1 is in these instances very subtle, and will be largely obscured by the experimental errors in measuring these small differences. In addition, an important approximation inherent in this approach is that hmax values are used for com- puting the wavelength shift Ah. The theoretical development of the relation between Ah (or, strictly, Av) and a actually requires that the shift be calculated from the (O-+O) transitions of the two bands concerned ( I ) . The errors resulting from this approximation

are particularly important if the bands are not symmetrical, o r if their oscillator strengths are greatly different.

Although the experimental errors and this approximation may be insignificant for systems where Ah is strongly dependent on the substituent, in the present case it would be unreasonable to expect other than a purely fortuitous linear correlation between Ah and a. It is apparent that, to observe linear correlations in a particular dye equilibrium system, the shift should have a pronounced dependence on the substituent a value. This condition is met most satisfactorily when the electronic transition of the visible band of one form of the dye is in the opposite sense to that of the other equilibrium form of the dye. In such cases the two forms of the dye have opposite substituent effects, and the wavelength shift varies markedly with a, unless both species are particularly insensitive to substituent effects. An example of two equilibrium forms of a dye with opposite substituent effects on their visible absorption maxima are the azo and hydrazone tautomeric forms of the 4-arylazo-1-naphthols. In this case a good linear correlation between A v and ts is observed experimentally (2).

An alternative approach for prediction of the colour change of dyes I and I1 accompanying metallisation was sought, and it was found that plots of Amax of the complexes versus Amax of the parent dyes gave good linear correlations, as shown in Figure 1 for the copper and nickel complexes of dyes I.

These plots can be represented by the general equation

Amax (complex) = A Amax (dye) + B . . . .(1)

where A and B are constants. The Table lists the experimental values of A and B calculated by the method of least squares for the various systems considered. The correlation coefficients r are also listed, and indicate the precision with which the results fit the calculated linear relation. An r value of unity depicts a perfect fit, and as shown in the Table good to excellent linear correla- tions are observed.

I-- ~ - 9 I

A80 i--. 2 ~ ~ - I . - 1-

390 400 A10 A20

AmoxValuer of dyer I .nm

Nickel camplexer 0 Copper complexes

Figure I-A,,x Values of complexes versus dyes I [Number (substituent): I ( H ) , Z(p-Me), jl(p-MeO), 4(p-CI) , 5 (m-CF,), 6(m-CN), 7(m-NOZ),

8 (p-MeCO), 9 (p-NO,)] Numerical Constants and Correlation Coefficients for the Linear Relation

Amax (deriv.) = A Amax (dye) + B Dyes Derivatives A B Correlation coefficient *

I Cu complexes 1.05 73 0.992a I Ni complexes 1 . 3 5 -24 0.983" I Co complexes 1.28 - 20 0.997b I Anions 3.50 -850 0.959" I1 Cu complexes 0.89 135 0*993c I1 Ni complexes 0-89 155 0.99.Y I1 Anions 0.17 230 O-9fXc *Computed values based on nine (a), five (b) and seven (c) derivatives

(4

From these linear relations one can predict with reasonable precision the colour of a complex from the Amax value of its parent dye. For example, dye I(R=m-CN) absorbs at 404nm,

402 JSDC NOVEMBER 1972; GRIFFITHS, MANNING AND RHODES

and from Figure I the nickel complex is predicted to absorb at 521 nm. Synthesis of the complex revealed that in fact it absorbed at 519 nm. As indicated in the Table, a similar linear relation exists between the hmax values of the dyes and their anions, although the correlation is poorer.

The explanation for these linear relations probably lies in the close similarity of the long-wavelength transitions of the dyes, complexes and anions. In a dye of structure I, one can consider that the effect of the substituent R is to perturb the transition energy of the visible band of the unsubstituted dye (I; R =H) by an amount hvR,H (dye) frequency units. In the complex the analogous perturbation of the visible band relative to the unsubstituted complex will be AVR,H (complex). Provided that the electronic transitions are the same in nature for both the dyes and the complexes, and differ only in the amount of charge transferred in the transition, one might expect the relation:

Thus

or . . . .(2)

where k and K‘ are constants, and K‘= VH (complex)-kvH (dye). Eqn 2 suggests a linear relation between the absorption frequencies of the complexes and their dyes. Over the limited frequency range examined in this work, to a reasonable approximation a linear relation should also exist for the hmax values. Thus Eqn 1, which is observed experimentally, can be understood qualita- tively.

It is of interest at this stage to consider the general effect of substituents on the colour of the complexes. Ernsberger and Brode (4) were the first to consider the spectroscopic properties of metal complexes of simple o-hydroxyazo dyes, and they concluded that the long-wavelength band produced by metallisation arose from an electronic transition localised on the metal atom. In agreement with this, they suggested that this band was insensitive to substituents in the dye chromogen. It is now evident, however, that their interpretations were based on spectra that had been recorded under conditions where decomposition of the complexes was a major factor, and thus their Amax values are not reproducible. It is now accepted that the long-wavelength band of a complex corresponds to the visible transition of the parent dye chromophore which has been bathochromically shifted by the perturbing influence of the metal ion. The bands are thus sensitive to substituent effects, as a consideration of Figures I and 2 will show.

AVR,H (complex) cc. AVR,H (dye)

VR (complex)- VH (complex) = k y ~ (dye) - k y ~ (dye)

VR (complex)= k y ~ (dye) + K’

7 4 s

-_L-- -. I L---l--.

360 380 400 420

A,,,xValuer of dyes E,nm

9 Nickel complexes 1 ) Copper complexes

Figure 2--AmBX Values of complexes versus dyes II [Number (substituent): I (MeCO), 2 ( H ) , 3 (CZ), 4 (Me), 5 (Ph), 6 (NHCOMe), 7 (OMe)]

The long-wavelength band of the dyes I is most probably due to the ‘intramolecular charge transfer’ transition, in which there is a drift of electrons from the hydroxyl group across the azoben- zene chromophoric system. For qualitative considerations, the

crude resonance approach for interpreting substituent effects may be employed (5, 6). In this the resonance structures 111 and IV (R1=Me) are regarded as contributing to the excited state. Although having no mathematical justification, the simple resonance approach is often qualitatively correct, and in this case suggests that factors which stabilise the resonance forms I11 and IV (R’=Me) will produce a bathochromic shift of the long- wavelength band. In agreement with this, substituents in I produce increasing bathochromic shifts as their electron-withdrawing properties increase, owing to stabilisation of resonance structure IV. Electron-withdrawing groups exert a similar effect in the complexes and anions of dyes I, confirming that their visible electronic transitions also entail charge migration from the oxygen atom.

( V W

In the anions, the bathochromic shift relative to the dyes is particularly large, since the resonance form V corresponding to 1V is greatly stabilised by the absence of charge separation. In the metal complexes VI a partial negative charge is developed on the oxygen atom, and thus for similar reasons a bathochromic shift results when the hydroxyl group is combined with a metal ion. An additional factor present in the complexes is the decrease in charge density at the (3-nitrogen atom (see VI) of the azo group, owing to complexing with the nitrogen lone-pair electrons. This perturbation results in stabilisation of the resonance form corresponding to 111, and thus in a further bathochromic shift. This phenomenon is well illustrated with the dye I1 (R=H), which absorbs at 380 nm in neutral solution but at 458 nm in concentrated sulphuric acid. The monoprotonated form VIT is present in the latter case.

Thus the shift of the visible band on metallisation is due to two simultaneous perturbations of the electron system, namely an increase in the charge density at the oxygen atom and a decrease at the (3-nitrogen atom of the azo group. A similar qualitative explanation of the bathochromic effect of the metal ion can be obtained if one considers the dye as a perturbed odd-alternant system (6).

The Table shows that the largest shifts are observed in the anions, which is to be expected in view of the stability of V. Of the metal complexes, nickel gives the largest colour change, whereas copper and cobalt have a smaller effect. It is interesting to note that nickel has a lower ionisation potential than copper and cobalt, and thus should provide a slightly higher charge density at the oxygen atom.

QUANTUM EFFICIENCY MEASUREMENTS OF FADING OF DISPERSE DYES 403

DYES I1 AND COMPLEXES

Seven dyes of structure I1 (R=OMe, Me, H, NHCOMe, Ph, C1 and COMe) were synthesised, and the spectra of the dyes, their copper and nickel complexes, and their anions were recorded in pyridine solution. Thedyes aresimilar incolour to series I, although complexes are generally redder in pyridine solution than those of I, having hmax values in the region 452-532 nm. The graphical plots of the hmsx values of the complexes versus those of the dyes are shown in Figure 2, and again reasonably linear correlations are found. The constants appropriate to Eqn 1 are listed in the Table, and include the corresponding relation for the anions of 11.

Figure 2 shows that, as in the previous series, nickel produces a more pronounced colour change than copper, although the maximum shifts are found in the anions of the dyes. The linear relations can again be used to predict the colour of complexes, as illustrated by the dye I1 (R=NH,), which has hmax=488 nm. The copper complex as prepared in situ is brilliant blue, which is unusual in this series, but is far too unstable for its Amax value to be recorded. The equation for the copper complexes gives the calculated hmax value of 569 nm, which is in general agreement with the observed colour.

The effect of substituents on the spectra of the dyes and complexes can again most easily be explained in terms of simple resonance theory. The resonance forms I11 and IV (R*=H) will be de-stabilised when R1 is electron-withdrawing and stabilised when R1 is electron-donating, and as a result electron-donating substituents exert a bathochromic effect. This is the reverse of the substituent effect in dyes of type I. The most pronounced shifts are thus observed with the copper and nickel complexes of I1 (R=NHa), which are blue in pyridine solution. However, consideration of Figure 2 shows that the wavelength shift is not closely related to the substituent Q constants. The observed order of substituents suggests that the transitions are far more sensitive to the bathochromic influence of the electron-donating mesomeric effect than the hypsochromic electron-withdrawing inductive effect.

Conclusions The hmx values of dyes I and I1 and their copper and nickel

complexes do not show usefullylinear relationswith the ts constants of substituents. In addition, the correlation between [Amax (complex) -Amax (dye)], (Ah) and cis also poor, which is at variance with results obtained for other dye equilibria. The failure ofAA to correlate with c may be attributed to the insensitivity of Ah to substituent effects. The small variation in Ah is difficult to measure sufficiently accurately, and approximations in the method of estimating Ah for absorption spectra render correlations extremely unreliable in these particular systems.

In the simple systems studied, the h m a values of the copper and nickel complexes of dyes I and I1 can be predicted from a know- ledge of the hmax values of the parent dyes, using the relations indicated in the Table. In both types of complex, nickel exerts the more pronounced bathochromic effect. In dyes and complexes of type I electron-withdrawing groups exert a bathochromic effect, whereas in dyes of type I1 the reverse phenomenon is observed. The shifts, however, are not simply related to the Hammett c constants of substituents. The linear relations can be understood qualitatively on the assumption that the visible electronic transitions of the dyes, complexes and anions are very similar in character, and differ only in the degree of charge transfer in the excited state.

(MS. received 4 February 1972; revised MS. received 7 July 1972)

References I Weller, in 'Progress in Reaction Kinetics', Ed. by Porter (London:

2 Griffiths, J.S.D.C., 88 (Mar 1972) 106. 3 Yagi, Bull. Chem. SOC. Japan, 36 (1963) 500,506,512; 37 (1964) 1878,

4 Ernsberger and Brode, J. Org. Chern., 6 (1941) 331. 5 Maocoll, Quarterly Reviews, 1 (1947) 16. 6 Dewar, Chemical Society Special Publication No. 4, (1956) 64.

Pergamon, 1961), p. 187.

1881.

Quantum Efficiency Measurements of Fading of some Disperse Dyes in Nylon and Polyester Films and in Solution

C. H. GILES, B. J. HOJIWALA AND C. D. SHAH

T. Graham Young Laboratory, The University of Strathclyde, Glasgow CI

Several anthaquinone and azo disperse dyes adsorbed on films of polyester and nylon 6.6 and dissolved in a mixture of acetone and water have been faded by various lines in the visiblespectrum, and the quantum yields (v) determined by means of the Parker-Hatchard potassium ferrioxalate actinometer. 'p for adsorbed dye lies between ca lo-* and decreasing with increase in exposure time, probably owing to the presence of associated dye, and with increase in wavelength of illumination. Values for dye in solution are 20-60 times as high as those for adsorbed dye. The low ,values for adsorbed dye are the .result of a combination of several factors, including association of dye, retardation of gaseous diffusion by the substrate, and the shield or cage effect of the surrounding polymer

substrate.

Introduction Although many reports of the photochemistry of vapour and

liquid phases have been published, relatively little is known of the photochemistry of solid phases, apart from that of the silver halide photographic process. Calvert and Pitts (I) attribute this apparent lack of interest in solid states to various experi- mental difficulties, including the very high light absorption of solids in their photoactive regions. However, absorption is not a problem with dyes, which can be used adsorbed in transparent film substrates. Even so, little is known of the fundamental

mechanisms of the fading of dyes, despite its intrinsic interest and high technical importance. In particular, the quantum efficiency 'p of the fading of dyes in textiles has not hitherto been determined.

The relevant 'p data that we have traced for other related systems, i.e. aromatic compounds, including dyes in solution and in unimolecular layers, and the photolysis of the solid sub- strates themselves, are given in Table 1 and are grouped ar- bitrarily into those of high (ca 10-l to > loo) and low (< lo-*) 'p values.