Materials Science and Engineering C 33 (2013) 3723–3729

Contents lists available at ScienceDirect

Materials Science and Engineering C

j ourna l homepage: www.e lsev ie r .com/ locate /msec

Reduction of Cr(VI) facilitated by biogenetic jarosite and analysis of itsinfluencing factors with response surface methodology☆

Zhihui Xu a,⁎, Bo Lu a, Jingyu Wu a, Lixiang Zhou b, Yeqing Lan a

a Department of Chemistry, College of Science, Nanjing Agricultural University, Nanjing 210095, PR Chinab Department of Environmental Engineering, College of Resources and Environmental Sciences, Nanjing Agricultural University, Nanjing 210095, PR China

☆ This is an open-access article distributed under the tAttribution-NonCommercial-No DerivativeWorks License,use, distribution, and reproduction in anymedium, provideare credited.⁎ Corresponding author. Tel.: +86 25 84396697; fax:

E-mail address: xuzhihui@njau.edu.cn (Z. Xu).

0928-4931/$ – see front matter © 2013 The Authors. Puhttp://dx.doi.org/10.1016/j.msec.2013.05.006

a b s t r a c t

a r t i c l e i n f o

Article history:Received 7 March 2013Received in revised form 23 April 2013Accepted 2 May 2013Available online 10 May 2013

Keywords:JarositeCr(VI) reductionSulfideResponse surface methodologyBox–Behnken Design

In this paper, the facilitating role of biogenetic jarosite in the reduction of Cr(VI) by sulfide and its mechanismwere investigated through batch experiments and analysis of X-ray photoelectron spectrum (XPS). To studythe effects of operational parameters on the reduction of Cr(VI) by sulfide, four operational parameters (pH ofsolution, operation temperature, loading of jarosite and reaction time) were considered as input variables forresponse surface methodology (RSM). Graphical response surfaces and contour plots were used to evaluate theeffect of interaction between operational parameters on the reduction of Cr(VI). The results suggest that a cycleprocess of converting Fe(III) to Fe(II) occurred on the surface of jarosite and markedly accelerated the reductionof Cr(VI) by sulfide. For example, the efficiency of Cr(VI) reduced by sulfide increased from 20.5% to 100% whenjarosite (1 g/L)was added to the homogenous reaction systemat pH = 8within 40 min. The analysis of variance(ANOVA) revealed a high coefficient of determination (p-value b 0.0001, R2 = 97.99%, Adj-R2 = 95.98%) be-tween experimental Cr(VI) removal efficiency and predicted one by RSM developed model. The Pareto analysisresults demonstrated that the pH of solution was the most significant term of the developed model. Operationtemperature, loading of jarosite and reaction time exhibited synergistic effects on the reduction of Cr(VI), andthe effect of interaction between independent factors on the response factor can't be ignored.

© 2013 The Authors. Published by Elsevier B.V. All rights reserved.

1. Introduction

Hexavalent chromium is toxic and mobile, causing particular envi-ronmental concern. Thus it is essential to be removed from industrialwastewater [1]. A strong oxidizing agent, hexavalent chromium iscarcinogenic and mutagenic and diffuses quickly through soil andaquatic environments. The most probable Cr(VI) species in aqueoussolution are Cr2O7

2−, CrO42−, H2CrO4, and HCrO4

−, the relative distribu-tion of which depends on the solution pH, on the Cr(VI) concentrationand on the redox potential [2]. However, none of these Cr(VI) speciesform insoluble precipitates making its separation impossible with adirect precipitationmethod [3], whereas Cr(III), being less toxic andmo-bile, can be readily precipitated out of solution in the form of Cr(OH)3[4,5]. To remove Cr(VI) from the aqueousmedia, it is necessary to reduceCr(VI) to Cr(III) first.

Many reductants, such as ferrous, sulfide and organic acids naturallyexisting in environments, can directly and effectively reduce Cr(VI) toCr(III) [6–10]. For instance, Buerge and Hug [8] reported that Cr(VI)

erms of the Creative Commonswhich permits non-commerciald the original author and source

+86 25 84395525.

blished by Elsevier B.V. All rights re

can be rapidly converted to Cr(III) in the presence of Fe(II) with a reac-tion rate decreasing from pH 1.5 to 4.5 and increasing from pH 5.5 to8.7. It is found that goethite and lepidocrocite can markedly enhancethe reduction of Cr(VI) by Fe(II) [11]. It is reported that hydrogen sulfidediluted by nitrogen gas can immobilize 90% to approximately 100% ofinitial Cr(VI) in soil columns [10]. Kim et al. [12] found that under anaer-obic condition elemental sulfur is a major product of sulfide oxidationby Cr(VI). It is first revealed that produced elemental sulfur serving asa catalyst greatly accelerates Cr(VI) reduction at the later stage of the re-action [6]. Recent studies show that activated Fe(III)/Fe(II) adsorbed onthe surfaces of soils or minerals can markedly facilitate the reduction ofCr(VI) by sulfide [13,14].

Jarosite (KFe3(SO4)2(OH)6) is a secondary iron sulphate mineraloften found in acid rock or acid mine drainage (ARD/AMD) environ-ments, particularly in mining wastes from polymetallic sulfide ore de-posits. Most studies have been done on the application of syntheticjarosite [15–20]. Based on the results by Lan et al. [14] and Xu et al.[20], it is speculated that activated Fe(III) as a constituent on jarositesurfaces might also play a facilitating role in the reduction of Cr(VI) bysulfide. In this study, biogenetic jarosite serving as a ferric source isintroduced into the reaction system to investigate the possible role ofjarosite in the reduction of Cr(VI) by sulfide and its mechanism, whichis beneficial to understand the transformation of Cr(VI) to Cr(III) inthe environment and to provide scientific evidence to search novelsoil remediation techniques for chromium decontamination.

served.

Table 1Actual and coded values of independent variables used for experimental design.

Variables (factors) Symbol Ranges and actual values ofcoded levels

−1 0 +1

pH of the solution X1 7.5 8.0 8.5Operating temperature (°C) X2 15 25 35Loading of jarosite (g/L) X3 0.8 1.0 1.2Reaction time (min) X4 20 30 40

3724 Z. Xu et al. / Materials Science and Engineering C 33 (2013) 3723–3729

Recently, response surface methodology (RSM) has been employedto optimize and understand the performance of complex systems[21–27]. By the application of RSM it is possible to evaluate the interac-tions of operational parameters on treatment efficiency with a limitednumber of planned experiments. The variables (independent factors)used in this study were pH of solution, operation temperature, loadingof jarosite and reaction time. Cr(VI) reduction efficiencywas consideredas the dependent factor (response). The application of RSM is to inves-tigate the effects of independent factors and their interactions on the ef-ficiency of sulfide reducing Cr(VI).

2. Materials and methods

2.1. Chemicals

All chemicals used in this study were of reagent grade. Sodiumsulfide crystal (Na2S·9H2O) was rinsed with deionized water two orthree times until the possibly oxidized surface layer was completelyremoved. In order to eliminate its loss through oxidation and volatil-ization, sulfide (Na2S) stock solution was prepared prior to use. Cr(VI)stock solution was prepared with K2Cr2O7 crystal, which was dried for2 h at 120 °C before weighing. 1, 5-Diphenylcarbazide (DPC), a colordevelopment for Cr(VI), was purchased from Sigma-Aldrich. 0.25 gDPC was dissolved in 100 mL acetone and kept in a refrigerator at4 °C prior to use. All solution pHs in the study were controlled by0.10 M phosphate buffer, obtained by dissolving NaH2PO4 in deion-ized water and adjusting with 0.1 M NaOH or HCl solution to the de-sired pHs (7.5–8.5). Such a narrow pH range was selected mainly tofacilitate the kinetic measurement in batch experimental systems. Ourpreliminary experiments and previous reports [6,14] demonstrated thatthe rate of Cr(VI) reduction was too fast at pH b 7.0 and too slow atpH > 9.0.

2.2. Preparation of jarosite

Jarosite was synthesized in the laboratory with a modified methoddescribed in our previous work [28]. In brief, the freshly preparedAcidithiobacillus ferrooxidans LX5 cell suspension was introducedinto an Erlenmeyer flask containing 240 mL of 0.16 M FeSO4 and53.3 mM K+ with pH 2.5. The flask was incubated in a reciprocalshaker at 180 rpm and 28 °C for 3 days and the resulting precipitateswere collected by filtration with a Whatman No. 4 filter film, washedwith 0.1 MH2SO4 and deionizedwater, respectively. Finally, the collect-ed minerals were dried at 60 °C for 6 h. The powder X-ray diffraction(XRD) patterns of the minerals obtained in this study demonstratedthe same mineralogy phase as described by Bai et al. and Xu et al.[28,29].

2.3. Experimental procedures

The desired amounts of phosphate buffer, jarosite, Cr(VI) and sul-fide stock solutions were sequentially introduced into a 40-mL amberbottle. The final total volume of suspension was 40 mL. The initialconcentrations of Cr(VI) and sulfide were 100 and 400 μM, respectively.Most of our experiments had a jarosite loading of 1.0 g/L. In our prelim-inary experiment, it was observed that the effect of dissolved oxygen onthe rates and extent of Cr(VI) reduction by sulfide was weak. Thus, solu-tions were not degassed with pure N2 to remove oxygen in this study.The amber bottle was closedwith a screw capwith a Teflon-lined siliconseptum and placed into a water bath at 25 ± 0.1 °C. Samples (about1.50 mL) were taken at required intervals with 3 mL plastic syringethrough the cap. Then, approximately 1.5 mL of each samplewas filteredthrough a 0.45 μmmembrane filter for Cr(VI) analysis.

Fe(III) (50 and 100 μM)was introduced into the homogeneous so-lution to examine the catalysis of externally added Fe(III) for the re-duction of Cr(VI) by sulfide. In contrast, NaF (50 and 100 μM) was

added to the heterogeneous system to investigate that F− suppressesthe catalysis of jarosite.

No strong electrolytes were adopted to control ionic strength inthis study since some literatures [9,14,30] indicated that the reduc-tion of Cr(VI) by sulfide was independent of ionic strength when itwas between 0.0 and 1 M. In this study, the ionic strength was essen-tially controlled by 0.1 M phosphate buffer.

2.4. Experimental design and data analysis

Four factors in the experiment process, viz. pH of the solution, oper-ating temperature, loading of jarosite and reaction time were selectedto be analyzed, aiming to figure out their influence on the reduction ofCr(VI) by sulfide in aqueous medium. The experimental ranges of thefour factors, introduced as RSM input variables, in coded and actualvalues are presented in Table 1. The number of the batch experimentalwas determined to be 29 by Box–Behnken Design (BBD) approach. Forthe evaluation of experiment data, the response variable was fitted by asecond-ordermodel in the formof quadratic polynomial equation givenbelow:

Y ¼ b0 þ∑biXi þ∑biiX2i þ∑bijXiXj ð1Þ

where Y is the predicted response (Cr(VI) reduction efficiency, %) usedas the dependent variable; Xi or Xj (i = 1, 2, 3 and 4; j = 1, 2, 3 and 4)is the independent variable or factor and b0, bi (i = 1, 2, 3 and 4), biiand bij (i = 1, 2, 3 and 4; j = 1, 2, 3 and 4) are the constant, linear,quadratic, and interaction coefficients, respectively. Experimental datawere analyzed by the use of Design-Expert 8.0.6 program includinganalysis of variance (ANOVA) to obtain the interaction between thevariables and the response. The probability values (p-value)were utilizedto check the important and effective terms of the developed model. Thesmaller the p-value, the more significant the corresponding parameterin the regression model is [31]. If the p-value was greater than 0.05, thecorresponding termwould be unmeaning and insignificant in the devel-oped model. Two-dimensional contour plots and three-dimensionalcurves of the response surfaceswere developed using the sameprogram.The extent of the influence of each variable on the response factor couldbe determined by calculating the percentage effect of each term on theresponse known as Pareto analysis [32,33]. The percentage effect ofeach term (Pi) could be calculated using the following formula:

Pi ¼b2i

∑ni¼1b

2i

!� 100 i≠0: ð2Þ

2.5. Analysis methods

A CyberScan pH2100 Bench Meter (Eutech Instruments), afterthree-point calibration, was applied to measure pH values. Cr(VI)concentration was determined by DPC colorimetric method, using a di-lute sulfuric acid solution (pH 2.0) to control pH for the color develop-ment. The absorbance was measured in a 1-cm cell at 540 nm on aUV-9100 Spectrophotometer (Beijing Ruili Corp.). Sulfide concentration

3725Z. Xu et al. / Materials Science and Engineering C 33 (2013) 3723–3729

in the stock solution was determined by the standard iodometric titra-tionmethod prior to use [34]. The concentration of total chromiumwasanalyzed by atomic absorption spectrometry (HITACHI 180-80). Pointof zero charge (pHPZC) for biogenetic jarosite was determined by titra-tion. Before titration, suspension was allowed to equilibrate and settlefor 20 h. Titration of 10 mL of precipitate suspension (1 g/L) with0.1 M NaOH or 0.1 M HCl was carried out in the range of pH 2–9. TheX-ray photoelectron spectrum was performed in a Kratos AXIS His,Mono Al Kα system (1486.71 eV, Kratos analytical, Japan).

3. Results and discussion

3.1. Facilitating role of jarosite in the reduction of Cr(VI) by sulfide

The concentration of Cr(VI) with reaction time under differentreaction conditions is illustrated in Fig. 1. Negligible concentrationchange of Cr(VI) was observed over 80 min in the absence of sulfideand jarosite. Eary and Rai [35] observed that 0.01 M NaH2PO4 couldeffectively displace Cr(VI) species adsorbed on the soil surfaces. Jamesand Bartlett [36] reported that phosphate displaces 75–100% of adsorbedCr(VI) species from Cr(VI)-treated soils, with a remaining 0–25% proba-bly being reduced to Cr(III) by soil. As shown in Fig. 1, only a little con-centration decrease of Cr(VI) was observed during the test time in thepresence of jarosite but the absence of sulfide, which suggests thatthe adsorption of Cr(VI) on jarosite surface in 0.1 M phosphate buff-er can be ignored when investigating the reduction of Cr(VI) facili-tated by jarosite. Oxidation–reduction reaction between Cr(VI) andsulfide proceeded slowly in the absence of jarosite. About 20.5%and 35.6% of initial Cr(VI) in the homogeneous system were convertedto Cr(III)within 40 and80 min, respectively. However, when the jarosite(1 g/L) was added to the reaction system, the reduction rate of Cr(VI)was markedly enhanced. It took approximately 40 min for Cr(VI) con-centration dropping to the detection limit.

In the heterogeneous system, the adsorption of reactants by solidparticles is important, because the reaction occurs on the solid sur-faces [37,38]. It has been known that jarosite is amphoteric in surfacecharge and the pHPZC of biosynthesized jarosite was measured to be5.6. This means that the surface of jarosite at pH b 5.6 is positivelycharged while it is carried with negative charges at pH > 5.6. There-fore, jarosite surface will hold negative charges at the tested pH,which is not conductive to the adsorption of Cr(VI) and sulfide onjarosite surface. However, rapid reduction of Cr(VI) took place. So, itcan be speculated that there are multiple pathways of Cr(VI) reduc-tion by sulfide in a heterogeneous system. Apart from the direct re-duction of Cr(VI) by sulfide in the solution, the activated Fe(III) onjarosite surfaces can be converted to Fe(II) by sulfide and then Fe(II)instantly reacts with Cr(VI), yielding Fe(III) again. Furthermore, theone-step reduction of Cr(VI) by sulfide seems to be slower than the

Fig. 1. The effect of sulfide and jarosite on the reduction of Cr(VI) under pH = 8.00 andtemperatureof 25 °C (c(Cr(VI)0 = 100 μM, c(Na2S)0 = 400 μM, jarosite loading = 1 g/L).

two-step process consisting of rapid Cr(VI) reduction by Fe(II) followedby Fe(III) reduction with sulfide [14]. During the rapid Cr(VI) reductionreaction, the surface of jarosite experienced a cycle process of convertingFe(III) to Fe(II). The possiblemain reactionsmentioned abovewere sum-marized as follows:

CrðVIÞ þ sulfide→CrðIIIÞ ð3Þ

Jarosite þ sulfide→FeðIIÞ ð4Þ

FeðIIÞ þ CrðVIÞ→FeðIIIÞ þ CrðIIIÞ: ð5Þ

The XPS results of jarosite after Cr(VI) reduction (present in Fig. 2and Table 2) disclosed the presence of two oxidation states of surfaceiron species, two doublets of Fe 2p with an intensity ratio of 2:1, and asplitting energy of about 13.3 eV in sample. The most interesting pointis that Fe(II) contributed to 54.51% of the total iron surface atoms, while45.49% of the total iron surface atoms were in the state of Fe(III). Thepresence of two oxidation states of surface iron species can supportthe speculation that Fe(III) species on jarosite surface play a role inthe reduction of Cr(VI). In essence, the Fe(III) species on jarosite surfaceserve as a “bridge” in the transportation of electrons between sulfideand Cr(VI), thus facilitating the reduction reaction of Cr(VI) [39].

3.2. Further evidence of the role of jarosite in promoting the Cr(VI) reductionby sulfide

In the previous discussion, it has been proposed that the facilitationof jarosite in the reduction of Cr(VI) by sulfide was ascribed to the acti-vated Fe(III) as a constituent of jarosite. To further confirm the assump-tion, 50 and 100 μM Fe(III) were introduced to the reaction system atpH 8.0 and 25 °C to examine the effect of externally added Fe(III) onthe reaction, respectively, and the results are illustrated in Fig. 3. It isfound in Fig. 3 that compared with control experiment (in the absenceof jarosite), the reduction of Cr(VI) was improved when Fe(III) was in-troduced and the rates and extents of the conversion of Cr(VI) to Cr(III)were close at different Fe(III) concentrations. However, the reactionrates were much slower than those in the presence of jarosite. This re-sults imply that externally added Fe(III) exists in the form of eitherFe(OH)3 or FePO4 at pH 8.0 buffered with 0.1 M phosphate, which re-sults in low activity of externally added Fe(III). In addition, these resultsalso demonstrate that the activity of Fe(III) on jarosite is higher thanthat on fresh precipitates of Fe(OH)3 or FePO4.

As it is known that F− can react with Fe(III) to form a stable complex(FeF63 −). Thus, it can be assumed that the introduction of F− to the reac-tion systemwould impede the role of jarosite in promoting the Cr(VI) re-duction by sulfide. The effect of NaF on the conversion of Cr(VI) to Cr(III)by sulfide in the presence of jarosite is illustrated in Fig. 4. It is clearly ob-served in Fig. 4 that the reduction of Cr(VI) markedly decreased when

720.0736.0 704.0712.0728.0

Binding Energy (eV)

Fe 2p3/2

Fe 2p1/2

Fig. 2. XPS spectrums of biogenetic jarosite after Cr(VI) reduction.

Table 2Binding energy of Fe 2p and Fe(II) surface concentration of jarosite after Cr(VI) reduction.

Binding energy (eV) Splitting energy(eV)

Fe(II) surfaceconcentration (%)

Fe(II) Fe(III)

2p1/2 2p3/2 2p1/2 2p3/2

708.98 722.29 710.78 724.09 13.31 54.51%

Fig. 4. The effect of externally added NaF on the reduction of Cr(VI) by sulfide.(c(Cr(VI))0 = 100 μM, c(Na2S)0 = 400 μM, pH = 8.00, T = 25 °C, jarosite loading =1 g/L.

3726 Z. Xu et al. / Materials Science and Engineering C 33 (2013) 3723–3729

NaF was added into the reaction system. Approximately 53% and 50% ofinitial Cr(VI)were converted into Cr(III)within 80 min in the presence of50 and 100 μMNaF, respectively. However, the reduction rates were stillhigher than those in the absence of jarosite. The results further provethat the role of jarosite in promoting the Cr(VI) reduction by sulfide isreasonable. Compared with control experiment (no jarosite), the reduc-tion rates of Cr(VI), even with an introduction of NaF, were obviouslyfaster, and close to the rates with externally added Fe(III) discussedabove. Furthermore, it is concluded that the introduction of NaF cannotcompletely eliminate the accelerating role of jarosite in the Cr(VI) reduc-tion by sulfide.

Table 3Applied Box–Behnken Design matrix and Cr(VI) reduction efficiency values (%).

Standardorder

pH of thesolution

Operatingtemperature

Loading ofjarosite

Reactiontime

Cr(VI) reductionefficiency

Exp.a Pred.b

1 −1 −1 0 0 100.00 102.442 1 −1 0 0 74.21 71.413 −1 1 0 0 100.00 101.944 1 1 0 0 95.83 92.535 0 0 −1 −1 83.19 83.446 0 0 1 −1 90.42 90.197 0 0 −1 1 95.17 94.548 0 0 1 1 97.56 96.45

3.3. Model development and validation

To investigate the effect of multiple variables, the most commonlyused method involves the variation of one variable while keeping theother variables constant, until all variables have been studied. Thismethodology has two disadvantages. First, a large number of experi-ments are required. Second, it is likely that the combined effect of twoor more variables may not be identified. In this work, a statistical ap-proach was employed based on BBD, which allowed us to infer aboutthe effect of variables with a relatively small number of experiments.The 4-factor BBDmatrix and experimental results obtained in reductionof Cr(VI) by sulfide runs were presented in Table 3. Based on these re-sults, an empirical relationship between the response and independentvariables was obtained and expressed by the following second-orderpolynomial equations:

Y ¼ −1236:99625þ 359:26833X1−6:03617X2 þ 140:76250X3

−2:23063X4 þ 1:08100X1X2 þ 18:37500X1X3 þ 0:76950X1X4

−0:94000X2X3−0:039325X2X4−0:60500X3X4−27:99833X12

þ0:00046667X22−117:64583X3

2−0:031712X42:

ð6Þ

Fig. 3. The effect of externally added iron(III) on the reduction of Cr(VI) by sulfide.(c(Cr(VI))0 = 100 μM, c(Na2S)0 = 400 μM, pH = 8.00, T = 25 °C, jarosite loading =1 g/L.

The corresponding regression equation with coded variables canbe presented as follows:

Y ¼ 99:03–10:11X1 þ 5:15X2 þ 2:16X3 þ 4:34X4 þ 5:41X1X2

þ1:84X1X3 þ 3:85X1X4–1:88X2X3–3:93X2X4–1:21X3X4

–7:00X12 þ 0:047X2

2–4:71X3

2–3:17X4

2:

ð7Þ

To ensure the adequacy of the employed model and avoid poor orambiguous results, an adequate fit of the model should be evaluated[40]. Table 4 shows the analysis of variance of regression parametersof the predicted response surface quadratic model for Cr(VI) reductionusing the results of all experiments performed. As it can be seen from

9 −1 0 0 −1 99.14 98.4810 1 0 0 −1 69.63 70.5611 −1 0 0 1 100.00 99.4612 1 0 0 1 85.88 86.9413 0 −1 −1 0 84.59 85.1714 0 1 −1 0 99.86 99.2415 0 −1 1 0 99.25 93.2616 0 1 1 0 100.00 99.8117 −1 0 −1 0 98.72 97.1118 1 0 −1 0 71.18 73.2119 −1 0 1 0 99.32 97.7620 1 0 1 0 79.13 81.2221 0 −1 0 −1 83.47 82.4822 0 1 0 −1 99.94 100.6523 0 −1 0 1 99.26 99.0224 0 1 0 1 100.00 101.4725 0 0 0 0 98.94 99.0326 0 0 0 0 98.76 99.0327 0 0 0 0 99.24 99.0328 0 0 0 0 99.03 99.0329 0 0 0 0 99.18 99.03

a Experimental values of response.b Predicted values of response by RSM proposed model.

Table 4ANOVA results for the response surface quadratic model for Cr(VI) reduction facilitatedby biogenetic jarosite.

Source d.f.a Analysis of variance

Sun of squares Mean square F-value Prob > F

Model 14 2536.25 181.16 48.71 b0.0001X1 1 1226.55 – 329.8 b0.0001X2 1 318.79 – 85.72 b0.0001X3 1 56.20 – 15.11 0.0016X4 1 226.03 – 60.78 b0.0001X1X2 1 116.86 – 31.42 b0.0001X1X3 1 13.51 – 3.63 0.0774X1X4 1 59.21 – 15.92 0.0013X2X3 1 14.14 – 3.80 0.0715X2X4 1 61.86 – 16.63 0.0011X3X4 1 5.86 – 1.57 0.2301X12 1 317.80 – 85.45 b0.0001

X22 1 0.014 – 0.003798 0.9517

X32 1 143.64 – 38.62 b0.0001

X42 1 65.27 – 17.55 0.0009

Residual 14 52.07 3.72 – –

Lack of fit 10 51.92 5.19 140.7 0.0001Pure error 4 0.15 0.037 – –

R2 = 97.99%, adjusted R2 = 95.98%.a Degree of freedom.

Fig. 5. Pareto chart showing the standardized effect of each model term on Cr(VI) re-duction efficiency.

Fig. 6. Response surface and contour plots of Cr(VI) reduction efficiency as a function ofpH and operating temperature (loading of jarosite = 1 g/L, reaction time = 30 min).

3727Z. Xu et al. / Materials Science and Engineering C 33 (2013) 3723–3729

the table, the model F-value of 48.71 and a very low probability value((Prob > F) b 0.0001) indicate that the regression model is significantfor Cr(VI) removal facilitated by biogenetic jarosite. Olmez [41] sug-gested that the correlation coefficient (R2) should be at least 0.80 fora good fit of a model. The obtained R2 value (97.99%) shows that theregression models explained the reaction well because 97.99% of thevariations for Cr(VI) reduction efficiency are explained by the chosenindependent variables and this also means that the model does not ex-plain only about 2.01% of the variation. Adjusted R2 (Adj-R2) is also atool for measuring the goodness of a fit, but it's more suitable for com-paring models with different numbers of independent variables [42]. Itcorrects the R2-value for the sample size and the number of terms in themodel by using the degrees of freedom on its computations. So, if thereare many terms in a model and not very large sample size, adjusted R2

may be visibly smaller than R2 [43]. Here, adjusted R2 value (95.98%)was very close to theR2 value. Hence, the response surfacemodel devel-oped in this study for predicting Cr(VI) reduction efficiencywas consid-ered to be satisfactory.

3.4. Effects of model terms and their interaction on Cr(VI) reduction

For the regression coefficients of coded variables, both the sign andmagnitude are important. The former indicates its effect direction,whereas, the latter determines the extent of the influence of the vari-able on the response factor. A positive sign of the coefficient representsa synergistic effect, while a negative sign indicates an antagonistic ef-fect. The operation temperature (X2), loading of jarosite (X3), reactiontime (X4), and the interactive terms viz. X1X2, X1X3, and X1X4 alongwith the quadratic term of X2

2 exhibited positive relationship with theCr(VI) reduction process, whereas, all the other terms showed negativeeffect on the process. The results obtained from Pareto analysis is illus-trated in Fig. 5. The linear terms exhibited the highest contribution(50.51%) followed by the quadratic terms (26.95%), whereas the inter-active terms had the lowest contribution (22.54%). This suggests thatthe first order independent variables have a direct effect on the Cr(VI)reduction efficiency, whereas the effect of interaction between factorson the response factor can't be ignored. The significance of these quadrat-ic and interaction effects between the variables would have been lost ifthe experiment was carried out by conventional methods. Table 4 clearlydemonstrates that all the linear and quadratic (except X2

2) terms are sta-tistically significant (p-value b 0.05), whereas the interactive terms ofX1X2, X1X4 and X2X4 are significant. However, the lack of the significance

of interactive terms (X1X3, X2X3 and X3X4) is also observed. For a singlemodel term, the pH of the solution had the most important effect(33.91%) on the Cr(VI) reduction efficiency followed by the quadraticterm of pH × pH (16.26%). Significant impact of pH predicted by Paretoanalysis is not surprising, because solution pH affects the chemical ther-modynamic driving force for the reduction of Cr(VI) [39].

Response surface plots provide a method to predict the Cr(VI) re-duction efficiency for different values of the tested variables, and thecontours of the plots help in the identification of the type of interactionsbetween these variables. In Fig. 6, the response surface and contourplots were developed as a function of initial solution pH and operationtemperature while the loading of jarosite and reaction time were kept

Fig. 7. Response surface and contour plots of Cr(VI) reduction efficiency as a function ofpH and reaction time (loading of jarosite = 1 g/L, operating temperature = 25 °C).

Fig. 8. Response surface and contour plots of Cr(VI) reduction efficiency as a function ofoperating temperature and reaction time (loading of jarosite = 1 g/L, pH = 8.0).

3728 Z. Xu et al. / Materials Science and Engineering C 33 (2013) 3723–3729

constant at 1 g/L and 30 min, respectively. As seen in Fig. 6, the reduc-tion efficiency of Cr(VI) increased when the operation temperaturewas increased and/or the pH of the solution was decreased. With tem-perature increasing and/or pH decreasing, the electrode potential ofCr(VI)/Cr(III) elevates, which increases the oxidation capacity of Cr(VI)and accelerates the chemical reduction of Cr(VI). Fig. 7 shows the inter-action effect of solution pH and reaction time on the reduction of Cr(VI)at constant jarosite loading (1 g/L) and operation temperature (25 °C). Itis clear that the reduction efficiency of Cr(VI) increased with increasingreaction time at the tested pH range of 7.5–8.5. As shown in Fig. 8, theresponse surface and contour plots illustrate the effect of operating tem-perature and reaction time on the removal efficiency of Cr(VI) at con-stant loading of jarosite (1 g/L) and pH (8.0). Both higher operationtemperatures and longer reaction time are conducive to the reductionreaction of Cr(VI).

4. Conclusions

In this study, the Cr(VI) reduction reaction by sulfide has been fa-cilitated by biogenetic jarosite. For instance, the efficiency of Cr(VI)reduced by sulfide increased from 20.5% to 100% when jarosite (1 g/L)was added to the homogenous reaction system at pH = 8 within40 min. The Fe(III) species on the surface of jarosite participate in thereduction reaction of Cr(VI), leading to a rapid removal of Cr(VI). Theeffects of four operational parameters such as the pH of solution, oper-ation temperature, loading of jarosite and reaction time on the reduc-tion of Cr(VI) were modeled using response surface methodology withBBD. The response surfacemodels developed for predicting Cr(VI) reduc-tion efficiency were considered to be quite satisfactory. The statisticalanalysis of the results suggested that solution pHwas themost significant

component of the model and the effect of interaction between factors onthe response factor can't be ignored. Operation temperature, loading ofjarosite and reaction time exhibited synergistic effects on the reductionof Cr(VI).

Acknowledgments

This studywasfinancially supported by theNational Natural ScienceFoundation of China (21077053, 40930738). The authors would like tothank Mr. Jinlong Geng and Mr. Chunyong Zhang for their excellenttechnical assistance.

References

[1] C.E. Barrera-Díaz, V. Lugo-Lugo, B. Bilyeu, J. Hazard. Mater. 223–224 (2012) 1.[2] R.M. Cespón-Romero, M.C. Yebra-Biurrun, M.P. Bermejo-Barrera, Anal. Chim. Acta

327 (1996) 37.[3] J. Nriagu, E. Nieboer, Chromium in the Natural and Human Environments, Wiley

Series, New York, 1988.[4] L.E. Eary, R. Dhanpat, Environ. Sci. Technol. 22 (1988) 972.[5] T.K. Yurik, A.K. Pikaev, High Energy Chem. 33 (1999) 208.[6] Y.Q. Lan, C. Kim, B.L. Deng, E.C. Thornton, H.F. Xu, Environ. Sci. Technol. 39 (2005)

2087.[7] X.F. Tian, X.C. Gao, F. Yang, Y.Q. Lan, J.D. Mao, L.X. Zhou, Geoderma 159 (2010)

270.[8] I.J. Buerge, S.J. Hug, Environ. Sci. Technol. 31 (1997) 1426.[9] M. Pittine, L. D'Ottone, L. Campanella, F.J. Millero, R. Passino, Geochim. Cosmochim.

Acta 62 (1998) 1509.[10] E.C. Thornton, J.E. Amonette, Environ. Sci. Technol. 33 (1999) 4096.[11] I.J. Buerge, S.J. Hug, Environ. Sci. Technol. 33 (1999) 4285.[12] C. Kim, Q.H. Zhou, B.L. Deng, E.C. Thornton, H.F. Xu, Environ. Sci. Technol. 35 (2001)

2219.[13] Y.Q. Lan, B.L. Deng, C. Kim, E.C. Thornton, Geochem. Trans. 8 (2007) 4.[14] Y.Q. Lan, J.X. Yang, B.L. Deng, Pedosphere 16 (2006) 572.[15] P. Asokan, M. Saxena, S.R. Asolekar, J. Hazard. Mater. B137 (2006) 1589.[16] H. Vu, J. Jandová, T. Hron, Hydrometallurgy 101 (2010) 1.

3729Z. Xu et al. / Materials Science and Engineering C 33 (2013) 3723–3729

[17] P. Asokan, M. Saxena, S.R. Asolekar, Mater. Charact. 61 (2010) 1342.[18] J.E. Dutrizac, T.T. Chen, Hydrometallurgy 102 (2010) 55.[19] M.P. Asta, J. Camaa, M. Martínezb, J. Giménez, J. Hazard. Mater. 171 (2009) 965.[20] Z.H. Xu, J.R. Liang, L.X. Zhou, J. Alloys Compd. 546 (2013) 112.[21] J.P. Wang, Y.Z. Chen, X.W. Ge, H.Q. Yu, Colloids Surf. A: Physicochem. Eng. Aspects

302 (2007) 204.[22] G. Sudarjanto, B. Keller-Lehmann, J. Keller, J. Hazard. Mater. B138 (2006) 160.[23] R.K. Goyal, N.S. Jayakumar, M.A. Hashim, J. Hazard. Mater. 195 (2011) 383.[24] R. Kumar, R. Singh, N. Kumar, K. Bishnoi, N.R. Bishnoi, Chem. Eng. J. 146 (2009) 401.[25] C.Y. Zhang, L.J. Yang, F. Rong, D.G. Fu, Z.Z. Gu, Electrochim. Acta 64 (2012) 100.[26] M.S. Bhatti, D. Kapoor, R.K. Kalia, A.S. Reddy, A.K. Thukral, Desalination 274 (2011) 74.[27] M.S. Shafeeyan,W.M.A.W. Daud, A. Houshmand, A. Arami-Niya, Fuel 94 (2012) 465.[28] S.Y. Bai, Z.H. Xu,M.Wang, Y.H. Liao, J.R. Liang, C.C. Zheng, L.X. Zhou,Mater. Sci. Eng. C

32 (2012) 2323.[29] Z.H. Xu, S.Y. Bai, J.R. Liang, L.X. Zhou, Y.Q. Lan, Mater. Sci. Eng. C 33 (2013) 2192.[30] M. Pettine, I. Barra, L. Campanella, F.J. Millero, Water Res. 32 (1998) 2807.[31] H. Meng, X. Hu, A. Neville, Wear 263 (2007) 355.

[32] A.K. Abdessalem, N. Oturan, N. Bellakhal, M. Dachraoui, M.A. Oturan, Appl. Catal.,B 78 (2008) 334.

[33] A.R. Khataee, M. Zarei, S.K. Asl, J. Electroanal. Chem. 648 (2010) 143.[34] APHA, AWWA, WPCF, Standard Methods for the Examination of Water and

Wastewater, 20th ed. American Public Health Association, Washington, D.C., 1998[35] L.E. Eary, D. Rai, Soil Sci. Soc. Am. J. 55 (1991) 676.[36] B.R. James, R.J. Bartlett, J. Environ. Qual. 12 (1983) 173.[37] J.W. Yang, Z.S. Tang, R.F. Gou, S.Q. Chen, Environ. Prog. 27 (2008) 302.[38] D.M. Zhou, H.M. Chen, C.R. Zheng, C. Tu, Environ. Pollut. 111 (2001) 75.[39] P. Zhou, Y. Li, Y.X. Shen, Y.Q. Lan, L.X. Zhou, J. Hazard. Mater. 237–238 (2012) 194.[40] R.H. Myers, D.C. Montgomery, Response Surface Methodology: Process and Product

Optimization Using Designed Experiments, second ed. John Wiley & Sons, USA,2002.

[41] T. Olmez, J. Hazard. Mater. 162 (2009) 1371.[42] M.B. Kasiri, A.R. Khataee, Desalination 270 (2011) 151.[43] S.C.R. Santos, R.A.R. Boaventura, Appl. Clay Sci. 42 (2008) 137.