J . Phys. Chem. 1985,89, 4565-4569 4565

scattering spectra were obtained on the filter difference spec- trometer (FDS) at the Los Alamos pulsed neutron Source. The FDS23 is an inverted geometry time-of-flight spectrometer, which utilizes the polycrystalline cutoffs of beryllium and beryllium oxide to select a band of final energies from the scattered neutron beam. A “white” neutron beam is incident on the sample, and the total time-of-flight for each event in the detectors is recorded. Sub-

(23) A. D. Taylor, E. J. Wood, J. A. Goldstone, and J. Eckert, Nucl. Inst.

(24) J. E. Stewart, J . Chem. Phys., 26, 248 (1957). Methods, 221, 408 (1984).

traction of the fixed final flight time determined by the band-pass filters then determines the energy transfer to the sample.

Acknowledgment. We are grateful to Professor E. W. J. Mitchell, of the Department of Physics, Clarendon Laboratory, University of Oxford, who kindly provided the single crystal of urea used in the single-crystal Raman studies. Work at Los Alamos was supported in part by the Division of Basic Energy Sciences, U S . Department of Energy.

Registry No. Neutron, 12586-31-1; urea, 57-13-6.

Spectroscopic Studies of Chemically Liberated “Free” -OH Groups in Aqueous N2H4,

NH,, and CH,NH2 Solutions

C. Austen Angel]* and Dana L. Fields

Department of Chemistry, Purdue University, West Lafayette, Indiana 47907 (Received: March 25, 1985)

As a contribution to understanding and identifying the nature of the disputed “broken” hydrogen bond in aqueous solutions we have sought to compare the spectroscopic character of “free” -OH groups produced by chemical means at constant temperature with that of the putative “free” -OH produced in pure water alcohols, etc. by temperature increases. “Free”, or at least very weakly bonded, -OH groups have been produced quasi-stoichiometrically at room temperature by displacing bound -OH groups from OH-0 bonds previously free with previously free -NH groups, the competitive status of which has been abruptly increased by protonation of the amine group. The sharp overtone bands of the unbonded -NH groups necessarily present in amine + H 2 0 solutions (due to the excess of protons over lone pairs) disappear stoichiometrically on protonation using HC104, and an equivalent number of weakly bonding -OH groups are produced. The spectroscopic signature of the chemically liberated -OH group is almost indistinguishable from that of the group liberated thermally, and a well-defined isosbestic point at 1442 nm (close to the approximate isosbestic point at 1440-1450 nm in the heated water spectra) implies that the same mechanism of “exciting across the centroid” (a (strong bond) - (weak bond) exchange within a continuum model) is involved in each case.

Introduction There has been much controversy in the literature on water and

other hydrogen-bonded systems concerning the appropriateness of the concept of “breaking” of hydrogen bonds. This is widely assumed by solution chemists, particularly in biological circles, to be a valid description of processes occurring in aqueous systems during composition and temperature changes. By contrast, all computer simulation studies on aqueous and alcoholic systems, irrespective of the form of the pair potential employed, have the common feature of denying the reality of any simple two-state “on”-“off“ picture of hydrogen bonds in these media. Instead of continuous and monotonic distribution of hydrogen bond energies is found.] The force of these latter studies, even allowing for the fact that most have not included non-pairwise-additive terms in the potential function, is sufficient to virtually rule out the simple two-state models of hydrogen-bonded systems (which, in any case, are incapable of dealing simultaneously with the observed responses to pressure and temperature variations2$).

One of the most plausible arguments for two-state descriptions of hydrogen bonding has been the existence of isosbestic points in the spectra of H-bonded systems undergoing temperature or composition variation^.^-^ These do not, however, withstand

(1) (a) W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey, and M. L. Klein, J. Chem. Phys., 79, 926 (1983); (b) D. L. Beveridge, M. Mezei, F. T. Marchese, G. Ravishan, T. Vasau, and S. Swaminat, Adv. Chem. Sec. 16, 204, 297 (1983).

(2) W. Kauzmann in “Water and Biological Systems”, Editions CNRS, Paris, 1976, p 37.

(3) C. A. Angell, J . Phys. Chem., 75, 3698 (1971). (4) W. A. P. Luck and W. Ditter, 2. Nuturforsch. E. 24, 482 (1969). (5) W. A. P. Luck and W. Ditter, J . Phys. Chem., 74, 3687 (1970). (6) (a) G. E. Walrafen, J. Chem. Phys., 52, 4176 (1970); (b) G. E.

Walrafen, “Water, a Comprehensive Treatise”, F. Franks, Ed., Plenum, New York, 1971, Chapter 6, and references therein.

detailed analysis insofar as the spectra cannot be described by the sums of two Gaussian components of varying intensities. Rather three or more are r e q ~ i r e d , ~ , ~ and a major one of these, as first shown by Luck and Ditter,s proves almost invariant under temperature change. A related phenomenon is seen in computer simulation studies, of both aqueous1’ and alcoholicI2 systems. In these, the population of hydrogen bonds of intermediate energy VH = 3.0 kcal/mol remains invariant, while populations of weaker and stronger bonds change. This led Stillinger and Rahmanll to the description “exciting across the centroid”, and Angell and Rodgers9 recently showed that the behavior of the overtone IR spectra of normal and supercooled water was consistent with this concept. The latter authors showed by short extrapolation how the overtone spectrum of water should appear when the weak bond population is minimized by vitrification. A broad band of fre- quencies extending from 6200 to 7200 cm-I (1 390 to 1610 nm) is identified out of which grows a relatively narrow band of frequencies a t the short wavelength edge 6900-7200 cm-’ (1390-1450 nm) as the system is excited.

The “weak-bond”-“strong bond” exchange, which can be sustained within the continuum model of water, thus seems a viable replacement for the evidently untenable two-state models. The present contribution is intended to strengthen this view by showing how the same relatively narrow band of “weak bond” frequencies can be grown systematically a t the expense of a narrow band of “free” -NH frequencies as a result of a titration procedure which

(7) W. C. MacCabe, S. Subramanian, and H. F. Fisher, J . Phys. Chem.,

(8) G. R. Chopin and K. Buijs, J. Chem. Phys., 39, 2042 (1963). (9) C. A. Angell and V. Rodgers, J . Chem. Phys., 80, 6245 (1984). (10) C. A. Angell, Annu. Reo. Phys. Chem., 34, 593 (1983). (11) F. H. Stillinger and A. Rahman, J . Chem. Phys., 57, 1281 (1972). (12) W. L. Jorgensen, J. Am. Chem. SOC., 103, 335 (1981).

74, 4360 (1970).

0022-3654/85/2089-4565$01.50/0 0 1985 American Chemical Society

4566 The Journal of Physical Chemistry, Vol. 89, No. 21, 1985 Angel1 and Fields

FREQUENCY (cm”)

7400 7000 6600 6200 I ”

1 -

1300 1360 1400 1440 1480 1520 1560 1600 1640

WAVELENGTH (nm) Figure 1. Spectra at 23 O C of aqueous solutions of hydrazine, ammonia, and methylamine, containing approximately equal concentrations of -NH groups, and for pure N2H4 (after scaling as indicated).

“activates” the -NH groups into a competitive status vis-a-vis -OH groups. In this condition, it will be argued, they replace -OH.-0 hydrogen-bonding interactions with the liberation of an equivalent number of -OH weak bonds. The spectral signature of these chemically induced weak bonds can then be compared with that of the thermally induced weak bonds.

The key to this study is that, in hydrazine, N2H4, and other ammonia-related molecules, there are more protons which are sufficiently acid to participate in hydrogen bonding than there are lone pairs to combine with them. This means that no matter how low the temperature, or efficient the packing, there must always be a population of -NH groups which do not participate in strong hydrogen bonding. This contrasts with the case of water in which the number of protons and lone pairs is the same. In a solution of water + amine-containing molecules, the nonbonded -NH groups will remain nonbonded because the protons attached to nitrogen are less acid than the H 2 0 protons. Their population may be monitored by the sharp -NH overtone band falling at or about 1520-1560 nm, (see Figure 1). If a proton is introduced into the solution, however, the situation changes. Since the amine (which is essentially unprotonated by water, in view of Kb - 10-4-10”) is the most basic group, the proton is accepted onto the nitrogen and the positive charge is distributed across the other protons, effectively activating them into a state in which they are now competitive with the -OH protons for lone pairs on either nitrogen or oxygen. Since the total number of lone pairs is conserved during proton additions, there may result a replacement of free -NH oscillators by “free” or at least weakly bonded -OH protons. Of course the proton cannot be introduced alone and the anion accompanying it will usually provide additional lone pairs. However, the stronger the acid the less attractive are these sites to the ”liberated” -OH groups, hence the higher the resultant “free” -OH frequency.

The number of protons which can be activated in this manner for each proton added to the solution depends on the particular

amine being titrated. In the case of methylamine (Kb = 4.4 X lo4) the addition of one proton results in the presence of three active protons as CH3NH3+, which could then attack existing -OH--0 bonds and displace three -OH’S from H-bonding in- teractions; hence, one proton can produce up to three “free” O H groups. In the case of ammonia, (& = 1.8 X 10-7, on the other hand, the addition of one proton results in the presence of four active protons, while in a solution of hydrazine, the protonated species is N2H5+ which would imply five active protons except for the fact that the proton will be localized on one of the two nitrogens. In any case, the study of the spectroscopic consequences of protonation of aqueous solutions of CH3NH2, NH3, and N2H4 offers the possibility of interesting insights into the spectroscopic signature of weak hydrogen bonds. The present paper reports the results of such studies.

Experimental Section Materials. CH3NH2 was obtained from Aldrich Chemical Co.

as an aqueous solution containing 40.0 wt % CH3NH2. N H 3 was obtained in the form of aqueous ammonium hy-

droxide, provided by Fisher Scientific. Comparison of density data with literature values listed in the CRC Handbood of Physics and Chemistry determined the solutions to be 23.69 wt % NH3.

N2H4 was purchased from Eastman Kodak. The lot assay concluded that this simple was 96.2 wt 7% N2H4.

HClO, was obtained from Fisher Scientific. Analysis by standard base titration showed it to contain 70.4 wt % HClO, (thus 24.2 M HClO,, 30.4 mol % HCIO, or --HC1O4.2H2O).

Spectroscopy. Solutions, made up volumetrically by using concentrated HC104, were added to standard photometric cells containing a spacer such that the path length was 0.5 mm except in the case of pure N2H4 in which a 0.3 mm spacer was used. Near-IR spectra in the frequency range 1300-1650 nm were obtained with a Cary-l7DHC spectrophotometer. All results reported in this paper were taken at room temperature, 23 O C .

Spectroscopic Studies of “Free” -OH The Journal of Physical Chemistry, Vol. 89, No. 21, 1985 4567

0 7 t

0 6 -

W 05 0 z

E 0 m m Q 0 3 -

-

0 4 -

0 2 -

0.648 0.757 0.847

,- 0.997

1538nm 1.99’ 01 -

00 I I 1 I

1300 1400 1500 1600

WAVELENGTH (nm) Figure 2. Spectra of aqueous solution of methylamine (initial concentration 5.8 M) during titration by HC104, showing the appearance of band at “free” -OH frequency coinciding with the disappearance of band at free -NH frequency. Note that the isosbestic point at 1442 nm is maintained up to the equivalence point.

Results In Figure 1 we show the spectra of solutions of CH3NH2, NH,,

and N2H4 in water a t concentrations such that the population of -NH groups is about the same, - 12 M. The features of interest are the sharp -NH overtone bands a t -1520-1560 nm.

In Figure 2 we show the spectral consequences of addition of HC104 to a 5.8 M solution of CH3NH2 (1 1.6 M in -NH bonds). The “free” -NH peaks (combined half-width = 120 an-’) are seen to diminish and, in fact, disappear shortly before the equivalence point has been reached. At the same time an increase in intensity of a band of only slightly greater half-width, 130 cm-’, centered about 1420 nm is observed. A precise isosbestic point is maintained up to the equivalence point.

The variations in the absorbances at the two -NH overtone frequencies, and at the “free” -OH frequency a t 1425 nm con- sequent on protonation, are shown in Figure 3. Here we plot the difference in absorbance of the initial CH3NH2 aqueous solution and that of each solution containing HC104, against the con- centration ratio [H+]/[CH,NH,]. These differences are seen to vary systematically, and in opposite directions at -NH and -OH frequencies, until a stoichiometric quantity of protons has been added, and thereafter assume a weak composition dependence consequent on the volume dilution, ( V , of C10; = 56 cm3 mol4 vs. H 2 0 , 18 cm3 mol-’).

To confirm that these differences, particularly in the “free” -OH region, are chemically induced rather than simply the consequences of addition of the perchlorate anion, we show a series of spectra of HC104 and NaC104 solutions in water in Figure 4. The increase in spectral intensity a t the “free” -OH frequency is compared with that of the methylamine- (or hydrazine-) containing solution in Figure 5 by using a simple [C104-] concentration axis. The [C104-] values when [H+]/[-NH2] = 1.0 are shown by arrows. [A simple scaling of [ClO;] and initial amine concen- tration is not expected.] The contrast is most marked for the case of methylamine in which the -NH protons should be the most competitive after protonation.

Titration spectra for NH, and NzH4 solutions similar to that shown for CH3NH2 in Figure 2 are summarized in Figure 6 . Note that in the N2H4 case intensity remains at the 1539-nm “free”

020 OK) 3 010

0 00

a Q -010

-020

-030

00 02 04 06 08 10 12 14 16 18 20

[Ht l /[CH3NH21

Figure 3. Variation of absorbance at “free” -NH and “free” -OH wavelengths during titration of aqueous solution of CH,NH, with HC10,. Note “breaks” in plots at the equivalence point.

-NH wavelength far beyond the equivalence point, showing some -NH protons remain uncompetitive. These are presumably protons on the second (unprotonated) nitrogen.

Discussion Figures 3 and 5 taken together make clear that there is a

reciprocal relationship between the existence of the unequivocal ”free” -NH overtone band and the somewhat broader band at 1425 nm. The origin of the latter can be simply accounted for as a result of the two sequential processes:

H+ + -NHz -+ -N(H6+)3 (1)

NH6+ + -OH--0 - -NH6+.-0 + -OH (2)

the validity of which is supported most directly by the breaks in the composition dependence of each band at the H+/amine equivalence point (Figures 3 and 5 ) . We might expect the equilibrium constant for process 2 to depend on the value of 6+,

4568 The Journal of Physical Chemistry, Vol. 89, No. 21, 1985 Angel1 and Fields

-

FREQUENCY (cm-’ O I 6 1 I 1 6600 ,

I ’ -7 T --P-

7400 7000 ----- 7

( a i 07 -

0 6 -

05 - W 0 z 2 0 4 -

0 a v,

0 3 -

0 2 -

HCI

1~ 1563

- LAP -1 ~- L - -1 1400 1500

00:- ’ ’ ’ ’

WAVELENGTH ( n m 1

FREQUENCY (cm-’1 6600 , T -1 0 7 ~ ~ - 7400 ~ - - . ~ 7 0 0 0 _ - -

fb l \ ,

No C 06 - 0 5 -

W 0 5 0 4 -

c r l 0 v) 0 3 -

m

m a 0 2 -

Figure 4. (a) Spectra of aqueous solutions of HCIO4 of various con- centrations in the absence of amine molecules. (b) Spectra of aqueous solutions of NaC10, of various concentrations.

hence to be largest for methylamine solutions and smallest for hydrazine at equal -NH concentrations. For a given amine we might further expect the equilibrium to lie further to the right when the amine concentration is higher. Figure 7, which sum- marizes results for the “free” -OH intensity increases for all the titrations of this study, shows that these expectations are largely borne out.

It should be recognized at this point that the presence of C104- alone leads to the generation of considerable intensity increases at 1425 nm, and that increases due to C104- additions also occur in association with an isosbestic point. This behavior has been studied in detail by Walrafed using Raman spectroscopy and McCabe et al.’ in the near-IR region of this study and is clearly due to the circumstance that those water molecules which are obliged to be the near neighbors of the dissolved C104- species cannot form as strong hydrogen bonds with the anion oxygens as they do with each other in its absence. The 1425-nm intensity due to weak bonds formed quantitatively as a consequence of process 2 above should, therefore, be assessed as the difference between the HC104-titrated solution and untitrated solution in-

014 1 I

012

t E 0‘0 t e .

P ,

MOLARITY [CI O;]

Figure 5. Change of absorbance at the “free” - O H wavelength, 1425 nm, from initial value, due to additions of HCIOl or NaC104 to water or amine solutions. Arrows mark equivalence points in those cases where titrations are involved.

FREQUENCY (cm”)

7000 -- - 66pO 7- 7400 -7-- 7 -

( a i 07 -

06 -

05 - W V z Q: 04-

LL

8 0 3 c

U

m

0 2

FREQUENCY (cm-’) 6200

r- -- 7400 70p0 6690 r - r --

( b ) 07 -

0 6 + 1442 I

05-

0 2 -

01 - 1566 w

1500 1600 00

1400

WAVELENGTH (nm) Figure 6. (a) Spectra of ammonia solution in water, initially 6.36 M, during titration with HCIO4. (b) Spectra of hydrazine solution in water, initially 3.02 M, during titration by 11.56 M HCIO4.

tensities a t the same C104- concentration. This is done in the inset of Figure 7 by subtracting the NaC104 solution spectral intensity

J. Phys. Chem. 1985,89, 4569-4574 4569

0 18

016

014

012

E 010

s 008

2 006

In (u

c 0

004

a02

aoy a2 04 06 ae 10

[H’l/[Xl Figure 7. Change of absorbance at 1425 nm from initial value, due to titration of different amine solutions by HC104, as a function of pro- ton:amine concentration ratio. For equal initial concentrations of -NH bonds, the buildup of ”free” -OH intensity is greatest for methylamine. Insert shows difference between AA in Figure 7 and corresponding AA for NaC104 solution, to normalize for the effect of perchlorate anions alone in disrupting the water structure.

at 1425 nm in Figure 4 (extrapolated where necessary) from the titrated solution intensities in the same figure. The value of AA at the equivalence point for the CH3NH2 solution, 0.085, is a little less than the value, 0.048 (1 1.616.0) = 0.093, expected from the N2H4 equivalence point value if only two of the four protons were activated by the first protonation, Le., the charge on one nitrogen weakly activates the protons on the other.

To add conviction to the notion that very weak essentially “broken” hydrogen bonds have their overtone intensity a t or near 1425 nm we make brief reference to the following study which will be the subject of a separate arti~1e.I~ When the protonation of the amine is performed with HCl rather than HC104 a less

(13) D. L. Fields and C. A. Angell, to be published.

well-defined isosbectic point is found at a longer wavelength 1490 nm, and there is a buildup of intensity at 1430-1450 nm-a broader band than in the case of HC104 protonation but still a well-defined one. Clearly the -OH groups released from OH--0 bonds in this titration prefer to form weak bonds with the C1- than to remain “free”. (Lone pairs about C1- can be invoked as the “trap” sites for the free -OH groups for illustrative purposes, although the electron density about the isolated CI- ion is actually spherically symmetric.)

In the case of the HClO, titration, the distinction between the -OH--OC103 interaction and the weak interaction between dis- placed -OH and already fully bonded H 2 0 molecules (e.g., the weak -OH bonds in hot water) is difficult to make. The inter- actions are almost equally weak and give rise to equally high- frequency -OH overtone frequencies. The isosbestic point at 1442 nm in Figure 2 is to be compared with the range, 1440-1450 nm, of the smeared isosbestic point for the pure water spectrum in the temperature interval 0-80 0C.499

Conclusion The technique we have described provides a relatively unam-

biguous way of producing and spectroscopically characterizing weak hydrogen bonds produced by a process of “breaking” or displacing initially intact well-formed hydrogen bonds without temperature change. Observing a close similarity with the spectroscopic consequences of heating water we suggest that the concept of (strong bond) - (weak bond) exchange for the tem- perature-induced disruption of the water structure is a sound one.

Acknowledgment. We are indebted to the Office of Naval Research, Grant No. N0014-78-C-0035, under whose auspices this work was initiated and to the National Science Foundation, Chemistry Grant No. CHE8318416, for support during its con- clusion. This work commenced with an attempt to titrate excess -NH protons with excess Hz02 lone pairs (at low temperatures). It was E.I. Cooper, now at IBM Research Laboratories, Yorktown Heights, NY, who pointed out that it would be necessary to protonate the amines to make them sufficiently acid to combine with the lone pairs in question. Out of the discussions which followed, the present project developed.

Registry No. N,H4, 302-01-2; NH40H, 1336-21-6; CH3NH2, 74- 89-5 .

Nonexponential Hole Burning in Organic Glasses

R. Jankowiak, R. Richert, and H. Bawler* Fachbereich Physikalische Chemie, Philipps- Universitat, 0-3550 Marburg, FRG (Received: March 26, 1985)

The time evolution of nonphotochemical holes in the absorption profile of tetracene doped into amorphous layers of 2,3- dimethylanthracene and 9,lO-diphenylanthracene has been investigated and compared with literature results on tetracene in an alcohol glass. The kinetics can be described in terms of a dispersive first-order reaction of noninteracting reaction centers assuming that site relaxation is a tunnel process, the tunnel parameter X being subject to a Gaussian distribution. Data for hole burning rates have been evaluated via model fits. Successful application of the concept to analysis of literature data for the recovery of photochemical holes suggests that a Gaussian distribution function for the rate-controlling parameter is superior to the conventionally constant distribution function with cutoff condition.

Introduction A distinguishing feature of a glass is the local fluctuation of

the potential energy. It reflects the disorder frozen in during quenching a melt or condensation of a vapor a t a cold substrate. Since the system is neither at minimum energy nor a t maximum entropy it is subject to a driving force which tends to establish thermodynamic equilibrium. Although macroscopic relaxation is often too slow to be observable on realistic time scales. mi-

0022-365418512089-4569$01.50/0

croscopic relaxation processes can be probed by various experi- ments. The example this paper deals with is the local motion of atoms or molecules or clusters thereof within two adjacent potential minima, usually referred to as two-level systems (TLS).1-3 They

(1) Anderson, P. W.; Halperin, B. I.; Varma, C. M. Philos. Mag. 1972,

(2) Phillips, W. A. J . Low Temp. Phys. 1972, 7, 351. 25, 1.

0 1985 American Chemical Society