1744
J. C. DAVIS,JR.,K. S. PITZERAND C. N. R. RAO
Vol. 64
NUCLEAR MAGNETIC RESONANCE STUDIES OF HYDROGEN BONDING. 11. ALCOHOLS BY JEFFC. DAVIS,JR.,’ KENNETH S. PITZER~ AND C. N. R. RAO Department of Chemistry and Lawrence Radiation Laboratory, University of California, Berkeley, California Received June 3. 1060
The roton magnetic resonance spectra of solutions of methanol, ethanol, 2-propanol and t-butyl alcohol in carbon tetrachgride and of ethanol in benzene have been measured over the temperature range 20-60’. The shifts 6 in very dilute solutions yield values of b y for the monomeric alcohols. Finite values of (Xi/az) at zero mole fraction are obtained which clearly indicate the presence of a dimeric species. The relationships between these data and the equilibrium constants and enthalpies of dimerization are discussed, but the probable presence of both cyclic and open dimers complicates the intergretation. For ethanol in benzene, where open dimers are believed to predominate, the enthalpy of dimerization is found to e 5.1 f 1 kcal./mole of dimer. Only qualitative comments are made concerning the results in concentrated solution where many polymeric species are present.
Many physical methods have been used to investigate association of alcohols by hydrogen bonding. A complete review of the very extensive literature is both impractical and unnecessary in view of the recent monograph by Pimentel and M ~ C l e l l a n . ~Although a number of n.m.r. investigations have concerned the alcohols, it is desirable to cover a wider temperature range in order to yield information about the various species present. The present investigation was undertaken as a systematic study of the chemical shifts at different concentrations and temperatures with the purpose of broadening the application of n.m.r. techniques to these systems and with the aim of elucidating the nature of the various species and the characteristics of the equilibria. Experimental The purification and handling of the reagents have been reviewed previously.4 All of the alcohols were distilled twice from calcium hydride to remove traces of water. All of the solutions were made by dilution of gravimetrically prepared stock solutions. Shifts were measured a t 60 Mcps. on a Varian V4300B High Resolution NMR Spectrometer utilizing special apparatus constructed for maintaining the desired sample temperatures as previously de~cribed.~Shifts were measured by the side-band technique.6 Some shifts were measured by means of known fine splittings of adjacent or superimposed peaks as, for example, when the ethanol OH peak was located inside the CHa triplet. Shifts were measured to an accuracy of st 0.5 c.p.5. and each reported shift is the result of a t least six separate measurements.
Experimental Results The chemical shifts of the hydroxyl protons of methanol, ethanol, isopropanol and t-butyl alcohol in CC14 and of ethanol in benzene are listed in Tables I-V. The symbol II: designates the total apparent mole fraction of alcohol in the solution assuming no association. Shifts are reported in cps measured a t 60 Mcps. A positive shift indicates that the hydroxyl resonance occurs a t a higher applied magnetic field than that, of the reference peak. I n all cases the centers of the methyl group resonance were used as references. These (1) Based in part on the Ph.D. thesis of J. C. D. whose present address is Department of Chemistry, University of Texas, Austin, Texas. (2) T o whom correspondence should be directed a t the above address. (3) G. C. Pimentel and A. L. McClellan, “The Hydrogen Bond,” W. H. Freeman and Co., San Francisco, Cal., 1960. (4) J. C. Davis. Jr.. and K. S. Pitaer, Tms JOURNAL, 64,886 (1960). ( 5 ) J. T. Arnold and M. E. Paokard, J . Chem. Phys., 19, 1608 (1951).
peaks were found to be independent of concentration by measurements with both external and internal standards. TABLEI CHEMICAL SHIFTSOF METHANOL IN CARBON TETRACHLORIDE SOLUTIONS 6 (c&p.s.at 60 Mc. f;om 23 35
X
1.00 0.742 .502 .249 .099 .077 .052 .027 .020 .016 * 010 .0082 .0077 .0066 .0052 .0040
-92.5 -87.5 -74.5 -48.0 44.8 79.1 115.4 145.5 152.8 157.0 163.4 165.2 165.8 166.7 168.1 169.2 172.9 950
bM
wax)o
-88.0 -79.6 -62.9 -28.7 72.0 106.1 135.6 145.2 160.4 163.0 166.8 167.8 168.1 168.5 169.5 170.0 172.2 540
methyl peak) 53.50
-76.1 -66.0 -40.1 8.7 133.7 147.0 158.9 167.6 168.0 168.9 170.0 170.5 170.5 170.9 171.O 171.3 172.1 200
TABLEI1 CHEMICAL SHIFTSOF ETHASOL IN CARBON TETRACHLORIDE SOLUTIONS 6 (c.p.s. a t 60 Mc. from methyl peak) 220 390 56’
X
1.00 0.505 .250 .loo .0775 .0525 .0270 .0206 .0165 .0104 .0083 .0067 .0052 ,0042 6M
(& / ~ S ) J
-250.0 -219.8 -200.3 -169.0 -160.1 -120.0 - 32.3 - 19.0
-
-241.5(37) -201.5 -167.8 -124.2 -118.1 - 77.8 4.2 12.1
9.0
16.0
4.0 7. I 13.0 16.7 18.9 28.2 2250
23.9 25.3 27.5 29.4 30.4 34.9 1070
-228.1(54) -176.0 -134.0 - 83.1 - 76.6 - 36.3 6.0 21.0 23.1 27.0 28.1 29.0 29.9 30.1 32.4 500
Monomer-Dimer Eaui1ibrium.-There is considerable evidence that & dilute solutions of alcohols
Kov., 1960
XUCLEAR
MAGNETIC RESONL4NCES T U D I E S OF HYDROGEN BINDING
TABLE' I11 CHEMICAL SHIFTS OF ISOPROPANOL IN CARBON TETRACHLORIDE SOLUTIONS 6 p.p.8.
X
1.00 0.746 .507 .254 .lo1 .078 ,053 .027 .021 .017 .011 .0084
.0078 .0067 .0053 8M
25
i
at 60 Mc. from methyl peak) 36' 53.50
-237.6 -235.1 -203.9 -166.0 -113.8 - 95.6 - 86.2 - 36.2 - 10.8 - 3.0 11.8 13.0 13.7 14.2 15.0 18 400
-227.0 -223.5 -185.9 -140.5 - 82.0 - 56.4 - 35.0 3.1 12.1 14.9 17.6 19.0 19.0 19.2 19.7 21.0 220
-211.6 -210.0 -161.5 -105.8 - 37.6 - 12.2 3.1 15.3 18.8 19.4 20.2 20.8 20.9 21 .o 21.1 21.9 130
-60 0
TABLE IV CHEMICAL SHIFTSOF &BUTYL ALCOHOLIN CARBON TETRACHLORIDE SOLUTIONS X
1.00 0.745 .505 .252
.loo
.078 .053 .027 .0206 .0165 .0104 .0083 .0078 .0067 .0053 8m (a6/ax)o
6 (c.P.s. a t 210
-210.1 -198.0 -168.8 -130.0 - 70.1 - 34.3 - 9.0 10.0 15.1 17.5 20.0 51.0
21.4 21.9 22.6 24.8 430
1745
3.0
0.02 0.04 0.06 Mole fraction alcohol. Fig. 1.
0.08
0.1
1
60 Mc.from methyl peakt 370 53
-190.1 -178.3 -143.0 - 96.1 - 27.5 - 2.0 10.2 17.1 18.9 20.1 21.3 22.2 22.4 22.6 23.1 24.5 270
-170.0 -156.8 -113.1 60.0 - 1.0 9.2 17.3 20.0 20.9 21.6 23.0 23.2 23.5 23.6 24.0 25.0 200
%
c;l
2.5 .
-
3.0
3.1
3.2 lOOO/T. Fig. 2.
3.3
3.4
Pimentel and Shoolerys have shown that for a system containing only monomer and dimer in equilibrium the observed chemical shift is given by the relation
TABLE v where m is the number of moles of monomer in the CHEMICAL SHIFTSOF ETHANOL IN BEXZEXE SOLUTIONSsolution, a is the total number of moles of alcohol 6 (c.P.s. at 60 hlc. f,om methyl peak) used to make up the solution, 6~ and 6~ are the 2 220 37 540
1.oo 0.504 .276 .145 .0742 .0373 .OR39 .0151 .0076 .0064 .0057 .0047 6M
(
-250.5 -229.0 -201.4 -156.7 - 87.0 - 4.8 9.8 15.3 23.9 24.2 25.0 26.4 31 .O 1020
-242.0 -208.8 -174.7 -129.0 - 58.2 8.1 18.0 21.0 25.9 26.9 27.2 28.2 31.2 680
-229.5 -185.0 -142.2 - 90.1 - 27.2 16.0 20.0 24.1 26.0 28.5 28.8 29.1 31.3 440
in non-hydrogen-bonding solvents the monomerdimer equilibrium is of importance. Huggins,
shifts characteristic of the monomer and dimer species, and AD = 6~ - 6 ~ f . If the equilibrium constant for the system is expressed in mole fraction units
then it can be shown that an infinite dilution (3)
Figure 1 shows a typical set of data in the low concentration region. It is clear that a satisfactory extrapolation can be made to yield both 6~ and the slope ( d S / b ~ ) ~Values . of these quantities are given in Tables I-V. Unfortunately the pres(6) C. M. Huggins, G . C. Pimentel and J. N. Shoolery, THISJOUR60,1311 (1956).
NAL,
1746
Vol. 64
J. C. DAVIS,JR.,K. S. PITZER AND C. N. R. RAO
ence of higher polymers makes it impossible to obtain 60, and thereby K z ,from the n.m.r. data. The assumption that the value of 6~ is independent of temperature, as it seems to be for certain carboxylic acids,4 makes possible a calculation of an apparent AH dimerization. From equation 3 we see that, if AD is constant (4)
Figure 2 shows the appropriate plot of log (d8/bzjo us. 1/T and Table VI collects the apparent AH values. Liddel and Becker' measured the infrared spectra of methanol, ethanol and tbutyl alcohol in carbon tetrachloride at' several temperatures. They assumed that the decrease in the intensity of the characteristic non-bonded 0-H st'retching band near 3630 cm.-' could be taken as a measure of dimerization in the dilute solutions, and from the temperature coefficient of this effect, they obtained the apparent AH values which are also listed in Table VI. The agreement between the two sets of values is excellent. But we shall indicate later that these treatment's are probably oversimplified and that the true AH of dimerization is somewhat smaller in some cases. TABLEVI DIMERIZATIOK OF ALCOHOLS IN SOLUTION Apparent" A H , kcal./mole This research L.B.7
CH30H in ccI4 -9.4 f 2 -9.2 f 2 . 5 C2H,0H in CC14 -7.6 f2 - 7 . 2 f 1.6 CnHbOHin CsHe -5.1 i1 .... GC3H70Hin CCla - 7 . 3 f 3b ....... .. t-C4H,0H in CC14 -4.4 f 2 -4.8 f 1 . 1 See section on Discussion for relationship to true AH of dimerization. The effect of higher polymers so interferes that this value is particularly uncertain.
Higher Polymers.-It is clear that higher polymers are formed in even moderately concentrated solution. Saunders and Hynes have treated their n.m.r. data for several alcohols on the basis of only a single type of polymer in addition to the monomer. They obtained rough agreement for a cyclic tetramer with methanol and ethanol and a cyclic trimer for t-butyl alcohol. However, the evidence from various sources for the presence of some dimer is incontrovertible. We tried extensions of the Saunders and Hyne t'reatment which considered the monomer, dimer and a single higher polymer. The inclusion of the dimer changed substantially the size of polymer yielding best agreement. But no case showed agreement within experimental error nor was this result surprising in view of the simplified assumptions. The dielectric constant datag on these systems clearly show that much of the polymeric alcohol must be open ended chains rather than rings, and in that case it is unreasonable to assume a single dominant' species. Consequently, we believe any realistic treatment of concent,rated solutions must (7) U.Liddel and E. D. Becker, Spectrochim. Acta, 10, 70 (1957). (8) M. Saunders and J . B. Hyne, J . Chem. Phys., 29, 1319 (1958), 31. 270 (1959); see a180 E. D. Becker, ibid.. 31, 269 (1959). (9) K. L. Wolf, H. Frahm and H. Harms, 2. phusik. Chem., B86, 237 (1937). gives n siinlnlarg- of s u c h data and references t o earlier papers a d to t1iesc.s.
consider an extensive array of associated species such as Coburn and Grunwald'O assumed in treating their infrared data on ethanol-CC4 solutions. Since n.m.r. data alone do not suffice to show the various species present, a proper interpretation must combine data from additional sources. It is beyond the scope of this paper to attempt such an interpretation for the higher polymers, but the present n.m.r. data should be useful in such a project.
Discussion The shift between 8~ for the monomer and the &value for fully hydrogen bonded polymer appears to be roughly 280 C.P.S.for each alcohol if we assume some dissociation in the pure liquids and that this dissociation increases with temperature. There appears to be most dissociation in t-butyl alcohol and successively less for the smaller molecules. The reference frequency for the CH3 protons is, of course, much lower in the case of methanol, and this increases the relative values of 6 for that substance as compared to the other alcohols. The values of 8hf show no significant change with temperature, and this result throws some additional doubt upon the large temperature change of the apparent 8~ values for certain carboxylic acids that we reported p r e v i ~ u s l y . ~The very much smaller dissociation made the 6~ values of the acids much less certain than the present values for the alcohols, but the acids may associate more strongly with the solvent and hence behave differently. There has been much discussion and speculation about the linear or cyclic nature of polymeric species of alcohols. For example, the dimer species may be either a symmetrical ring, -4, with two nonlinear hydrogen bonds, or an unsymmetrical open chain, B, with a single linear hydrogen bond. R-0-
1
- -H
H- - -A-R A
R
I
0-H-
R - -0-H
B
While there is good reason to believe that the linear hydrogen bond in B is stronger than one bent bond, it is not obvious whether one linear bond gives lower energy than the two bent bonds in A or not. The cyclic form would be expected to have no infrared absorption at the monomeric 0-H frequency whereas the open dimer (B) would have one 0-H vibration which would still absorb a t approximately the monomer frequency. Likewise the n.m.r. chemical shift for the non-bonded proton in the open form would be expected to be about that of the monomer while in the cyclic form both alcohol protons would have shifts affected by the hydrogen bonding. However, only the average shift is observed for the alcoholic protons of all species and there is no independent evidence for this average value in either the open or cyclic form. In the vapor phase the entropy and enthalpy of formationll of the dimer indicate the open struc(10) W. C ' . C'oburn and E. Grunnald, J . A m Chem S O P 80, , 1318 (19.58). (11) W. Ueltner and K. S. I'iteer, %bid, 73, 2606 (1951), C. B. Iiretschmer and R Wieb?, zbzd , 7 6 , 2579 (1951), G. AI. Barron, J . Chem. Phys., 20, 1739 (1952).
Nov., 1960
COPPERCHELATE POLYMERS DERIVED FROM TETRAACETYLETHANE 1747
ture, whereas a t very low temperatures in inert matrices, the spectral datal2favor cyclic structures. There seems no doubt that cyclic tetramers and possibly pentamers, hexamers, etc., have special stability. Such a molecule can have linear hydrogen bonds and still close the ring. Coburn and GrunwaldIo fit their infrared data on dilute ethanol-CCL solutions to a model of open dimers and trimers together with ring structures for higher polymers. Xevertheless, the dielectric constant datag for alcohol solutions show that open chain polymers of high dipole moment dominate in moderately concentrated solutions. There is a deep minimum in the molar polarization curve for systems such as methanol or ethanol in cyclohexane a t approximately 0.03 mole fraction. For ethanol in CC14 the minimum is less deep and lies a t approximately 0.05 mole fraction. The results of Coburn and Grunwald indicate that a t this concentration the cyclic polymers have become very important species, and they have, of course, small or zero dipole moments. The rise in molar polarization indicates, however, that the still larger polymer species formed a t higher concentration must have large dipole moments and are presumably of open structure. (12) M. Van Thiel, E. D. Beoker and G. C. Pimentel, J . Chem. Phye., 27, 95 (1957).
It will require dielectric constant values of extreme precision for very dilute solutions to give a clear decision between open and cyclic dimers. We have not found such data in the literature. It seems to us entirely plausible that the dimeric species may be partly cyclic and partly open chain a t room temperature but with the proportion of the cyclic form decreasing with rise in temperature. I n this case the assumption of a constant AD in the calculation of the AH values in Table VI is subject to question. The AH values calculated from the infrared data of Liddel and Becker' are also based upon the assumption of a single type of dimer and are likewise subject t o some question. In either case, however, the difference between the true and apparent AH values is not likely to be very large. In the case of ethanol-benzene solutions the molar polarization datag show no minimum corresponding to cyclic polymers, hence, it seems safe to assume that the dimer is predominantly open chain and the apparent AH value may be adopted as reliable. Acknowledgments.-We thank Dr. George C. Pimentel for interesting discussion and many helpful suggestions. This work was performed under the auspices of the Atomic Energy Commission.
COPPER CHELATE POLY-MERS DERIVED FROM TETRAACETYLETHANE' BY ROBERT G. CHARLES Westinghouse Research Laboratories, Pittsburgh 55, Pennsylvania Received June 8, 1060
Chelate polymers have been prepared by the reaction of tetraacetylethane with copper(I1) acetate in aqueous tetrahydrofuran solution. The products are crystalline and are of relatively low molecular weight Tith an average chain length of about five units. The polymers decompose rapidly in the temperature range 250 to 350 . They have about the same thermal stability as the structurally analogous cupric acetylacetonate. Part of the copper is reduced to the metal during the pyrolysis with the concurrent formation of tetraacetylethane. In addition to metal, the non-volatile portion of the degradation products appears to consist of highly conjugated polymeric organic materials in which some of the metal chelate ring systems are retained.
Previously reported work from this Laboratory has concerned t,heheat st,abilitiesof metal acetylacetone ~ h e l a t e s . Because ~~~ of current int'erest in metal chelate polymers as possible heat stable materials4-* it has seemed wort'hwhile to extend these studies to metal chelates derived from tetraacetylethane (I). Tetraacetylethane (TAE) is capable, in its (1) Presented, in part, a t the Inorganic Polymer Symposium, 136th National A.C.S. Meeting, Atlantic City, N. J., Sept. 1959. (2) R. G. Charles and M. A. Pawlikowski, T H IJOURNAL, ~ 62, 440 (1958). (3) J. yon Hoene, R. G. Charles and W. M. Hickam, ibid., 62, 1098 (1958). (4) W. C. Fernelius, Wright Air Development Center Report WADC 56-203,Part I (Oct. 1956); b., Part I1 (Sept. 1957); c., Part I11 (Feb. 1958). (5) K.V. Martin, J . A m . Chem. Soc., 80, 233 (1958). (6) J. C. Bailar, Jr., W. C. Drinkard, Jr., and M. L. Judd, Wright Air Development Center Report WADC 57-391 (Sept. 1957). (7) C. S. Marvel and J. €I. Rassweiler, J . A m . Chem. Soc., 80, 1197 (1958). (8) C. S. Marvel and A I . M. Martin, ibid., 80, 6600 (lY.58).
dienolic form, of reacting with divalent metal ions to form polymeric materials with the recurring unit shown in 11. The recurring unit is closely similar in structure to the (non-polymeric) divalent metal acetylacetonates 111. Metal-containing polymers derived from TAE, as well as from other bis-8-diketones, have been reported p r e v i o ~ s l y , ~ ~ ~ ~ ~ but there has been very little information publisbed regarding the properties of these materials. I n the present work we have prepared copper chelate polymers derived from TAE and have studied their composition and some of their properties, with particular attention to their thermal stabilities. Results and Discussion Preparation and Characterization.-Copper-containing polymeric materials are formed as finely (9) J. P. Wilkins and E. L. Wittbeoker, U. S. Patent 2,659,711(Nov. 17 1953).
