Environ. Sci. Techno/. 1993, 27, 1924-1929
Modeling Ion Adsorption on Aluminum Hydroxide Modified Silica Xiaoguang Mengt and Raymond D. Letterman' Department of Civil and Environmental Engineering, Syracuse University, Syracuse, New York 13244-1 190
The effect of component hydroxide and oxide interaction on Cd2+and so42-adsorption in suspensions of Al(OH)3/ Si02 was modeled using two kinds of surface site distributions incorporated in the triple-layer surface complexation model. The mixed site distribution assumes that Al(0H)B surface sites are uniformly distributed on the Si02 surface with the overall surface potential contributed by both Al(OH)3 and Si02 surface sites. The patch site distribution assumes that two kinds of surfaces are present on the particle and that the double layer associated with one surface does not affect surface reactions on the other. Enhanced Cd2+adsorption and reduced S042-adsorption a t low A1 to Si02 ratios were modeled using the mixed site distribution. The modeling results suggest that the change in surface potential caused by the formation of Al(0I-i)~ on Si02 surface has significant effects on Cd2+and S042adsorption.
Introduction Surface complexation models have been used to describe the adsorption of ions on pure oxides (1-4); however, our ability to quantitatively predict ion adsorption on multicomponent oxides is still limited (5-7). Ion adsorption on mixed oxides has been calculated using the adsorption properties of the component oxides (8-10). However, it has been demonstrated that in some multicomponent oxide systems the overall adsorption of ions is not the sum of the uptake by independent adsorbents (5,7,11). Recently, Anderson and Benjamin (12) incorporated a surfacecoating interaction in the constant capacitance surface complexation model to predict ion adsorption in an Al(OH)3/Fe(OH)3system. They also modeled ion uptake in a Fe(OH)3/SiOzsuspension in the presence of competition for Fe surface sites by silicate dissolved from silica (13). Honeyman (5) and Anderson and Benjamin (12) stated that the change in surface potential of base particles caused by the formation of surface precipitation may be an important factor influencing the adsorption properties of multicomponent oxides. However, no surface model has been developed to investigate this effect. In this study, the triple-layer surface complexation model (TLM) and two surface site distribution assumptions were used to simulate the physical-chemical properties of the Si02 surface partially covered by A1 hydrolysis products. The effect of the change in electrical potential caused by interaction between Al(OH)3 and Si02 on the adsorption properties of the mixed oxides was studied by comparing the modeling results obtained with each of the site distribution assumptions. The modeling results were evaluated and interpreted using Cd2+and Sod2-adsorption data. Both cation and anion adsorption were studied to test the significance of electrostatic effects in adsorption. t Present address: Center for EnvironmentalEngineering, Stevens Institute of Technology, Hoboken, NJ 07030. 1924
Environ. Sci. Technol., Vol.
27, No. 9, 1993
Models for Surface Modified Oxide In a previous paper (14),it was demonstrated that Al(OH)3precipitate forms on Si02 surfaces when a suspension containing Si02 and Al(N03)3 was neutralized with KOH solution. It has been suggested that Al(OH)3precipitate is formed on Si02 by surface condensation of monomeric or polynuclear aluminum hydroxide species or by adhesion of colloidal hydroxides formed in solution (15-1 7). When the size of the aluminum hydroxide entity on the Si02 surface is small, the electrical potentials associated with the Al(OH)3 and Si02 surface groups may overlap. Therefore, the surface potential associated with one type of surface group affects the uptake of ions by the other type of surface group. In this study, two kinds of site distributions were assumed to simulate the Si02 surface modified by the Al(OH)3. In the mixed site distribution (MSD) assumption, Al(OH13 sites are uniformly distributed on Si02 surface. An overall surface potential created by both Al(OH)3and Si02 surface sites is assumed for the modified surface. The ions bound a t the surface are affected by one set of electrical double-layer properties. In the patch site distribution (PSD) assumption, two kinds of surfaces [Al(OH)3and Si021 are present on the modified Si02 particles. The electrical double-layer properties of each type of surface are the same as those of pure Al(OH)3and SiOz. The PSD corresponds to the formation of relatively large homogeneous areas of Al(OH)3sites on the Si02 surface. The PSD assumption is similar to the surface-coating assumption discussed by Anderson and Benjamin (12). The MSD and PSD represent the diametrical extremes for the distribution of Al(OH)3sites on the Si02 surface. The MSD and PSD assumptions were used with the triple-layer surface complexation model (TLM) of Yates et al. ( I ) , Davis et al. (2), and Davis and Leckie ( 3 ) . According to Hayes and Leckie ( 4 ) and Hayes et al. (18), in the TLM, H+, OH-, and other strongly binding ions are located at the innermost layer (as an inner-sphere surface complex). They contribute to the surface charge uo and experience the surface potential $0. Separated from the surface by a region of capacitance, C1 is the inner Helmholtz plane (IHP). Weakly binding ions are held at the IHP and are labeled outer-sphere surface complex. They contribute to the IHP charge UP and experience the potential at IHP, &. The surface parameters used in the TLM calculations of this study for pure Al(OH)3and Si02 are summarized in Table I. The surface acidity (kZt, K Z ) and electrolyte binding (Kzt,K G ) constants were determined by fitting the calculated surface charge and diffuse-layer potential to measured values of the surface charge and {potential. All the model simulations in this work were performed with the TLM as implemented in the computer program MINTEQAB (19). The K? and K F for Si02 are comparable to those obtained by James and Parks (20) using their double-extrapolation method. The surface acidities 0013-936X/93/0927-1924$04.00/0
0 1993 American Chemical Society
Table I. Surface Parameters and Constants for Si02 and
AI(OH)a Si02
AI(OH)3
site concentration4 1.67 mmol/g 5 mmol/5 mmol of A1 site density (site/nm2)b 8.0 5.0 125 capacitance, C1 (pF/cm2) 125b 20 capacitance, CZ(pF/cm2) 20b -1.5 8.0 log rb;" -5.5 -10.2 log r;t -7.0 -10.4 log rnt -1.5 7.8 log K& 4 Measured with deuterium isotopic exchange method (14). From James and Parks (20).
and the Helmholtz capacitances for A1(OH)3 agree with the values used by Westall and Hohl (21) for T-Al203. An important step in calculating ion adsorption on the Si02 modified b y A l ( 0 H ) is ~ to determine the total number of Al(0H)B and Si02 surface sites. In this study, the Al(OH)3 site fraction was estimated using p potential and point-of-zero charge, ~ H P Z C measurements. , I t has been demonstrated that the pHpzc of an Al(OH)3/Si02 co-gel is the weighted average of the ~ H P Z Cof ' S its component gels (22, 23). In the PSD assumption, there is no electrostatic interaction between Al(OH)3and Si02 surface sites. The overall surface charge density, UT (C/cm2),can be calculated with UT
=c
~ i f+ ~ iUSifSi
AI/S io2 (mmol/g) Figure 1. Experimental pHpzcas a function of AI to SiOz ratio, and fAloHcalculated using eq 4 for the PSD assumption.
(1)
where I
Od
f ~ l A A ~ / ( A A+~A s i )
and
AI + fsi = 1 UAl and USi are the charge densities and A Aand ~ A s i are the surface areas of the Al(OH)3and Si02 patches, respectively. In the diffuse layer, the charge-potential relationship is described by the Gouy-Chapman equation
= -(8~ek7')'/~ sinh(zeqd/2k~
(2) where c is the electrolyte concentration, 6 is the permittivity, k is the Boltzmann's constant, T is the absolute temperature, z is the ionic charge, and e is the electron charge. Combining eqs 1, 2, and f ~ l fsi = 1 yields bd
+
sinh(zeqdT/2k7') - sinh(zeqdsi/2k7') - sinh(zeqdsi/2kn
f ~ =l sinh(zeqdM/2kn
(3)
The fraction of the total area covered by Al(OH)3, AI, was calculated from the diffuse-layer potentials of the pure AI(OH13, q u i , pure si02, $'dSi, and the Al(OH)3-modified Si02, $dT, using eq 3. {potentials measured in a previous study (14) were used as estimates of q d . Site densities of 8 and 5 site/nm2 for pure Al(OH)3 and Si02 (20), respectively, were used with eq 4 to calculate the Al(OH)3 site fraction, ~ A ~ O H . fMoH
= 8fA1/(3fA1 + 5,
(4)
The experimental ~ H P Zand C the calculated f ~ l are o ~ plotted in Figure 1 as a function of the A1 to Si02 ratio. The total site concentration in each Al(OH)3/SiO2 sus-
1
I
I
I
2 3 4 AI/Si02 (mmol/g)
I
I
5
6
Figure 2. Total surface sites determined by deuterium exchange method as a function of AI to SiOz ratio. Si02 concentration is always 1 g/L.
pension was measured by a deuterium-exchange method ( 1 4 ) , and the points are plotted in Figure 2. The concentrations of Al(OH)3 and Si02 sites in the PSD o ~total site assumption can be estimated using the f ~ l and concentration data in Figures 1 and 2. In the MSD assumption, the speciation of Al(OH)3and Si02 surface sites is interactive; therefore, eqs 3 and 4 are not appropriate. Interaction between Al(OH)3 and Si02 surface sites changes the surface charge density relative to noninteracting surfaces. According to eq 2, the diffuse layer potential, q d , is directly related to the surface charge density, Crd. For the MSD assumption, the Al(OH)3 site fraction was estimated by fitting the pHpzc of Al(OH)3/ Si02 suspensions using the TLM and the surface parameters listed in Table I. The MSD assumption is incorporated in the TLM by the use of both Al(OH)3and Si02 o ~ obtained sites on a surface. Best-fit values of f ~ l were by varying the amount of total Al(OH)3 (and Si02) sites until the calculated pHpzc was the same (fl%) as the experimental value. The calculated ~ H ~ isz compared c with the experimental value in Figure 3. The best fit f ~ l is also shown in Figure 3 as a function of the A1 to Si02 ratio. A comparison of the results in Figures 1and 3 indicates that when the A1 to Si02 ratio is less than 2 mmol of Al/g of Si02, a 1owerfMoHis obtained with the MSD assumption than with the PSD assumption for the same A1 to Si02 ratio. However, it is expected that f ~ l will o ~ be higher Environ. Sci. Technol., Vol. 27, No. 9, 1993
1925
o ~
Table 11. Surface Reactions
surface reactions
equilibrium expressions
Surface Ionization (Inner-Sphere) SOH + H+ = SOH2+ SOH = SO-+ H+ Electrolyte Adsorption (Outer-Sphere) K K = exp(F(+g+o)/RT)[SO-Kl [H+l/ [SOHI [K+l (3) SOH + N O s + H+ = K N O=~exp(F(+o-+g)/ SOH2 - NO3 R T ) [SOHz-NOsI/ [SOH1tH+I[NOal (4) Inner-Sphere Surface Complexes SOH + Cd2+= SOCd+ + H+ K = exp(F~0/RT)[S0Cd+l[H+l/ [SOH] [Cd2+l(5) llSOH + SO42- + H+ = K = exp(-F$o/RT)[SSO4-]/ [SOHI [S04'-1 [H+l (6) (SOH)loSS04- + HzO llSOH + HdSi04 = K = exp(-F+0/RT)[SH2SiO4-](SOH)loSH2Si04- + tH+l/[SOHl[H4Si041 (7) H+ + HzO llSOH + H4Si04 + K+ = K = exp(F(+b--+o)/RT)(SOH)loSHzSi04[SH&04-K1 [H+1/ K + H+ + HzO [SOHI [H4SiOrl[K+l (8)
SOH + K+ = SO - K + H+
AI/Si02 (mmol/g) Flgure 3. fAMHfor the MSD assumption, estimated by calculatingthe pHPzcof the AI(OH)&lOp particles using the TLM, and the calculated pHpzcas a function of the AI to Si02 ratio, [H4SiO4Ieq = (0-5) X M.
with the MSD than with the PSD assumption for the same A1 to Si02 ratios. Based on the fAloH results in Figures 1 and 3 and the total surface site concentration in Figure 2, a higher Al(OHi3surface site concentration is calculated for the PSD assumption than for the MSD assumption. For instance, a t 0.5 mmol of Al/g of Si02, the calculated Al(OH)3surface site concentrations are 0.65 and 1.19mmol for the MSD and PSD assumptions, respectively. The Al(OH)3surface site concentration of the pure A1(OH)3, measured with the deuterium isotopic exchange method, is 1 mmol site/mmol of Al(OH)3 (Table I). The much higher Al(OH)3surface site concentration calculated using the PSD assumption suggests that the PSD assumption is inappropriate for the Al(OH)3/SiO2suspensions at low A1 to Si02 ratios.
Surface Complexation Reactions In addition to the surface parameters listed in Table I, the type of surface complex and the magnitude of the formation constants for the site binding reactions are required for the model computation. Several general types of surface complexes have been suggested in the literature, including inner-sphere and outer-sphere complexes (4, 18), and hydrolytic complexes of metal ions (3). The innersphere surface complex, SOCd+, has been used by Hayes and Leckie ( 4 ) to model Cd2+adsorption on goethite. The reactions considered in this study are summarized in Table 11. SOH in Table I1 denotes AlOH or SiOH sites. Two reactions (reactions 7 and 8 in Table 11)were used to model H4Si04 adsorption. Hayes et al. (18) have used similar reactions to model selenite adsorption on ferric oxide. Initial calculations indicated that an AlOH site coverage of 11 sites per adsorbed sod2-was required to model the S042- adsorption in suspensions containing different amounts of Al(OH)3. Hingston (24)reported that a sorbed sod2-covers an area of 1.2 nm2 on the gibbsite surface. Thus, according to the site density for Al(OH)3 (Table I), one adsorbed S042-ion covers approximately 10 AlOH sites. In the surface reactions involving sod2and H4Si04 (reactions 6-8 in Table 11),a coefficient of 11was used for surface sites. In reactions 6-8, S042-or H4Si04 forms a surface complex with one SOH and physically covers 10 SOH. Therefore, the mass law equations corresponding 1928
Environ. Sci. Technol., Vol. 27,
No. 9, 1993
Table 111. Formation Constants
reaction" log K&d 1% %qo4 log KkTzSi04 log KSHzSi04K
(5) (6) (7) (8)
surface Al(OH)3 Si02 -2.4 9.0
-9.1
-4.1 -2.2
The numbers denote the reactions labeled in Table 11.
to reactions 6-8 were written in terms of [SOH] and not [SOHI1l. A similar approach has been used by Davis and Leckie (25). The magnitude of the formation constants listed in Table I11 were determined by fitting the results of TLM computations to the experimental adsorption data for pure Al(OH)3 or Si02 using the surface parameters and constants listed in Table I and the reactions listed in Table 11. The uptake of Cd2+,so42-, and H4Si04 calculated using the constants in Table I11 agreed quite well with the experimental data. The best-fit results for Cd2+adsorption are compared with the experimental values in Figure 4.
Results and Discussion The parameters and constants estimated above were used with the MSD and PSD assumptions in the TLM to predict ion adsorption on the Al(OH)~-rnodifiedSi02. The amount of dissolved silica measured in the suspensions was used as the equilibrium concentration in the model calculations. Cd2+ adsorption results are shown in Figure 5 as a function of the A1 concentration in the presence and absence of Si02 at pH 7.0. The experimental procedures are described in a previous paper (14). According to Figure 5, Cd2+removal increases significantly as the A1 to Si02 ratio increases from 0 to 0.5 mmol of Al/g of Si02 and decreases as the ratio increases from 0.5 to 5 mmol of Al/g of Si02. I t is apparent that the amount of Cd2+adsorbed by the suspension containing 0.5 mmol of Al/g of Si02 is greater than the sum of the Cd2+adsorbed by the 0.5 mM
100 U
80
Q, n
Smmol %(OH)3/g
8 60
I
v)
'0
a
u
0
8
I
20
Figure 4. Model simulation of Cd2+adsorption in AI(OH)3suspension and SiOz suspension, [Cd2+IT= 1.78 X M, [KNOs] = 0.04 M.
100 I
I
I
1
2
3
4
5
6
Al(mM) Flgure5. Percentage sorption of Cd2+as a function of AI concentration In the presence and absence of 1 g of SiOz/L, pH = 7.00 f 0.03, [Cd2+]~= 1.78 X M, [KNOB] = 0.04 M.
80 60 40 20 n
"0
1
2
3
I
6
7
8
9
PH
PH
"0
I I
a:
04 -
80
I
40
4
5
6
[All (mM) Figure 8. Cd2+ adsorption calculated using the MSD and PSD assumptions as a function of the AI to Si02 ratio, [KNOB] = 0.04 M, [Cd2+]~= 1.78 X M, [H4S/04]eq= (0-5) X M.
Al(OH13 suspension and the 1 g of SiO2/L suspension. These results were reported in a previous paper (14). Modeling results for the MSD and PSD assumptions are shown in Figure 6. According to these plots, modelpredicted Cd2+adsorption for the MSD assumption agrees reasonably well with the experimental observations at low A1to Si02 ratios. The MSD assumption predicts maximum
Flgure 7. Cd2+ adsorption calculated using the MSD and PSD assumptions as a function of pH in a suspension of 0.75 mmol of Al/g of Si02, [KNOB] = 0.04 M, [Cd2+IT= 1.78 X M, [H4Si0& = 3.3 x 10-5 M.
Cd2+ adsorption at 1 mmol of Al/g of Si02 while the observed maximum is at 0.5 mmol of Al/g of Si02. However, at higher A1 to Si02 ratios, the MSD assumption predictions are significantly higher than the observed adsorption. Results obtained with the MSD assumption suggest that when the A1 to Si02 ratio is greater than 0.2 mmol/g, Cd2+ uptake by Si02 sites is negligible a t pH 7. Apparently, the negative potential created by SiO- sites enhances Cd2+ adsorption on Al(OH)3 sites. As the A1 to Si02 ratio increases, the number of positive AlOH2+ sites increases. The more positive surface potential of the Al(OHI3modified Si02 at high A1 to Si02 ratios inhibits the formation of the AlOCd+ surface complex. In contrast to the MSD results, the predictions obtained with the PSD assumption (Figure 6) do not show a maximum in the adsorption curve. The PSD results show adsorption increasing continuously with increasing A1 to Si02 ratio. On the other hand, the PSD predictions are very similar to the observed Cd2+ uptake in the pure Al(OH)3suspensions. This result is expected because the AlOCd+ surface species is not influenced by the surface potential of the Si02 in the PSD assumption. Also, the number of Al(OHI3 surface sites increases continuously with increasing A1 to Si02 ratio (Figures 1 and 2). In Figure 7 ,the percent Cd2+uptake by 0.75 and 5 mmol of Al/g of Si02 suspensions is plotted as a function of pH. For pH C7.5, Cd2+removal by the 0.75 mmol of Al/g of Si02 suspension is higher than by the 5 mmol of Al/g of Si02 suspension. The solid and dashed lines in Figure 7 are the calculated Cd2+adsorption results for a 0.75 mmol of Al/g of Si02 suspension and the MSD and PSD assumptions, respectively. According to Figure 7 ,the MSD assumption yields a model-predicted adsorption edge which compares well with the observed results for pH