Proton magnetic resonance study of aluminum chloride and aluminum

(Received August Z5, 1969). Solutions of AICI3 in acetonitrile were studied from —40 to 80° by proton magnetic resonance. The apparent solvation nu...
0 downloads 0 Views 434KB Size
P M R OF

743

AlCl3 AND Al(ClO4)3 in ACETONITRILE

Proton Magnetic Resonance Study of Aluminum Chloride and Aluminum Perchlorate in Acetonitrile1 by James F. O'Brien and Mohammed Alei, Jr. University of CalifoTnia, Los Alamos Scientific Laboratory, Los Alamos, New Mexico 87644

(Received August 26, 1969)

Solutions of A1C18 in acetonitrile were studied from -40 to 80" by proton magnetic resonance. The apparent solvation number of 1.5 was verified by integration of free and bound proton resonances. The value of the rate constant for exchange of solvent from the Al(II1) coordination sphere was 8.1 sec-1 at 25". The enthalpy and entropy of activation were 19.1 +Z 0.4 kcal mol-' and 9.7 & 1.3 cal deg-l mol-', respectively. Proton nmr was also used to show that the apparent solvation number of Al(II1) in acetonitrile when perchlorate is the anion is 2.9 z?= 0.3. The result has been observed before and did not seem to vary significantly with temperature. Evidence was found for several different species in the aluminum perchlorate solutions, indicating that 2.9 is an average value for the solvation number.

Introduction Al(II1) has been shown by pmr to have solvation numbers of 6 in water,* dimethyl sulfoxide (DMSO),3 dimethylformamide (Di\/IF),4and a mixed solvent of ~ a t e r - D h I S O . ~In all of these cases the anion, C1- or Clod-, is completely displaced by solvent from the metal ion coordination sphere. In acetonitrile, however, the Al(II1) species in solution appear to depend upon the anion. In solutions of A1(C104)3in acetonitrile, the Al(II1) has been reported6 to have an average solvation number of 2.8. The suggestion has been made, supported by evidence from infrared spectra,6 that the perchlorate ion competes with CHSCN for coordination shell positions. A recent study7 of AlC13 in CH3CN using 27Aland 'H magnetic resonance pointed out that two 27Alresonances exist in a 3 : 1 area ratio, that there is only one A1 containing species that also exhibits a proton resonance, and that the apparent solvation number, determined by integration of the proton peaks, is 1.5. The author proposed7

A ~ ( C H S C N ) ~(1) ~+ This is consistent with the observation of Raman lines due to A1C14-.* I n addition to the apparent anion effect on the Al(111) solvation number in acetonitrile, it is also possible that both the solvent exchange rate and the chemical shift of bound protons are affected by the anion. Table I contains rate constants and chemical shifts for DMF solutions of aluminum salts. The rate constants are for exchange of one solvent molecule and have been adjusted to 25" for easy comparison. The solution containing C1- has a rate constant larger than the others by a factor of about 30. In all cases listed in Table I,9r10 the exchanging species is A1(D?(/IF)G3+.Movius and

Matwiyoff lo have concluded that anions in the second shell affect solvent molecules in the first sphere. The purpose of the present work was to confirm the reported values of the solvation numbers and to investigate the possibility of an anion effect on the solvent exchange rate and chemical shift of bound acetonitrile molecules. Table I : Nmr Data for AI(II1)-DMF Solutions Salt

AlCls AlBra A113 A1(C101)s

ka, seo-1

11.O sec-1 0.36 0.38 0.20

db,

HZ

49

41 31 17

Reference

9, 10 9, 10

9, 10 4, 10

a 25'. b S is the shift of the bound formyl proton from the free formyl resonance.

Experimental Section Acetonitrile was purified by successive distillations from PtOs under argon, followed by a simple distillation. The starting material for preparation of solutions of aluminum chloride in acetonitrile was AlC13 prepared by reaction of 99.998% A1 metal with Clz at 400". Care (1) Work done under the auspices of the United States Atomic Energy Commission.

(2) R. E. Schuster and A. Fratiello, J . Chem. Phys., 47, 1554 (1967). (3) S.Thomas and W. L. Reynolds, ibid., 44,3148 (1966). (4) W.G.Movius and N. A. Matwiyoff, Inorg. Chem., 6,847 (1967). (5) A. Fratiello, R. E. Lee, V. M. Nishida, and R. E. Schuster, J . Chem. Phys., 47,4951 (1967). (6) L. Supran and N. Sheppard, Chem. Commun., 832 (1967). (7) J. F.Hon, Mol. Phys., 15,57(1968). ( 8 ) C.D.Schmulbach, J . Inorg. Nucl. Chem., 26,745 (1964). (9) A. Fratiello and R. E. Schuster, J . Phys. Chem., 71, 1948 (1967). (10) W.G.Movius and N. A.Matwiyoff, ibid., 72,3063 (1968).

Volume 74, Number 4 February 10, 1970

JAMES F. O'BRIENAND MOHAMMED ALEI, JR.

744 was taken to exclude water from the resulting solid, and acetonitrile was then distilled onto it in vacuo. The resulting solutions was then evaporated until A1Ch 2CH3CN crystallized. This solid was analyzed for A1 by the 8-hydroxyquinoline gravimetric method, and for C1 by AgCl gravimetric. Anal. Calcd for A1C13.2CH3C N : Al, 12.52; C1, 49.37. Found: Al, 12.59; C1, 50.18. Nmr samples were prepared by dissolving this solid in acetonitrile in vacuo and sealing the nmr tubes. Aluminum perchlorate in acetonitrile was prepared by reaction of aluminum chloride and silver perchlorate in acetonitrile under anhydrous conditions. Silver perchlorate was dried at 80" in vacuo and dissolved in anhydrous acetonitrile. This solution was added to an anhydrous solution of A1C13 in CH3CN. After heating to 70" under argon, AgC1, which has a solubility product of 10-12.4in CH3CN,11 was allowed to separate out. After 2 days 99.5% of the AgCl expected on the basis of AlC13,the limiting reagent, had been collected. The remaining solution was then degassed, concentrated, and cooled. A white solid, soluble in both HzO and CH3CN, resulted. This solid was tested for C1- and Ag+ and contained neither. An aluminum analysis showed that the solid contained 5.2% Al. Assuming acetonitrile to be the only other constituent of the solid, the molecular formula would be Al(CHaCN)4.,(C104)3. This formula was confirmed by comparing the proton resonance areas of p-dioxane and acetonitrile in an aqueous solution containing known amounts of the aluminum solid and p-dioxane. The result was 4.8 acetonitrile molecules per aluminum atom. Pmr line width and shift measurements were made in normal fashion on a Varian DA-60A instrument. Samples were contained in standard 50-mm od nmr tubes, which were sealed to prevent possible uptake of water from the atmosphere. The aluminum chloride samples contained ThlS as an internal standard.

+ CH3CN* =

A1(CHaCN)s3+

Al(CH&N)b(CH3CN*)

+ CHaCN

(2)

was performed on an IBM 7094 computer. First-order rate constants were calculated for both solutions at ten temperatures from 30 to 81". Figure 2 shows the results of a least-squares analysis of the rate constants performed on a CDC 6600. The enthalpy and entropy of activation were AH* = 19.1 f 0.4 kcal mol-' and AS* = 9.7 h 1.3 cal deg-l mol-'. The rate constant for the exchange of one solvent molecule at 25" was 8.1 sec-I. The above measurements were made on the free solvent and coalesced resonances. A study12of the bound acetonitrile resonance in a 0.70 rn solution of A1C13 in CHSCN found AH* = 18.8 f 1.1 kcal mol-' and AS' = S.4 3.5 cal deg-1 mol-'. The value of 8.1 sec-l is similar to the A1C13-DMF exchange rate shown in Table I. The fact that all Al(II1) solvent exchange rate constants in H20, DMSO, and DMF are essentially the same when C1- is not present suggests that chloride ion catalyzes Al(II1) solvent exchange when acetonitrile is the s01vent.l~ We prepared an aluminum perchlorate-acetonitrile solid as described in the Experimental Section. This solid was used to make an acetonitrile solution that was

*

L

"I 5.5f

3.5t

Results and Discussion Exchange rate and solvation number studies were made on two solutions of AlCla in CHSCN, one of which was 2.01 m in Al(II1) and the other 3.71 m. At apprecbbly lower Al(II1) concentrations, insufficient linebroadening makes it difficult to obtain accurate kinetic data. Integration of bound and free acetonitrile peaks in these solutions were performed at 30, - 29, and - 38". The resulting apparent solvation numbers showed no trend with temperature and were 1.4 f 0.1 and 1.5 0.1 for the 3.7 and 2.0 m solutions, respectively. These values are in agreement with those previously reported by Hen.' Figure 1 illustrates the effect of temperature on the principal resonance in the 3.7 m solution. The chemical exchange reaction affecting the proton line width is that shown in eq 2. A complete line shape analysis using the McConnell equations for the magnetization

*

The Journal of Physical Chemistry

I

iI

2.9

,

I

1

3.3

I 3.7

I

I

4. I

1

IOYT Figure 1. Temperature dependence of the full line width a t half-height for the principal acetonitrile proton resonance in a 3.71 m solution of AIClr in CH,CN.

(11) D. C. Luehrs, R. J. Iwamoto, and J. Kleinberg, Inorg. Chem., 5 , 201 (1966). (12) J. F. O'Brien, Ph.D. Thesis, University of Minnesota, (1968). (13) Al(C1On)a in DMSO has a rate constant for the exchange of one solvent molecule of 0.31 sec-1 at 25O; S. Thomas, M.S. Thesis, Univ. of Minnesota, 1966. Al(C1Oa)a in H20 has a rate constant of 0.13 sec-1 a t 2 5 O ; D. Fiat and R. E. Connick, J . Amer. Chem. Soc., 90, 608 (1968).

745

PMROF AlCb AND A1(C104)3in ACETONITRILE

Figure 3. Bound CHaCN resonance at in 0.09 m Al(ClO&.

3.00

3.20

10%

Figure 2. Least-squares plot of solvent exchange rate constant for AlCla-CH&N system: V, values for 3.71 m AlCla; 0, values for 2.01 m AICla.

0.09 m in Al(II1). I n addition to comparing rate constants, we wanted to confirm the reported6value of the apparent solvation number of AI(CIO& in CH3CN. The low solubility of AI(C104)3in acetonitrile restricted measurements to solutions in which the area of the bound solvent resonance was very small compared to that for free solvent. Consequently, we found it more convenient and precise to compare the area of the bound proton resonance with that of the 13C satellite of the free solvent peak. The integrations were made difficult by the low intensity of the signals and their position near the very intense free solvent peak. The ratio of the bound CH&N area to the 13Csatellite area, determined at 30" with the aid of a time averaging device (CAT), was 1.95 f 0.18. This corresponds to an apparent solvation number of 2.9 f 0.3. Attempts to measure the solvation number a t temperatures down to -30" resulted in similar values. There does not seem to be a trend with temperature, although the accuracy of the integrations was such that this possibility cannot be ruled out. The value 2,s has been reported previously by Supran and Sheppard,6who attributed the low solvation number to competition by the perchlorate ion for first coordination sphere sites. Other evidence from infrared and magnetic mea~urements*4-~~ also indicates that the perchlorate group can indeed act as a ligand. The infrared evidence has been ques-

+lo"

tioned.ls However, Hon has found that addition of LiC104to solutions of A1Cl3 in CH3CN reduces the number of acetonitrile molecules in the Al(II1) coordination sphere.lB Our work and that of Supran and Sheppard6 indicates that in anhydrous solutions of Al(C104)3 in CH3CN, the perchlorate ion does compete with acetonitrile for coordination sites. The solvation number of 2.9 f 0.3 can be explained by the presence of all aluminum in the form Al(CH3CN)s(C104)3or by a mixture of species containing various ratios of acetonitrile and perchlorate in the coordination sphere of the Al(II1). I n the latter case, one might expect different chemical shifts for the various species. Figure 3 is a trace of the bound CH3CN resonance. There are clearly a t least four different kinds of bound acetonitrile, with resonances separated by about 0.8 He. That the multiplet is not the result of 27Al-1H coupling can be seen by the fact that such coupling would result20 in six peaks of equal intensity and peak height ratios of 138:90:115:115:90:138. Obviously then, there is a mixture of species in the solution. Evaluation of the rate constant for solvent exchange was not possible in the Al(ClO& system. The solution was too dilute to cause appreciable broadening of the free solvent peak. Measurement of the bound solvent line width was complicated by the splitting shown in Figure 3. It is clear, however, that solvent exchange is slower than in the chloride case since the bound resonance persisted to 80°, a temperature at which, as shown in Figure 1, the C1- solution gives a narrow coalesced line. Since the exchanging species in the AICla solutions has only acetonitrile in the first sphere, while in the perchlorate case it is likely that a given AI(II1) first coordination shell will contain both clodand CHICN, it is probably not valid to consider the difference in solvent exchange rate to be due to an anion effect in the same sense as in the DMF case. I n the (14) B. J. Hathaway, 2444 (1962).

D,J. Holah, and A. E. Underhill, J. Chern. Soc.,

(15) B.J. Hathaway and A. E. Underhill, ibid., 3091 (1961). (16) A. E.Wickenden and R. A. Krause, Inorg. Chem., 4,404 (1965). (17) L. E. Moore, R. B. Gayhart, and W. E. Bull, J , Inorg. Nucl. Chem., 26,896 (1964). (18) N.A. Matwiyoff, P. E. Darley, and W. G. Movius, Inorg. Chern., 7 , 2173 (1968). (19) J. F.Hon, private communication. (20) J. A. Pople, Mol. Phys., 1, 168 (1958). Volume 74, Number 4

February 19, 1970

W. R. GILKERSON

746

DMF systems anions in the second shell affect the first shell solvent molecules. There did not appear to be an anion effect on the chemical shift of bound CHaCN protons. The bound resonance was 35 Ha downfield from free acetonitrile in A1C13 solutions and 33 to 37 Hz in the A1(Clo4)3solutions. The position in the perchlorate solutions is a bit surprising in view of the mixed species in the coordination sphere* There is a spectrum6 Of Al(C104)3in H20-CH3CN which shows the bound CHB-

CN at about 57 Hz downfield from free CHBCN. It is interesting to note, however, that the bound proton peak for acetonitrile when (NH3)2Pt(C104)2 is the solute is also 35 Hz downfield from bulk CH3CN.21

Acknowledgment. The authors wish to acknowledge the valuable assistance of A. E. Florin in obtaining spectra and discussing the results. (21) J. F. O'Brien, G. E. Glass, and W. L. Reynolds, Inorg. Chem., 7, 1664 (1968).

The Importance of the Effect of the Solvent Dielectric Constant on Ion-Pair Formation in Water at High Temperatures and Pressures by W. R. Gilkersonl Department of Chemistry, University of South Carolina, Columbia, South Carolina 89808 (Received J u n e 84,1969)

The contention of Marshall, Quist, and coworkers that ion-pair formation of a number of electrolytes in water at high temperatures and pressures does not depend on changes in the solvent dielectric constant but only on changes in the solvent density and temperature is examined and rejected. Their data are analyzed in terms of the Gilkerson modification of the Fuoss equation for the ion-pair dissociation constant. The data reported by Marshall and coworkers is uniquely suited to experimental separation of the effects of specific ion-solvent and ion-pair-solvent interaction and the effects of changes in solvent dielectric constant. Distances of closest approach are obtained which are somewhat smaller than interionic distances in crystals. Marshall and coworkers12 in reporting a series of studies of the electrical conductances of aqueous electrolytes up to 800" and 4000 bars, have proposed that changes in ion-pair formation in solution at constant temperature are independent of changes in the solvent dielectric constant and depend only on changes in the concentration of a specifically solvating solvent species.a This view is quite contrary to that taken by most workers in the field of electrolyte solution^.^ The low densities of water (and consequent low values of its dielectric constant) at high temperatures under high pressure result in measurably low values of the ion-pair dissociation constant, Kd, for the process (for NaCl for instance) Na+, C1-

Na+

+ Cl-,

Kd

=

[Na+][C1-]y2*/ [Na,+C1-]

(1)

where [Na,+C1-] represents the molar concentration of the ion pair, Na,+ C1-, and y& is the mean ionic activity coefficient calculated using the Debye-Huckel equation. I shall be concerned in this report with the The J O U T of ~ Physical Chemistry

correlation of the values of Kd obtained by Marshall and coworkers with the physical properties (dielectric constant, density, and temperature) of the solvents used. I shall not be concerned with the treatment of the conductance data (equivalent conductance, concentration) to yield limiting equivalent conductances and the ion-pair dissociation constant15Kd. The magnitudes of the changes in the values of Kd as the solvent system changes can be illustrated by one example;2bfor NaCl a t 40", -log K d = 4.43 at d = 0.30 g/ml, dielectric (!) This work has been supported in part by a grant from the National Science Foundation, GP-6949. (2) (a) A. S. Quist and W. L. Marshall, J. Phys. Chem., 70, 3714 (1966); (b) ibid., 72, 684 (1968); (c) ibid., 72, 2100 (1968); (d) ibid., 72, 1545 (1968); (e) L. A. Dunn and W. L. Marshall, ibid., 73, 723 (1969). (3) A. S.Quist and W. L. Marshall, ibid., 72, 1536 (1968). (4) (a) M. Szwarc, Accounts Chem. Res., 2, 87 (1969); (b) D. F. Evans and P. Gardam, J . Phys. Chem., 73, 158 (1969); (c) J. C. Poirier in "Chemical Physics of Ionic Solutions," B. E. Conway and R.G . Barrades, Ed., John Wiley & Sons, New York, N. Y., 1966, p 9; (d) J. E. Prue, ref 4c, p 163. (5) Called the conventional ionization constant by Marshall, et d.