894
JACK N. FINCH AND ELLISR. LIPPINCOTT
VOl. 61
HYDROGEN BOND SYSTEMS-TEMPERATURE DEPENDENCE OF OH FREQUENCY SHIFTS AND OH BAND INTENSITIES BYJACK N. FINCH' AND ELLIS R. LIPPINCOTT Department of Chemistry, Kansas State College, Manhattan, Kansas Department of Chemistry, University of Maryland, College Park, Maryland Received November 1 . 1066
The temperature dependence of OH stretching hydrogen bond frequencies and apparent integrated band intensities has been studied in the range 30 to -80" for methyl, ethyl, isopropyl, n-butyl, isoamyl and n-hexyl alcohols, phenol, ochlorophenol and 2-methyl-4-methoxy-2- entanol. Studies were made on the pure liquids and in suitable solvents. The variation of OH hydrogen bonded gequencies and ap arent integrated band intensities with concentration was also studied for a number of these compounds. For alcohoi capable of intermolecular hydrogen bonding only, the decrease in OH stretching frequency and increase in intensity was directly proportional to the absolute temperature. The experimental data have been interpreted in terms of a potential function model of hydrogen bonding. The explanation for the shifted OH stretching frequencies with temperature in alcohols is based on a Boltzmann distribution of hydrogen bond energies, resulting from excitation of the 0--0 mode of vibration to higher energy levels.
Introduction
and intensity of the bonded OH frequency with A considerable number of studies have been changes in temperature, we have conducted a made of OH- - -0 hydrogen bond systems correlat- number of experimental studies to check predicing OH frequency shifts and OH band intensities tions from the model. The position and apparent with the strength of the hydrogen b ~ n d . ~ I-n~ integrated intensity of the bonded OH frequency general temperature dependence studies have been in several of the lower molecular weight aliphatic made to determine the temperature dependence of alcohols, phenol, 2-methyl-4-methoxy-2-pentanol the equilibrium constant of the monomer-dimer and o-cholorophenol have been determined a t suitpolymer equilibria. Considerations from a one- able intervals from 30 to -80" with examples, 2dimensional model of OH- - -0 hydrogen bonding methyl-4-methoxy-2-pentanol and isopropyl alwhich has quantitatively correlated and predicted cohol, above room temperature. Similar informaa number of hydrogen bond propertieslo suggest tion was obtained for isopropyl alcohol dissolved in that the shifts in frequency and changes in inten- carbon tetrachloride, diethyl ether and triethylsity of OH bands cannot be explained solely on the amine as a function of concentration. A prelimbasis of a shift in monomer-dimer-polymer equi- inary report of this has been published.l* Experimental librium. Other inve~tigatorsll-'~have observed The infrared spectra were obtained with a Perkin-Elmer such changes for large intervals in temperature, 12C single beam, double pass, automatic recording but only recently have any systematic studies model spectrophotometer. The spectra, recorded from 2500 to been made on the temperature dependence of 4000 em.-', were obtained with a LiF prism while using bonds in pure liquid^'^^'^ or upon compounds exhibit- NaCl windows. A low temperature cell similar to one described by Powling and Bernstein'Q but using silver chloride ing OH- - -0 hydrogen bonding. 16*17 windows was found adequate to obtain spectra a t suitable Since the one-dimensional model'" of OH- - -0 intervals from 30 to -80". Spectra were also obtained to hydrogen bonding predicted changes in position +65" with this same cell by first heating the sample holder (1) A dissertation presented in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Chemistry a t Kansas State College. (2) Several reviews on hydrogen bonding have been published: M. Davies, Chem. SOC.Ann. Rep., 48, 1 (1946); L. Hunter, ibid., 48, 141 (1946); L. Kellner, Rept. Prog. Phys., 16, 1 (1952). (3) R. E. Rundle and M. Parasol, J . Chem. Phys., 20, 1487 (1952). (4) R. C. Lord and R. E. Merrifield, ibid., 21, 166 (1953). (5) G. M. Barrow, THISJOURNAL, 69, 1129 (1955). (6) V. I. Malyshew, Bull. Acad. Sci. U.R.S.S., Ser. phys., 9, 198 (1945). (7) K. Nakamoto, M. Margoshes and R. E. Rundle, J . A m . Chem. SOC.,7 7 , 6480 (1955). (8) G. C. Pimentel and C. H. Sederholm, J . Chem. Phys., 24, 639 (1956). (9) A . F. Stuart and G. B. B. M. Sutherland, ibid., 24, 559 (1956). (IO) E. R. Lippincott and R. Schroeder, ibid., 23, 1099 (1955); J . Phys. Chem., 61, 921 (1957). (11) P. C. Cross, J. Burnham and P. A. Leighton, J . A m . Chem. Soc., 69, 1134 (1937). (12) J. Errera, G. Gasport and H. Sack, J . Chem. Phys., 8 , 63 (1940). (13) V. N. Nikitin and N. G. Yarvalashii, DokEady Acad. Nauk, U.B.S.R., 77, 1015 (1951). (14) I. A. Yakovlev, Bull. Acad. Aci. U.R.S.S., Ser. Phys., 9, 196 (1945). (15) E. 6. Slowinski and G . C. Claver, J. Opt. SOC.A m . , 46, 396 (1955). (16) R. H. Hughes, R. J. Martin and N. D. Coggeshall, J . Chem. Phya., 24, 489 (1956). (17) U. Liddel and 9. D. Becker, ibid., 26, 173 (1956).
and sample to the desired temperature and then recording spectra while the sample and sample holder cooled in the above cell. Temperatures were measured with a calibrated copper-constantan thermocouple. The alcohols and phenols used were of reagent grade. A Fenske column having about ten theoretical plates was employed for distillation and only the middle cut was used for measurement. Considerable care was taken to protect the materials from atmospheric moisture and other contaminants. The carbon tetrachloride and diethyl ether used as solvents in the investigation were found adequate for use without further purification but were dried over sodium and Drierite, respectively. The triethylamine used as a solvent consisted of a good grade product which had been redistilled with a Fenske column and dried over potassium hydroxide pellets. I t s spectra showed the abaence of any secondary or primary amines. As the associated OH band is not completely symmetrical with respect to an axis perpendicular to the frequency axis at the frequency associated with maximum absorbancy, i t was decided to determine the band center frequency by a method that would yield a frequency related to the "center of gravity" of the band. Such measurements were effected by drawing a line parallel to the background trace through the mid-point of a vertical line which passed through the point of maximum absorbancy of the spectrophotometer trace. The frequency associated with the mid-point of this parallel line was selected as the frequency of the center of
(18) J. N. Finch and E. R. Lippincott, ibid., 24, 908 (1956). (19) J. Powling and H. J. Bernstein, J . A m . Chem. Soc., 75, 1815 (1951).
July, 1957
TEMPERATURE DEPENDENCE OF OH FREQUENCY SHIFTS
the band. The method was found superior to locating the frequency associated with the center of gravity of a plot of log Zo/Z against wave length, but both methods gave essentially the same results. If the band consisted of two components which overlapped such as the inter-bonded and intra-bonded OH frequencies in o-chlorophenol, the band center was taken to be that fre uency where absorbancy was maximum for the intermolecgar bonded OH frequency. Since one objective was to calculate the apparent integrated intensity for the bonded OH frequency a t different temperatures and concentrations, the spectra were plotted in a manner that would yield curves that easily could be integrated graphically. T o effect this, the background trace, using the HzO bands at 2.679 and 2.630 p for trace calibration, was superimposed upon the trace of the bonded OH frequency and a point by point measurement of log 10/1was made a t intervals of about 0.025 p over the region 2.7 to 3.3 p . It was found convenient to set the short wave length side of the band to zero a t 2.7 to 2.8 p . These data were then plotted against wave length or frequency on paper ruled 10 X 10 per square inch and the area under the curve determined. Fortunately, there was little overlap between the CH stretching frequency and bonded OH frequency at about 3.3 p for the thin films of the hydroxy compounds examined. It was assumed that these groups contributed equally to the absorbancy in this region and the integration was carried out to the wave length or frequency corresponding to the mid-point of the overlapped region. For the strong hydrogen bonding existing in triethylamine solutions the broad OH band overlapped the CH bands considerably and the concealed portion was estimated as suggested by the work of Barrow.6 The usual expression for integrated intensity, A = (2.31 c l ) flog l o / dw, l where c is concentration in moles per liter, 1 is path length in centimeters, and w is the frequency in cm. -1 was utilized to calculate the apparent integrated intensity for o-chlorophenol a t various temperatures and for isopropyl alcohol in diethyl ether and in triethylamine at various concentrations. Since i t was desired to study the shift of the bonded OH frequency with temperature, band center data were used to calculate the apparent integrated intensity for all the pure alcohols in the liquid state and for solutions of phenol in carbon disulfide and isopropyl alcohol in carbon tetrachloride. The integrated intensity was expressed in the usual units moles-' liter cm.-z. Cell thickness and concentration were selected such that the most precise instrumental measurements could be realized. Consequently transmittancy was maintained at approximately 37%. The effect on density of decreasing the temperature from 300 to 200°K. was determined to influence the integrated intensity by not more than 8%. This was estimated from an average coefficient for alcohols and from the linear coefficient of expansion of the spacer material. A linear correction reflecting the increase in concentration with decreasing temperature for the temperature interval studied has been applied to the intensity data. The band center data are considered to be reliable to better than &3 cm.-1 and only in the case of o-chlorophenol and in the dilution data for isopropyl alcohol in diethyl ether and triethylamine is i t possibly greater. The intensities are considered to be internally reliable to better than 5%. Band center data for the temperature de endence of pure methyl, ethyl, isopropyl, n-butyl, isoamyfand n-hexyl alcohols are plotted in Fig. l and 2 . The data for the first member of this series are listed in Table I. Similar data for 2-methyl-4-methoxy-2-pentanoland o-chlorophenol, which are also capable of intrabonding, are illustrated in Fig. 3. It is observed that the bonded OH fre uency for all the pure compounds in the liquid state varies jirectly with the temperature. As there is similarity among the band center data for all the aliphatic alcohols capable of only inter-bonding, they are discussed together. Figures 4 and 5 are included to illustrate the change in intensity and frequency with temperature of the bonded OH frequency of isopropyl alcohol, capable of only interbonding, and for o-chlorophenol which is also capable of intra-bonding.
Discussion and Interpretation of the Temperature Dependence of Bonded OH Frequencies Alcohols Which Are Capable of Only Intermolecular Bonding.-It is observed that at room temperature (300°K.), the bonded OH frequency
895
300
@ E;
250
200 3300 3320
3300 3320 3300 3320 3340 Cm.-1. Fig. 1.-Band center data for methyl, ethyl and isopropyl alcohols as a function of temperature. 300
!$ 250
6
3300 3320
3300 3320 3340
3300 3320
Cm.-'. Fig. 2.-Band center data for n-butyl, isoamyl and n-hexyl alcohols as a function of temperature. 350
300
-
250 -
200 3380 3400 3420 3440 3460 3420 3440 3460 Cm.-l. Fig. 3.-Bonded OH band center as a function of temperature for 2-methyl-4-methoxy-2-pentanol and o-chlorophenol.
JACKN. FINCHAND I
896
,LISR.
Vol. 61
LIPPINCOTT
n
I
361 2700
-
900-
Isopropyl Alcohol
80
E)
z w
Crn:l.
5
0-
-900
-
-1800
2.6
2.6 R
0-
80
Chlorophenol I I SO00 3200
I
I
3400
3600
Cm;l,
Fig. 4.-(Top) Spectra of bonded OH band in isopropyl alcohol illustrating changes in shape and intensity for three temperatures. Fig. 5.-(Bottom) Spectra of the inter- and intra-bonded OH band in o-chlorophenol illustrating change in position and intensity for three temperatures.
-
-
TABLE I BANDCENTERDATAFOR METHYLALCOHOL Cell = 0.0015 cm. T,OK. Crn.-l T,OK. Cm.-1 243.0 302.0 3335 3317 296.0 3333 240.0 3317 3333 231.5 295.0 3314 230.5 289.0 3331 3313 222.5 3310 3330 283.0 278.0 3329 220.0 3310 3327 218.0 275.5 3309 3328 213.0 274.5 3309 210.0 3307 3327 273.0 3324 202.0 263.5 3305 260.5 3324 194.0 3304 251.0 3320 183.0 3302
for methyl, ethyl, isopropyl, n-butyl, isoamyl and n-hexyl alcohols is about 3333 cm.-l. When the temperature is changed from 300 to 200"K., the bonded OH frequency in methyl, ethyl, isopropyl, n-hexyl alcohols decreases 30, 35, 44, 49, 44 and 50 cm.-', respectively. The order of the alcohols in terms of their frequency shifts corresponds to the order in terms of
IN
3.0
i.
Fig. 6.-The lower curve gives the quantitative dependence of hydrogen bond energy on 0-0 distance as calculated from the model of reference 10 in text. The potential minimum corresponds to a hydrogen bond having a R, value of 2.75 A. and a OH frequency shift of near 500 cm.-l. The upper curve qualitatively gives the 0-0 potential curve when the OH bonded mode has been excited by one quantum. v is the OH stretching vibrational quantum number. From these curves one notes that since dissociation of the hydrogen bond corresponds to a OH frequency shift of near 500 cm.-l, a value of 1.85 cal./degree (the value of dE/dT) corresponds to approximately 0.2 cm. -'/degree for dw/dt, a result in agreement with calculations from equation 3.
basicity given by well known chemical and physical evidence. Chemically, the increase of basicity of the normal alcohols with an increasing number of carbon atoms in the chain has been attributed to the increased electron density on the oxygen atom. A similar effect has been observed for the branched chain alcohols such as isopropyl, but it has been noted that the basicity is not appreciably increased by increasing the length of the chain as in isoamyl alcohol. When the electron density on an oxygen atom in a pure alcohol in the liquid state is increased, the 0---0distance will decrease because of the greater attraction of the oxygen atoms to the hydrogen atom in the OH---0 bond. Consequently, the decrease in the 0-0 distance due to the increase in electron density on the oxygen atoms is accompanied by an increase in the energy of the bond and an increase in the frequency shift as predicted by the one-dimensional model of hydrogen bonding.10 The order of the normal alcohols in terms of their frequency shifts would be expected to be: methyl < ethyl < butyl < hexyl alcohol. The frequency shift for isopropyl and isoamyl alcohols would be expected to be similar and to be greater than the shift for ethyl alcohol
.
1
and probably less than that of n-butyl alcohol. The frequency shift of 44 cm.-l is commensurate with these predictions. Calculation of the Temperature Dependence of Bonded O H Frequencies from a Potential Function Model.-A semi-quantitative calculation of these frequency shifts as a function of temperature will be made by using a model of 0-H---0 hydrogen bonding based on a potential function. The potential function for the model consists of four terms. (1) V = VI V z Va (repulsive) V4 (electrostatic) where VI = Do[l - exp(-nAr2/2R)] VZ = DO*exp[-n*(R - r - r0*)~/2(R- 1.11 Va
=
-B/Rm
A diagram illustrating some of these quantities is r