J . Phys. Chem. 1989, 93, 1733-1737 the other hand the /3 values for both these carbon and hydrogen atoms are relatively low, for symmetry reasons. The average anisotropy values in Table I1 for the carbon classes are ordered as /3c,!p~ > /3c,sp3 > /3c,sp, the same ordering observed for their associated hydrogens. IV. Conclusions For the extensive set of APT’s examined in this work the following conclusions can be made. Comparisons of APT’s for different molecules can be made by using p and /3 values.instead of values for all five tensor invariants with only a small loss of statistical information. Effective charge values are strongly correlated with the absolute values of the mean dipole moment derivatives. The use of p values for comparisons with molecular structural parameters is suggested since these values contain algebraic sign information. For the Mulliken charge parameters obtained from ab initio molecular orbital calculations the p in-
1733
variant has the additional advantage of summing to zero for all the atoms in a molecule. However comparisons of effective charge values with more commonly used molecular parameter values can be useful if care is taken to restrict studies to groups of molecules for which sign changes of the effective charge invariant are not expected. The mean dipole moment derivative values correlate well with electronegativity parameter values for various subsets of the APT set studied here. Differences in the anisotropy values observed for the APT’s can be related to the electronic environment of atoms and bonds adjacent to the atom being studied. Larger anisotropy values are encountered for’APT’s of atoms adjacent to an unsaturated bond or to an atom with a polarizable lone pair than for APT’s of atoms in a completely saturated environment. The p and /3 values allows some class discrimination for APT’s of the same kind of atom which can be useful as a guide for polar tensor transference procedures.
Evidence from Alkali-Metal NMR Spectroscopy for Ion Pairing in Alkaline Silicate Solutions A. V. McCormick, A. T. Bell,* and C. J. Radke Center f o r Advanced Materials, Lawrence Berkeley Laboratory, and Department of Chemical Engineering, University of California, Berkeley, California 94720 (Received: March 4, 1988; In Final Form: July 15, 1988)
Interactions between alkali-metal cations and silicate anions in silicate solutions were investigated by cation NMR spectroscopy. The concentration of cation-anion pairs was determined from the chemical shift and the resonance line widths. The concentration of ion pairs involving large anions increases with increasing cation size. For silicate solutions in which there is a distribution in cation size, the concentration of ion pairs exhibits a maximum with increasing cation size. This maximum is interpreted in terms of the selectivity of ion-pairing reactions.
Introduction Alkali-metal cations are known to influence the chemistry of silicate anions in alkaline aqueous solution. Information obtained from 29SiN M R , Raman spectroscopy, and trimethylsilation/ chromatography has shown that, for a fixed silica concentration and silicate ratio, the molecular weight of the silicate anions increases with the size of the cation.’” Cation size has also been observed to influence the reactivity of silicate anions with aluminate anions“ and with this the selectivity of zeolite ~ynthesis.~ While the exact means by which cations affect the chemistry of silicate anions is not known, there is increasing evidence that cation-anion pairing plays an important role.’,* Because of this, there is a strong interest in learning more about the pairing between alkali-metal cations and silicate anions. The chemical environment of alkali-metal cations can be probed by N M R spectroscopy. Each alkali metal has a high abundance of an NMR-active isotope. Interactions of the cations with water (1) McCormick, A. V.; Bell, A. T.; Radke, C. J. In Perspectives in Molecular Sieue Science; Flank, w. H., Ed.; American Chemical Society: Washington, DC, 1988; ACS Symp. Ser. No. 386, p 222. (2) Dutta, P. K.; Shieh, D. C. Zeolites 1958, 5 , 135. (3) Ray, W. H.; Plaisted, R. J. J . Am. Chem. SOC.1986, 108, 7159. (4) Durouane, E. G.; Nagy, J. B.; Gabelica, Z.; Blom, N . Zeolites 1982, 2, 299. (5) Dent Glasser, L. S.; Harvey, G. J . Chem. SOC.,Chem. Commun. 1984, 1250. ( 6 ) Gabelica, Z.; Blom, N.; Derouane, E. G. Appl. Catal. 1983, 5 , 227. (7) Barrer, R. M. The Hydrothermal Chemistry of Zeolites; Academic Press: London, 1982. (8) Lok, B. M.; Carman, T. R.; Messina, C. A. Zeolites 1983, 3, 282.
0022-3654/89/2093-1733$01.50/0
and anionic species are reflected by the chemical shift and the width of the cation line.g This paper describes the use of N M R spectroscopy to characterize the interactions of alkali-metal cations with hydroxide and silicate anions in alkaline solutions. The experimental data provide direct evidence of cation-anion pairing and the effects of silicate anion size on its effectiveness in forming ion pairs. Experimental Section Alkali-metal silicate solutions were prepared from Baker analyzed Si02 gel, reagent grade alkali-metal hydroxides, and deionized water. Stock solutions at about 3 mol 76 Sios, [Si02]/[M20] = R = 3.0 ( M = alkali metal), were aged in polypropylene bottles for periods of several weeks to assure that all the SiO2 dissolved. Samples were then formulated with extra water and base to achieve a desired composition and were further aged for several days. The N M R spectra of 7Li, 23Na, 39K, 87Rb, and 133Cswere recorded on a Bruker AM-500 spectrometer. Care was taken to utilize sufficiently large spectral widths to avoid foldback of peaks. Chemical shifts were externally referenced to dilute salt solutions, and no fieldlfrequency lock was employed. Results Alkali-metal N M R spectra were taken of both silicate and hydroxide solutions at room temperature. Only a single resonance (9) Deverell, C. In Progress in NMR Spectroscopy; Emsley, J. W., Feeney, J., Sutcliffe, L. H., Eds.; Pergamon: Oxford, 1969; Vol. 4, p 235.
0 1989 American Chemical Society
1734 The Journal of Physical Chemistry, Vol. 93, No. 5, 1989
McCormick et al.
3 0-
- 25-2 2 0 E
-
-
Is i l i c a t e hydroxide
-
0
[ Na20]
I
I
i
I
I
I
I
i
2
3
4
5
1
I
I
1
I
7
8
9
IO
Figure 4. 13'Cs NMR chemical shift as a function of the base concentration with (D) and without ( 0 )about 3 mol % Si02.
I
I
I
I
6 [ C S ~ O ] (mol%)
(mol%)
Figure 1. 23NaNMR chemical shift as a function of the base concentration with (D) and without (0) 3 mol % SO2. I
I
P
LzlF 2
6-
~~
I
-
0
Y
n
Stability
hydroxide
IV I
4 k A i
'j
2
8
;
1
2
3
4
IO
Figure 2. 39K NMR chemical shift as a function of the base concentration with (D) and without (0) about 3 mol % Si02.
-
1
I
2
I
i
3
4
I
I
5 6 7 [ R b 2 0 ] (mol%)
I
I
i
8
9
1
1
0
1
2
3
4
5
6
fwhm, Hz cation 7Li 23Na 39K 87Rb '33Cs
i
0 6 Y
TABLE I: Cation Line Widths
[K20] ( m o l % )
silicate hydroxide
5
[ No,O] (mol%) [ C S , ~ ] (mol%) Figure 5. Hydroxide anion concentration as a function of base concentration in silicate solutions with 3 mol % SiOz. The straight line indicates the hydroxide concentration expected in the absence of SO2.
-
Is i l i c a t e
I
0
-
a
2 0I 0
0
Figure 3. *'Rb NMR chemical shift as a function of the base concentration with (D) and without ( 0 )about 3 mol % Si02.
line was observed for each sample, indicating that the exchange between the individual chemical environments is fast compared to the frequency separation of the isolated chemical shifts.1° Figures 1-4 show the effect of alkali-metal oxide concentration on the chemical shift of the alkali-metal cation, for alkaline solutions with and without the presence of dissolved SO,. The chemical shift for 7Li is not shown since it was too small to observe reproducibly without an internal standard. All chemical shifts are referenced to that for the nearly fully hydrated cation observed in a 0.1 M M O H solution. The chemical shift for the hydroxide solutions is a linear function of base concentration, which passes through the origin. ( I O ) Lindman, B.; Forsen, S. In NMR and the Periodic Table; Harris, R. K., Mann, B. E., Eds.; Academic Press: London, 1977; p 129.
0.1 mol 5% M,O 1 .o 8.0 5.0 95.0 1 .o
1.5 mol %
1.o 9.3 7.2 260.0 1 .o
M,O
1.5 mol % M20, 3.0 mol % SO, 6.8 71.5 53.6 556.0 7.0
At low M 2 0 concentrations, the addition of S O 2 to the basic solution deshields the cation nucleus. With increasing M 2 0 concentration, the chemical shift approaches that observed for the hydroxide solutions and eventually drops somewhat below it. Finally, at high MzO concentrations, the slope of the curve for each silicate solution approaches that of the line for the corresponding hydroxide solution. For a constant M 2 0 concentration the cation shift was found to be only a weak function of the S i 0 2 concentration in the range 1-3 mol %. The minimum MzO concentration to produce a stable solution free of gel is indicated in Figures 1-4. While cation chemical shifts were not measured at this critical M 2 0 concentration, a value can be estimated by extrapolating in data obtained at higher M 2 0 concentrations. The error bars in Figures 1-4 indicate the uncertainty in the chemical shift at the critical M 2 0 concentration. Figure 5 illustrates the variation in OH- concentration as a function of M 2 0 concentration for the Na and Cs silicate solutions. The data shown in this figure are based on measured values of pH corrected for ion interference at the measuring electrode surface.' It is apparent that the OH- concentration is only a weak function of base composition. At low base concentrations, the OH- concentration rises slowly, but above 2 mol 3'% M20, the curve rises toward the values expected for hydroxide solutions in the absence of S O z . Notice that at the MzO concentration corresponding to the stability limit, the OH- concentration is very low. As a consequence, virtually all of the negative charge resides on highly condensed silicate anions. The effects of base concentration on the fwhm (full width at half-maximum) of the cation peak are presented in Table I. In the absence of Si02, the fwhm increases with increasing base
The Journal of Physical Chemistry, Vol. 93, No. 5, 1989 1735
Ion Pairing in Alkaline Silicate Solutions concentration. This concentration dependence of the fwhm is well-documented for alkali-metal cations in salt solutions9 and is attributed to cation-anion pairingg Table I shows that the fwhm increases considerably upon addition of S i 0 2 to a fixed solution of base, indicating that in the presence of silicate anions the cation spin relaxes more rapidly than in the absence of these ions.
0.5
- 04 ?
3 0.3 (D
Discussion Chemical Shift. The influence of the cation environment on the chemical shift of a cation can be interpreted by using the theory of Kondo and Yamashita.Io For a cation participating in rapid exchange between different environments, the total chemical shift, 6, is given by
E
-- 0.2 X
4
m
5 0.1 m
r
C (
I
6
l
0.8
l
l
1.0
1
I
I
I
~
1.2
I .4 C a t i o n Radius
1
1
l
1
2.0
1.8
I 6
(8)
Figure 6. Silicate contribution to the cation shielding,f3(X3- A,) of eq
where S is a property of the alkali-metal nucleus representing its susceptibility to magnetic shielding?J is the fraction of the cations occupying the chemical site i, and Xi is a measure of the overlap integral for the outer orbitals of the cation and the complexing species. Three cation environments are of interest in the present study. These are designated as i = 1 for the fully hydrated cation, i = 2 for the cation interacting with a hydroxide anion, and i = 3 for a cation interacting with a silicate anion. If chemical shifts are referenced to the shift for the fully hydrated cation, then eq I can be rewritten as
A6 = S[@2- X d f 2 +
0 3
- XJf31
(11)
In the absence of dissolved S O 2 , the cations (M’) can exist in either a fully hydrated state or as part of an ion pair with OHanions. The equilibrium between these two states is represented by the reaction M+.6H20
+ OH- = M+OH-.5H20 + H 2 0
(1)
In the limit f2 Lfi(X2
- 1l)lhyd
(Iv)
where the subscript si1 indicates the silicate solution and subscript hyd indicates the hydroxide solution. The relationship between and cf2)tyd can be established by taking into account the following factors. First, since the SiOH group is more acidic than H20,12we expect S i O - to be less electron donating than OH-, and consequently that X3 < X2. Second, the observation that for [M20] I3.0-3.5 mol % the pH of the silicate solution is lower than that of the hydroxide solution of the same ionic strength Cf2)hyd; in other words, at a constant ionic strength, the cations bind to a greater extent with silicate anions than with hydroxide anions. (1 1) Abragam, A. The Principles of Magnetic Resonance; Oxford University Press: London, 1963. (12) Iler, R. K. The Chemistry o f s i l i c a ; Wiley: New York, 1979.
IV, at the solution stability limit ( R = 3.0) as a function of cation radius.
a
4t-LuhJ 0 0
1
2
[CS,O]
3
4
5
6
(mol%)
Figure 7. Determination of the silicate contribution to the cation shielding (As,) of 23Naand I3’Cs: (a) hydroxide contribution to the cation shielding as a function of base concentration; (b) silicate contribution to the cation shielding as a function of base concentration.
The relative affinity of different cations for silicate anions can be determined from the value of (A6)sil, obtained by extrapolation of the data in Figures 1-4 to the solution stability limit. For this limiting condition, the contribution of OH- to the total chemical shift in minimal since the concentration of OH- approaches zero. A value of f 3 ( X 3 - XI) can then be determined by dividing the extrapolated value of (A& by S. Using values of S reported in ref 9, we obtain the results presented in Figure 6 . Since X3 is roughly independent of cation size (p 290 in ref 9), Figure 6 can be interpreted as a plot of f3 versus cation size. Therefore, we conclude that the value off3 tends to increase with increasing cation size. This trend contradicts Bjerrum’s model of ion pairing,I3 which predicts that the number of ion pairs should decrease monotonically with increasing cation size. One reason why Bjerrum’s model may not hold is that large, polarizable cations interact in a delocalized fashion with large, rigid silicate anions.I4 Therefore, one might expect Cs+ ions to bind more effectively with the large silicate anions present a t the stability limit. The contribution of silicate anions to the total chemical shift at M 2 0 concentrations other than zero can be determined and will be illustrated for the cases of Na+ and Cs+. As indicated by eq 11, the total chemical shift is comprised of contributions attributable to hydroxide and silicate anions. Since the OHconcentration in the silicate solutions is known (see Figure 5), the contribution of OH- to the chemical shift can be determined from the known dependence of A6 on OH- concentration for hydroxide ( 1 3) Bockris, J. O’M.; Reddy, A. K. N. Modern Electrochemistry; Plenum: New York, 1970. ( 1 4) Liebau, F. The Structural Chemistry of Silicutes; Springer-Verlag: Berlin, 1986. (15) Weldes, H. H.; Lange, K. R. Ind. Eng. Chem. 1969, 61(4), 29.
McCormick et al.
1736 The Journal of Physical Chemistry, Vol. 93, No. 5, 1989
solutions. The resulting values for A62 are shown in Figure 7a. The OH- contribution can then be subtracted from the total chemical shift to isolate the contribution due to the silicate anions, Ah3. Figure 7b shows the results. The behavior of the curves shown in Figure 7b can be ascribed to the dependence off3 on [M20],since the value of X3 was found to be roughly independent of the silicate ratio. (The slope of the chemical shift with total solute concentration was constant for different silicate ratios.) It is evident that for Na', and to a lesser extent for Cs',f3 passes through a maximum as the concentration of M 2 0 increases. To understand this trend, we must first establish the relationship betweenf3 and the concentrations of silicate anions in solution. By definition
4c
4
?-. 3 -
1
N \ U
rum
U
Y
10
2-
r
I -
0 016 ' 01 8 '
I1O
'
11 2
'
11 4
Cation R a d i u s
" 16
'
11 8
'
2J0
(1)
Figure 8. Effect of cation size on the silicate contribution to the cation fwhm enhancement in silicate solutions with R = 2.0.
where [M+IHz0is the concentration of the hydrated cation, [M+IOH-is the concentration of cations in ion pairs with OH-, and [M+]siO,- is the concentration of cations in ion pairs with silicate anions. For the conditions of the present experimentsf3
