J. Phys. Chem. 1981, 85, 3542-3543
3542
substances, including the F- anion by Stillinger,2 the Li+ cation by David,3ammonia by Turner and David; as well as the phosphate anion (PO4*) and its associated anions HPO:-, H,PO,, and technically phosporic acid: and CO,, COZ-, HC03-, and, again technically, carbonic acid.6 The latter calibration provides a polarization model picture of C02 which is linear with zero dipole moment, and CO2which is perfectly triangular. The relative energies of the carbonate anion and carbon dioxide were calibrated to be in accord with experiment. Since the Li+ cation, in interacting with COl-, would interact with the carbon (C4') of CO2- through a coulomb interaction, and with the oxide anions of COZ- through previously calibrated functions, it follows that we can compute the energy and geometry of the LiC03- anion. This anion is of special interest because it has been implicated in the process of transporting the Li+ cation across cell membranes' in the treatment of manic depressives. Thus, not only does Li+ transport, in competition with Na+, K+, and H+, but, it also transports surreptitiously by imitating HC03- and travelling through the anion route to membrane transport.* It is of some interest, therefore, to compare the predicted structures of the LiCOy and HCO; anion, to see if there is any energetic or structural similarities or contrasts which might be suggestive or interesting. Of course, these computations will have to be checked by ab initio computations. The polarization model has been calibrated to produce a Cot' anion which is triangular, with a C-0 bond length of about 1.4 A and an energy of -4036.5 kcal/mol. The computed energy of the HC03- anion is -4559.3 kcal/mol. Two of the C-0 bond distances have shrunk slightly, to 1.27 A, while the third has grown to 1.78 A. The elongated C-0 bond is the one which connects C-0-H to bond the C032-to the proton. The LiC032-anion is quite differrent. The energy is computed to be -5200.8 kcal/mol. The anion has a perfect triangular CO," arrangement,with the Li-C distance equal to 0.76 A, while the C-0 distances remain constant at 1.37 A. The Li+ cation is almost perfectly equidistant from the oxide anions, with Li-0 distances of 1.50, 1.68, and 1.50
A.
The polarization model further predicts structures of hydrated ions. In this case, employing still the simplistic coulomb term between the Lit and the C4+in the model, we obtain two widely different structures for the monohydrate of LiCOy. With the water on the same side of the COS2-as the Li+, the energy is computed to be -6336.2 kcal/mol, but the Li-C distmce has skunk to 0.3 A. With the water on the opposite face from the Li+, the situation is even more bizarre. One oxide from C032-is pulled toward the water, forming On---C---O---H---O--H
I
L I+
The energy has dropped further, to -6369.0 kcal/mol. It would appear that the Li-C potential is not sufficiently repulsive, i.e., more than just a coulomb repulsion (1)F.H.Stillinger and C. W. David, J. Chem. Phys., 69,1473(1978). (2)F.H.Stillinger, Int. J. Quant. Chem., in press. (3)C. W. David, ACS Symp. Ser., no. 86,(1978). (4)P.Turner and C. W. David, J. Chem. Phys., 74,512 (1981). (5)F.H.Stillinger, T. Weber, and C. W. David, manuscript in preparation. (6)C.W. David, P. Turner, and E. Eisler David, manuscript in preparation. (7)J. Funder, D. C. Tosteson, and J. 0. Wieth, J. Gen. Physiol., 71, 721 (1978). (8) B.F. Becker and J. Duhm, J.Physiol. (London),282,149 (1978). 0022-3654/81/2085-3542$01.25/0
is required to maintain the Li-C distance in the acceptable range during hydration. For this reason, high-quality ab initio computations are now required for both the anion, and its various hydrated structures. Since anions are quite difficult to handle quantum mechanically, we feel that only a laboratory with sufficient experitise should attempt such a computation. Even if further tinkering with the Li-C potential is ultimately shown to be needed, the polarization model computations presented here will have served their heuristic purpose. Ultimately, one expects to obtain a coherent picture of this quite important substance in aqueous solution. Acknowledgment. This work was supported in part by a grant from the National Institutes of Health under grant no. GM-26525, for which I am extremely grateful. Department of Chemistry University of Connecticut Storrs, Connecticut 06268
Carl W. Davld
Received: April 2, 198 1; In Final Form: September 4, I98 1
Existence of Tetrachloroironate(111) in Hydrochloric Acid Solutions
Sir: Although several inve~tigatorsl-~ have found tetrachloroironate(II1) to be the predominant species in solutions prepared from iron(II1) chloride and hydrochloric acid, Magini and Radnai4 report this finding to be incorrect. Using solution X-ray diffraction experiments followed by extensive indirect computer calculations, they conclude that FeC14- is present only at 20-30% in aqueous FeC13-6H20-HC1solutions (but at the level of 50% in some aqueous solutions). They also propose that Fe(H2O)&l2+, Fe(H20)4C12+,and/or Fe(H20)3C13are present in all of these solutions, with trichlorotriaquoiron(II1) the most important species in the solutions prepared from FeC136H20 and hydrochloric acid. We have examined, by X-ray diffraction followed by the direct evaluation of the resulting atomic radial distribution function (ARDF),5a a solution prepared from FeC13 (Baker Analyzed, Reagent) and hydrochloric acid (Mallinckrodt Analytical Reagent) in order to determine if the ARDF method is suitable for identifying solute species present in this ionic solution, and, if so, to identify the average species in the solution. This solution, with density = 1.415 g/mL, X F ~= 0.021 and x C I / x F e = 5.2, is similar in composition to Magini solution ClA2,4 which was prepared from FeC13-6H20. The ideal peak was calculated" for potential Fe-0 and Fe-Cl atom pairs by TFe-L(r) = (2?r)-1'2unFe-LSfFe(s)fL(S) [COS (sr1) COS (sr,)lM(s) ds
+
where u = 2, r1 = r dFeL and r2 = r - dFeC1,and n is the number of nearest-neighbor ligands of type L surrounding the Fe3+. The ideal peak area (IPA) for each potential interaction was calculated by IPA = JTFeL(r) dr to be IPAFd1 y 29.0 e2 and IPAFd N 13.6 e2for this solution. Shown in Figure 1is the atom-pair correlation function obtained from our aqueous FeCl,-HCl solution. The (1)C.L.Standley and R. F. Kruh, J. Chem. Phys., 34, 1450 (1961). (2)S. K. Sharma, J. Chem. Phys., 60,1368 (1972). (3)G.A. Gamlen and D. 0. Jordan, J. Chem. SOC.,1435 (1953). (4)M.Magini and T. Radnai, J. Chem. Phys., 71,4255 (1979). (5)D. L. Wertz and G. W. Hicks, J. Phys. Chem., 84, 521 (1980). (6)D. L. Wertz and M. L. Steele, Inorg. Chem., 19,1652 (1980).
0 1981 American Chemical Society
J. Phys. Chem. 1981, 85, 3543-3545
-1 &
M
1
:: ,
--
1
O
*. .-. :: ;\ - _ _ :. : ., :. .,; ...w......... .. .. . .
~
d
3543
e
s
*,
; i’
0.0
’
-*-
C..?’..
’
**.
-
1
a-
W
*
8
6
r, A
Flgure 1. The pair correlation function, calculated by g ( r ) = 1 -t ( 2 n 2 r p J 1 ~ , b si ( s ) M ( s ) sin sr ds5.
300
5
1
Flgure 3. (a) The experimental interference curve calculated by i ( s ) = I&) - C X , ( ~from ( S )the X-ray scattering data (solid line) compared to the interference curve, j ( s ) calculated from the Fe-CI and the nonbonded CI-CI atom pairs of tetrahedral FeC.,I (b) The experimental interference curve, i ( s ) compared to the interference curve calculated from FeCI,- and three CI-0 atom pairs.
i
1.2..
16
10
S
..,,,*..
2.0
2.5
3.0
Flgure 2. D ( r ) , calculated by D ( r ) = 4a2r2p r ) , compared to the ideal peak calculated for four Fe-CI pairs per°F$+ with dFe4 = 2.26
A.
ARDF contains three well-defined peaks; a sharp symmetrical peak centered at 2.26 A (Pl),and peaks centered at 3.08 A (P2) and at 3.72 A (P3). P1 is due to nearest-neighbor Fe-ligand interactions. The location of P1 is consistent with the Fe-C1 bond distances in tetrahedral complexeP but not the Fe-0 or Fe-Cl bond distances found in octahedral complexes.1° The area under P1 in D(r) (Figure 2) is 113 f 8 e2and may be correlated with either FeC14- (tetrahedral) or ca. FeC12(H20)4+(octahedral) as the average species in this solution or any combination to these species, but T(r) calculated for four Fe-C1 pairs per Fe3+,with the Fe-CI distance being 2.26 A, is in excellent agreement with D(r) in the region of the Fe-ligand bonding. T(r) calculated from the octahedral model, i.e. four Fe-0 and two Fe-Cl pairs per Fe3+,is not in good agreement with D(r) in the 2.0-2.5-A region. P2 is due primarily, if not exclusively, to hydrogenbonded 0-C1 atom pairs12-14 and cannot be utilized to establish (or eliminate) various coordination models for Fe3+ in this solution. (7) R. R. Richards and N. W. Gregory, J. Phys. Chem., 69,239 (1965). (8) M. J. Benne, F. A. Cotton, and D. L. Weaver, Acta Crystallogr., 23, 581 (1967). (9) N. C. Baenzler, Acta Crystallogr. 4, 216 (1951). (IO) M. D. Lind, J. Chem. Phys., 45, 2010 (1967); Acta Crystallogr. Sect. E , 26, 1058 (1970). (11) J. Waser and V. Schomaker, Rev. Mod. Phys., 25, 671 (1953). (12) J. N. Albright, J. Chem. Phys., 56, 3783 (1972). (13) S. C. Lee and R. Kaplow, Science, 169, 477 (1970). (14) L. S. Smith and D. L. Wertz, J. Am. Chem. Sac., 97,2365 (1975). 0022-3654/81/2085-3543$01.25/0
P3 is attributed to Cl-Cl atom pairs. That P3 occurs at (8/3) X 2.26 A indicates that FeC14- is “tetrahedral”. P3 can be accounted for in no other manner. In Figure 3a is a comparison of the experimentally obtained interference curve, i ( ~ )and , ~the ~ interference curve calculated from the structural details (model) of FeC14-, i.e., j ( s ) where j ( s ) = CxiSf;fj47rr2pij[sinsrlsr] dr. The correlation is, at best, only general. Significantly better agreement between i(s) and j ( s ) is found when the hydrogen-bonded C1-0 atom pairs (at ca. 3.08 A) are included in the model (Figure 3b). The best correlation, Le. minimizing Ci(s)- j ( s ) and minimizing C[i(s)- j ( s ) ] : is obtained when Ncl-0 CI 3. In conclusion our results show that tetrahedral FeC14is the average and probably the only important solute species in this solution prepared from anhydrous FeC13and concentrated hydrochloric acid and that the ARDF completely identifies the one-dimensional structural details of this complex as well as the hydrogen-bonded 0-C1 interactions. No evidence of the various octahedral species reported by Magini and Radnai is found. (15) i(s) = Icoh(s)
-
cXif?(s).
Deparfment of Chemistry University of Southern Mississippi Hattiesburg, Mississippi 3940 1
Mlckey D. Luter David L. Wertz’
Recelved: July 1. 1981; In Flnal Form: September 9, 198 1
Penta- and Hexacoordlnated Sllicon Sites on Silica Surfaces
Sir: In a recent series of papers Morrow et al. described extensive and detailed infrared spectroscopic studies of silica which provided evidence that a new active site was formed when silica was degassed above 400 0C.1-6 The (1)Morrow, B. A.; Devi, A. J. Chem. Sac., Faraday Trans. I 1972,68, 403. ( 2 ) Morrow, B. A.; Cody, I. A. J. Phys. Chem. 1975, 79, 761.
0 1981 American Chemical Society