J. Phys. Chem. 1981, 85, 76-78
76
The Problem of Rotational Entropy Contributions in Carbamates and Thiocarbamates. NMR Multicoalescence and Saturation Transfer Experiments M. L. Martln," F. Mabon, and M. Trlerweller Laboratoire de Physicochlmie Mol6culaire, ERA 315 CNRS, U.E.R. de Chimie, 44072 Nantes Cedex, France (Received: June 18, 1980)
Owing to the interest in determining enthalpy and entropy of activation parameters rather than the less meaningful AGT* parameters, a great deal of effort has been devoted to performing variable-temperature experiments on rates of rotation. Large entropy contributions are currently reported and such was the case for carbamates and thiocarbamates. These compounds have been reinvestigated by resorting to multicoalescence and saturation transfer 13C (I3C) experiments, and a critical analysis of various types of kinetic experiments has been carried out. It is shown that the entropy term is nearly zero in carbamates and this situation is likely to hold for most unimolecular isomerization processes in the usual conditions. The results further emphasize the interest of multicoalescence methods and, eventually, of complementary saturation transfer experiments, in limiting the risk of systematic errors.
Introduction Although it has been regularly suggested, in the last 15 years of literature, that the entropy of activation of first-order processes associated with rotational isomerization, about a CN bond for example, is likely to be near zero, the possibility of noticeable entropy contributions is still frequently discussed. Indeed many studies do not restrict themselves to determining free energies of activation at a given temperature, AG*, but attempt to estimate the activation parameters AH*,AS*, or E,, log A. High values of AS*, e.g., > 8 J mol-' K-l, are still reported, even when the investigations have been carried out by means of the total line shape analysis (TLSA) method which benefits from the reputation of being more accurate than the one-parameter methods. Such AS* values which largely exceed the specified limits of error are sometimes published without any particular comment. In other cases they are the subject of an interpretation which usually rests upon the existence of differences in intermolecular steric or polar effects between the ground and transition states of rotation. Thus high AS* values have been reported in carbamates and thio~arbamates.l-~These effects are attributed to a conformational equilibrium which should render the rotational transition state more favorable to efficient solvation than the ground state.2 In the course of investigations concerning the influence of S-Fe or 0-Fe bonds upon the rotational barriers about the C-N bond and upon the nitrogen chemical shift of organometallic structures involving carbamates and thiocarbamates moieties we found it interesting to confirm the existence, and subsequently to determine the magnitude, of an eventual entropy contribution: We have therefore studied the activation parameters of compounds 1 and 2
+
1
2
for which high values of AS* have been published.'-3 In order to obtain valuable insight into the actual accuracy of the dynamic parameters derived, we have performed in conjuction coalescence and saturation transfer (1) A. E. Lemire and J. C. Thompson, Can. J. Chem., 63,3732 (1975). (2) P. T. Inglefield and S. Kaplan, Can. J. Chem., 50, 1594 (1972). (3) I. D. Kalikhman, N. A. Kryukova, V. A. Pestunovich, N. I. Ivanova, S. V. Amosova, and B. A. Trofimov, Zzu. Akad. Nauk SSSR, 1388 (1976). (4) F. Mabon, E. Roman, D. Astruc, and M. L. Martin, results to be
published.
0022-3654/81 /2085-0076$01.OO/O
experiments in both 'H and 13C spectroscopies, and we have applied several methods of data analysis. Experimental Section Solutions of 1 and 2 in several solvents were prepared. A given solution was placed in a 5-mm 0.d. tube for proton and in a 10-mm 0.d. tube for 13C investigations. The samples were degassed by the freeze-pump-thaw technique and sealed. The spectra were obtained by using several spectrometers of different nominal frequency: Varian A-60-A (lH, 60 MHz, CW); Varian XL-100-12 (lH, 100 MHz, CW, FT; 13C,25.18 MHz, FT); Bruker WH 90 (I3C,22.63 MHz, FT); and Cameca 250 (lH, 250 MHz, CW and I3C, 62.86 MHz, FT). In lH spectroscopy, the signals from CH2C12,-OCH3,or Me4Siwere observed to control the resolution in the course of the variable-temperature experiments. Small sweep widths (1-2 Hz/cm) and low sweep rates were used to record the lH CW spectra. Similarly, in order to ensure a good digital resolution, we selected the smallest 13Csweep widths which guarantee the absence of folding back (1000-1500 Hz). Several procedures were employed to measure the temperature. The sample tube and a platinum thermometer were exchanged several times in order to associate a given line shape to a given temperature. A thermocouple in contact with the NMR tube was also previously calibrated. With the Cameca spectrometer, the temperature device was calibrated with respect to ethylene glycol and methanol standards, and the relation was checked before and after each series of experiments. Due to the large shift variations observed at 250 MHz a high degree of accuracy is reached. In general, it can be estimated that the precision, expressed in terms of reproducibility in temperature determinations, is about f0.2 K for 'H and half as good for 13Cmeasurements, and, taking into account systematic errors, the accuracy is about 0.4 and 1.0 K, respectively. The technical requirements of the "variabletemperature" method are described in ref 5. Modifications of the WH 90 DS spectrometer have been performed for working out saturation transfer experiments without resorting to an auxiliary frequency synthesizer. The second radiofrequency required to irradiate the I3C ( 5 ) (a) G. J. Martin, M. Berry, D. Le Botlan, and B. Mechin, J. Magn. Reson., 22, 523 (1976); (b) D. Le Botlan, M. Berry, B. Mechin, and G. J. Martin, J. Phys. Chem., 84, 414 (1980).
0 1981 Amerlcan Chemical Society
The Journal of Physical Chemistry, Vol. 85,No.
Rotational Entropy in Carbamates and Thiocarbamates
I, 1981 77
TABLE I: Kinetic Parameters Derived from Multicoalescence Measurementsa vo 1
T,,K nucleus MHz
1 1
[ 9002+02 [
AF
0, obswotion
~
Dqoupling COI
mixer
offset
amplifiers
0 2 decoupllng offset
~~~
341 348.5 360.5 371
'H
277.6 299 288 310
'H "C
'H 'H I3C
Figure 1.
nuclei is derived by mixing the basic 90.02 MHz + 0 2 frequency with the frequency vx = 90.02 MHz - v(iaC) delivered by the synthesizer. Two frequencies 22.63 MHz O1 and 22.63 MHz O2 are therefore available to observe and irradiate the 13Cspectrum. In order to be able to control the offset frequency for efficient proton decoupling the frequency 90.02 MHz + O2 is mixed with an AF frequency of about 3000 to 4000 Hz which depends on the value O2selected to irradiate the 13Csignals. A scheme of the device is given in Figure 1. In both types of experiments designed to measure the saturation factors, 9,and the apparent relaxation times, broad-band proton decoupling is continuously applied. Homonuclear 13C (13C)double irradiation is switched off during the acquisition time (3-4 s) and applied otherwise. A delay of 30 s is selected to ensure recovery of magnetization.
+
+
Results Compounds 1 and 2 provide typical examples of exchange phenomena involving sites separated by very large and very small chemical shifts. Thus at 263 K, Av('H) = Y A - VB = 22.8 Hz for 1 and Av = 1.8 Hz for 2 (YO = 100 MHz). Chemical shift behaviors as a function of temperature have been investigated in the slow exchange limit. solution (25% vol/vol) the Thus for 1 in a CH2C12/CD2C12 difference in resonance frequencies vA - vB exhibits negligible variation over a temperature range of 30 K. For 2 in CDCl, (50% vol/vol) variations of about 0.008 Hz/K are observed. Several methods have been used to derive the results: ( a )Line-Shape Determinations. The temperature-dependent spectra have been obtained both by the conventional procedure and by using our system of programmed temperature, "varytemp", which ensures a better control of temperature variation^.^ The curves resulting from a given temperature experiment have been studied by total line-shape analysis. In this treatment account is taken of eventual variations in the line widths and in the differences between the resonance frequencies of the exchanging sites as a function of t e m p e r a t ~ r e . Moreover, ~ owing to the persistent interest in one-parameter methods, we have also determined the lifetimes by resorting to the separation between the maxima AV,,~ the half-height line widths Av1/28 and two types of line intensity ratio^.^,^ ( b ) Multicoalescence Experiments. We have already emphasized that the joint use of proton and carbon spectroscopies may be advantageous in dynamic measurements.1° Indeed the extension of the active temperature range often accessible in multicoalescence experiments is expected to enhance the accuracy of the activation (6) M. L. Martin, J. J. Delpuech, and G. J. Martin, "NMR Practical Spectroscopy", Heyden, 1980, p 315, and references herein. (7) R. R. Shoup, E. D. Becker, and M. L. McNeel, J. Phys. Chem., 76, 71 (1972). (8) H.S.Gutowsky and C. H. Holm, J. Chem. Phys., 25,1228 (1956). (9) V. S.Dimitrov, Org. Magn. Reson., 8, 132 (1976).
A v , Hz
AGT*,
kJ mol-'
k,, s-'
~
'H I3C
Compd 1 60 13.65 30.32 100 22.8 50.65 250 57 126.62 22.63 113 251.02
74.10 74.16 74.23 74.35
Compd 2 100 1.8 25.18 14 250 4.8 62.86 36
64.48 64.57 64.65 64.61
3.998 31.15 10.663 79.97
a A v is the difference in resonance frequencies o f sites A and B corresponding to the coalescence temperature for 1 and 2, respectively.
TABLE 11: Results o f Saturation Transfer Experiments Involving Thiocarbamate la A%*, TIA,
T,K
9A
283 288 293
0.49
PB
s
Til33
s
TB,
s
kJ mol-'
s
3.70 0.55 0.69
TA,
7.55 2.4 1.7
4.36 2.46
73.81 73.85 73.80
a The saturation transfer factors q A and qB, and the apparent relaxation times, T , A , T ' B , used to derive the life times in sites A and B and the free energy of activation hGT* are measured with '3Cspectroscopy at 22.63 MHz.
parameters significantly.l0J1 We have therefore determined the coalescence of the proton and carbon signals in different magnetic fields and the results are given in Table I. ( c ) Saturation Transfer Experiments. A much greater improvement in the range of investigated lifetimes may be obtained by resorting to saturation transfer experiments which are more simply interpreted in carbon spectroscopy! Thus we have determined the factor of saturation transfer, vi = (Ii- Ia)/I~,at several temperatures by measuring the intensity of a given methyl carbon signal (i = A or B) successively in the presence (Ii)and in the absence (Ioi)of 13C (13C)saturation transfer. While broad-band proton decoupling is continuously applied a coherent irradiation field is gated with the appropriate selectivity, either at the resonance frequency of a given carbon ( I A , IB) or well outside the range of methyl resonance frequencies, IOA, IOB (Table 11). Besides the 9.4 and factors, the apparent relaxation times 71A and 7 1 were ~ obtained by performing 13Cinversion-recovery experiments in the presence of both proton decoupling and homonuclear 13Cdouble irradiation at a given methyl site. In order to avoid recourse to an auxiliary source of radiofrequencya technical modification described in the Experimental Section has been implemented to perform these experiments. Since the dynamic range of the relaxation time measurements is limited by the degree of saturation transfer, only three temperatures correspondingto moderate values of the saturation factors have been investigated in order to preserve reasonable accuracy of the results (Table 11). Discussion When the results derived separately from total lineshape analysis, one-parameter treatments, and multi(10) M. L.Filleux. F. Mabon. and G. J. Martin. Tetrahedron Lett., 3907 (1974). (11) H.S.Gutowsky and H. N. Cheng, J. Chem. Phys., 63,2439 (1975).
78
Martin et al.
The Journal of Physical Chemistty, Vol. 85,No. 1, 1981
TABLE 111: Analysis of t h e Kinetic Results Obtained b y Various Experimental Methods and Theoretical Treatments' measurement method CVTb
analysis method
TLSA f
l-Pg vary tempC MCd
1-P CF
CVT t MC t STe
C F t ST TLSA t C F + S T
solution A T , K 1-a 1-b 1-c 2-c 1-a 2-c 1-a 1-a 2-c 1-a 2-a
45.9 34 20 21.6 37.9 21.6 33.2 30 33 88 88
N
A H * , k J mol-'
AS*,J mol-' K-'
R2
18 14 I 10 48 10 20 4 4 7 25
74.82 i 0.50 72.39 f 0.5 84.35 i 2.2 86.77 f 1.25 14.9 i 0.5 95.89 i 4.10 73.72 i 1.61 72.39 f 1.80 64.90 f 0.84 72.10 i 0.30 12.34 +. 0.40
0.84 i 1.25 -4.09 t 1.21 30.55 i 6.6 78.6 i 5.0 0.84 i. 2.5 1 1 2 . 2 r 15.1 -2.93 f 1.67 -6.65 f 5.10 0 . 7 1 i 3.55 -6.32 t 0.93 -4.50 i 2.6
0.999 0.999 0.996 0.998 0.998 0.986 0.998 0.998 0.997 0.999 0.998
a A T is the temperature range investigated; N the number of points considered in the regressional analysis. ( a ) Solvent CH,Cl, CD,Cl,, ( b ) C,H,,, ( c ) CDCl,. b Conventional variable temperature. Programmed temperature. Multicoalescence. e Saturation transfer. Total line-shape analysis. One-parameter method. Coalescence formula.
coalescence methods are subjected independently to a linear regression analysis, k = f(l/T), the correlation coefficients R are usually very high PO.99) and the standard deviation on the estimation of AH* and AS* remains in the limits of f1.8 kJ mol-' and f 5 J mol-' K-', respectively. These figures characterize in fact the reproducibility of the measurements but have no real meaning as far as accuracy is concerned. Indeed, large discrepancies in the values obtained from the different kinetics experiments may occur, and the results given in Table I11 call for the following conclusions. When all the experimental parameters including the results of saturation transfer measurements are simultaneously considered a large temperature range and a high confidence level are reached. It can therefore be safely concluded that the entropy term is nearly zero for the rotational process involving the C-N bond of the carbamate and thiocarbamate 2 and 1 in the considered solvents. The occurrence of an entropy contribution associated with a strongly polar conformation in the transition state2 can therefore be disregarded. The determination of rotational barriers through total line-shape analysis for two exchanging sites may be subject to large systematic errors. In the case of 2 in a CDCl, solution, for example, a AS* value of up to 78 (f5) J mol-' K-' is deduced from a TLSA treatment corresponding to a restricted temperature range. In fact the actual accuracy of the results is far from being reflected by the precision derived from regression analysis (Table 111). The risk of errors increases when the active temperature range decreases and it becomes very important when the differences in site frequencies are small. It should be noted that such errors are expected to be attenuated when more complex, and therefore "more sensitive", spectra are concern;d.'2 When one-parameter methods are used in which the only approximation involved is the equality of the tran-
sverse relaxation rates in both sites the accuracy of the results may be comparable to that of total line-shape analysis methods. We also further confirm the advantages of the multicoalescence methodl0J' which limits the danger of systematic errors, especially for exchanging signals relatively near one another. In fact, it becomes increasingly evident, and our results on carbamates further support this idea, that the entropy of activation of unimolecular isomerization processes is likely to be very near zero in most cases. However, the fact that total line-shape analysis and other determinations of lifetime variations as a function of temperature are still currently performed in such cases means that this idea is not firmly adopted. Since intramolecular contributions to AS* are expected to be srna11l3a clear confirmation of the existence of differential solvation effects in cases where they may be reasonably suspected would certainly be of interest. In practice unless experimental methods are implemented which allow a sufficiently large range of active temperatures to be investigated the entropy parameters, even when determined through line-shape analyses, are susceptible to high systematic errors. Time wastage should then be avoided by restricting the determination of the dynamic parameters to the free energy of activation AGT* which is recognized to be accessible with a reasonable accuracy. If we keep in mind that a t 300 K an activation entropy of 8 J mol-' K-' corresponds to a variation in energy as high as 2.4 kJ mol-' the fact that many experiments do not attain this accuracy on AS* explains that, in spite of the considerable efforts developed to perform and analyze variable-temperature experiments, nearly all the discussions and comparisons of rotational processes make use solely of AGT* values.
(12) G. Binsch, "Dynamic Nuclear Resonance Spectroscopy", L. M. Jackman and F. A. Cotton, Ed., Academic Press, N.Y., 1975, p 45.
(13) C. Piccinni-Leopardi, 0. Fabre, D. Zimmermann, and J. Reisse, Can. J. Chem., 55, 2649 (1977).
Acknowledgment. The authors thank Dr. E. Roman and who prepared compounds and 2* Dr. J. p*