J. Phys. Chem. 1982, 86, 2516-2522
2516
the C2 model is the most probable to occur in red-colored zeolite A. Its structure is three-dimensional, and the low-energy bands are shifted to lower energy. The interaction between the two Ag, units is strong; i.e., the formation of the Ag, cluster is determined by metal-metal interactions rather than by metal-lattice interactions. The zeolite lattice imposes only the geometry (”steric effect” of the zeolite). On the other hand, a very weak cluster-support interaction exists. The shift of the bond orders of the zeolite tetrahedra is very weak, the charge delocalization is only noticeable by substituting Si for Al, the Ag-0 coordination bonds are weak, the character of the molecular orbitals responsible for the electronic spectra have predominately Ag character, and the energy-level diagrams are similar to the MO level scheme of the isolated Ag, molecule (Figure 5).
The substitution of Si for A1 makes the interaction with the support somewhat stronger. The Ag-Ag overlap decreases and an electron transfer toward the oxygens of the frameworks occurs, resulting in a positive charge of Ag,. However, the framework Si-0 and A1-0 bonds are virtually unaffected and the Ag-0 bond orders are still very small. Acknowledgment. L.R.G. and W.J.M. thank the Belgian National Fund for Scientific Research (NFWO) (Belgisch Nationaal Fonds voor Wetenschappelijk Onderzoek) for research positions as Research Assistant (“Aspirant”) and Research Associate (“Bevoegdverklaard Navorser”) and one of us (L.R.G.) acknowledges the NFWO (Belgium) and the D616gation aux Relations Universitaires Internationales (DRUI, Ministere de 1’Education Nationale, France) for a traveling grant to the Universit6 de Rennes, France.
Conformational Equilibria of Normal Alkanes, I-Alkenes, and Some ( E ) - and (2)-2-Alkenes in Neat Liquids and in Some Selected Solvents. A Carbon-I3 Nuclear Magnetic Resonance Study L. J. M. van de Ven, J. W. de Haan,” Laboratory of Instrumental Analysis, Eindhoven University of Technology, Eindhoven, The Netherlands
and A. BuElnskl Department of Petroleum, Faculty of Chemical Technology, Slovak Technical University, Bratislava, Czechoslovakia (Received: April 28, 198 1; In Final Form: September 8, 198 1)
13C NMR chemical shifts have been measured for the homologous series of propane through n-octane and of n-dodecane as neat liquids and as dilute solutions in methanol, carbon tetrachloride, acetone, cyclohexane, and cyclopentane. The resulting shift differences, corrected for bulk susceptibilities,were compared with those of a number of di-, tri-, and tetramethylcyclohexanes. Similar measurements were carried out for the series 1-butene through 1-nonene and for 1-dodecene as well as for some ( E ) -and (2)-2-alkeneswith the exception of acetone solutions. The results for the substituted cyclohexanes indicate that solvent effects and thus site factors do not differ significantly for methyl groups with or without steric y interactions. The same applies to methylene carbons. Subsequently, the results for the alkanes and for the aliphatic part of the alkenes show that the conformational equilibria in methanol, carbon tetrachloride, and cyclohexane solutions do not differ by more than 1 or 2%. By means of a few additional experiments it is shown that the same is true for neat hydrocarbons. These conclusions are compared with recent results on hydrocarbon conformational equilibria obtained with other physical measurements like vibrational spectra, SANS, light scattering, etc. A number of recent results from thermodynamic experiments and from different types of calculations are also reviewed in light of the conclusion drawn in the present paper. For some alkenes for which anomalous GC behavior was reported in the literature, anomalies also are observed in the present study.
Introduction The conformational behavior of n-alkanes in different media is not only of intrinsic interest. Small n-alkanes can also serve as model compounds for the behavior of larger molecules like lipid membranes and aliphatic amino acid side chains on the one hand and polymeric molecules or segments on the other hand. This is also true for the time scales of interconversion processes between stable conformeric states. Mainly for the above reasons a rather large number of papers on those subjects appeared during the past few years. These studies can be roughly divided into the following four categories: (I) chemical evidence from, 0022-365418212086-2516$01.2510
e.g., relative ring closure probabilities1 or intramolecular excimer formation;2 (11) theoretical calculations such as Monte Carlo procedures with or without conformational weighing,’ molecular mechanic^,^ Brownian dynamics,, and, finally, ab initio SCF methods4 (this latter only for (1) (a) D. S. Saunders and M. A. Winnik, Macromokcules, 11, 18 (1978); (b) ibid., 11,25 (1978); (c) R. Breslow, J. Rothbard, F. Herman, and M. L. Rodriguez, J.Am. Chem. Soc., 100,1213 (1978); (d) N. C. Den0 and E. J. Jedziniak, Tetrahedron Lett., 1259 (1976). (2) A. M. Halpern, M. W. Legenza, and B. R. Ramachandran,J . Am. Chem. Soc., 101, 5736 (1979). (3) For leading references, see G. T. Evans, J . Chem. Phys., 72, 3849 (1980).
0 1982 American Chemical Society
13C NMR Study of Conformational Equilibria
“unperturbed” molecules, i.e., in the gas phase); (111) large numbers of thermodynamical experiments like heats of m i ~ i n gheat , ~ capacities: excess volumes, and closely related phenomena like solubilities via viscosity measurements,’ measurements of partial molal volumes,s free energies of ~ o l u t i o n calculations ,~ of boiling points,1° and determination of surface tensions;l’ (IV) spectroscopic measurements like optical anisotropies:J2 X-ray diffractions, electron diffractions,13small angle neutron scattering (SANS),14and vibrational spectra, both IR and Raman.15 Not all of the above papers refer exclusively to conformational equilibria, as already noted. In particular, some of the theoretical studies are mainly concerned with the time scales of interconversion and/or interconversion rates.16J7 The results are of interest for NMR TImeasurements rather than for chemical shifts. In those papers dealing with conformational equilibria of neat, liquid, n-alkanes, no uniform conclusions regarding the number of gauche forms per unit of chain length as a function of the lengths of the alkane chains are presented (”coiling”or “kinking”). At present, the majority of papers seem to point to increasing coiling with increasing chain length (see also Discussion). Although 13CNMR chemical shifts are known to be rather sensitive probes for testing conformational equilibria, remarkably little has been published.l8 The aim of the present paper is to contribute to solution of the above-cited intriguing and challenging problems by means of precise measurements of 13CNMR chemical shifts in some selected solvents, partially as a continuation of older work.laa,b A number of 1- and 2alkenes is also included in these measurements in order to investigate the effects of a polarizable end group and also because in gas chromatography alkenes containing the pentene-1 fragment have been found to exhibit anomalous retention times. This was ascribed to the conformational (4) M. R. Peterson and I. G. Csizmadia, J. Am. Chem. Soc., 100,6911 (1978). (5) (a) M. Couchon, P. Nguyen Hong, and G. Delmas, Can. J. Chem., 56,2472 (1978); (b) R. Philippe, G. Delmas, and P. Nguyen Hong, ibid., 56,2856 (1978); (c) H. Phuong Nguyen and G. Delmas, Macromolecules, 12,740 (1979); (d) ibid., 12,746 (1979); (e) R. Philippe, G. Delmas, and P. Nguyen Hong, Can. J. Chem., 57, 517 (1979). (6) For leading references, see S. N. Bhattacharyya and D. Patterson, J. Phys. Chem., 83, 2979 (1979). (7) (a) D. Filiatrault and G. Delmas, Macromolecules, 12,65 (1979); (b) ibid., 12, 69 (1979). (8) (a) G. Mann, Tetrahedron, 23,3375 (1967); (b) J. T. Edward, P. G. Farrell, and F. Shadidi, J. Phys. Chem., 82, 2310 (1978); (c) J. T. Edward, P. G. Farrell, and F. Shadidi, Can. J. Chem., 57, 2887 (1979). (9) M. H. Abraham, J . Am. Chem. Soc., 101, 5477 (1979). (10) R. Thomas Myers, J . Phys. Chem., 83, 294 (1979). (11) (a) F. M. Fowkes, J. Phys. Chem., 84,510 (1980); (b) B. Lemaire and P. Bothorel, Macromolecules, 13, 311 (1980). (12) (a) P. Tancrede and P. Bothorel, J. Chem. SOC.,Faraday T r a m . 2, 73, 15 (1977); (b) D. Patterson, J. Chem. Phys., 69, 3250 (1978). (13) S. Fitzwater and L. S. Bartell, J. Am. Chem. Soc., 98,8338 (1976). (14) M. Dettenmaier, J. Chem. Phys., 68, 2319 (1978). (15) (a) P. E. Schoen, R. G . Priest, J. P. Sheridan, and J. M. Schnur, J. Chem. Phys., 71,317 (1979); (b) J. R. Scherer and R. G. Snyder, ibid., 72, 5798 (1980); (c) L. Colombo and G. Zerbi, ibid., 73, 2013 (1980). (16) (a) T. A. Weber, J. Chem. Phys., 70,4277 (1979); (b) D. Chandler and B. J. Berne, ibid., 71, 5386 (1979). (17) R. M. Levy, M. Karplus, and J. A. McCammon, Chem. Phys. Lett., 65, 4 (1979). (18) (a) A. R. N. Wilson, L. J. M. van de Ven, and J. W. de Haan, Org. Magn. Reson., 6,601 (1974); (b) J. W. de Haan, L. J. M. van de Ven, A. R. N. Wilson, A. E. van der Hout-Lodder, C. Altona, and D. H. Faber, ibid., 8, 477 (1976); (c) H.-J. Schneider and W. Freitag, J . Am. Chem. Soc., 98,478 (1976); (d) L. J. M. van de Ven and J. W. de Haan, presented at the Sixth International Symposium on Magnetic Resonance, Banff, Canada, May, 1977; (e) F. W. Vierhapper and R. L. Willer, Org. Magn. Reson., 9, 13 (1977); (0S. H. Grover, J. P. Guthrie, J. B. Stothers, and C. T. Tan, J. Magn. Reson., 10, 227 (1973); (9) H. N. Cheng and F. A. Bovey, Org. Magn. Reson., 11,457 (1978); (h) G. Mann, E. Kleinpeter, and H. Werner, ibid., 11, 561 (1978); (i) D. M. Grant and B. V. Cheney, J. Am. Chem. Soc., 89, 5315 (1967).
The Journal of Physical Chemistry, Vol. 86, No. 13, 7982 2517
properties of the pentene-1 fragment in the stationary phase, which are different from those of the other alkene~.~~
Experimental Section Normal alkanes, 1- and 2-alkenes, and the substituted cyclohexanes used as solutes were obtained from various sources. All samples were a t least 95% pure. The chemical shifts at infinite dilution were, in most cases, obtained by extrapolation of results from three to five measurements a t different concentrations. With the notable exception of 1-alkenes in methanol, linear shiftvolume fraction plots were obtained. The spectra were measured under proton noise decoupling at 39 “C in 10-mm sample tubes with acetone-& as an external lock. The line positions were measured with respect to the carrier frequency and, after extrapolation to infinite dilution (see above), corrected for bulk magnetic susceptibilities. The instrument used was a Bruker HX90R spectrometer interfaced to a Digilab FTS-NMR-3 pulsing and data system. Usually, spectral bandwidths of 4000 Hz were accumulated in 8K data points, resulting in a digital resolution of 0.97 Hz, corresponding to 0.04 ppm. For n-hexane we repeated the measurements with a bandwidth of 1000 Hz, resulting in a digital resolution of 0.01 ppm. Discussion In a previous paper1” we ascribed a part of the downfield shifts which are observed for 13CNMR chemical shifts of short n-alkanes like n-pentane and n-hexane upon dilution in higher homologues (e.g., n-hexadecane) to changes in conformational equilibria. It was argued that the freedom of, e.g., n-pentane to assume conformations with one or two gauche forms (excluding g+g- combinations) would decrease when dissolved in the higher homologues and vice versa. Afterward, Tiffon et al. stated that the observed chemical shift differences could be explained by taking into account only changes in the bulk properties of the solvent, multiplied by a suitable constant (“site factor”) which would be different for each carbon atom in the solute molecule.20 The basic assumption inherent in this approach is, that upon dilution of “apolar” (in the sense of “not easily polarizable”) solutes in apolar solvents, only van der Waals interactions would determine the differences in 13CNMR chemical shifts. (Throughout this paper we will use the terms polarizable and nonpolarizable for solutes. In a number of papers the expressions polar and nonpolar are used in this sense. This can lead to confusion.) The solvent properties were expressed in terms of the function
in which n is the refractive index of the solvent. Linear dependencies were found for 13CNMR chemical shifts of nonpolarizable solutes or parts of solutes in a large variety of The reasons for using g 2 plots are given in ref 20a and 21. Although the use of g 2 plots is probably perfectly legitimate for proton NMR, there seem to be no a priori reasons why the same should be true for carbon chemical shifts of flexible molecules. Carbon NMR (19) L. Soja, J. Janik, and J. A. Rijks, J. Chromatogr., 135,71(1977), and references cited therein. (20) (a) B. Tiffon and J. P. Doucet, Can. J. Chem., 54, 2045 (1976); (b) D. Cans, B. Tiffon, and J.-E. Dubois, Tetrahedron Lett., 2075 (1976); (c) J. Magn. Reson., 30,l (1978); (d) B. Tiffon and J.-E. Dubois, Org. Magn. Reson., 11, 295 (1978).
2518
The Journal of Physical Chemistty, Vol. 86, No. 13, 1982
chemical shifts are strongly dependent on conformational equilibria.’8b*c (In a more recent paper, Rummens21d mentioned the possibility that terms other than -BF2 would contribute to the van der Waals shifts.) Instead of n also other bulk properties like 6 have been used recently to develop linear relationships between 13CNMR chemical shifts and solvent property functions.23 Until more certainty is reached concerning the applicability of theoretical models in predicting gauche-anti equilibria in gas and condensed phases with more con~istency,~ it seems best to use a more pragmatic approach. Perhaps it is reasonable to consider that I3C NMR chemical shifts in flexible hydrocarbons are primarily determined by conformational equilibria and solvent effects. Conformational Equilibria. Direct evidence for the existence of differential shieldings between anti and gauche conformers of n-alkanes would be obtained by measuring 13CNMR spectra under slow-exchangeconditions. Since the barrier to internal rotation is relatively small4 this would require measurements at rather low temperatures. Such experiments have, to our knowledge, not been performed. On the other hand, there are a number of indirect indications. Comparison of the 13Cchemical shifts of C1 of propane and of n-butane in the gas phase reveals a shielding of -2.7 ppm for n-butane with respect to propane.lsd This differential shielding stems almost completely from the ca. 50% of the n-butane molecules4existing in the gauche conformation. This becomes clear by comparison with results obtained by Vierhapper and Willer for equatorial and axial methylcyclohexane.’” In the equatorial isomer, the methyl induces a substituent effect of -0.25 ppm on the (anti) y carbons. In the axial conformer, however, the y carbons (gauche with respect to the methyl group) are shielded by -6.33 ppm. Conversely, the methyl group in the axial conformer of methylcyclohexane is shielded by -5.00 ppm with respect to the methyl group in the equatorial position. Similar arguments as presented here for the outer carbons in a four-carbon fragment of a chain apply also to the inner carbons, albeit with smaller differences. Theoretical rationales have been presented by Grant and co-workers in terms of sterically induced electron redistributions leading to shielding of the four carbons in a gauche fragment.18’ In other sterically hindered conformationslarge deshieldings have been reported by Stothers and co-workers. These conformations, however, are comparable to g+g- in n-alkanes and thus do not contribute measurably to the conformational equilibrium, due to prohibitively high conformational energies.lab Other indirect evidence for differential shieldings in anti and gauche n-alkane fragments can be found in the variable-temperature studies of SchneiderlsCand the group from this laboratorylsb and more recent work from the groups of Bovey’@ and Mann.lah As expected, relative shielding was observed upon heating the n-alkanes. It should be kept in mind, however, that part of the observed shielding could, in some cases, have been caused by changing solvent properties, unless appropriate model experiments have been performed. Summarizing,we estimate that the differential shielding between gauche and anti fragments amounts to ca. -5 ppm for outer methyl carbons and -2 ppm for inner methylene carbons (possibly the value of a methylene carbon gauche to a methyl group should be rated at ca. -4 ppm). A shift in the conformational equilibrium of 2% should therefore produce differential shieldings of the order of -0.1 ppm, (21) (a) F. H. A. Rummens, Chem. Phys. Lett., 31, 596 (1975); (b) J . Chim:Phys., 72,448 (1975); (c) Can.J. Chem., 54,254 (1976); (d) F. M. Mourits and F. H. A. Rummens, ibid., 55, 3007 (1977).
van de Ven et ai.
well within the limits of the present measurements. Moreover, one gauche conformation induces shielding at four carbons simultaneously. Solvent Effects. Solvent effects, mainly van der Waals interactions, arise when sufficient precautions are taken concerning the polarities of the solvents and the polarizabilities of the solutes. It remains to be seen whether solvent effects depend also on conformations, for instance via different “site factors” for methyl carbons undergoing different numbers of gauche steric interactions. The theory describing solvent influences on proton and/or carbon shielding was developed primarily for spherical solute molecules surrounded by small solvent molecules.mJ1 A linear dependence of 13C NMR shieldings on the function g2,as described above, should be defined by using a large number of solvents.mp21In this work, only a limited number of “model” solvents was used together with the neat n-alkanes and n-alkenes. Therefore the graph represented in Figure 1does not have quite the same slope as a “genuine”g2 plot. In view of the very constant differential shieldings observed for model alkanes between CH,OH and CCl, solutions (vide infra) we are convinced that it does serve to define differential solvent effects (CH30H vs. CC14 or other combinations). Conformational equilibria of flexible hydrocarbons depend on (a) intramolecular steric (b) intermolecular packing effects,16a,22and (c) dielectric phenomena (dipole-dipole interactions). The latter interactions will presumably not change appreciably when comparing members of homologous alkanes or alkenes. The same is true for intramolecular steric interactions per bond. Still, with varying packing possibilities and the above-cited sources for chemical shift changes, the problem of relating shift changes with conformational equilibrium changes remains undetermined. Therefore we measured the 13C NMR chemical shifts of some di-, tri-, and tetramethylcyclohexanes in carbon tetrachloride (CC14)and methanol (CH30H). These media have rather large differences in refractive indices n or dielectric constants c.m,23 A number of these cyclic solutes remain essentially in one conformation with two or three methyl groups equatorial irrespective of the solvent. Other solutes will undergo rapid interconversions between two equally populated conformers, also without solvent dependence. The methyl groups in these solvents will thus have a known number of gauche interactions (none, one, or two). The measurements should therefore provide evidence regarding the sensitivity of site factors with respect to the number of gauche interactions, i.e., the steric environment of the methyl groups. The results are summarized in Table I. The difference between solvent effects in CC14 and CH30H hardly depends on the equatorial or axial position of the methyl groups, i.e., on the number of gauche interactions. Only for cis- and trans-1,3-dimethylcyclohexaneare the differences larger than the experimental error. The differential solvent effects in the solvents CH30H and CCl, are from 1.22 to 1.29 ppm. The spread of 0.07 ppm represents only 1 or 2% of the chemical shift ranges of comparable carbons in cis-trans isomers, in a given solvent. Similar results are reached by comparing the differential solvent effects (CH30H-CC4) of the several ring carbon signals, although the spread is now somewhat larger at 3 or 4% (22) (a) L. R. Pratt, C. S. Hsu,and D. Chandler, J. Chem. Phys., 68, 4202 (1978); (b) T. A. Weber, ibid., 69, 2347 (1978). (23) (a) W. Freitag and H.J. Schneider, J.Chem. SOC.,Perkin Trons. 2,1337 (1979); (b) J. L. M. Abboud and R. W. Taft, J . Phys. Chem., 83, 412 (1979).
-
13C NMR Study of Conformational Equilibria
The Journal of Physical Chemistry, Vol. 86, No. 13, 1982 2519
13.0
13.5
14.0
n C6
14.5 I
25
n CB
I
30
I
I
n C12
c c5
n C7
MeOH
c CB
cc14
I I
1
35
SOLVENT
I
40
b
45
g2
x
103 '
Flgure 1. Example of the dependence of 13C shielding on g2 for the C1 of n-alkanes.
TABLE I: 13C NMR Chemical Shift Differences in ppm of Cyclohexane and Some Di-, Tri-, and Tetramethylcyclohexanes Neat or Dissolved in Carbon Tetrachloride, Acetone, Cyclopentane, and Cyclohexane with respect to a Solution in Methanol compd cyclohexane
solv CCl, (CH3)2C0
C ,H IO C6H
1 ,trans-4-dimethylcyclohexane
I2
cc1,
1,trans-3-dimethylcyclohexane
+0.45 +0.02 + 0.23 t0.23
+ 1.26 + 0.34
+0.18
neat
+0.16
+ 0.24
+0.32 +0.42 +0.79
cc1, (CH,),CO C,HlO
t0.19 -0.10 +0.12
neat
+0.12
+0.47 0.00 +0.20 +0.22 +0.29
+1.28 +0.32 +0.33 + 0.43 +0.69
CCl,
+0.25
+ 0.41
t0.13
+0.43 +0.31 +0.26
+0.50
neat
+0.26 +0.21
CCl,
+0.29 t0.27 t0.13
+0.32 + 0.23 +0.16
+0.38 + 0.27 +0.21
+1.21 +0.48 +0.73
+0.29
+0.32 +0.24 +0.24
CCl, neat
1,l,cis-3,trans-5-tetramethylcyclohexane
Me
+0.17
C6H12
1 ,l,cis-3,cis-5-tetramethylcyclohexane
c,
+0.15
neat 1,cis-3,cis-5-trimethylcyclohexane
c,
C,HlO
C6Hl2
1 ,cis-3-dimethylcyclohexane
C2
+0.30 -0.10 t0.12
(CH3)2C0
1 ,cis-4-dimethylcyclohexane
Cl
cc1,
~0.18
+ 0.29
+0.18
i-0.38
-0.05
CP,,
+0.40 +0.15
cc1,
+0.48
CP,,
of the chemical shift range. (There are remarkably consistent differences in differential solvent effects for isomeric 1,3- and 1,4-dimethylcyclohexanes.The number of measurements and the differences are, however, too small to warrant further discussion at this time.) Next we consider the solvent effects in CH30H and CC14 with n-butane through n-octane plus n-dodecane as solutes.
+ 0.18
t1.25
+ 0.73
+1.21
+ 0.50 +0.87 +0.33
+0.15
+0.32 +0.17
+1.21 +0.40
t0.42 +0.17
+0.22 +0.12
+1.22 +0.42
The results are summarized in Table 11. The differential shieldings are rather constant for comparable C atoms throughout the series. In most cases the differences are smaller than the experimental error. Moreover, the numerical values for the differential solvent effects of the methyl signals are practically the same for the substituted cyclohexanes (Table I) and for C1 and C, in n-alkanes and
2520
van de Ven et at.
The Journal of Physical Chemistry, Vol. 86, No. 13, 1982
TABLE 11: 13C NMR Chemical Shifts of n-Alkanes Neat or Dissolved in CCl,, (CH3),C0, C,H,,, and C,H,, with respect to a Solution in C H 3 0 H comud propane
solv
cc1, C6H
n -butane
12
cc1, neat
n-pentane
cc1, (CH3)2C0
CSHIO C6H
12
neat n-hexane
cc1, (CH3),CO 5 H IO
neat n-heptane
CCl, (CH3)2C0
+1.29 + 0.44
+ 0.26
+ 1.24 +0.42 -0.21
+0.42 +0.20 -0.14
+ 1.29 +0.37 t0.37 + 0.45 t0.24
+ 0.34
t0.19 +0.19 i-0.11
-0.06 t0.15 + 0.1 5 t0.16
+ 0.46 +0.02 +0.24 + 0.24
+1.29 +0.33 +0.33 +0.46 +0.36
+0.36 -0.03 t0.19 t0.19 +0.24
+0.42 -0.01 + 0.20 +0.25 + 0.30
+0.46 +0.02 10.24 t0.24 +0.21
+ 0.37
+ 0.41
-0.03 t0.19 +0.15 +0.26
+0.02 +0.24
+ 1.28
+ 0.46
+0.32 +0.37 +0.45 10.71
+0.02 +0.24 + 0.24 + 0.36
t0.36 -0.03 t0.19 +0.19 +0.31
CCl, (CH3)2C0
CSHIO neat
C,
C,
+ 0.41
+ 0.41
+0.23 +0.23 + 0.40
t0.06 +0.23 + 0.23 + 0.44
+0.37 -0.02 +0.20 t0.20 +0.19
t0.18
neat
C8IO
C*
+0.46
+ 0.02
+1.29 + 0.33 +0.33 + 0.41 +0.53
CCl,
C,
+0.54
+0.23 + 0.23 +0.24
(CH3)2C0
n-dodecane
C,
+1.29 +0.33 +0.33 + 0.41 +0.47
neat n-octane
C,
t0.45
t0.01
+ 0.19 +0.12
+ 0.41 + 0.02
+ 0.06
+ 0.24 + 0.20 +0.40
TABLE 111: I3C NMR Chemical Shifts of 1-Alkenes Neat or Dissolved in CCl, and C,H,, with respect to a Solution in CH,OH compd 1-butene
solv CC1, neat
C, +1.39 t0.03
C, -0.15 -0.62
1-pentene
CC1, C,H,, neat
-1.40 +0.29 t0.20
-0.10
CC1, neat
+1.35 +0.30 +0.35
1-heptene
CC1, C,Hl2 neat
1-octene
1-nonene
1-hexene
c5
6'
c,
c9
C8
t1.31 +0.43 +0.27
-0.14 -0.62 -0.45
+0.43 +0.22 t0.20
+0.26 +0.17 +0.13
+0.47 +0.21 +0.12
t1.26 +0.43 +0.41
+1.37 +0.31 +0.44
-0.10 -0.56 -0.45
+0.41 t0.21 +0.22
+0.25 t0.22
+0.18
+0.42 t0.22 +0.23
+0.41 t0.26 +0.23
+1.29 +0.43 t0.47
CC1, C,H,, neat
+1.34 +0.29 +0.48
-0.07 -0.54 -0.49
+0.42 +0.25 10.26
+0.24 +0.25 +0.22
+0.41 +0.20 +0.26
+0.37 +0.20 +0.21
+0.46 +0.25 +0.26
t1.29 +0.47 +0.61
CCl, C,H,, neat
+1.33 +0.30
t0.55
-0.11 -0.57 -0.46
+0.42 +0.21 +0.29
+0.38 t0.22 +0.34
+0.42 a +0.34
+0.46 a +0.38
t0.37 t0.17 +0.29
+0.50 +0.26 +0.29
+1.27 +0.43 t0.68
CC1,
+1.33 t0.27 10.63
-0.11 -0.56 -0.42
1-0.42 +0.23 +0.33
+0.32 +0.18 +0.32
+0.41 b +0.41
+0.41 b +0.41
+0.41 b +0.41
+0.46 b +0.59
+0.54 b t0.49
C,H,, neat a
c4
+1.13 t0.06
+0.36 +0.19 +0.10 + 0 . 0 5
C,H,,
1-dodecene
c3 +0.45 -0.03
-0.58 -0.54
+0.39 t0.22
Signals of C,-C, coincide in C,H,, solution.
CIO
CI,
c,,
+0.37 +0.19 +0.32
+0.50 +0.23 +0.37
+1.29 +0.40 +0.73
Signals of C,-C, coincide in C,H,, solution.
C, in 1-alkenes (except 1-pentene, see below), (E)-2heptene, (E)-2-octene, and the (Z)-2-&enes with five, six, and eight carbon atoms. This means that the conformational equilibria about C&3 and C,,-C,2 in n-alkanes are the same within 1or 2% in both solvents, irrespective of the chain length. The same applies to Cwl-Cws in the 1-alkenes from 1-hexene onwards and to some 2-alkenes, see above. If, on top of the differential solvent effects,
changes in conformational equilibria of the n-alkanes would also occur, other differential shifts should be observed. One example might serve to illustrate this point. Suppose n-octane would change its conformational equilibrium on going from CH30H to CC14solution by having 4% more gauche forms about C&3 and C&,. The observed differential shift for C1/Cs should then be +1.28 - 0.04 X 5 = +1.08 ppm, in contrast with the observations
The Journal of Physical Chemistry, Vol. 86, No. 13, 1982 2521
I3C NMR Study of Conformational Equilibria
TABLE IV: I3C NMR Chemical Shifts of 2-Alkenes Neat or Dissolved in CCl, and C,H,, with respect to a Solution in CH,OH
compd (E)-2-pentene
solv
c,
c*
c3
c,
c,
CCl, C,H., ” .*
+1.43 +0.53 +0.43
+0.29 -0.18 -0.18
+0.37 +0.01 -0.05
+0.50 +0.26 i-0.16
+1.12 +0.40 t0.35
CCl, C6H,,
+1.44 +0.51 +0.47
+0.28 -0.17 -0.16
+0.37 -0.07 -0.07
+0.45 t0.27 +0.23
+0.37 +0.19 +0.15
+1.33 +0.45 +0.45
+1.41 +0.49 +0.53
+0.32 -0.12 -0.08
+0.37 -0.07 -0.03
+0.46 +0.24 +0.28
+0.28 +0.15 +0.23
+0.50 +0.23 +0.27
+1.28 +0.49 +0.58
t1.43 +0.50 t0.63
+0.29 +0.17 +0.27
+0.38 -0.07 +0.02
t0.57 40.28 +0.33
+0.33 +0.27 +0.27
+0.41 +0.18 +0.32
+0.40 +0.17 +0.17
+1.29 +0.45 +0.63
+1.49 +0.41
+0.40 -0.16
+0.40 -0.06
+0.62 +0.19
+1.27 +0.41
cc1,
t1.42 +0.46
+0.46 -0.13
+0.41 -0.04
+0.54 t0.22
+0.46 +0.18
+1.28 +0.40
CC1,
+1.46 +0.64
+0.40 -0.06
t0.46 t0.08
+0.59 +0.36
+0.37 +0.30
+0.41 t0.30
+0.50 t0.30
+1.29 +0.60
neat
(E)-a-hexene
neat (E)-2-heptene
CCl, C6H,,
neat (E)-a-octene
cc1, C6H
12
neat (2)-2-pentene
CC1,
neat (Z)-2-hexene
neat
(2)-2-octene
neat
(vide infra). Similar conclusions can be reached by comparing differential solvent effects of the methylene signals in n-alkanes, the 1-alkenes from 1-hexene on, and the already mentioned 2-alkenes. For propane, in which no anti-gauche conformations occur, we measured also a +1.29 ppm difference in solvent effects for C1 between CH30H and CC14. Apparently, this number is hardly dependent on the degree of substitution of the neighbor atom. This is in agreement with the results of the differential solvent shift of C1 in 2,4-dimethylpentane: +1.27 PPm. The above conclusions regarding conformational equilibria of n-alkanes in CH30H and CCl, are qualitatively in agreement with the result of Abrahameg Linear relationships between the energies of solution of the n-alkane series C3to C8 in a number of solvents, including CH30H and CC14, would hardly be expected when the solutes would assume different conformational equilibria in the two solvents. This applies perhaps even more to the linear relationships between AG,” and a solute parameter related to solute radii, as postulated by the same a ~ t h o r . ~ Next, the neat hydrocarbons will be discussed. The signals of C1 and C, of n-alkanes, C, of 1-alkenes,and C1 and C, of 2-alkenes in the neat liquids are always shielded, compared with their expected positions based on the chemical shifts in CH30H and CC,, combined with a linear g2 dependence (see Figure 1). The deviations tend to increase with increasing chain length. At first sight, it seems attractive to attribute this to increasing amounts of “coiling’bnear the chain ends, in accordance with a number of theoretical predictions and also with some physical measurements like vibrational spectra.15 In order to check this point experimentally, we measured also the 13CNMR chemical shifts of propane in the same series of solvents. The signals of C1 and C3 of propane dissolved in n-C, ( x = 5-8,12) deviate from the line drawn through the shifts of C1 and C3of propane dissolved in CH30Hand CCl, in a linear g2 plot in the same way as C1 and C, measured in neat n-C,. A posteriori our results published in 1974’” can be explained similarly after corrections for differences in magnetic susceptibilities. It appears that the C, to C4 and the C, to Cw3 parts of n-alkanes do not change conformational equilibrium by more than 2-3%, irrespective of whether solutions in CH30H and CCl, or neat liquids are considered. Similar conclusions can be
6‘
c,
c*
reached for C, to Cw3in the 1-alkenes studied here, as well as the C, to Cws parts of the E isomers of hexene-2, heptene-2, and octene-2. For the n-alkanes this conclusion is consistent with the results of Abrahamgwho found similar regularities as cited above for CCl, and CH30H as solvent and also for the solvents n-hexane and n-decane. There is also overall agreement with Halpern et a1.2who noted that ”n-hexane, itself an n-alkane, would not change the conformational equilibria of n-alkane chains with 4-20 methylene groups by means of solvent-solute interactions”. Our conclusion is also not necessarily at odds with the results achieved by thermodynamic means.&’ Patterson et a1.6 noted already that increasing orientational ordering, often referred to in the explanations of trends in heats of mixing5or optical anisotropies,12bmay well be due to increasing densities of packing of longer n-alkanes rather than to increasing molecular anisotropies. Patterson et al. concluded also that x12contains contributions from dispersion forces as well as correlation forces.6 Similar remarks were made by Schoen et al. who referred to “dielectric effects” and ”hard-core repulsive interactions”, re~pective1y.l~~ When one component of a binary mixture consists of (nearly) spherical molecules, then the experimental xlz or heat of mixing hE reflects the degree of order in the second componenta3That means that differences in xI2or hE values between different solutes in the same globular solvent would be a measure of differences in the order in neat solutes. In many cases cyclohexane was used as an “order destroyer” with the homologues n-alkanes or linear polymers as partners.”’ Now “order” itself may consist of two contributions: intramolecular terms (anti vs. gauche conformers) and intermolecular interactions (see above). The heats of mixing of n-pentane and n-hexadecane with cyclohexane are ca. 240 and ca. 800 cal/mol, respectively.6 If the difference of ca. 560 cal/mol would reflect only intramolecular (conformational) differences between neat n-pentane and neat n-hexadecane, then the effect would be clearly visible in the I3C NMR spectra of the neat liquids, in contrast with our results, see above. In fact, changes of 1-2% in the conformational equilibria about a certain bond would result in shift changes of the order of 0.1 ppm in the 13C NMR spectrum, well within our limits of detection especially since three to six signals should shift independently (see above). Moreover, the
2522
The Journal of Physical Chemistry, Vol. 86, No. 13, 7982
resulting “disordering energies” should be approximately linearly dependent on the number of internal bonds within the solute molecules that can undergo anti-gauche isomerism. This is not the case. On the other hand, if intermolecular contributions dominate, the energy required to “unzip” a pair of solutes should depend on chain lengths, in qualitative agreement with the results of Patterson.6 Our results cannot easily be reconciled with those of Thomas Myers.’O Unless systematic errors in the polarizabilities and/or ionization energies are present, his results point to increasing deviations from all-anti conformations for increasing chain lengths. There is also a lack of agreement with the vibrational spectra results of Schoen et al.,15awho do not give quantitative i n f ~ r m a t i o n . ~ ~ Although the gross results of molecular dynamics calculations regarding gas phase vs. condensed phases for n - a l k a n e ~have, ~ ~ ~as , ~yet, ~ not been disputed, it seems certain that the method will fail quantitatively for small solutes in large solvent molecule^.^ For “pragmatic applications” like NMR, Brownian dynamics were recommended.3 Remarkably enough, one of the groups working on that particular methodz5concluded very recently that “the distribution of each dihedral angle within a n-alkane system is insensitive to increasing chain lengths”. This is in complete agreement with the 13CNMR results presented in this paper. Using skeletal alkane models (molecular mechanics), Weber already came to a similar, though less explicit, statement.22bOn the other hand, Karplus et al. concluded that slightly more anti conformations would be present in n-heptane than in n-butane.17 Quite recently, however, the entire matter of correlation of molecular orientation in liquid n-alkanes was questioned, based on computer simulations.26 In one case partial molal volumes of, e.g., n-alkanes were compared with calculated numbers of gauche interactions per solute molecule.& So that consistent results for neat n-alkanes and for n-alkanes dissolved in CC4, C6Hs, C2H50H, and C6H12 could be obtained, it had to be assumed that for the latter solvent the number of gauche interactions would increase more rapidly with chain length of the n-alkanes than in the other solvents. In other words, a (24) C. S.Hsu, L. R. Pratt, and D. Chandler, J. Chem. Phys., 68,4213 (1978). (25) M. R. Pear and J. H. Weiner, J. Chem. Phys., 72, 3939 (1980). (26) M. Vacatello, G. Avitabile, D. Corradini, and A. Tuzi, J . Chem. Phys., 73,548 (1980). (27) After this work was finished, a Raman study of neat hexadecaiie and of hexadecane dissolved in CDC1, was published.2s In contrast with earlier work by another groupm using the same technique, it was concluded that, upon dissolving n-hexadecane in CDC13or another spherical solvent like CC4, the chain would adopt about 5% more gauche conformers. The conclusions described by us are in complete agreement with those of FischeP and thus at odds with those of Wunder and Merajver.% A conformational change of 5% would certainly have resulted in different Ab’s for, e.g., n-dodecane. At present we are not able to explain the different results, assuming that n-dodecane and n-hexadecane behave similarly. We do question, however, the validity of the calibration curves presented in ref 28. (28)S. L. Wunder and S. D. Merajver, J . Chem. Phys., 74,5341 (1981). (29) E. W. Fischer, G. R. Strobl, M. Dettenmaier, M. S t a ” , and N. 26 (1980). Steidle, Discuss. Faraday SOC.,
van de Ven et al.
larger degree of “coiling” of n-alkanes in cyclohexane was assumed. On the other hand, Abrahamg included cyclohexane in his series of solvents without taking recourse to extra coiling. Another explicit reference to n-alkane solutions in cyclohexane can be found in the SANS work of Dettenmaier14 who also concluded that there were only minor differences between n-C32 in the melt and cyclohexane solution. Our 13CNMR results support this latter conclusion. Although sizeable shielding deviations are found in cyclohexane solutions for the shift of C1 of the n-alkanes from the “CH30H-CC14 line”, these deviations were also found for C1 in propane and for the methyl signals of the di- and trimethylcyclohexanes studied here. The deviations of internal methylene carbon signals were matched by those of neat cyclohexane and also by the signals of the ring carbons of substituted cyclohexanes dissolved in cyclohexane. Finally, similar results were obtained for the alkyl fragments of the alkenes. Remarkably, the deviations of, e.g., all methyl signals in conformationally homogenous compounds like the substituted cyclohexanes and also in flexible chain molecules are the same. This indicated once more that no conformational changes are detected by 13C NMR when dissolving long chain alkanes in cyclohexane (see above). Finally, the results for alkenes will be discussed in view of the observed anomaly of pentene-1 containing alkenes in GC retention indices.lg The reason for these anomalies was supposed to be related to preferential ringlike conformations of these alkene parts in the stationary phase, although quantitative data were not mentioned. It seems best to compare first shift differences of the CH30H-CC1, solutions. The results show that C5 of 1-pentene shows a shift difference which is larger than that for any of the n-alkanes and also larger than that of C, in the other 1-alkenes studied here. The difference is, however, only just outside the experimental uncertainty. On the other hand, C5 (methylene) in 1-hexene through 1-nonene and in 1-dodecene does not show this deviation, but it is present for c6 in (E)-2-hexene with respect to the other (E)-2-alkenes measured here and it is absent in (2)-2hexene (see also Tables I11 and IV). These regularities correspond to the anomalies in GC. At present, we are unable to either confirm or refute the proposed cyclic conformations for alkenes containing a 1-pentene fragment. The deduction of gauche-anti equilibria itself of n-alkanes in the condensed phase or in solution will require additional experiments, such as 13C labeling of alkyl chains combined with measurements of the 13Cto H spin coupling over two and three bonds. Acknowledgment. The authors are indebted to Mr. F. den Otter who performed most of the measurements concerning substituted cyclohexanes. This investigation has been supported by The Netherlands Foundation for Chemical Research (SON) with financial aid from The Netherlands Organization for the Advancement of Pure Research (ZWO). A.B. thanks the Scientific Exchange Agreement (SEA) for financial support.
