Colloids and Surfaces A: Physicochemical and Engineering Aspects 207 (2002) 13 – 24 www.elsevier.com/locate/colsurfa Surface complexation modeling of cadmium adsorption on gibbsite R. Weerasooriya a,*, H.K.D.K. Wijesekara a, A. Bandara b a Chemical Modeling Project, Institute of Fundamental Studies, Kandy 2000, Sri Lanka b Department of Chemistry, Uni6ersity of Peradeniya, Peradeniya, Sri Lanka Received 8 February 2001; accepted 18 December 2001 Abstract Cadmium adsorption on gibbsite was examined as a function of pH, background electrolyte concentration, and adsorbate loading. The adsorption data was quantified by charge distribution multi-site ion complexation (CD MUSIC) model using reaction stoichiometry given; 2( \ AlOH − 1/2) +Cd2 + = (\ AlOH)x2 Cdy; log K= 6.727. The charge distribution factor, f is treated as an adjustable parameter in model fitting yielding x =1.2 valence units (vu) and y= 0.8 vu. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Cadmium; Gibbsite; Charge distribution multi-site ion complexation model; Langmuir model 1. Introduction Reactions at the particulate-water interface are one of the key processes in determining transport and fate of trace elements in natural environment [1]. When particulate materials are enriched with natural organic moieties such as humic substances (HS), the fate of metal ions is determined to a large extent by chemical complexation [2]. However, there are situations in which inorganic particulates such as metal hydrous oxides are equally important or even dominate the overall sorption process [3]. This is the case for environments in which very little organic matter is present, as is * Corresponding author. Tel.: +94-8-232002; fax: + 94-8232131 E-mail address: [email protected] (R. Weerasooriya). encountered for example in many aquifers or in clay liners or landfills [4,5]. Unfortunately, mechanistic understanding of metal ions binding on natural soils, sediments or aquifers is an arduous task, partly due to inherent complexity of these solids [6]. Over the past three decades or so, much attention was, therefore paid to quantify metal ions binding on well-characterized mineral phases that dominate in nature. The rationale here is to consider these discrete mineral phases as ‘building blocks’ of natural solids in that they could be combined in such a way to identify dominant adsorptive surfaces. From this viewpoint substantial chemical data on trace element binding on iron hydrous oxides has become available (Ref. [7] and references therein). However, similar data on gibbsite (and other Al, Si, or Mn) hydrous oxides that are also common in nature is comparably low 0927-7757/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 7 - 7 7 5 7 ( 0 2 ) 0 0 0 0 4 - 3 14 R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 [8 –16]. Further, none of these studies address the issue of cadmium adsorption mechanism on gibbsite. The aim of this work was therefore, to determine the behavior of cadmium adsorption on gibbsite over a range of experimental conditions that are important environmentally. This experimental system was selected primarily due to following factors. (a) Cadmium is a nonessential element; it is released into the environment largely from various anthropogenic activities. It poses serious threat to plants, animals and humans because of their toxicity and persistence. Hence, the US EPA has categorized it, as a priority pollutant designating maximum contaminant level goal is 0.005 mg l − 1 [17]. (b) Cadmium exhibits low tendency for polymerization; therefore it is particularly suited to probe nature of active surface sites. (c) Cadmium typically forms weaker aqueous and surface complexes than do with other heavy metals; therefore Cd should exhibit high mobility particularly under slightly acid conditions. (d) Moreover, gibbsite is a useful analog of naturally occurring aluminum oxides, hydroxides and aluminous silicate minerals; it offers reactive surface sites when in contact with surface and ground waters. (e) Gibbsite is a simple solid that has \Al2OH0 and \AlOH − 1/2 surface sites; only \ AlOH − 1/ 2 is found to be reactive when the pH is ranged 4– 10 [12]. Therefore, it provides a convenient way for a model testing. (f) Recently, the chemical data on gibbsite-solution interface over a range of experimental conditions has become available [13–16]; hence, it provides a convenient way to model the interfacial chemistry of gibbsite– cadmium system. In this work, a series of experiments were conducted to assess the cadmium adsorption as a function of pH, adsorbate loading, and electrolytic concentration. Particular attention was paid to determine the degree of site heterogeneity of cadmium adsorptive centers by non-calorimetric method. After a critical assessment of existing mechanistic models in light of current requirements, we have selected a charge distribution multi-site surface complexation (CD MUSIC) model to quantify cadmium adsorption data [18]. This model emphasizes the importance of the surface structure and of the charge distribution at the interfaces. Further, the CD MUSIC model is capable to account for the finite size of adsorbing ions, and the site heterogeneity based on crystal faces. The charge of the adsorbing ions is specifically distributed between two electrostatic planes according to the modified Pauling’s bond valence theory. Such an approach is not possible in widely used 2-pK triple-layer model (TLM) or Frumkin isotherm based models [19–23]. However, in the CD MUSIC modeling the surface is treated planer. Moreover, the lateral interactions of surface sites have not been explicitly accounted for. Further, the CD MUSIC model formulation requires explicit selection of an electrostatic model. Presently, we have selected the 1-pK three-plane model (TPM) since by making several assumptions it can readily be degenerated into constant capacitance (CCM), Stern layer (SLM) or diffuse layer models (DLM). However, the 1-pK TPM suggested by Venema et al. [24] differs from the classical 2-pK TLM with respect to following factors; outer-layer capacitance, C2 (TLM C2 = 0.2 F m − 2; TPM C2 = 5 F m − 2), and ion pairs placement (TLM, ionpairs in SLM, TPM, in diffused layer). The following additional factors merit the selection of CD MUSIC model for our calculations. When compared to 2-pK modeling, the 1-pK approach significantly reduced the total number of adjustable parameters (e.g. 2-pK TLM, 7 parameters; 1-pK TPM 5 parameters). The intrinsic acidity constants derived by 2-pK approach are dependent on electrostatic model used; however such a limitation is not present in the 1-pK modeling technique; the acidity constant is simply derived experimentally as pHzpc of the solid. Further assumptions based on pKi (i= 1 or 2) are no longer needed (2-pK models are often constrained by a ZpKi ). The temperature (T) dependence of surface charging can simply be predicted by assessing pHzpc. Finally, placing them in two different planes distinctly separates the outer- and inner-sphere complexes [18,24]. R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 2. Materials and methods 2.1. Materials Gibbsite was obtained from ALCOA (Australia). All reagents were prepared in de-ionized, distilled water that was de-gassed by boiling and then purging with N2 (purity 99.996%). All pH adjustments were made either with 0.541 M HNO3 or 0.600 M NaOH. Particular attention was paid to store NaOH under N2 to prevent possible CO2 contamination. All experiments were performed in a glove box that was purged with 99.996% N2 to minimize atmospheric contamination of CO2 according to following flushing cycle; thrice prior to use; twice after sample change, and twice daily to minimize any slowly diffused CO2 into the glove box. The gas samples inside the glove box was checked after each flushing cycle with the GC/TCD detector, and was always below 5 pM. The 1000 mg l − 1 (8.897 mM) cadmium stock solution (AAS grade) used was obtained from Mercks, Germany. Table 1 shows the parameters (e.g. specific surface area, site density, intrinsic acidity and electrolyte binding constants, formation constants of cadmium species used in calculations. 2.2. Methods mined as a function of pH, ionic strength, and adsorbate loading. Unless otherwise mentioned, the system temperature was always kept at 298 K. A 500 ml of 2 g l − 1 gibbsite suspensions was prepared in 0.001 M NaNO3, and let it equilibrated for 2 h. Always the suspension pH ranged between 7 and 8. Before the addition of calculated aliquot of 0.890 mM Cd2 + to reach desired metal ion concentration in the range of 0.77–7.74 mM, the system pH was first adjusted to 3.5 with 0.541 M HNO3. Ten milliler aliquots were then sampled into 40 ml capacity polystyrene cen- Table 1 Surface parameters and reaction stoichiometries used in CD MUSIC modeling of cation adsorption Parameter/reaction Value Source (Reference) Parameter Surface area (m2 g−1) Site density (sites nm−2) 13 8.5 [15] [15] Stern layer capacitance C1 (F m−2) C2 (F m−2) 2.58 5 [12] Protolysis reaction \AlOH+1/2 2 8.70 [15] \ AlOH −1/2 2.2.1. Adsorption kinetics Chemical kinetics of cadmium adsorption on gibbsite was determined in 0.01 M NaNO3 at pH 5.5. The pH adjustment was made with HNO3. Relevant experimental details are well documented [16], hence only an outline is given. Briefly, a 250 ml of batch solution was prepared with 2 g l − 1 gibbsite, and it was spiked with stock cadmium solution to reach a final concentration at 0.78 mM. At pre-defined time intervals, samples (5 ml aliquots) were withdrawn into a syringe, which was fixed, to a 0.45 m disposable filter cartridges for immediate solid-solution separation. The filtrate was acidified at pH B1 for cadmium analysis. 2.2.2. Adsorption edges Cadmium adsorption on gibbsite was deter- 15 +H+ s Ion pair formation − \AlOH−1/2 +H+ s s +NO3 7.975 [15] \AlOH+1/2 −NO− 2 3 \AlOH−1/2 +Na+ s s [15] −0.396 \ AlOH−1/2−Na+ Aqueous phase reactions Cd2++2NO3− =Cd(NO3)20 0.00 Cd2++2OH− =Cd(OH)20 7.70 Cd2++3OH− =Cd(OH)3− 8.99 Cd2++4OH− =Cd(OH)4−2 8.71 Cd2++5OH− =Cd(OH)5−3 8.07 Cd2++6OH− =Cd(OH)6−4 7.19 2Cd2++2OH− =Cd2(OH)+2 7.60 4Cd2++4OH− =Cd4(OH)4 28.08 Cd2++NO3− =CdNO3+ 0.31 Cd2++OH− =CdOH+ 3.90 H++OH− =H2O −1.40 Na++OH− =NaOH0 −0.20 [25] [25] [25] [25] [25] [25] [25] [25] [25] [25] [25] [25] 16 R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 trifuge tubes for successive addition of variable volumes of 0.600 M NaOH to adjust the pH between 3 and 10 at 0.5 pH unit intervals. Once the upper pH value was reached (normally 10), addition of 5 M NaNO3 was made to adjust background ionic strength to 0.01 M. Soon after the system pH was decreased to 3.5 with 0.541 M HNO3 and samples were taken again after adjusting the pH between 3.5– 10. The whole process was repeated to obtain sample aliquots at 0.1 M NaNO3. Samples were equilibrated in a temperature regulated water bath orbital shaker (Eyela, B 603, Japan). The pH of the samples was measured after an equilibration period of 24 h. After phase separation by centrifugation (14 000 rpm) at regulated temperature for 5– 10 min., and followed by membrane filtration (0.45 m), the filtrates were acidified for metal analysis. The degree of reversibility of cadmium adsorption was also examined. The methodology followed remained same except following changes. The ionic strength was kept at 0.01 M. The spiking of 0.77 mM Cd was made for a well-equilibrated gibbsite suspension at pH 8. The pH adjustments were done with 0.541 M HNO3 before sampling 10 ml of solid suspension into centrifuge tubes for 24 h equilibration. 2.2.3. Adsorption isotherms Cadmium adsorption isotherms were constructed in 0.01 M NaNO3 as a function of pH. Experimental protocols remained essentially same as stated in the section under adsorption edges except following changes. The 2 g l − 1 gibbsite suspension was equilibrated for 2 h at a desired pH. Afterwards, a 10 ml portion of samples was transferred into centrifuge tubes. The addition of different quantities of cadmium into these tubes was made after matching the solution pH. The total concentration of cadmium was varied between 0.1 and 1000 mM. At the end of 24 h equilibration period, the pH of the stirred suspension was measured, and regulated at desired value typically for 0.5–1 h till a stable pH reading is achieved. Subsequently solid– solution phases were separated to determine solute concentration. 2.2.4. Analyses All pH measurements were made using Ross combination pH electrode and auto chemistry analyzer (Orion 960) in a well-stirred solid suspension. Cadmium concentration was measured either with flame (Cd \2 mM) or graphite furnace (GF) (Cd B 2 mM) atomic absorption spectroscopy (AAS) using an atomic absorption spectrometer (AA) equipped with a GF and an auto sampler (AA GBC 933AA, GF GBC GF 3000, and GBC PAL 3300). Both flame and GF measurements were made at 228.8 nm. In the GF-AA pyrolitic graphite coated tubes were used with L’ van platforms. The atomization program consisted of following steps: drying (two steps): step1 : temperature 90 °C ramp time 10 s, hold time 10 s; step2 : (120 °C, 20 s, 10 s); pyrolysis (400 °C, 1 s, 5 s); and atomization (1800 °C, 0.8 s, 0.8 s) with the inert purge gas flow arrested and flushing (2000 °C, 1 s, 5 s) (in atomization step, the inert purge gas flow was arrested. Air flow rate was 4.5 ml min − 1 in other steps). 2.2.5. Data modeling The surface of gibbsite consists of several structurally different functional groups. These groups differ due to the surface oxygens on gibbsite being singly and doubly coordinated to aluminum. The structure of gibbsite forms stacked sheets of linked octahedrons of Al(OH)3. The octahedrons are composed of Al3 + coordinated to six OH ions yielding + 0.5 (vu) according to Pauling’s theory. Two aluminum ions are needed to neutralize one OH− in the gibbsite structure. Thus in the crystal structure all hydroxyls are doubly coordinated. The planar faces of gibbsite have only Al2OH0 groups whereas the edge faces have either Al2OH0 or AlOH − 1/2 groups. As calculated by Hiemestra et al. [12] the protonation of \Al2OH occurs when pH 0. Similarly the de-protonation of \ Al2OH sites occur when pH 12. In the pH range that we considered the \ Al2OH are inert. Hence, only \ AlOH − 1/2 sites were treated as active. All calculations were performed using ECOSAT-FIT utility [25,26]. Very recent release of ECOSAT has incorporated FIT optimization algorithm to extract binding and other constants from the adsorption measurements. R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 17 3. Results and discussion 3.1. Adsorption kinetics Interactions of solids with water involve a combination of chemical processes; i.e. surface hydration and hydroxylation, adsorption and desorption, diffusion. It is known that surface hydration is a slow process [7,21], which gradually adds adsorption sites to the system that result an enhanced ion adsorption over prolong periods of equilibration. Although it is difficult to infer information about a given process from a simple kinetic experiment such data is used to determine the time required for first appearance of ion equilibration. Therefore, the adsorption kinetics was studied in order to determine the time required for equilibrium to be achieved. It is evident from the data in Fig. 1 (A) that the cadmium uptake by gibbsite was sluggish reaching an apparent plateau within first 6– 7 h. Previously, a two-step process has been used to characterize the retention of metal ions on hydrous oxides; a rapid adsorption preferably due to the external surfaces followed by a slow retention due to intra-particle diffusion along the micro pore walls utilizing internal surface area [27]. In the slow adsorption process, with time the amount of metal ions sorbed to the oxide gradually increased due to inter-particle surface diffusion and surface hydration over extended equilibrium periods. In our experiments, within the first 24 h equilibration period, such an observation was not observed for cadmium adsorption on gibbsite. Within this equilibration period, as shown in Fig. 1 (A), the time-dependent sorption curve is essentially flat. This indicates that the surface sites present in external surfaces of gibbsite govern the cadmium adsorption within first 24 h equilibration period. 3.2. Hysteresis The hysteresis between cadmium adsorption and desorption curves was also examined. As shown in the Fig. 1 (B), the cadmium adsorption on gibbsite was completely reversible with pH within 12 h; in other words, within the limits of experimental error, decreasing the pH yielded a Fig. 1. (A) Variation of GCd (cadmium adsorption density) as a function of equilibration time. Initial conditions used were; [Cd2 + ]= 7.78 mM; pH 7; T= 298 K; 0.01 M NaNO3. Similar trends were observed both at system pH 8 or 9 (data not given). (B) Variation of GCd as a function of pH. In desorption, gibbsite – cadmium complex was first formed at pH 8.0 after equilibration for 24 h. Subsequently, the metal – gibbsite complex was treated with 0.01 M NaNO3 at different pH for 24 h. Relevant data is shown by opened symbols. Filled symbols represent adsorption data obtained by varying pH from 4 to 10. Measurements were made in triplicate to determine error bars. Experimental conditions employed were; gibbsite = 2 g l − 1 and T= 298 K. 18 R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 desorption curve that follows the adsorption curve. Based on the data given in Fig. 1 (A and B), therefore, it was concluded that cadmium adsorption on gibbsite was dominated by a sluggish, reversible-surface mechanism when 24 h equilibration period was employed. 3.3. Adsorption isotherms Cadmium adsorption density, GCd, was determined for different pH in 0.01 M NaNO3. The results obtained are shown as adsorption isotherms in Fig. 2 (A). At all instances, the GCd increased with equilibrium cadmium concentration approximating to a slope one. If cadmium were polymerized, then the slope would be steeper than one. Slopes greater than one were not observed in any of the isotherms investigated [23]. This implied that only monomeric cadmium species reacts with the surface. Solution chemistry of cadmium suggests that it can form several chemical species. Relative abundance of these species was calculated using thermodynamic data (Table 1). In the calculation, the precipitation of cadmium was not taken into account due to following factors. The solubility product, Ksp of Cd(OH)2 is 10 − 14.35 at 298 K [28]. Cd(OH)2 precipitation occurs when [Cd2 + ]\ 1 mM. In our system, the total [Cd2 + ]B 1 mM which shows an unsaturated condition with respect to Cd(OH)2 precipitation. Fig. 2 (B) shows the distribution of various cadmium species as a function of pH. It is evident from the diagram, that only free cadmium ion is abundant in solution when pH ranges 4–9. Both CdOH+ and Cd(OH)20 become significant when pH \ 10. Fig. 2. (A) Variation of Gads with equilibrium [Cd2 + ] in 0.01 M NaNO3 at 298 K as a function of pH. Symbols represent experimental data. Error analyses were done on triplicate experimental data. Solid/broken lines show CD MUSIC/TPM model calculations carried out by using parameters given in Tables 1 and 2. The slope, m was calculated by linear regression of log[Gads] vs. log[Cd] curves. When m : 1 adsorption data can be interpreted in terms of Langmuir equation. Otherwise (m 01), adsorption is governed by Freundlich equation. B: Variation of cadmium species as a function of pH in 0.01 M NaNO3. Initial [Cd] =8.7 mM. The system was unsaturated with respect to cadmium. Hence, no solid precipitates were assumed. R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 Table 2 Langmuir isotherm constants for cadmium adsorption at pH 7.5, 7.1, and 6.0 pH K 9 error (m3 mol−1) Gmax 9 error (mol m2−) Rr 2 7.5 7.1 6.0 140 9 64 30.0 9 15 21.8 913 4.2E−049 8.6E−05 4.9E−049 8.5E−05 7.3E−069 4.2E−07 0.97 0.98 1.00 In all cases, the temperature was fixed at 298 K. Isotherm data were calculated from a non-linear least square fit method and the related correlation coefficients (R 2) were also given in the table. R 2 values were almost 0.98 or higher which indicate a very good mathematical fit. However, the error estimates are high, hence no attempt was made to report any data below pH 6. However, our working pH window was ranged from 4 to 9.5. Therefore, it can be concluded that under the experimental conditions imposed in this study, only free cadmium ion is present in a significant proportion for surface interactions. Further, as shown in Fig. 2 (A) the general shape of the isotherms can be modeled with a Langmuir equation: Gads = k[M]Gmax (1+ k[M]) (1) where Gads and [M] are the concentrations of adsorbate at the surface (mol m − 2) and in solution (M), respectively, k is the equilibrium constant for the adsorption– desorption process, and Gmax is the amount of adsorbate required to form a mono-layer. The Langmuir model assumes that the energy of surface sites is equal (e.g. homogeneous surface). The parameters k and Gmaxwere estimated simultaneously, using a non-linear least square fit of the experimental data, and the optimized values are given in Table 2. Within the limits of error estimates, both k and Gmax have shown an apparent increase with the pH. The k represents an empirical affinity constant that depends on Gads. The increase in Gmax with the pH is attributed to an increase in the fraction of surface sites that favor cadmium adsorption. The pHzpc of gibbsite is 8.7. When pHBpHzpc, the majority of surface sites are charged positively. As the pHzpc is approached, a significant proportion of these sites 19 will be neutralized and a small number will acquire a negative charge before the pHzpc is reached. Given the adsorbing species (e.g. Cd2 + ) is positively charged, adsorption will occur most readily on neutral or negative surface sites. The increase in Gmax with pH is therefore due to an increase in the fraction of neutral sites available for adsorption, as pHzpc [29]. However, dominance of surface sites of single class (\ AlOH − 1/ 2) does not provide conclusive evidence for homogeneous surface sites. Because of surface roughness, on an atomic scale there will be a difference in reactivity among sites with the same coordination [30]. These effects when combined should generate a surface with a wide range of adsorption energies that can subsequently result in a heterogeneous surface. Accordingly, we have assumed that at least for the experimental conditions imposed in the study, that the \ AlOH − 1/2 are homogeneous. 3.4. Adsorption edges Fig. 3(A, B and C) shows the variation of cadmium adsorption density (GCd) as a function of ionic strength and pH at different initial cadmium concentrations. The GCd has increased (from 1% to almost 100%) with the pH increase from 4 to 9.5. Within the experimental error, no difference in cadmium adsorption is observed for values below pH 6. Above this pH value, the adsorption edge occurs within a narrow pH range (less than 2 pH units) when [Cd]initial 0 8 mM. When initial [Cd]=77.4 mM, the broadening of adsorption edge occurs between pH 6.2 and 8.5. In all cases, it illustrates a typical sigmoid curve characteristic of d-group metals; similar trends were observed for lead and copper adsorption on gibbsite [15,16]. The inflection point of the cadmium adsorption edges, which correspond to the pH of 50% adsorption (pH50%) as determined by d2G/dpH2 was always found to be within 7.50–8.00 for a 100fold variation of [Cd]initial. The pH of the 50% adsorption (i.e. pH50%) was 7.40 for Cd 0.77 mM, 7.70 for Cd mM and 7.80 for Cd 77 mM. As shown in the data, the position of the adsorption edge shifts slightly to a higher pH as the initial metal 20 R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 concentration increases. At a given experimental condition, when the initial cadmium concentration is set so that the adsorption density data falls Fig. 3. in the linear region of the isotherms (shown in Fig. 2 (A)), the position of the adsorption edge (as determined by pH50% values) showed virtually no shift with the adsorbate loading. When the relative position of the edge shifts with the adsorbate loading, it implies that the initial experimental conditions are set to fall in the non-linear region of the isotherm. In all cases, whether the initial cadmium concentration falls in the linear or nonlinear region, a maximum adsorption density, Gcdmax, was reached when pH \7.5. Further, the Gcdmax value was directly proportional to the initial [Cd]initial. Such proportionality cannot, however, be deduced when Gcd B Gcdmax. Therefore, it is concluded that cadmium adsorption on gibbsite over the entire pH range examined exhibits a non-proportional behavior for the variation of adsorbate loading and pH. Relative shifts of the adsorption edges with respect to a given reference point for variation of ionic strength have often been used to infer bonding mechanism(s) of surface complexes [31]. The adsorption density of weakly bonded complexes has showed a marked effect for the variation of ionic strength, yielding a distinctly separated adsorption edges. Such an observation cannot however be made when ions are strongly bonded. Further, demarcation of a boundary between weak vs. strong surface complexes is somewhat arbitrary yielding ambiguous results for ions that exhibit ‘intermediate’ affinity for surfaces. Furthermore, it is important to note that the relative shifts of adsorption edges always do not yield Fig. 3. Variation of Gads as a function of pH, ionic strength and adsorbate loading. In all cases 2 g l − 1 gibbsite was used. (A) Initial [Cd2 + ]= 0.77 mM; (B) 7.74 mM; (C) 77.4 mM. Replicate measurements were made only for data set (B). All measurements were made at 298 K. All lines (both solid and broken) show calculated data by using CD MUSIC model. Only cadmium – gibbsite bidentate complex was considered. Charge distribution of Cd2 + between 0- and 1-planes was calculated according to following equations; Zz0 =nHzH + fzMe and Dz1 =(1 −f )zMe + Smizi. The nH is the change of protons on the surface ligands involved in the surface reactions; zH is valence of protons zMe charge of central ion in surface complex; mj is number of ligands positioned in 1-plane; zi is the charge of the ligands in 1-plane; f is the charge distribution factor calculated according to the modified bond valence theory. R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 conclusive evidence into the binding mechanism. For example, species with large log K will have a smaller effect on ionic strength than with a low log K, irrespective of the way of bonding. In the present case, as shown in Fig. 3, the pH50%, (the pH corresponds to 50% adsorption) differs only by 0.4 pH units for a 100-fold variation of NaNO3 concentration. This points to a strong interaction with the surface, possibly yielding an inner-sphere complex losing the sheath of H2O molecules present in the hydration shell. Molecular scale information provided by in situ spectroscopic methods is needed to precisely define bonding mechanism and surface speciation of adsorbed ions at a mineral–water interface. Such data for our cadmium– gibbsite system is in progress. In the meantime, a-Al2O3 was selected as an analogous solid for gibbsite due to following points. Both solids contain AlO8 octahedra; the oxygens are closest-packed hexagonally and two thirds of interstitial sites occupied by Al atoms. In-situ extended X-ray absorption fine structure spectroscopy (EXAFS) measurements have shown that cadmium formed inner-sphere complex at the a-Al2O3 –water interface [32]. The macroscopic adsorption data presented here for gibbsite– cadmium system also supports this notion. 3.5. Proton exchange ratio, rH/Cd This is defined as a measure of surface protons released as a result of cadmium adsorption. The exact mechanism of surface protons release is, however a controversy to date. It is assumed that 21 a part of the rH/Cd is attributed to direct release of surface proton(s) upon the metal ion adsorption that subsequently increased the surface charge. Release of additional protons through coulombic repulsion compensates any changes of surface charge [7]. Recently, Rietra et al. [33] have shown that the electrostatic interactions are dominant factor in governing this ratio and not the stoichiometry of the reaction. Table 3 shows the rCd/H data that was determined with a pH-stat approach. When [Cd]initial/[Cd]absorbed 1 and pH \ 5, the rCd/H always converges to a non-integer value between 1.33 and 1.5 (Table 3), which shows the release of, more than a proton per metal ion adsorbed. 3.6. Data modeling The cadmium adsorption data presented in Fig. 2 (A) and Fig. 3 was modeled using CD MUSIC model. The specific reasons for the selection of this model were discussed under the introduction section. The CD MUSIC model requires seven parameters to be determined. As listed in Table 1, following data of gibbsite-solution interface is available in the literature; C2, outer-layer capacitance was assigned a value of 5 F m − 2. The inner-layer capacitance C1 is an adjustable parameter, which was optimized to give best description of surface charge data. As outlined under the Materials and Methods section, gibbsite possess only AlOH − 1/2 sites that are active. Site homogeneity is assumed, at least for the stringent conditions imposed in the study, as a result of Table 3 Release of protons upon cadmium adsorption on gibbsite at different fixed pH values pH 7.10 7.40 8.00 a GCd, mmol m−2 0.32 0.63 1.58 [H+]released/[Cd+2]adsorbed Experimental 9S.D.a Nb R 2c Calculatedd 1.3290.09 1.51 9 0.01 1.5590.30 3 4 3 0.90 0.89 0.99 1.40 1.60 1.65 Standard deviation. Number of samples used. c Linear correlation coefficient. d All calculations were done using CD MUSIC model. b 22 R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 Table 4 Optimized surface complexation constants of Cd(II) determined by pCd titration data NaNO3, M Log KCda b R2 c 0.001 0.01 0.1 7.118 6.476 6.326 0.99 0.91 0.99 6.569E−09 3.331E−08 7.369E−09 Weighted av. Log KCd 6.727 RMSE All experiments were conducted in 0.01 M NaNO3 at 298 K. Model calculations were carried out using CD MUSIC model. Weighted averages were calculated using log K= Swi (log K)I where wi = (1/| log K)I and s log K is the standard deviation calculated from ECOSAT-FIT for i h data set.[7,25,26]. a Surface complex [(\AlOH)2x Cdy]2+ = 2(\AlOH−1/2)+ Cd2+ log KCd. b R 2 equals to one in a perfect fit. c RMSE: residual mean of sum of squares. satisfactory fits by Langmuir model. Further, ionic strength dependence data of cadmium adsorption suggest an inner-sphere bonding mechanism. In theory, cadmium can form both monodentate and bidentate surface complexes with active \ AlOH − 1/2. Bargar et al. [34] em- ployed an empirical model based on the modified bond valence concept to predict the relative stability of different kinds of adsorbed species on the aluminum surface in an aqueous solution. According to this model, the charge saturation of the central atom (presently the oxygen) is calculated (Sl0…m ). A given surface complex is thermodynamically stable when Sln…m B z, where z is the charge of the central atom. If the sum is greater than 2.0 vu, desorption of ion from the surface should occur. Presently, the charge of the central ion (e.g. oxygen) in the bidentate cadmium complex is unsaturated leading to a stable configuration. In the case of mondentate complex, the charge of the surface oxygen is over-saturated, which means that this complex is thermodynamically unstable. Hence the reaction stoichiometry corresponding to bidentate cadmium complex was chosen for CD MUSIC modeling. In order to account for the finite size of adsorbing ion, the CD MUSIC model requires the charge of the surface complex is distributed in the interface between o- and b-planes. Presently we have considered the charge distribution factor as a fitting parameter. As shown in Table 4, the log K value Fig. 4. CD MUSIC model calculations of different surface species of cadmium and \ AlOH − 1/2 in 0.01 M NaNO3. In the calculation the following initial conditions were imposed; solid content = 2 g l − 1. (A) Initial [Cd2 + ]= 0.77 mM (B) [Cd]=7.74 mM; (C) [Cd]=77.4 mM NaNO3 = 0.01M. All other parameters used were given in Table 1 and Table 2. R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 of bidentate cadmium surface complex is optimized using the cadmium binding data given in Fig. 3 (B). Intrinsic binding constants of cadmium surface complexes were obtained from the adsorption edge data shown in Fig. 3 (B). These model parameters were subsequently utilized to quantify adsorption edges constructed at initial Cd2 + 0.77 and 7.78 mM, and adsorption isotherms shown in Fig. 2 (A). As shown in the Fig. 2(A) and Fig. 3 by solid and broken lines, the model calculations agree reasonably well with experimental data. Finally, different surface species of gibbsite and cadmium were calculated using CD MUSIC model, and the results obtained are given in Fig. 4. When compared to [(\ AlOH)21.2Cd0.8], free [\ AlOH − 1/2] and [\AlOH1/2] represent a major portion of the surface sites. According to crystallographic data, the maximum site density of gibbsite was estimated as 14 mmol m − 2. In all experiments, maximum Gcdnever exceeded 3.0 mmol m − 2. Therefore, gibbsite surface is unsaturated with respect to the adsorbate loading when initial [Cd2 + ]B7.78 mM. 4. Conclusions Surface sites of gibbsite can be treated homogeneous for cadmium ion binding. Only cadmium bidentate surface complex, \Al(OH)21.2Cd0.8 was used in CD MUSIC modeling. When [Cd]initial/ [Cd]absorbed 1 and pH \ 5, the rH/Cd was ranged between 1.2 and 1.6. This is attributed to a release of more than a proton per metal ions adsorbed. Acknowledgements We thanked International Foundation for Science (IFS, Sweden) for offering travel supports under research grant no. AF 2407-2 and gifting Orion Auto-chemistry Analyzer to our group (Grant No. A.F.2407). Meindert (WAU, The Netherlands) is thanked for supplying ECOSATFIT (ver. 4.7) to our group. The comments made by anonymous reviewer have enhanced manuscript quality. 23 References [1] W. Stumm, Chemistry of the Solid – Water Interface, Wiley, New York, 1992. [2] J. Buffle, Complexation Reactions in Aquatic Systems: An Analytical Approach, Ellis Horwood, London, 1988. [3] J.A. Davis, J.A. Coston, D.B. Kent, C.C. Fuller, Environ. Sci. Technol. 32 (1998) 2820. [4] K. Spitz, J. Moreno, A Practical Guide to Groundwater and Solute Transport Modeling, Wiley, New York, 1996. [5] B.K. Schroth, G. Sposito, Environ. Sci. Technol. 32 (1998) 1404. [6] A.W.P. Vermeer, J.K. Mcculloch, W.H. van Riemsdjik, L.K. Koopal, Environ. Sci. Technol. 33 (1999) 3892. [7] D.A. Dzombak, F.M.M. Morel, Surface Complexation Modeling: Hydrous Ferric Oxide, Wiley, New York, 1990. [8] F.J. Hingston, A.M. Posner, J.P. Quirk, J. Soil. Sci. 23 (1972) 176. [9] M.B. McBride, Soil. Sci. Soc. Amer. J. 49 (1985) 843. [10] M.B. McBride, Clays Clay Min. 38 (1982) 21. [11] T. Hiemstra, W.H. van Riemsdijk, M.G.M. Bruggenmert, Neth. J. Agric. Sci. 36 (1987) 281. [12] T. Hiemstra, W.H. van Riemsdijk, Coll. Surf. 59 (Section A) (1991) 7. [13] W.N. Rowlands, W. O’Brien, R.J. Hunter, V.J. Patrick, J. Coll. Interf. Sci. 188 (1997) 325. [14] T. Hiemstra, H. Young, W.H. van Riemsdijk, Langmuir 15 (1999) 5942. [15] R. Weerasooriya, B. Dharmasena, D. Aluthpatabendi, Coll. Surf. 170 (Section A) (2000) 65. [16] R. Weerasooriya, D. Aluthpatabendi, H.J. Tobschall, Colloid Surf., Section A, 189 (2001) 131. [17] US Environmental Protection Agency, Drinking Water Standards, 1995. Available from http://www.epa.gov. [18] T. Hiemstra, P. Venema, W.H. van Riemsdjik, J. Coll. Interf. Sci. 181 (1996) 45 – 59. [19] L.K. Koopal, W.H. van Riemsdjik, M. Groffery, J. Coll. Interf. Sci. 116 (1987) 197. [20] J.A. Davis, D. Kent, in: M.F. Hochela, A.F. White (Eds.), Mineral – Water Interface Geochemistry, American Geophysical Union, Washington DC, 1990. [21] J. Westall, in: W. Stumm (Ed.), Aquatic Surface Chemistry, Wiley, 1987. [22] E. Gileade, E. Kirowa-Eisner, J. Penciner, Interfacial Electrohemistry: An Experimental Approach, AddisonWesley, MA, 1972. [23] H. Tamura, N. Katayama, R. Furuchi, J. Coll. Interf. Sci. 195 (1997) 241. [24] P. Venema, T. Hiemstra, W.H. van Riemsdjik, J. Coll. Interf. Sci. 183 (1996) 515. [25] M. Keizer, W.H. van Riemsdijk, ECOSAT version 4.7, Equilibrium Calculations of Reactions and Transport, Department of Soil Science, Wagenigen Agricultural University, The Netherlands, 1999. [26] D. Kinniburgh, FIT User Guide Technical Report WD/ 93/23, British Geological Survey, Nottingham, 1993. 24 R. Weerasooriya et al. / Colloids and Surfaces A: Physicochem. Eng. Aspects 207 (2002) 13–24 [27] P. Trivedi, L. Axe, Environ. Sci. Technol. 34 (2000) 2215. [28] A.E. Martell, R.M. Smith, Critically Selected Stability Constants of Metal Complexes, National Institute of Standard Technology, USA, 1998 database version 5. [29] B.B. Johnson, Environ. Sci. Technol. 24 (1990) 112. [30] K.C. Hass, W.F. Schneider, A. Curioni, W. Andreoni, Science 282 (1998) 265. [31] K.F. Hayes, J.O. Leckie, J. Coll. Interf. Sci. 115 (1987) 564. [32] C. Papelis, G.E. Brown Jr., G.A. Parks, J.O. Leckie, Langmuir 11 (1995) 2041. [33] R.P.J.J. Rietra, T. Hiemstra, W.H. van Riemsdjik, Geochim. Cosmochim. Acta 63 (1999) 3009. [34] J.R. Bargar, S.N. Towle, G.E. Brown Jr., G.A. Parks, J. Coll. Interf. Sci. 185 (1997) 473.
© Copyright 2026 Paperzz