NOTEIS value of ‘76 cm-’, although from a theoretical point of view its value may be slightly lower in complexes, like other parameters. After this, preliminary values of B and C are determined from the two Dq independent transitions to f4&(G), 4E(G)]and “E(D)states and that for Dq isi determined from the position of 4T1(G) band which is most sensitive to the value of Dq. After obtaining this trial set, the parameters are varied over a Elmall range to give a good fit for all bands, particularly the lowenergy ones. I n most of the cases, the trial Bet itself is found to be quite good, which supports our procedure. I n Tables J and I1 results are compared with and without thle inclusion of CY term for different cases, The calculated energies without the a term for Mn(C104)2.Aq arid n/rrtO are taken from references mentioned in the tables.1s-2z For MnFz and MnC12, calculations were also performed without the CY term and reasonable set of parameters, by inspection, were chosein for the calculated energies given in the tables. The agreement is seen to be improved for most of the bands on including a and in particular for the *TI(P) band where the deviation is reduced from about 2000 cm-1 to about 260 cm-l. Regarding the relative improvement among different complexes, it is found that the energy levels in complexes with lesser amount of covalency are better described by this theory in comparison to those with higher covalency. The values of parameters were obtained in this way for all the complexes. These are given in Table I11 with the positions of the Dq-independent [4A1(G), 4E((3;3]and 4E(D) ’lev . A comparison of these and value9 of parameters and G shows the validity of nephelauxe tic series. Similarly, a comparison of Dq values shows the validity of spectrochemical series. It is found for all the MnZ+spectra that the deviations between c a l c ~ l ~ md t ~ dexpbrimental energies are larger for the higket energy bands. The theory used in the present work neglwts the interactions with higher configurations and since the higher lying energy states are closes to excited configurations, the departures from the theory re more pronounced for the bands involving these stateEl. A Icasl-squares fitting is not tried in the present work sinoe, in view of the theory being less accurate tt,t, higher energies, such a fitting makes the (13) 6. Moore, ]Vat. Bur. Stand. (U.S.), Circ., No. 467, 2 (1952). (14) A. Mehra .tnd 1’. Venkateswarlu, J . Chem. Phys., 47, 2334 (1967). (15) A. Mehra a,nd P.Venkateswarlu, ibid., 48, 4381 (1968). (16) 1%. Pappalardo, PhG. Mag., 2, 1397 (1957). (17) R. Pappdardo, 1. Chem. Phys., 31, 1050 (1959). (18) FL. Pappalardo, ibid., 33, 613 (1960). (19) A. Mehra, ibid., 48, 1871 (1968). (20) AL. Mehra, Rhyoica Statu8 Solidi, 29, 847 (1968). (21) 6.W.Pratl and R. Coelho, Phys. Rev., 116,281 (1959). (22) D.a. Muffman, R. L. Wild, and M. Shinmei, J . Chem. Phys., 59, 4092 (1969).
437 set of parameters equally bad for all the levels and is not very useful for interpretation. In our fitting, which gives preference to the lower levels, the errors from configurational mixing are minimized and the parameters may be evaluated for physical meaning in terms of the environment of the ion,
A Gas-Phase Density-Dependent Directly Bonded Coupling Constant by A. Keith Jameson* and John P. lteger Department of Chemistry, Loyola University of Chicago, Chicago, Illinois 60688 (Received September 14, 1970) Publication costs assisted bg Loyola University of Chicago
Solvent shifts of nuclear spin-spin coupling constants have previously been noted.’ Of particular interest are the relatively large shifts of 1J(29Si-’*F)in SiF4in various solvents.2 The shifts are all to larger magnitudes of J relative to the gas and are as large as 5%. One might be tempted to explain solvent shifts in J on the basis of some property of the solution (“reaction field” for instance). If this is so, then there should be no significant dependence of the coupling constant in the gas phase. If such a dependence on density could be found, it would most likely be in cases where large solvent shifts have been observed if the gas-phase shifts are caused by the same mechanism(s) as t,he liquid-phaae shifts. Also J itself should be relatively large in order that changes in J be large enough to be experimentally measurable. On the basis of these considerations, we decided to search for density dependence of coupling constants in SiF4 gas, although Goyle, et ul., found no measurable difference between the coupling constants in SiFe at 30 and 110 atm. The samples were sealed in borosilicate tubing having an inside diameter of 1.2 mm and an outside diameter of 3.9 mm. The volume was usually about 0.1 ml. When the tubes were carefully prepared and properly annealed we could safely obtain pressures as high as 200 atm. A Varian A-56/60 high resolution spectrometer was used. For SiF4,calibrated audio side bands of the 2aSi19Ffluorine singlet were impresse on both the upfield and downfield sides of each member of the 2eSi19F doublet and a linear interpolation was made. Several spectra of each sample were taken. I n all cases the standard deviation in the measured J was about k0.15 Hz. This corresponds t o measurement. of the position (1) For an extensive review see P. Laszlo in ^‘Progress in Nuclear Magnetic Resonance Speotroscopy,” Vol. 3, J. W. Emsley, J Feeney, and L H. Sutcliffe, Ed., Pergamon Press, Oxford, 1967. (2) T. D. Coyle, R. B. Johannesen, F. E. Ebrinckman, and T. C. Farrar, J . Phys. Chern., 70, 1682 (1966).
The Journal of Physical Chemistrgt, Val. 76, No. 41, 1971
438
NOTE0
A
GlLLESPlE
0 CQYLE, ET. AL.
DENSITY OF SIF, (IDEAL AMAGATS) Biguse 1. The gas-phse density dependence of pure $33'4.
/
T GAS DENSITY(IDEAL AMAGATS) Xi'igure 2. Density dependence of ?r (SiFa) in various solvent gases.
of each member of the doublet to 1 0 . 1 He. All line widths were about 0.3-0.5 HZ at, half-height, except in the Ctnzrnixtuyes in which they were about 1.5 Hz. A combination of experimental factors (low filling factor, 5% natural abundance of 29Sj)prevented our proceeding to SiFs densities lower than about 100 ideal amagats. At this density we could obtain a signal to noise ratio of about 5 with very careful tuning. In Figure 1 we see that the values obtained extrapolate nicely to a value consistent with the low density values obtained by uaiP and by Coyle, et aL2 %regihown in Figures 1 and 2. Clearly phase density-dependent coupling constant (even in the case of SiF4 itself, contrary to the findings of Coyle el ai.). If we assume.that binary collisions cause t s effect, then *re can write for a
(A
+
(s)
(B
B A-B
where Ja(0) = 0oup.tin.gconstant for the isolated mole-
cule, ( W / b c A ) A - A = density dependence of the coupling constant because of binary collisions between A molecules, ( ~ J / ~ ( B ) A=- B density dependence of the coupling constant because of A-B collisions, (A and (B being the densities of molecules A and B in the gas mixture. The results can be interpreted reasonably well in these terms as shown in Figures 1 and 2. The value obtained for (bJ/b{)aiFA-SiF, is used in determining ( a J / a ( B ) BiF+B. The effect of the second gas is independent of the effect of SiFl as was demonstrated by data on CH4 a t two different densities of SiF4 (ca. e nature of the of SiF4). We see in Figure 2 th second molecule clearly affects ( d J / d c B ) S i F * - B = It is interesting to note that when the density of the solvent gas is extrapolated to the densities of the liquids, we obtain J s i F 2 172, 174, and 175 Hz for SiF4 in COS, €IC], and CH4 a t densities of 560,730, and 590 ideal amagats, respectively. Although these gaseous molecules are different from the liquids used by Coyle, et I C ~ . the ,~ extrapolated hypothetical liquid phase values fall nicely into the range of values (170.5-178.5 Hz) they found. Therefore, it appears likely that it will not be necessary to invoke an effect peculiar to the liquid phase to explain solvent shifts in directly bonded coupling constants. We are beginning to search for other molecules in whioh there m%ybe density dependent directly bonded coupling constants. We have found no density dependence in CH4,%He, or PFs. We are in agreement with other workers on the magnitudes of IJ howeverah We looked a t CH4 alone a t densities from 20 to 200 ideal amagats and in the presence of SiF4 a t 100 and 150 ideal amagats. We covered a density range from 20 to 200 ideal amagats in SiHc and from 100 to 190 ideal amagats in PF3. In all cases except PR our line widths were 0.3-0.5Hz a t half-height. In PF, they were about 3.5 Hz a t half-height. As a consequence, in PFs our precision was not so good. Nevertheless, within our experimental error we can see no defin dence in the coupling constant for These results are summarized in Tab At the present time we are not in a position to do much more than speculate as to the cause for most directly bonded coupling constants having no density dependence. It may be that the F and spin-dipolar terms are aff binary collisions. If this is so, might give us an experimental ~ e a s ~ i of r e the terms affected most highly. We are extending our observations to more binary fluorides i an attempt to find more than a single case of denaity (3) R. J. Gillespie and J. W. Quail, J . Chem. Pkgs., 39, 2555 (1963). (4) (a) CH4: N. Muller and D. E. Pritchard, {bid., 31, 768 (1959); (b) SiHa: E.A. V. Ebsworth and J. J. Turner, %%id., 36, 2628 (1962); (0) PFa: H.8. Gutowsky and C. J. Hoffman, &id., 19, 1259 (1951).
NOTEM
439
Solute Solqen; anoIecul
