Surface complexation modeling of Cd(II) sorption to montmorillonite, bacteria, and their composite

Abstract. Surface complexation modeling (SCM) has emerged as a powerful tool for simulating heavy metal adsorption processes on the surface of soil solid components under different geochemical conditions. The component additivity (CA) approach is one of the strategies that have been widely used in multicomponent systems. In this study, potentiometric titration, isothermal adsorption, zeta potential measurement, and extended X-ray absorption fine-structure (EXAFS) spectra analysis were conducted to investigate Cd adsorption on 2 : 1 clay mineral montmorillonite, on Gram-positive bacteria Bacillus subtilis, and their mineral–organic composite. We developed constant capacitance models of Cd adsorption on montmorillonite, bacterial cells, and mineral–organic composite. The adsorption behavior of Cd on the surface of the composite was well explained by CA-SCM. Some deviations were observed from the model simulations at pH

Soil solid components, which include bacteria and clay minerals, often exist as composites in natural environments.The component additivity (CA) approach has been widely used in the application of surface complexation models for describing heavy metal adsorption on mixtures and on the composites of soil solid components.The CA-diffuse layer model (CA-DLM) has been successfully applied to model the adsorption of Co 2+ (Landry et al., 2009) and Pb 2+ (Re-Published by Copernicus Publications on behalf of the European Geosciences Union. ich et al., 2010), on mixtures of hydrous ferric oxide (HFO) and quartz with kaolinite.Dong and Wan (2014) indicated that the CA-SCM can provide excellent predictions for uranium (VI) adsorption on quartz-sand-dominated sediments.Fowle and Fein (1999) concluded that the CA approach can be used to model the adsorption of Ca 2+ , Cu 2+ , Cd 2+ , and Pb 2+ on mixtures of Bacillus licheniformis and B. subtilis.Additionally, CA could successfully describe Cd, Co, Sr, and Zn adsorption onto the mixtures of 10 species of Gram-positive bacteria (Yee and Fein, 2001).For the mixtures of minerals and organic systems, the extended constant capacitance model was used to depict Cd 2+ adsorption onto goethite, kaolinite, and Muloorina illite in the presence of citric acid (Lackovic et al., 2004a, b).The linear additivity model has succeeded in predicting Cu 2+ adsorption but has failed to simulate Ca 2+ adsorption on goethite in the presence of fulvic acid (Weng et al., 2008).The adsorption of Cd 2+ onto Comamonas spp.-ferrihydrite composite could be accurately predicted by the addition of the reaction between end-member ferrihydrite and bacteria (Song et al., 2009).However, the amount of Cd 2+ adsorption onto the composite was up to 10 % lower than that predicted by the additive approach.In the study of Alessi and Fein (2010) the CA approach was applied to describe the extent of Cd adsorption on the surface of two-and three-component mixtures of B. subtilis cells, HFO, and kaolinite at various mass ratios in the presence of acetate.The models matched well with experimental data in the absence of acetate, whereas they underestimated Cd adsorption in the presence of acetate.Moon and Peacock (2013) found that Cu adsorption behavior on ferrihydrite-B.subtilis composites with ferrihydrite as the major component can be modeled by a CA approach.By contrast, composites with a majority of bacteria component cannot be simulated by a CA approach, and significant deviations were observed.
Previous studies concerning surface complexation models have mainly focused on heavy metal adsorption by individual minerals or bacteria or mineral-organics and metal oxide-bacteria composites.To the authors' knowledge, no SCM investigation using 2 : 1 clay mineral-bacteria composites, has been performed.Montmorillonite, a representative of the 2 : 1 phyllosilicates, exhibits large specific surface area and high cation exchange capacity, which make it an effective adsorbent for heavy metals.The clay mineral is widely distributed in a variety of soils in temperate zone.The model developed in clay mineral-bacteria composites will provide insights into the behaviors of heavy metals in contaminated soils of these regions.Thus, in this study, we first develop surface complexation models for Cd adsorption on end-member 2 : 1 clay mineral montmorillonite, Grampositive bacteria Bacillus subtilis, and finally their mineralorganic composite by using the CA approach.Moreover, linear combination fitting of the extended X-ray absorption finestructure (EXAFS) spectra was applied to verify the distribu-tion ratio of Cd between different components derived from CA-SCM.

Adsorbent and reagent
Montmorillonite (Sanding Group Co. Ltd., Shaoxing, Zhejiang, China) was prepared following previously outlined procedures (Hong et al., 2012(Hong et al., , 2013)).The clays were oxidized using H 2 O 2 (30 %) to remove any residual organic matters.The < 2 µm colloidal fractions were isolated through sedimentation.The specific surface area, determined by the Brunauer-Emmett-Teller (BET) method (N 2 adsorption), was found to be 60.2 m 2 g −1 .Previous study has shown that a particle was estimated to be formed of nearly 20 stacked layers in the dehydrated state used in the gas adsorption experiment (BET' method), whereas it was estimated to be composed of only 1 or 2 layers in aqueous suspension (Tournassat et al., 2003).Thus, the values derived from BET' method usually underestimate the actual total surface area in aqueous suspension.Herein, we set the specific surface area of montmorillonite to be 800 m 2 g −1 , and the detailed processing procedures are shown in the Supplement.
The test bacteria used in this study came from the aerobic Gram-positive species B. subtilis, which were obtained from the State Key Laboratory of Agricultural Microbiology.The microorganisms were inoculated into 10 mL lysogeny broth (LB) medium and cultured for 7 h at 28 • C to reach the exponential phase with an optical density of 1.4 at 600 nm (OD 600 nm ).Subsequently, 2 mL of the resulting bacterial suspension was transferred to 200 mL LB medium for another 18 h inoculation (OD 600 nm = 2.7) to reach the midexponential phase.Cells were pelleted by centrifugation at approximately 7200 g for 5 min and rinsed three times in ultrapure water.The ratio of the wet weight to dry (60 • C in drying oven) weight of the biomass is 9 : 1.The biomass mentioned in this paper all refer to dry weight.
Montmorillonite and the bacterial cells were suspended in 0.01 M NaNO 3 to produce the suspensions that were used in the experiments.The montmorillonite-bacteria composite was prepared following a montmorillonite : B. subtilis mass ratio of 7 : 1 to produce suspensions (4 h, 28 • C).The resultant montmorillonite-bacteria composite suspensions were used for subsequent experiments.Preliminary isothermal adsorption (pH 8) of bacteria to montmorillonite results showed that the maximum adsorption predicted by Langmuir equation was approximately 0.14 g (dry cell)/g (clay), which was equal to a mass ratio (clay to bacteria) of 7 : 1 (Supplement Fig. S1).Clearly adsorption of bacteria on montmorillonite decreased with the increase of pH; thus bacteria can be completely adsorbed by montmorillonite to a large extent in all experimental conditions (pH 3-8.5) at this ratio.The bacte-rial cells remained intact under the experimental condition (see AFM images of Fig. S2).
Analytical reagent grade chemicals were used throughout the study.Cd(NO 3 ) 2 was used to prepare 178.0 µmol L −1 Cd(II) stock solution.Water was Milli-Q reagent grade.

Zeta potential analysis
Zeta potential analysis was conducted to study the surface charge properties of montmorillonite, B. subtilis, and montmorillonite-B.subtilis composite.Suspensions were prepared the same as in Sect.2.1, over the pH range of 3 to 8. Samples were injected into a zeta potential analyzer (Ze-taplus90, Brookhaven) and measured in triplicate.

Potentiometric titrations
Titration experiments were performed (in duplicate) according to the procedures described by Fein et al. (2005).The suspensions (1 g L −1 , 40 mL) of montmorillonite and B. subtilis and the composite in 0.01 M NaNO 3 were placed in a sealed titration vessel maintained under a positive pressure of N 2 .Titrations were conducted using an automatic potentiometric titrator (Metrohm titrator 836, Switzerland).Each suspension was first titrated to pH 3 by using 0.1136 M HNO 3 solutions and then to pH 8.5 by using 0.0976 M NaOH.At each titration step, a stability of 0.01 mV S −1 was attained before adding the next drop of the titrant.
Gran plot was adopted to determine the specific volume of titrant added at the equivalence point (V e ) and to derive the total H + concentration (TOTH; Chu et al., 2002;Du et al., 1997;Liu et al., 1999Liu et al., , 2013)).
The values for the Gran function (G) were calculated as follows: × 100 on the acidic side, and (13.8−pH)  × 100 on the alkaline side.
Gran plots for titration data on montmorillonite and B. subtilis are illustrated in Fig. S3, where V 0 stands for the initial volume of the titrimetric suspension; and V a and V b represent the total volume of acid and alkaline solution added at each titration point, respectively.In the Gran plots, the OH − added to a suspension firstly neutralizes the excess H + before V eb1 , then reacts with the functional group on the surface of the sorbent between V eb1 and V eb2 , and finally changes the pH after V eb2 .No apparent surface reactions consuming OH − before V eb1 were noted.The total concentration of reacting H + at each titration point (TOTH) was calculated using the following equation, where C b stands for NaOH concentration:

Adsorption experiment
The adsorption edge experiments were conducted in 0.01 M NaNO 3 solutions to investigate the adsorption of Cd onto montmorillonite, B. subtilis, and the composite; 15 mL 178.0 µmol L −1 Cd(II) solution and 15 mL 2 g L −1 adsorbent suspensions were added to 50 mL centrifuge tubes.
A suitable volume of concentrated HNO 3 or NaOH solutions was added to adjust the pH of the suspensions from 3 to 8. The final concentration of Cd and adsorbents were 89.0 µmol L −1 and 1 g L −1 , respectively.Fixed pH isotherm experiments were performed at pH 6 and 8 in 0.01 M NaNO 3 electrolyte.The initial concentrations of Cd ranged from 8.9 to 89.0 µmol L −1 with a sorbent concentration of 1 g L −1 .The mixtures were allowed to equilibrate for 2 h at 28 • C.After equilibration, the final suspensions were centrifuged at 12 557 g for 5 min, and the clear supernate was analyzed using atomic adsorption spectroscopy (AAS; Varin AAS240FS) to determine the residue Cd.Solution speciation of 89.0 µmol L −1 Cd(II) in 0.01 M NaNO 3 was calculated as a function of pH using the geochemical software Visual MINTEQ 3.1.Cd 2+ was determined to be the main species at the experimental conditions.Cd(OH) 2 , CdOH + , and Cd 2 OH 3+ were also found at pH > 7.8, but in relatively small percentages.

X-ray absorption spectroscopy
We selected pH 6 sample of the adsorption edge experiments in Sect.2.4 as representative to perform the X-ray absorption spectroscopy (XAS) experiments.The solid phases after centrifugation were transformed immediately to collect XAS spectra in 12 h.The sample holders were packed together with moist tissues (stored at ∼ −4 • C) to minimize sample drying during shipment to the Shanghai Synchrotron Radiation Facility (SSRF), China.The X-ray absorption data at the Cd K-edge (∼ 26 700 keV) of the samples were recorded at room temperature in the fluorescent mode with silicon drift fluorescence detector.The station was operated with a Si(III) double crystal monochromator.During the measurement, the synchrotron was operated at energy of 3.5 GeV and a current between 150 and 210 mA.EXAFS spectra were collected from 26511 to 27551 eV.Background subtraction and normalization of the XAS data were conducted in ATHENA, a program in the IFEFFIT package (Ravel and Newville, 2005).The background subtraction was performed using PySpline.The pre-edge was fit to a linear function and the post-edge background to two second-order polynomial segments.To obtain information on the partitioning of Cd between the different fractions of the composites, the spectra were fitted using a linear combination fitting procedure (LCF).Cd-adsorbed montmorillonite, Cd-bacteria were used as an end-member for the two-component fitting of the binary composites, with linear combination performed over the k 3 range of 2.3-10 Å −1 (force weight to sum to 1).The goodness of fit was evaluated by the best-fit criterion values (R; Du et al., 2016).

Surface complexation modeling
The program FITEQL 4.0 was utilized to determine functional group site concentrations, proton, and metal binding stability constants (Herbelin and Westall, 1999).A constant capacitance model was implemented (Tertre et al., 2006).The activity coefficient correction was made within FITEQL using the Davies equation.An indicator of the goodness of fit is the overall variance, V , in Y (Herbelin and Westall, 1999).
DF , where WSOS is the weighted sum of squares of the residuals and DF is the degrees of freedom.Lower V Y values signify better fits.The systematic process of developing the SCM was started by fitting the model to the experimental data for montmorillonite and B. subtilis.
The potentiometric titration data was fit to generate the acid-base constants and surface site densities (Fig. S4).Acid-base reactions on the surface of montmorillonite are represented by the following equations: where SOH and WOH are aluminol and silanol sites on the surface of montmorillonite, respectively.X − is the negative exchange site that can react with both H + and Na + .The total concentration of exchange site and the binding stability constants of XH and XNa were used as fixed input parameters when the titration data were fitted.These three values were 8.7 × 10 −4 mol g −1 , −2.2, and 1.4, respectively (Table S1 in the Supplement), which come from Tertre et al. (2006) and Soltermann et al. (2014).The acid-base constant and the site density of SOH and WOH were set as iterated parameters.In addition, these parameters were iterated simultaneously.
The acid-base reactions on the surface of B. subtilis were represented by where R represents the bacterium to which each functional group is attached.Carboxyl sites, phosphate sites, and hydroxyl sites were considered to be the main functional groups on the surface of B. subtilis (Yee and Fein, 2001).L i H 0 represents these three groups.Acid-base equilibrium constants and site densities (Tables S1 and S2) were used as fixed input parameters in the modeling of the Cd pH adsorption edge and fixed pH isotherm data.The constants of Cd complexes with the functional groups on each sorbent were iterated simultaneously.The following equilibria were used to describe Cd(II) adsorp- tion onto montmorillonite: Due to their higher pK a , hydroxyl sites are not deprotonated over the range of pH 3 to 8. To simplify the model, hydroxyl sites were not invoked.Cd(II) adsorption on B. subtilis was described with the following reaction equilibria: Finally, we combined the model of Cd(II) adsorption on montmorillonite and B. subtilis to simulate the adsorption behavior of Cd(II) on the composite.End-number montmorillonite and B. subtilis models were combined in FITEQL 4.0 using the CA approach.

Zeta potential
The zeta potentials of montmorillonite, B. subtilis, and montmorillonite-B.subtilis composite are given in Fig. 1.Over the whole pH range studied, both montmorillonite and the composite had negative surface charge with very low pH pzc around pH 3.1.The bacteria displayed a negative surface charge at pH > 3.5 and had a higher pH pzc than the value reported by Fein et al. (2005) (pH pzc < 2).The negative electrokinetic charge of all sorbents increased with pH and remained constant when pH was higher than 5.The zeta potentials of montmorillonite and bacteria were −2.29 and 12.27 mV at pH 3, −26.70, and −19.46 mV at pH 8.The surface of the montmorillonite-bacteria composite was more negatively charged than that of montmorillonite and B. subtilis at pH > 5; the zeta potential of the composite was −20.30 mV at pH 3.

Cadmium adsorption
Figure 2 shows the pH dependence of Cd adsorption on montmorillonite, B. subtilis, and montmorillonite-B.subtilis composite.Adsorption onto montmorillonite increased slowly over the pH ranges, and about 55 % of the total Cd(II) was adsorbed at pH 8. Adsorption onto B. subtilis increased sharply between pH 3 and 7 and then reached a plateau at pH 7-8.Nearly 90 % of total Cd(II) was adsorbed at pH 8.In comparison to bacteria, the uptake of Cd on montmorillonite-B.subtilis composite increased more slightly over the range of pH 3 to 8. The extent of adsorption onto the composite increased swiftly from pH 3 to 5 and varied slightly at pH 5-8.About 60 % of the total Cd(II) ions were adsorbed onto the composite at pH 8.

Modeling of Cd adsorption onto montmorillonite and B. subtilis
The titration data for montmorillonite and B. subtilis are presented in Fig. S4.Bacteria exhibited significant buffering capacity over the range of pH 3 to 8.5, which was assigned to the deprotonation of functional groups on cell walls (Borrok et al., 2004;Moon and Peacock, 2011;Yee and Fein, 2001).Montmorillonite presented strong buffering capacity only at pH < 4 and pH > 7.5.This result was in line with the report of Ikhsan et al. (2005), which also showed similar buffering be- a Fixed at those determined for montmorillonite-bacteria composite (obtained from Tables S1   and S2).b Calculated using the surface site densities or surface areas or the specific capacitance of the end-member montmorillonite and B. subtilis weighted to the mineral : bacteria mass ratio of 7 : 1.
haviors of a calcium montmorillonite over these pH ranges.
The titration data of montmorillonite and B. subtilis were successfully modeled with Eqs. ( 1)-( 5) and ( 6).The protonation constants and site densities of sorbents calculated from titration data are indicated in Tables S1 and S2.
Adsorption of Cd onto the surface of montmorillonite and B. subtilis was successfully modeled with Eqs. ( 7)-( 10).The best fit for cadmium adsorption on montmorillonite is depicted in Fig. S5a.The adsorption of Cd(II) onto montmorillonite was modeled using two adsorbed species: X 2 Cd and SOCd + .The calculated complex stability constants of these two species were 8.30 and −1.68, respectively (Table S1).The complexation of Cd(II) onto two X − sites (X 2 Cd) dominated at pH < 4.5 while monodentate complexation onto SOH sites of the crystal edges (SOCd + ) predominated at pH > 4.5.The V Y value of the model which describes the Cd adsorption on montmorillonite was 5.6.The best-fit model for the Cd adsorption onto B. subtilis is presented in Fig. S5b.The optimized logK R−COOCd+ and logK RPOCd+ values were 2.54 and 2.87, respectively (Table S2).Surface complexation model successfully simulated the adsorption edge of B. subtilis over the pH ranges of 3 to 8. The model predicted that one inner-sphere monodentate complexation (R-COOCd + ) dominated at low pH while another inner-sphere monodentate complexation (R-POCd + ) dominated at high pH.The V Y value of the model which described the Cd adsorption on bacteria was 3.8.

Modeling of Cd adsorption onto montmorillonite-B. subtilis composite
The parameters for the model of Cd adsorption onto montmorillonite-B.subtilis composite are shown in Table 1.
As per the CA approach, the constants for protonation and the acid-base reactions on the surface of the composite were the same as those for the end-member montmorillonite and B. subtilis.The operational site densities of functional groups used in the model of Cd sorption on the composite were calculated based on the end-member site densities weighted to the montmorillonite : bacteria mass ratio (Moon and Peacock, 2013).The specific capacitance and surface area were calculated in the same way.Four Cd complexation species were considered in the model of Cd sorption on the composite, i.e., X 2 Cd and SOCd + on montmorillonite; two innersphere complexes on bacteria.The complexation constants of the four species were the same as those optimized by the models of Cd adsorption on montmorillonite and B. subtilis.
The adsorption behavior of Cd on montmorillonite-B.subtilis composite was well described by CA-SCM, except some deviations were observed at pH < 5 (Fig. 3, V Y = 15.4).The complex X 2 Cd dominated below pH 4. At pH 4.5, the SOCd + was the major species with deprotonation of the SOH surface sites.At pH 3-5.5, the inner-sphere complex R-COOCd + on the surface of bacteria contributed 10-20 % of the total Cd adsorption.At pH above 5.5, the inner-sphere complex on the bacterial surface (R-POCd + ) played an important role and amounted to 10-20 % of the total Cd adsorption.The CA-SCM provided an adequate fit to the adsorption isotherms of Cd on the composite at pH 6 and 8, with V Y values of 4.3 and 2.4, respectively (Fig. S6). Figure 4 depicts the distribution of the total adsorbed Cd between the composite fractions.Montmorillonite played a dominating role in Cd sorption on the composite over the whole pH range studied.The percentage for the adsorption of Cd on montmorillonite and B. subtilis in the composite at pH 6 was 71.4 and 28.6 %, respectively.The V Y value ranging from 0.1 to 20 can generally be considered a good fit.Overall, within the constructs of FITEQL SCM, CA-SCM is adequate to describe the Cd adsorption behavior on the surface of the composite.

Distribution of Cd on montmorillonite-B. subtilis composite by XAS
Since the model successfully described Cd adsorption on bacteria-clay composite at pH > 5, herein we selected pH 6 samples to investigate whether model predictions match well with the EXAFS prediction.Linear combination fitting (LCF) procedure has been successfully applied to obtain the partitioning of metal adsorption on bacteria-mineral composites (Du et al., 2016;Moon and Peacock, 2012;Templeton et al., 2003).This approach is based on the intermediate adsorption behavior of metals between the end-member components.The Cd K-edge (EXAFS) spectra of montmorillonite, Bacillus subtilis, and their composite are shown in Fig. 5.A few differences between the spectrum of mineral and bacteria are pronounced at k values of 6-10 Å −1 .We were able to produce excellent fits to the Cd-composite spectra with a linear combination of Cd-adsorbed montmorillonite and Cd-adsorbed B. subtilis clusters (R factor < 0.1).Linear combination results for the distribution of total adsorbed Cd by the composite yielded a ratio of 69.1 % : 30.9 % for montmorillonite and bacterial cells, respectively.This ratio closely approaches the value obtained from model speculation, which is 71.4 % : 28.6 %.

Discussions
A number of investigators have employed CA-SCM to describe metal adsorption onto mineral-bacteria composites (Alessi and Fein, 2010;Moon and Peacock, 2013;Song et al., 2009).However, to the authors' knowledge, this is the first information on the modeling of Cd adsorption by 2 : 1 clay mineral-bacteria composites.The species and distribution of Cd on the surface of montmorillonite-B.subtilis composite were modeled using CA-SCM in this study.Cadmium was found to be adequately described by adsorption to the nominally assigned RCOOH and POH sites, and the X − and SOH groups on the surface of montmorillonite-B.subtilis composite.Our results imply that all these groups have to be considered for a successful simulation of Cd adsorption on 2 : 1 clay mineral-bacteria mixtures, compared to ferrihydrite-bacteria composites where, of the bacterial com- ponent, only carboxyl (RCOOH) groups were found to be responsible for heavy metal adsorption over the entire pH range studied (Moon and Peacock, 2013).The model of Cd adsorption on montmorillonite-B.subtilis composite in our study suggested that X 2 Cd was the major species on the composite at pH < 4. Interestingly, montmorillonite played a dominant role in Cd sorption on the composite (the mass ratio of montmorillonite to bacteria was 87.5 % : 12.5 %) over the entire pH range studied.However, the contribution to Cu adsorption from the components of a ferrihydrite-bacteria mixture with a mass ratio of 82 % : 18 % was dependent on pH (Moon and Peacock, 2013).Bacteria and ferrihydrite governed Cu adsorption at low and high pH values, respectively.The montmorillonite has a permanently charged XH site, which provides a certain level of adsorption capacity independent of pH, whereas the ferrihydrite-bacteria composites to which the authors are contrasting their work possess only amphoteric sites (Moon and Peacock, 2013).Our data revealed that the percentage of total adsorbed Cd on montmorillonite in the composite was in the range of 87.3-69.1 % from pH 3 to 8 (Fig. S7).These results suggest that the bacteria fraction in the composite plays an increasingly important role in the retention of cadmium ions from pH 3 to 8, although slight decreases in Cd attachment were observed for the bacteria from pH 4 to 5. Some of the Cd ions may have been transferred from the montmorillonite surface to the bacteria in the composite with the rise of pH.
The modeled distribution of Cd on individual components in our montmorillonite-B.subtilis composite was supported by the EXAFS investigations at pH 6.The ratio of Cd adsorption onto montmorillonite to that of the bacterial cells present in the composite as obtained from EXAFS spectra analysis closely coincides with the value obtained from model speculation.Both the model simulation and EXAFS observation from our study suggested that the fate of Cd is largely affected by the presence of montmorillonite in soils and sediments, particularly in locations where the mineral exists as the major clay mineral component.Our 2 : 1 claybacteria system is different from the 1 : 1 clay-bacteria system reported by Alessi and Fein (2010), where they found that the majority of Cd is predicted to occur onto the permanent structural site of the kaolinite component only below pH 4 in a kaolinite-B.subtilis mixture dominated by the mineral fraction (75 %), and B. subtilis contributes nearly 58 % of the total reactive site.Templeton et al. (2003) reported that 71 and 29 % of Pb ions were adsorbed by goethite and bacteria in a goethite-bacteria composite with a mass ratio of 70 % : 30 % at pH 7. Moon and Peacock (2012) found that the distribution of Cu adsorption between ferrihydrite and bacteria fractions in a ferrihydrite-bacteria composite (with a mass ratio 65 % : 35 %) was 80 % : 20 % at pH 6.4.Our work thus corroborates the findings of Moon and Peacock (2012), who predict that, in mineral-organic composites dominated by the mineral component, the mineral will be responsible for the majority of the heavy metal adsorption and the composites will behave approximately additively.Moreover, this study has extended their work, which was based on iron (hydr)oxide-organic composites, to clay mineral-organic composites.The established models from these investigations may provide a universal rule to predict the behavior of trace metals in more complicated ecosystems bearing mineral-organic complexes.
Our study provides useful insights on clay-bacteria interactions in the biogeochemical interface of complex soil and its environmental consequence.Although SCM has been widely applied to describe trace metal adsorption behaviors www.biogeosciences.net/13/5557/2016/Biogeosciences, 13, 5557-5566, 2016 Under this premise, mineral-organic composite is a suitable analogue for a good predictor of the mobility and thus eventual fate of Cd in natural environments.Lastly, our study has implications for the remediation of Cd contaminant soils and sediments where 2 : 1 phyllosilicates dominate the mineral composition, in particular, when various organic agents are applied, e.g., straw, manure, or engineering microorganisms.These substances are easily bound to soil mineral surface.In these cases, it is vital to consider the fate, mobility of heavy metals with respect to mineral-organic composites, in addition to pure mineral phases.

Conclusions
In this study, we developed surface complexation model for Cd(II) adsorption on montmorillonite, B. subtilis, and their composite.The Cd adsorption behavior onto the surface of the composite can be described by a CA approach (CA-SCM).Underestimation was observed for the model prediction at pH < 5.The X 2 Cd, SOCd + , R-COOCd + , and R-POCd + were the predominant species of Cd adsorbed on the surface of montmorillonite-B.subtilis composite.Montmorillonite played a more important role for the binding of Cd ions in the mineral-bacteria composite over the ranges of pH 3 to 8. The distribution of adsorbed Cd between the montmorillonite and bacteria fractions in the montmorillonite-B.subtilis composite as predicted by CA-SCM closely coincided the results obtained from the EXAFS spectra analysis.
The Supplement related to this article is available online at doi:10.5194/bg-13-5557-2016-supplement.
N.Wang et al.: Surface complexation modeling of Cd(II)

Figure 1 .
Figure 1.Zeta potential of montmorillonite, B. subtilis, and montmorillonite-B.subtilis composite at different pH values.

Figure 2 .
Figure 2. Adsorption of Cd(II) to montmorillonite, B. subtilis, and montmorillonite-B.subtilis composite as a function of pH in the presence of 0.01 M NaNO 3 (symbols are data points and lines are model fits).The sorbent and initial Cd concentration is 1 g L −1 and 10 ppm, respectively.

Figure 4 .
Figure 4. Distribution of the total adsorbed Cd between montmorillonite and B. subtilis fractions in the composite predicted by CA-SCM (symbols are data points, lines are model fits) at 28 • C in the presence of 0.01 M NaNO 3 .Cd(II) : 10 ppm, composite : 0.875 g L −1 Mont +0.125 g L −1 B. subtilis.

Figure 6 .
Figure 6.Schematic cartoon illustrating Cd binding on clay-bacteria composite.

Table 1 .
Surface complexation model parameters for Cd(II) sorption on montmorillonite-B.subtilis composite.