6234
J. Phys. Chem. 1994,98, 6234-6236
Solid Phases and Phase Transitions of Cycloheptane David M. Snyderman, J. Mark Adams, Andrew F. McDowell, and Mark S. Conradi' Department of Physics, 1105, Washington University, One Brookings Drive, St. Louis, Missouri 63130-4899
William H. Bunnellet Department of Chemistry, University of Missouri. Columbia, Missouri 6521I Received: February I , 1994"
The phases and phase transitions of solid cycloheptane are examined with differential thermal analysis, with two new findings. Phase I samples cooled below -200 Kdirectly transform to phase 111, not phase I1 as reported earlier. Phase I11 readily supercools through the 111-IV transition, yielding a glassy crystal 111, or transforming into a previously unknown metastable phase, 111', depending on details. These findings impact the identification of the phases studied in previous reports. Deuterium NMR of phase I11 shows that the reorientations are not isotropic, indicating that no molecules reside in sites of cubic symmetry. The deuterium spectrum suggests several inequivalent sites in phase 111.
Introduction There are four stable solid phases of cycloheptane (at essentially zero pressure), as determined by a thorough adiabaticcalorimetry study.' Only the low-temperature phase, IV, is ordered; this can be deduced from the IV-I11 transition entropy, which exceeds the 111-11, 11-1, and I-liquid transition entropies by factors of 5 and more. Thus, the phases 111, 11, and I are disordered and are reported to be rotor solid^^.^ by NMR4and by an IR and Raman study.5 One suspects that the multitude of phases is due in part to the conformational flexibility of the seven-membered ring.6J Glass transitions have been observed by differential thermal analysis (DTA) in the supercooled rotor phases of cycloheptane.a10 Similar behavior has been reported for cyclohexanol,lOJ'cycloheptano1,"Jcyclooctanone,I2and several other cycloalkane derivatives. An IR and Raman study of cycloheptane indicated that both phases I1 and I11 can be supercooled to yield glassy crystals, 11, and 111,, respectively.' The original calorimetric study mentioned that phase I1 cycloheptane readily supercooled.' The NMR investigation of solid cycloheptane reported that supercooling of some rotor phase(s) was common but did not specify which phases.4 We report here thorough DTA results that yield different conclusions concerning the supercooled phases. These conclusions impact the identification of the phases in previous work. In addition, deuterium NMR spectra shed light on the nature of molecular reorientation, particularly in phase 111.
Deuterium NMR at 13.05 MHz was used to determine the extent of molecular reorientations. Ninety-degree pulse times of 6 ~s were adequately short for the relatively narrow deuterium lines encountered. Spectra were obtained by Fourier transformation of the quadrupole echo, formed with a 90,-t-9Ort4ho pulse sequence.
Results and Discussion DTA. A broad survey of the extensive DTA measurements on
cycloheptane is presented in Figure 1. The best way to understand the data is by reference to Figure 2, showing the proposed relationship of the Gibbs free energy as a function of temperature for the various phases. The DTA data in Figure 1 are labeled with the phases (I, 11, etc.) according to Figure 2. DTA data upon cooling are displayed in Figure la. Freezing of the benzene is followed by freezingof the somewhat undercooled cycloheptane (compare to melting, in Figure 1b). Figure 1band IC presents data upon warming, after being cooled as in Figure la. A sample was cooled to 77 K as in Figure la, warmed to 125 K, and then recooled to 77 K; subsequent warming yielded the data of Figure Id. The following observations are important guides in the assignment of phases and in the construction of Figure 2. First, the product of the 206 K endothermic transition upon warming in Figure l b has the correct melting point (264 K; ref 1 lists 265 K), so it is phase I. Second, the product of the exothermic conversion near 118 K (see Figure 1b and IC)is always phase IV, Experimental Section because the large IV-I11 endotherm always follows at 137 f 1 Cycloheptane from Aldrich had a stated purity of 98%. The K (1 35 K in ref 1). Further, as shown in Figure Id, the product material was further purified by repeated fractional distillation, of the -1 18 K exotherm shows no phase transitions below 137 resulting in a purity of 99.6% as determined by gas chromatoK, as expected for phase IV. The lack of any thermal signatures graphic analysis. The reference sample in the DTA work was in the 110-120 K region of Figure Id demonstrates the low noise benzene (99.9%), used as received from EM Sciences. level of the apparatus and confirms that the features near 110 The home-built DTA has been described previ~usly.'~J~ K in Figure la-c are real. Finally, the product of the IV-I11 Briefly, it consistsof a massive copper block containingthe sample endothermictransition at 137 K (Figure lb-d) is indistinguishable and reference in glass tubes, each with a thermocouple. The from the material obtained by cooling below 199 K (Figure la), block warms from 77 to 270 K in about 2 h on the laboratory as repeatedly tested by phase transitions upon subsequent heating table. The rate of temperature rise continuouslydecreases as the or cooling. Thus, this material is phase 111. temperature increases toward room temperature; this effect At our normal rates of heating (-2 K/min average from 80 reduces the relative amplitude of the high-temperature peaks. to 250 K) and cooling (- 15 Klmin), we always observed a single We note that this is only a factor of 2 effect over the 100-200 transition between phases I11 and I, without the appearance of K temperature interval. phase 11. In particular, the 111-1 transiton always occurred at 206 f 1 K upon warming, indicating this is the true 111-1 phase Now at Abbott Laboratories,ChemicalandAgriculturalProductsDivision, transition temperature (i.e.,not increased by rate effects). Given Dept. 54-P,Bldg. R8,North Chicago, IL 60064. the 111-11 and 11-1 latent heats and transition temperatures,' the e Abstract published in Adounce ACS Absrracrs, May 15, 1994. 0022-365419412098-6234$04.50/0
0 1994 American Chemical Society
The Journal of Physical Chemistry, Vol. 98, No. 24, 1994 6235
Phases and Phase Transitions of Cycloheptane
137K
A
ll2K
1\
Y
137K
N
206K
264K
1
LIC
\r W
ll8K
i 20K
TEMPERATURE
Figure 1. DTA thermograms for cycloheptane: (a) DTA data during cooling; the arrows show the direction of time. Freezing of the benzene reference and then the cycloheptaneis apparent, with the benzene peak off scale. (b) Typical warming data. A small endotherm is evident at 108 K. (c) DTA warming data with pronounced endotherm at 111 K. (d) Warming data for sample cooled to 77 K after the exotherm at 120 K is traversed. Once formed, phase IV is evidently stable below 137 K. In all cases, the phases are labeled according to the scheme of Figure 2. The warming and cooling rates of the four traces differ, so the peak heights cannot be compared. The temperatures shown are accurate to f 2 K.
-1
I
2 TEMPERATURE
Figure 2. Sketch of Gibbs free energy as a function of temperature, showingthe four stable phases and the metastable phase 111’. The arrows show the path of the sample in the DTA run of Figure lb. The temperatures of the transitions are from ref 1 for the stable phases and from the present workotherwise. The dashed line indicatesan exothermic transformation.
111-1 transition is predicted to be 206 f 1 K (see Figure 2). The direct 111-1 transition upon warming has been observed previously.5~*The 1-111 transition upon cooling (Figure l a ) is rate depressed (i.e., it supercooled slightly), we believe. By holding a sample between 198 and 212 K for 25 min, we were able to transform -50% of the sample to phase 11, as evidenced by a subsequent 11-1 transition at 212 K. The present finding of a direct 1-111 transition upon cooling is important for thecorrect phase identification in previous works. Phase I1 is difficult to form; data reported from phase I1 should include the specific evidence for assignment of the phase I1 label. For example, glassy crystalline phases prepared by direct cooling of phase I (assumed to yield I1 and eventually 11,) and by cooling the product of the IV-I11 transformation (111,) have essentially indistinguishable IR and Raman spectra.5 In accord with our finding of a direct 1-111 transition upon cooling, the material in both cases is likely 111, (or 111’; see below). Thus, reports of data from phase I1 or 11, must be re-examined in light of the present results. We now turn to the behavior below 125 K. Upon cooling (Figure la), phase I11 transforms a t 110 K. The resulting phase
Figure 3. DTA data, all obtained on warming from 77 K, showing the diversity of behaviors near 95-1 12 K. (a) A pronounced endotherm at 112 K, followed by a 122 K exotherm into phase IV. (b) A weak rise, interpreted as a glass transition,at 98 K. (c) The 11 1 K endotherm on
the shoulder of the exothermic signature. cannot be IV, because of the behavior upon subsequent warming (Figure lb; compare to Figure Id). This new phase is labeled III’in Figure 1and Figure 2. The endothermic 111’-I11 transition is evident near 110 K in Figures lb,c and 3a. This observation determines that the Gibbs free energies of I11 and 111’ cross a t 110 K in Figure 2. Because phase IV was reported’ stable a t all temperatures below 135 K, 111’is always metastable with respect to IV (hence the primed designation). In actuality, the thermal behavior between 95 and 112 K is rather variable. In Figures IC and 3a, pronounced endotherms (111’-111) are observed. The immediately subsequent exothermic conversion into phase IV is typical of such systems, having been documented in cyclooctane13J4 and cyclooctanone.12 The exotherm appears in Figure 2 as a dashed line. In the run of Figure 3b, a step increase in the heat capacity a t 98 K is observed, indicativeof a glass transition (111,-HI), as previously reported.&lo In Figure 3c, the 111 K endotherm appears as a bump on the shoulder of the exotherm; presumably a portion of phase 111’ is transforming directly to IV. Despite the multiplicity of observed behaviors, the sharp transitions near 110 K in Figures la,c and 3a are strong evidence of the 111’-I11 transition. We note that the 111-111’ transition was rate depressed to as low as 94 K in some cooling runs. NMR. Pulsed proton N M R measurements of T1and T2 were performed at 85.03 MHz. The results are in good agreement with a previous C W N M R study4 and are not shown here. Proton N M R confirms the phase behavior revealed by DTA. Furthermore, phase IV is devoid of reorientational motion, as indicated by the short proton Tz. Deuterium NMRspectra wereexamined to further understand molecular reorientation in the solid phases.15 A sample of C7H14 was -50% deuterided by bubbling 02 gas through solution for 1 week, with a palladium-on-charcoal catalyst. In phase I, a single sharp line was observed. This is the result expected for isotropic (spherical) reorientation of the molecules. It is also the result expected for rapid diffusion between sites of anisotropic rotation, such as occurs in the Pm3n structure of cyclooctane phase 1.16 From the present and previous4 proton NMR (Tz and line width), it is clear that diffusive jumps in phase I are rapid (>lOss-1). Thus, thedeuteriumNMR cannot differentiate these two explanations. In phase IV, a 128-kHz splitting was observed, as expected for nonreorienting molecules. Deuterium N M R spectra from phase I11are presented in Figure 4. The line is considerably narrower than in phase IV, indicating substantial orientational averaging. However, a sharp central
6236 The Journal of Physical Chemistry, Vol. 98, No. 24, 1994
Snyderman et al. in the sketch of Gibbs free energy us temperature, Figure 2. The results for the stable phases are in accord with previous thermal measurements.' However, we also report the existence of a previously undetected metastable phase 111' which is obtained by cooling phase I11 to near 77 K. This phase 111' is always higher in free energy than IV. Occasionally, cooling of phase I11 to 77 K resulted in the formation of glassy crystal III,.s.*-lO Thus supercoolingof I11 can result in either phase III'or 111,, depending on details.
Figure 4. Deuterium NMR spectra of partially deuterated solid cycloheptane in phase 111. In addition to the Pake powder doublet, there is considerable intensity across the center of the spectra.
line, the signature of isotropically reorienting molecules (see spectrum of cyclooctane, ref 16), is not present. Thus, all of the molecules in phase I11 are anisotropic rotors. The spectra of phase I11 in Figure 4 have a nearly temperatureindependentshape and splitting. This indicates that any molecular motion is either too slow (105 s-1). As suggested by the nearly constant 27-ps value of proton T2 from 140 to 200 K and in agreement with previous work: diffusion is slow in phase I11 and anisotropic rotation is rapid. Furthermore, because of the small energy barrier, rapid pseudorotation is e ~ p e c t e d . ~ . ~The .'~ expected deuterium NMR line shape for these assumptionswould be temperature independent, in agreement with experiment. The line shape should be a Pake powder doublet characteristic of uniaxial symmetry. The cusp-to-cusp splitting will be 128 p2 kHz, where & = (3 cos2 0 - 1 )/2 with 0 the angle between the symmetry axis of the site and the C-D bond. The rapid pseudorotation will makeequivalent all 14 D siteson themolecule. However, the spectra in Figure 4 have too much intensity in the center to be simple Pake powder doublets. Given that the dipolar broadening is small (12kHz) on the frequency scale of the figure, dipole broadening does not explain the shape. Instead, we propose three explanations. First, there may exist several inequivalent molecules per unit cell. Second, the angular average pz may be unusually sensitive to strains and impurities in the structure. Over several different experimental runs, we noticed small but real changes in the deuterium line shape. Third, pseudorotation may not be rapid on the NMR time scale (les s); the 14 D sites on the molecule would not be equivalent and would have different quadrupolar splittings. Of these possibilities,the first seems most likely.
Cooling of phase I through -200 K resulted in a direct 1-111 transformation, "skipping over" phase 11. The phase I11 so obtained had thermal behavior identical to that of phase I11 obtained by warming phase I V through the IV-I11 transition. Thus, in contrast to previous work, the present DTA results indicate that rapid cooling of phase I results in phase 111. Previous reports of the formation of supercooled phase I1 and subsequently 11, should be re-examined. We note that the conditions of our DTA are more similar to those of most previous work than the painstaking, slow conditions of adiabatic calorimetry.' Deuterium NMR of phase I11 demonstrates that anisotropic reorientation occurs throughout the temperature range. Thus, none of the molecules are located in sites of cubic symmetry. The spectra are not good Pakedoublets, suggestingseveralinequivalent molecules in the unit cell.
Acknowledgment. The authors appreciate helpful conversations with D. F. R. Gilson and thank him for sending a cycloheptane sample. S.Gilbertson generously directed the purification of the cycloheptane. The support of this work through NSF Grant DMR 90-24502 is acknowledged. References and Notes (1) Finke, H. L.;Scott,D.W.;Gross, M. E.;Messerly,J.F.;Waddington, G. J. Am. Chem. Soc. 1956, 78, 5469. (2) Parsonage, N. G.; Staveley, L. A. K. Disorder in Crystals; Clarendon: Oxford, U.K., 1978. ( 3 ) Timmermans, J. J. Phys. Chem. Solids 1961, 18, 1. (4) Brookeman, J. R.; Rushworth, F. A. J . Phys. C Solid Store Phys. 1976, 9, 1043.
(5) Haines, J.; Gilson, D. F. R. J. Phys. Chem. 1990, 94, 3156. (6) Bocian, D. F.; Strauss, H. L. J. Am. Chem. SOC.1977, 99. 2876. (7) Elser, V.; Strauss, H. L. Chem. Phys. Le??.1983, 96, 276. (8) Adachi, K.; Suga, H.; Seki,S.Bull. Chem. SOC.Jpn. 1970,43,1916. (9) Suga, H.; Seki, S.J . Non-Crys?.Solids 1974, 16, 17 1. (10) Suga, H.; Seki, S.Furaduy Discuss. Chem. Soc. 1980,69,221. (1 1) Adachi, K.; Suga, H.; Seki, S.Bull. Chem. Soc. Jpn. 1968,41,1073. (12) Adams, J. M.; Snyderman, D. M.; Conradi, M. S. J. Phys. Chem. 1993, 97, 11092. (13) Keller, R. C.; Coffey, M. S.; Lizak,M. J.; Conradi, M. S. J. Phys. Chem. 1989, 93, 3832. (14) Huang, Y.; Gilson, D. F. R.; Butler, I. S.J . Phys. Chem. 1991, 95, 5051.
Conclusions The results of the present DTA measurements are summarized
(15) Boden, N. In The Plusricully Crysralline Srure; Sherwood, J. N., Ed.; Wiley: New York, 1979; chapter 5. (16) Lizak, M. J.; Keller, R.C,; Coffey, M. S.;Conradi, M. S.; Bunnelle, W. J. Phys. Chem. 1990, 94,992.
