Accepted Manuscript

Sulfate-reduction, Sulfide-oxidation and Elemental Sulfur Bioreduction Proc‐

ess: Modeling and Experimental Validation

Xijun Xu, Chuan Chen, Duu-Jong Lee, Aijie Wang, Wanqian Guo, Xu Zhou,
Hongliang Guo, Ye Yuan, Nanqi Ren, Jo-Shu Chang

PII: S0960-8524(13)01181-4
DOI: http://dx.doi.org/10.1016/j.biortech.2013.07.113
Reference: BITE 12158

To appear in: Bioresource Technology

Received Date: 22 June 2013

Revised Date: 21 July 2013
Accepted Date: 24 July 2013

Please cite this article as: Xu, X., Chen, C., Lee, D-J., Wang, A., Guo, W., Zhou, X., Guo, H., Yuan, Y., Ren, N.,
Chang, J-S., Sulfate-reduction, Sulfide-oxidation and Elemental Sulfur Bioreduction Process: Modeling and
Experimental Validation, Bioresource Technology (2013), doi: http://dx.doi.org/10.1016/j.biortech.2013.07.113

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 1  Sulfate-reduction, Sulfide-oxidation and Elemental Sulfur
2  Bioreduction Process: Modeling and Experimental Validation
3  Xijun Xu1, Chuan Chen1, Duu-Jong Lee1,2,3*, Aijie Wang1, Wanqian Guo1, Xu Zhou1,
4  Hongliang Guo1, Ye Yuan1, Nanqi Ren1*, Jo-Shu Chang4
1
5  State Key Laboratory of Urban Water Resource and Environment, Harbin Institute of
6  Technology, Harbin 150090, China
2
7  Department of Chemical Engineering, National Taiwan University, Taipei 106,
8  Taiwan
3
9  Department of Chemical Engineering, National Taiwan University of Science and
10  Technology, Taipei 106, Taiwan
4
11  Research Center for Energy Technology and Strategy, National Cheng Kung
12  University, Tainan, Taiwan
*
13  Corresponding authors: djlee@ntu.edu.tw (DJL); rnq@hit.edu.cn (RNQ)
14 
15 
16  ABSTRACT

17  This study describes the sulfate-reducing (SR) and sulfide-oxidizing (SO) process

18  using Monod-type model with best-fit model parameters both being reported and

19  estimated. The molar ratio of oxygen to sulfide (ROS) significantly affects the kinetics

20  of the SR+SO process. The S0 is produced by SO step but is later consumed by

21  sulfur-reducing bacteria to lead to “rebound” in sulfide concentration. The model

22  correlated well all experimental data in the present SR+SO tests and the validity of

23  this approach was confirmed by independent sulfur bioreduction tests in four

24  denitrifying sulfide removal (DSR) systems. Modeling results confirm that the ratio of

25  oxygen to sulfide is a key factor for controlling S0 formation and its bioreduction.

26  Overlooking S0 bioreduction step would overestimate the yield of S0.

27  Keywords: Kinetic model; sulfur bioreduction; microaeration; data fitting

28 

29  1. INTRODUCTION

30  Sulfate-bearing wastewaters are produced by pulp and paper manufacturers,

31  petrochemical plants, mineral processes and acid mine drainage from mining

32  activities (Knobel and Lewis, 2002). Under anaerobic environment with the presence

33  of chemical oxygen demand (COD), the sulfate-reducing bacteria (SRB) can convert
1 
 
34  sulfate in wastewaters to sulfide, which is toxic to living being and is corrosive to

35  metals. Biological conversion of sulfide to elemental sulfur (S0) using

36  sulfide-oxidizing bacteria (SOB) is a promising process (Xu et al., 2012;

37  Lohwacharin and Annachhatre, 2010) since the formed S0 can be recovered as a

38  renewable resource for fertilizer industries, sulfuric acid production and as substrates

39  for bioleaching processes (Celis-Garcia et al., 2008).

40  Integration of the sulfate-to-sulfide conversion step by SRB and

41  sulfide-to-S0conversion by SOB into one single reactor is of practical interest (van der

42  Zee et al., 2007). The level of dissolved oxygen (DO) has been proposed as an

43  effective process parameter to regulate the activities of SRB and SOB (Okabe et al.,

44  2005). To have SRB+SOB working in the same reactor faced difficulty of low S0

45  conversion. Xu et al. (2012) showed that the activities of SOB were enhanced by

46  limited oxygen to peak recovery of S0 from sulfate. The sulfide was oxidized by free

47  oxygen at increased DO so conversion of S0 declined. At high DO concentration the

48  activities of SRB were inhibited so the sulfate reducing (SR) + sulfide oxidizing (SO)

49  reactor failed. A few SRB strains can utilized the formed S0 as electron acceptor at the

50  expense of COD in the wastewaters (Chen et al., 2008a, 2008b). Certain

51  methanogenic bacteria could also form sulfide from oxidized states of sulfur via

52  dissimilatory sulfur reduction pathway (Zhou et al., 2011). The excess assimilatory

53  sulfur metabolism was claimed for reduced S0 yield in the SR+SO process (Thauer et

54  al., 1977).

55  Quantitative determination of the formed S0 was mainly by mass balance

56  calculation (Chen et al., 2009) or by sulfite method (Jiang et al., 2009). Kinetic

57  models can assist development and optimization of SR+SO reactor with maximum

58  sulfur recovery. This paper describes a mathematical model for SR+SO reactions on

2 
 
59  microaerophilic treatment of sulfate-bearing wastewaters. In particular, the process

60  parameter, ROS (molar ratio of oxygen to sulfide), is used for process optimization.

61  Experiments were conducted to validate the model outputs.

62 

63  2. MATERIALS AND METHODS

64  2.1 Experimental setup

65  Activated sludge was collected as inoculum from an anaerobic reactor for sulfur and

66  nitrogen-containing wastewater (Chen et al., 2008a). All batch tests were performed

67  in 250 mL anaerobic media bottles sealed with butyl rubber stoppers. The medium

68  consisted of 600 mg L-1 sulfate and 2000 mg COD L-1 with sodium lactate as carbon

69  source. Macro-nutrients were added in the following amounts (g L-1): NH4Cl, 0.575;

70  CaCl2, 0.070; MgSO4.7H2O, 0.100; K2HPO4, 0.22. 1 mL L-1 of trace solution was

71  added as described elsewhere (Chen et al., 2008b). The pH of suspensions was

72  adjusted to 8.0 using bicarbonate. Before experiments, nitrogen was sparged into the

73  bottles for 5 min to remove oxygen from both the aqueous phase and the headspace.

74  Experiments, investigating elemental sulfur production through coupling

75  microaerophilic sulfate-reduction, sulfide-oxidation and elemental sulfur bioreduction,

76  were carried out. Pure oxygen was added depend on the molar ratios of oxygen to

77  sulfide (ROS) indicated in Table 2 and the volume of oxygen to be added to the

78  headspace of each bottle was calculated according to Johnston and Voordouw (2012).

79  Assuming the concentration of gaseous oxygen at 23 oC and 1 atm being 41.2 mM,

80  the volume (V) of 100% (v/v) oxygen to add was calculated as

81  V = (mM O2 wanted in solution 50 mL headspace)/ 41.2 mM

82  The concentration of gas phase H2S was ignored in the calculation of ROS. For the

83  abiotic experiments, the inoculums were autoclaved and cooled before oxygen was

3 
 
 84  added.

85  In order to calibrate the model and its parameter values, independent experiments

86  were conducted with initial conditions as follows. 23.0 mL oxygen were added to the

87  headspace of anaerobic media bottle to generate ROS=1.0. For model evaluation, six

88  more experiments (Table 2) were carried out. All tests were run in triplicate and the

89  averaged results were reported.

90 

91  2.2 Analysis

92  An ion chromatography (Dionex ICS-3000) measured the concentration of sulfate

93  (SO42-), thiosulfate (S2O32-) in the collected liquor samples following 0.45-µm

94  filtration. Sample separation and elution were performed using an IonPac AG4A

95  AS4A-SC 4mm analytic column with carbonate/bicarbonate eluent (1.8 mmolL-3

96  Na2CO3/1.7 mmolL-1 NaHCO3 at 1 mL min-1) and a sulfuric regeneration (H2SO4, 25

97  mmolL-1 at 5 mL min-1). Sulfide concentration (including H2S, HS- and S2-) was

98  determined according to the methylene blue method (Truper and Schlegel, 1964).

99  Both volatile suspended solids and suspended solids were measured according to

100  Standard Methods. The dissolved oxygen in liquid samples was measured by DO

101  meter (pH/Oxi 340i, WTW, Germany) and the oxygen in the headspace was

102  determined by gas chromatography (GC-6890, Agilent, Foster City, CA, USA).

103  Elemental sulfur production was calculated according to:

104  [S0] = [Influent S]-[SO42-]-2*[S2O32-]-[HS-]

105 

106  2.3 Modeling approach

107  The total S0 production is combination of sulfate-reduction, sulfide-oxidation and

108  elemental sulfur bioreduction (eq 1)

4 
 
 dS S 0 dS S 2− dS S 0
109  net
= ox
+ re
                         (1)
dt dt dt

110  Four main processes associated with sulfur-species cycle were incorporated in this

111  model with all parameters listing in Table 1. The Monod-type kinetics for substrate

112  utilization is adopted. Si stands for the concentration of component i; XSRB, XSOB and

113  XS0 are biomass responsible for SR, SO and S0 bioreduction process respectively. We

114  consider excess COD presented in the suspension so its effect on the sulfate reducing

115  process is ignored. The pH and H2S inhibition effects were included in the ADM1

116  model (Batstone et al., 2002), but are not considered in this model since the

117  suspension pH was at around 8.0.

118  The Process 1 considers SO42- reduction to S2- (R1), mediated by SRB,

119  consuming SO42- and COD (electron source) and yielding biomass (eq. S1). Its kinetic

120  expression is stated as follows:

121  SO42− + 8 H + + 8e − → S 2− + 4 H 2O                       (S1)

dSSO 2− µ SSO2−
122  4
=- SRB SRB 4 X SRB                       (2)
dt YSRB K SO 2− +SSO2−
4 4

123  The Process 2 considers S2- oxidation to S0 (R2), mediated by SOB, consuming S2-

124  and O2 and yield biomass (eq. S2). The kinetic expressions for this process for S2- and

125  O2 are as follows:

126  2 HS − + O2 → 2S 0 + 2OH −                         (S2)

dS S 2 - µ S 2− SO2
127  ox
=- SOB SOBS X SOB                   (3)
dt YSOB K S 2− +SS 2− KO2 +SO2
SOB

dSOSOB 1 µ S 2− SO2
128  2
=- ⋅ SOB SOBS X SOB                 (4)
dt 2 YSOB K S 2− +SS 2− K OSOB
2
+SO2

129  The Process 3 considers S0 reduction to S2- (R3), mediated by sulfur bioreduction

130  bacteria, consuming S0 and COD and yield biomass (eq. S3). The kinetic expression

5 
 
131  for S3 is eq. 5.

132  S 0 + 2e − → S 2 −                             (S3)
dS S 0 µeSRE SS 0 KO2
133  re
=- X S0                           (5)
dt YeSRE K eSRE
S0
+ SS 0 K O2 + S S 0

134  The Process 4 considers aerobic oxidation of COD by heterotrophs, consuming

135  oxygen and COD in eq. S4. The corresponding kinetic expression is stated in eq 6.

136  O2 +4e − + 2 H + → 2OH −                         (S4) 

dS H
O2 µX SO2
137  =- H
XH                       (6)
dt YX H K XH
O2 + SO2

138  The Process 5 takes into account the biomass for SO42--reduction, S2--oxidation and S0

139  bioreduction, yield and decay. The kinetic expressions for the biomass as follows:
0
dX bS S0 0
140  =(µeSRE eSRES -kdeSRE )X bS                   (7)
dt K S 0 + SS 0

dX SRB SSO 2−
141  =（µ SRB SRB 4 -k dSRB ) X SRB                                                       (8)
dt K SO 2− +SSO 2−
4 4

dX SOB S 2− SO2
142  =（µ SOB SOBS -k dSOB ) X SOB       (9)
dt K S 2− +SS 2− K O2 +SO2
SOB

143  Since aerobic COD oxidation by facultative organisms can consume part of

144  oxygen and since S0 is the main end-product of sulfide oxidation under

145  limited-oxygen condition (Janssen et al., 1997), S2- oxidation to SO42- step with O2 as

146  electron acceptor is ignored. (This assumption is justified in experimental findings

147  reported latter). The competition of chemical oxidation on bioreduction of elemental

148  sulfur is described by the inhibition function of O2 (eq 4). Equation 5 is the kinetic

149  equation of oxygen consumption for sulfide oxidation. The kinetics of oxygen

150  consumption for aerobic COD oxidation by heterotrophs is described by eq. 6 and the

151  parameters for this equation are taken from the published literature (Koch et al., 2000).

6 
 
152  Equation 7 presents the kinetics of active biomass. According to Zhou et al. (2011),

153  we assume that methanogenic bacteria were responsible for sulfur bioreduction

154  process and so when modeling the sulfur bioreduction process we applied a new

155  parameter XS0 for sulfur bioreduction biomass. The input values for XSRB, XSOB and

156  XH were estimated based on the results of experiments using the baseline endogenous

157  OUR level prior to substrate addition (Ni et al., 2012). In the present work, the initial

158  concentrations of total active biomass, SRB, SOB and heterotrophic biomass in the

159  batch tests were measured as 32000, 10000, 6000 and 15000 mg L-1, respectively, and

160  thus the initial concentration of active sulfur bioreduction biomass XS0 was calculated

161  to be 1000 mg L-1.

162  The model parameters in eqs. 2–9 were estimated. For sulfate reduction process,

163  the parameters were extracted from Moosa et al. (2002). For sulfide oxidation process,

164  we estimated YSOB, µSOB, KS2-SOB, kdSOB and KO2SOB by fitting eqs. 3, 5, 6 and 9 with

165  the S2- and O2 data provided in this study. For sulfur bioreduction process, YeSRE, µeSRE,

166  KS0eSRE, KO2, kdeSRE were estimated by fitting eqs. 4 and 7 with the experimental data

167  for S2-. The sulfide rebound phenomenon (as shown latter in the experimental section)

168  was considered to be attributed to sulfur bioreduction. Since the kdSOB is low

169  (10-5–10-6 h-1) in value, this parameter was ignored in model fitting.

170  The weighted nonlinear least-squares analysis was applied to determine the

171  kinetic parameters by fitting the experimental data using criterion in eq. 10 (Ni et al.,

172  2012):

n
173  SSWE = ∑ ( Si0 measured − Si0 predicted )2                     (10)
i =1

174  where Si0measuredand Si0predicted are the i-th measured and predicted concentrations of

175  specific substrates listed in eqs. 2–7, respectively. Modeling and simulations were

7 
 
176  performed using the software package AQUASIM (Reichert, 1998).

177 

178  3. RESULTS AND DISCUSSION

179  3.1 Elemental sulfur production at different ROS

180  At ROS=0 (anaerobic environment), the dosed SO42- was converted to S2- in 2 hr (Fig.

181  1). The sulfate reduction data at ROS=0.25–2.5 were also shown in Fig. 1 for

182  comparison sake. The DO level investigated had minimal effects on sulfate reduction,

183  correlating with the findings of Xu et al. (2012). Experimental data for concentrations

184  of S2-, S0 and O2 at ROS=0.25–2.5 are shown in Fig. 2. No SO42- was detected after its

185  exhaustion at 2 hr and onward (Fig. 1), hence supporting the assumption that S2-

186  oxidation to SO42- step with O2 as electron acceptor is negligible in the modeling steps.

187  Both SO42- and S2O32- concentrations were low in the batch tests. These observations

188  supported the assumption that S0 was the main oxidation product in the SO process

189  and chemical sulfide oxidation could be ignored during the present work based on the

190  fact that thiosulfate was the main product of chemical sulfide oxidation (Nielsen et al.,

191  2005; Chen et al., 2012). Meanwhile, the variation of substrates (sulfate, oxygen and

192  COD) was less than 5% under abiotic conditions, which showed that the degradation

193  of substrates was mostly attributed to biotic function.

194  Three phases of S2- profiles were observed. The first phase was related to the S2-

195  production governed by simultaneous SO42- reduction and S2- oxidation. The second

196  phase represented the S2- oxidation only, which was followed by a slow S2- oxidation

197  lag by S2- utilization and O2 uptake by both SOB and heterotrophs. The third phase of

198  S2- profile was associated with the S0 bioreduction, which emerged right after the

199  depletion of O2.

200  The impact of different oxygen concentrations (ROS) on methanogenic activity

8 
 
201  was depicted (data not shown) and no obvious difference was observed. Although

202  methanogens were related to extremely oxygen-sensitive organisms, they could create

203  their own anaerobic environment to survive and oxygen, which was harmful to these

204  strict anaerobes, could be removed from their biotopes by non-enzymatic reduction of

205  O2 by H2S formed by the sulfate-reducing bacteria; this might have contributed to the

206  extensive distribution of the methanogens in nature (Stetter and Gaag, 1983).

207 

208  3.2 Model fitting

209  The model parameters in eqs. 2–9 were estimated. For SR process, the parameters

210  were extracted from (Moosa et al., 2002). For SO process, parameters YSOB, µSOB,

211  KS2-SOB, kdSOB and KO2SOB were fitted with eqs. 3, 5, 6 and 9 using the S2- and O2 data

212  provided in this study. For sulfur bioreduction process, YeSRE, µeSRE, KS0eSRE, KO2, kdeSRE

213  were estimated by fitting eqs. 4 and 7 with the experimental data for S2-. The sulfide

214  rebound phenomenon (as shown latter in the experimental section) was considered to

215  be attributed to sulfur bioreduction. Since the kdSOB is low in value (10-5–10-6 h-1),

216  this parameter was ignored in model fitting.

217  As the maximum specific growth rate (µSRB), decay coefficient and yield

218  coefficient (YSRB) in SR process were not dependent on the SO42- concentration, and

219  the half-saturation constant (KSO42-SRB) showed a linear increase with increase in initial

220  substrate concentration, their values were obtained as 0.061 h-1, 0.035 h-1, 0.584

221  g-bacteria/g-SO42- and 0.02 kg m-3, respectively.

222  The experimental data at ROS=1.0 for S0 utilization for bioreduction were used to

223  fit eq. 5 for estimating maximum specific growth rate (µeSRE), half saturation constant

224  ( ), decay coefficient ( ) and yield coefficient (YeSRE). The rate coefficients

225  and KO2 for S0 bioreduction were estimated by fitting eq. 5 to the S0 production.
9 
 
226  Table 1 lists the values and units of the kinetic and stoichiometric parameters

227  used in the present model.

228 

229  3.3 Model validation

230  The model predictions based on the so-fitted data were used to calculate the time

231  course curves for SO42-, S2-, S0 and O2 (solid curves in Figs. 1 and 2). The model

232  predictions correlate well with experimental results with no systematic deviations.

233  The proposed model was further validated using the experimental results by Zhou et

234  al. (2011) which also observed and studied sulfur bioreduction process in denitrifying

235  sulfide removal system and by Chen et al. (2010). The inoculums by Zhou et al. (2011)

236  were denitrifying sulfide granules and the media were: 200 mg L-1 S2-, 240 mg L-1

237  acetate, and 194, 387.5, 542.5, or 620 mg L-1 nitrate giving sulfide/nitrate molar ratios

238  of 5/2.5, 5/5, 5/7, 5/8. Figure 3 shows that the simulation results agree well with the

239  measured S0 concentrations, and sulfide and oxygen consumption profiles. The

240  capability of the present model with the best-fit parameters using SR+SO data to

241  describe kinetic behaviors of sulfur bioreduction in a denitrifying sulfide removal

242  system confirmed the validity of the proposed model.

243 

244  3.4 Parametric sensitivity

245  Uncertainty analysis was performed according to Ni et al. (2012). High

246  sensitivity of certain parameter suggests that this parameter is easy to be assigned by a

247  unique value. The surface plots of the objective functions (SSWE in eq. 8) for the

248  levels of correlation between parameters were evaluated (Fig. 4). The surface plots of

249  the objective function for maximum growth rate (µeSRE) versus half saturation constant

250  ( ), and maximum growth rate (µeSRE) versus decay coefficient ( ) show a

10 
 
251  well-defined valley, with the fitted values of these parameters residing in. These

252  best-fitted parameters are regarded well defined as listed in Table 1. Conversely, the

253  sensitivity of µeSRE to the model is higher than that of . There is a greater

254  change in SSWE on the µeSRE compared with the axis. The output sensitivities

255  of model parameters for S0 bioreduction were shown in Fig. 5. Effects of maximum

256  growth rate of bacteria, µeSRE (h-1), on the substrate profiles were shown in Fig. 5a. At

257  high µeSRE, its effect on S0 bioreduction is minimal. As µeSRE was decreased, the S0

258  bioreductionrate was significantly changed. The sensitivity analyses of affinity

259  constant Ks and yield coefficient YH were performed. The bioreduction rates of S0

260  were increased with reducing Ks and YH (Figs. 5b and 5c).

261 

262  SR+SO processes

263  An integrated model including SR+SO and S0 bioreduction processes was

264  established. The production of S0 depended greatly on ROS (Fig. 2). The S0

265  concentrations was increased to 79, 109, 194 mg L-1 at around 7 hr at ROS=1.0, 1.5

266  and 2.0, respectively. The formed S0 was reduced by MPB to S2- in the latter stage of

267  all tests, leading to low S0 yield for the SR+SO process (Belyakova et al., 2006;

268  Mogensen et al., 2005; Nakagawa et al., 2005; Thabet et al., 2004). The “rebound” in

269  S2- concentration was in agreement with those reported by Chen et al. (2008b).

270  From experimental data and model simulation, the peak points of S2- profiles

271  correspond to complete removal of SO42- and the bending point of S0 profiles

272  correspond to the exhaustion of O2 for S2- oxidation and occurrence of S0 bioreduction.

273  Thus, the time course of S2- indicates the transition from S0 production to

274  bioreduction.

275  In DSR studies the excess organic substances stimulated the activities of
11 
 
276  Methanobacterium sp. to lead to reduction of S0 (Zhou et al., 2011). In the present

277  study, the formed S0 was also consumed by bioreduction; however, the consumption

278  rate was reduced in the presence of trace oxygen. This observation is attributable to

279  the fact that SRB (Belyakova et al., 2006; Mogensen et al., 2005) and methanogenic

280  bacteria (Stetter and Gaag, 1983) have DO-sensitive activities. Although S2- could

281  also be produced abiologically by disproportionation of S0, since no SO42- or S2O32-

282  was accumulated, the S0 noted in the present study should be formed via biological

283  pathway and S2- production was exponential rather than linear shape.

284  To the authors’ best knowledge, no kinetic parameters are available in the

285  literature for bioreduction of S0. The parameters reported herein are hence valuable

286  for model development involving S0 kinetics. The maximum specific growth rate

287  (µeSRB) of 0.035 h-1 determined for S0 bioreduction was close to those obtained by

288  Zavarzinaet al. (2000) and Escobar et al. (2007), although these authors worked with

289  pure culture. The YeSRE estimated herein, 0.712g biomass/g S0, is close to that by

290  Escobar et al. (2007). The half-saturation constant and decay coefficient for S0

291  bioreduction were 0.024 kg m-3 and 5.75× 10-6 h-1, respectively.

292  The µSOB for SOB was low (0.028 h-1) when compared with those reported in

293  other studies (Alcantara et al., 2004; Gadekar et al., 2006). The value of

294  half-saturation constant for SOB, KSOB (0.011 kg m-3), estimated in this work is

295  generally in accord with those reported in (Alcantara et al., 2004; Gadekar et al.,

296  2006). The YSOB estimated in this study, 0.0029g biomass/mmol sulfide, is

297  significantly lower than those reported by Gadekar et al. (2006) and McComas and

298  Sublette (2001). In general, the present SR+SO system has lower growth rates and

299  yields than the literature results considering only SR or SO systems.

300  Since S0 bioreduction takes place in the SR+SO process, overlooking S0

12 
 
301  bioreduction step would overestimate the yield of S0. On the other hand, S0

302  bioreduction is a primitive means of energy conservation and for mesophilic bacteria

303  at moderate temperatures, S0 appears to be an unfavorable oxidant due to the

304  relatively few energy yield in its reduction (Stetter and Gaag, 1983; Thauer et al.,

305  1977; Belkin et al., 1985). However, the presence of S0 could enhance final cell yield

306  by a factor of up to 4 (Belkin et al., 1985). The present model indicates that the S2-/S0

307  cycles by SRB and SOB contribute to this increased cell yield, with the side benefits

308  of consuming excess COD in the pathways using S0 as a mediator. So the proposed

309  model in this work including sulfate-reduction, sulfide-oxidation and elemental

310  sulfur-bioreduction would help us better understand the procedure of elemental sulfur

311  formation and enhance its production by minimizing the likelihood of S0 bioreduction

312  reaction.

313 

314  4. CONCLUSIONS

315  The SR+SO process considering the formation and bioreduction of S0 was

316  modeled. The Monod-type kinetic equations were used to fit the model parameter

317  with experimental data. The validity of the best-fit parameters was confirmed using

318  the present SR+SO data and the DSR data from literature. The kinetic parameters for

319  bioreduction of S0 were for the first time reported. The effects of DO on the dynamic

320  behavior of the studied SR+SO process were presented. The present model can be

321  applied for process design and optimization that involve biological sulfur (SO42-/S2-)

322  cycle.

323 

324  ACKNOWLEDGEMENTS

325  This research was supported by the National Natural Science Foundation of China

13 
 
326  (Grant No.51176037), National High-tech R&D Program of China (863 Program,

327  Grant No.2011AA060904), Project 51121062 (National Creative Research Groups),

328  the State Key Laboratory of Urban Water Resource and Environment (2012DX06)

329  and NSC 102-3113-P-110-016.

330 

331  REFERENCES

332  1. Alcantara, S., Velasco, A., Munoz, A., Cid, J., Revah, S., Razo-Flores, E. 2004.

333  Hydrogen sulfide oxidation by a microbial consortium in a recirculation reactor

334  system: sulfur formation under oxygen limitation and removal of phenols.

335  Environ. Sci. Technol. 38, 918–923.

336  2. Batstone, D.J., Keller, J., Angelidaki, I., Kalyuzhnyi, S.V., Pavlostathis, S.G.,

337  Rozzi, A., Sanders, W.T.M., Siegrist, H., Vavilin, V.A. 2002. The IWA anaerobic

338  digestion model no 1 (ADM1). Water Sci. Technol. 45, 65–73.

339  3. Belkin, S., Wirsen, C.O., Jannasch, H.W. 1985. Biological and abiological sulfur

340  reduction at high temperatures. Appl. Environ. Microbiol. 49, 1057–1061.

341  4. Belyakova, E.V., Rozanova, E.P., Borzenkov, I.A., Tourova, TP., Pusheva, M.A.,

342  Lysenko, A.M., Kolganova, T.V. 2006. The new facultatively

343  chemolithoautotrophic, moderately halophilic, sulfate-reducing bacterium

344  Desulfover miculushalophilus gen. nov., sp. nov., isolated from an oil field.

345  Microbiol. 75, 161–171.

346  5. Celis-Garcia, L.B., Gonzalez-Blanco, G., Meraz, M. 2008. Removal of sulfur

347  inorganic compounds by a biofilm of sulfate reducing and sulfide oxidizing

348  bacteria in a down-flow fluidized bed reactor. J. Chem. Technol. Biotechnol. 83,

349  260–268.

350  6. Chen, C., Ren, N.Q., Wang, A.J., Yu, Z.G., Lee, D.J. 2008a. Microbial community

14 
 
351  of granules in expanded sludge bed reactor for simultaneous biological removal

352  of sulfate, nitrate and lactate. Appl. Microbiol. Biotechnol. 79, 1071–1077.

353  7. Chen, C., Ren, N.Q., Wang, A.J., Yu, Z.G., Lee, D.J. 2008b. Simultaneous

354  biological removal of sulfur, nitrogen and carbon using EGSB reactor. Appl.

355  Microbial. Biotechnol. 78, 1071–1077.

356  8. Chen, C., Wang, A.J., Ren, R.Q., Lee, D.J., Lai, J.Y. 2009. High-rate denitrifying

357  sulfide removal process in expanded granular sludge bed reactor. Bioresour.

358  Technol. 100, 2316–2319.

359  9. Chen, C., Ren, N.Q., Wang, A.J., Liu, L.H., Lee, D.J. 2010. Enhanced

360  performance of denitrifying sulfide removal process under micro-aerobic

361  condition. J. Hazard. Mater. 179, 1147-1151.

362  10. Chen, C., Zhou, C., Wang, A.J., Wu, D.H., Liu, L.H., Ren, N.Q., Lee, D.J. 2012.

363  Elementary sulfur in effluent from denitrifying sulfide removal process as

364  adsorbent for Zinc(II). Bioresour. Technol. 121, 441–444.

365  11. Escobar, C., Bravo, L., Hernandez, J., Herrera, L. 2007. Hydrogen sulfide

366  production from elemental sulfur by Desulfovibrio desulfuricans in an anaerobic

367  bioreactor. Biotechnol. Bioeng. 98, 569–577.

368  12. Gadekar, S., Nemati, M., Hill, G.A. 2006. Batch and continuous biooxidation of

369  sulfide by Thiomicrospira sp. CVO: reaction kinetics and stoichiometry. Water

370  Res. 40, 2436–2446.

371  13. Jiang, G.M., Sharma, K.R., Guisasola, A., Keller, J., Yuan, Z.G. 2009. Sulfur

372  transformation in rising main sewers receiving nitrate dosage. Water Res. 43,

373  4430–4440.

374  14. Johnston, S.L., Voordouw, G. 2012. Sulfate-reducing bacteria lower

375  sulfur-mediated pitting corrosion under conditions of oxygen ingress. Environ.

15 
 
376  Sci. Technol. 46, 9183–9190.

377  15. Klok, J.B.M., van den Bosch, P.L.F., Buisman, C.J.N., Stams, A.J.M., Keesman,

378  K.J., Janssen, A.J.H. 2012. Pathways of sulfide oxidation by haloalkaliphilic

379  bacteria in limited-oxygen gas lift reactors. Environ. Sci. Technol. 46,

380  7581–7586.

381  16. Knobel, A.N.; Lewis, A.E. 2002, A mathematical model of a high sulfate

382  wastewater anaerobic treatment system. Water Res. 36, 257–265.

383  17. Koch, G., Egli, K., van der Meer, J.R., Siegrist, H. 2000. Mathematical modeling

384  of autotrophic denitrification in a nitrifying biofilm of a rotating biological

385  contactor. Water Sci. Technol. 41, 191–198.

386  18. Lohwacharin, J., Annachhatre, A.P., 2010. Biological sulfide oxidation in an

387  airlift bioreactor. Bioresour. Technol. 101, 2114–2120.

388  19. McComas, C., Sublette, K.L. 2001. Characterization of a novel biocatalyst for

389  sulfide oxidation. Biotechnol. Progr. 17, 439–446.

390  20. Mogensen, G.L., Kjeldsen, K.U., Ingvorsen, K. 2005. Desulfovibrioaerotolerans

391  sp. nov., an oxygen tolerant sulphate reducing bacterium isolated from activated

392  sludge. Anaerobe 11, 339–349.

393  21. Moosa, S., Nemati, M., Harrison, S.T.L. 2002. A kinetic study on anaerobic

394  reduction of sulfate, part I: effect of sulfate concentration. Chem. Eng. Sci. 57,

395  2773–2780.

396  22. Nakagawa, S., Inagaki, F., Takai, K., Horikoshi, K., Sako, Y. 2005.

397  Thioreductormicantisoli gen. nov., sp. nov., a novel mesophilic, sulfur-reducing

398  chemolithoautroph within the ε-proteobacteria isolated from hydrothermal

399  sediments in the Mid-Okinawa tough. Int. J. Syst. Evol. Microbiol. 55, 599–605.

400  23. Ni, B.J., Fang, F., Xie, W.M., Xu, J., Yu, H.Q. 2012. Formation of distinct soluble

16 
 
401  microbial products by activated sludge: kinetic analysis and quantitative

402  determination. Environ. Sci. Technol. 46, 1667–1674.

403  24. Nielsen, A.H., Hvitved-Jacobsen, T., Vollertsen, J. 2005. Kinetics and

404  stoichiometry of sulfide oxidation by sewer biofilms. Water Res. 39, 4119–4125.

405  25. Nielsen, A.H., Vollertsen, J., Hvitved-Jacobsen, T. 2003. Determination of

406  kinetics and stoichiometry of chemical sulfide oxidation in wastewater of sewer

407  networks. Environ. Sci. Technol. 37, 3853–3858

408  26. Okabe, S., Ito, T., Sugita, K., Satoh, H. 2005. Succession of internal sulfur cycles

409  and sulfur-oxidizing bacterial communities in microaerophilic wastewater

410  biofilms. Appl. Environ. Microbiol. 71, 2520–2529.

411  27. Reichert, P. 1998. AQUASIM 2.0-User manual, computer program for the

412  identification and simulation of aquatic systems. Dubendorf, Switzerland: Swiss

413  Federal Institute for Environmental Science and Technology (EAWAG).

414  28. Stetter, K.O., Gaag, G. 1983. Reduction of molecular sulfur by methangenic

415  bacteria. Nature, 305, 309–311.

416  29. Thabet, O.B., Fardeau, M.L., Joulian, C., Thomas, P., Hamdi, M., Garcia, J.L.,

417  Ollivier, B. 2004. Clostridium tunisiense sp. nov., a new proteolytic,

418  sulfur-reducing bacterium isolated from an olive mill wastewater contaminated

419  by phosphogypse. Anaerobe, 10, 185–190.

420  30. Thauer, R.K., Jungermann, K. Decker, K. 1977. Energy conservation in

421  chemotrophic anaerobic bacteria. Bacteriol. Rev. 41, 100–180.

422  31. Truper, H.G., Schlegel, H.G. 1964. Sulfur metabolism in Thiorhodaceae:

423  Quantitative measurements on growing cells of Chromatiumokenii.

424  AntonieLeeuwnhoek 30, 225–238.

425  32. van der Zee, F.P., Villaverde, S., Garcia, P.A., Fdz.-Polanco, F. 2007. Sulfide

17 
 
426  removal by moderate oxygenaion of anaerobic sludge environments. Bioresour.

427  Technol. 98, 518–524.

428  33. Xu, X.J., Chen, C., Wang, A.J., Fang, N., Yuan, Y., Ren, N.Q., Lee, D.J. 2012.

429  Enhanced elementary sulfur recovery in integrated sulfate-reducing,

430  sulfur-producing reactor under micro-aerobic condition. Bioresour. Technol. 116,

431  517–521.

432  34. Zavarzina, D.G., Zhilina, T.N., Tourova, T.P., Kuznetsov, B.B., Kostrikina, N.A,

433  Bonch-Osmolovskaya, E.A. 2000, Thermanaerovibriovelox sp. nov., a new

434  anaerobic, thermophilic, organotrophic bacterium that reduces elemental sulfur,

435  and emended description of the genus Thermanaerovibrio. Int. J. Syst. Evol.

436  Mocrobiol. 50, 1287–1295.

437  35. Zhou, X., Liu, L.H., Chen, C., Ren, N.Q., Wang, A.J., Lee, D.J. 2011. Reduction

438  of produced elemental sulfur in denitrifying sulfide removal process. Appl.

439  Microbiol. Biotechnol. 90, 1129–1136.

440 
441 

18 
 
Table 1. Kinetic and stoichiometric parameters of the model
Parameter Definition Values Unit
Kinetic parameters
µ SRB   Maximum specific growth rate of SRB  0.061 h-1 
kg m-3 
SRB
K SO2−   Sulfate affinity constant for SRB  0.02
4

k dSRB   Endogenous decay rate of SRB  0.035 h-1 

µ SOB   Maximum specific growth rate of SOB  0.028 h-1 
K SSOB
2−   Sulfide affinity constant for SOB 0.011 kg m-3 
KOSOB
2
  Oxygen affinity constant for SOB  0.2 kg m-3 
µeSRE   Maximum specific growth rate of sulfur reduction bacteria  0.035 h-1 
K SeSRE
0   Elemental sulfur affinity constant for sulfur-reduction  0.024 kg m-3 
kdeSRE   Endogenous decay rate of sulfur-reduction bacteria 5.75 × 10-6 h-1 
K O2   Oxygen inhibiting coefficient constant for sulfur-reduction  0.0016 kg m-3 
Stoichiometric parameters
YSRB Yield coefficient for SRB  0.584 g VSS g-1 SO42- 
YSOB Yield coefficient for SOB  0.090 g VSS g-1 S2- 
YeSRE   Yield coefficient for sulfur bioreduction bacteria 0.712 g VSS g-1 S0

19 
 
Table 2. Initial conditions of the eight experiments for model evaluations
Condition I II III VI V VI VII
Sulfate concentration
600 600 600 600 600 600 600
(mg L-1)
Oxygen injected
0 5.75 11.5 23.0 34.5 46.0 57.5
volume (ml)
ROS 0 0.25 0.5 1.0 1.5 2.0 2.5

20 
 
Figure Captions

FIGURE 1. Model fitting results of the sulfate reduction equations to the substrate
utilization data at all ROS and sulfide production data at ROS=0. The model (solid line)
was fitted to data at ROS=1.0, which resulted in the parameter values. Model curves
obtained with the same parameter values were shown for other six data sets for
comparison.

FIGURE 2. Modeling fitting results of sulfide oxidation, elemental sulfur

bioreduction and oxygen consumption equations to the experimental data at different
ROS (except ROS=0). The model (solid line) was fitted to data at ROS=1.0, which
resulted in the parameter values. Model curves obtained with the same parameter
values were shown for other six data sets for comparison.

FIGURE 3. Comparison between the model simulations and the experimental data
from Zhou et al. (2011) (A) and Chen et al. (2010) (B~C) for the verification of the
approach. (A) Different initial S0 concentrations; (B) 740 mg L-1 initial S2-
concentration with S/N=5:6; (C) 540mg L-1 initial S2- concentration with S/N=5:6.
FIGURE 4. Surface plots of the objective function used for elemental sulfur
bioreduction parameter estimation (SSWE) as a function of different parameter
combinations: kd vs Ks; µm vs kd; µm vs Ks; Y vs kd; Y vs Ks; Y vs µm. The plots
were drawn using the optimal parameters (Table 1) as midpoint of intervals with 1
order of magnitude change (except Y, which was always lower than 1) on both sides
of intervals. The detailed information could be seen in Ni et al. [27].

FIGURE 5. Output sensitivity of parameters to elemental sulfur bioreduction: (A)

µeSRE ; (B) K SeSRE
0 ; (C) YeSRE .

21 
 
 0 2 4 6 8 10

600 600

Sulfate
450 450

(mg/l)

300 300

Sulfide at Ros=0

150 150

0 0
0 2 4 6 8 10
Time (h)
FIGURE 1. Model fitting results of the sulfate reduction equations to the substrate
utilization data at all ROS and sulfide production data at ROS=0. The model (solid line)
was fitted to data at ROS=1.0, which resulted in the parameter values. Model curves
obtained with the same parameter values were shown for other six data sets for
comparison.

22 
 
 0 10 20 30 40 50 0 10 20 30 40 50
400 400 300 400

Ros=2.5 Ros=2.0
250
300 300 O2 2- 300
0 S
200 S
O2 0 2-
S S
200 200 150 200

100
100 100 100

50

0 0 0 0
0 10 20 30 40 50 0 10 20 30 40 50
0 10 20 30 40 50 0 10 20 30 40 50
250 400 400

2-
Ros=1.5 200 Ros=1.0
200
S
2-
O2 300 S 300
150
150 0
S
(mg/l)

200 200
100 O2
100
0
S
100 100
50
50

0 0 0 0
0 10 20 30 40 50 0 10 20 30 40 50
 
0 10 20 30 40 50 0 10 20 30 40 50
400 400

200 Ros=0.5 200

2- 2-
S 300 S Ros=0.25 300
150 150

200 200
100 100

O2 100 100
50 50
0
S O2 0
S
0 0 0 0
0 10 20 30 40 50 0 10 20 30 40 50

Time (h) Time (h)

FIGURE 2.Modeling fitting results of sulfide oxidation and elemental sulfur

bioreduction equations to the experimental data at different ROS (except ROS=0). The
method for estimating parameter values was the same as described in Figure 1.

23 
 
  

0 10 20 30 40 50 60 70 80
0 10 20 30 40 50
160 160 750 1000

(A) (B)
0 -1 600 2- 800
Elemental Sulfur (mg/l) 120 S/N=5/2.5, S =145 mg L S
120

Concentration(mg/L)
0 -1
S/N=5:5, S =100 mg L O
2
0 -1 450 600
S/N=5:7, S =61.6 mg L
80 0 -1
S/N=5:8, S = 6.2 mg L 80

300 400

40 40
150 200

0
0 0 0
0 10 20 30 40 50 0 10 20 30 40 50 60 70 80

Time (h) Time(h)

0 10 20 30 40 50
600 1000

(C)
500 2-
S 800
Concentration(mg/L)

O
400 2
600

300

400
200

200
100

0 0
0 10 20 30 40 50
Time(h)
 

FIGURE 3. Comparison between the model simulations and the experimental data from Zhou et al. (2011) (A) and Chen et al. (2010)
(B~C) for the verification of the approach. (A) Different initial S0 concentrations; (B) 740 mg L-1 initial S2- concentration with S/N=5:6;
(C) 540mg L-1 initial S2- concentration with S/N=5:6..

24 
 
 4
x 10
12

10

0
250
200 6
150 5
4
100 3
50 2 -5
1 x 10
0 0
Ks kd

5
x 10
2.5

1.5

0.5

0
6
5
0.35
4 0.3
3 0.25
-5 0.2
x 10 2 0.15
1 0.1
0.05
0 0
kd u

5
x 10
2.5

1.5

0.5

0
250
200 0.35
150 0.3
0.25
100 0.2
0.15
50 0.1
0.05
0 0
Ks u

25 
 
 5
x 10
3.5

2.5

1.5

0.5

0
6
5
1
4 0.9
0.8
3 0.7
0.6
-5
2 0.5
x 10 0.4
1 0.3
0.2
0 0.1
0
kd Y

5
x 10
4

3.5

2.5

1.5

0.5

0
250
200 1
0.9
150 0.8
0.7
100 0.6
0.5
0.4
50 0.3
0.2
0 0.1
0
Ks Y

5
x 10
5

0
0.35
0.3
0.25 1
0.9
0.2 0.8
0.7
0.15 0.6
0.5
0.1 0.4
0.3
0.05 0.2
0 0.1
0
u Y

FIGURE 4. Surface plots of the objective function used for elemental sulfur
bioreduction parameter estimation (SSWE) as a function of different parameter
combinations: kd vs Ks; µm vs kd; µm vs Ks; Y vs kd; Y vs Ks; Y vs µm. The plots
were drawn using the optimal parameters (Table 1) as midpoint of intervals with one
order of magnitude change (except Y, which was always lower than 1) on both sides
of intervals. The detailed information could be seen in Ni et al. [27].

26 
 
 200

180

160
u
140

Elemental Sulfur (mg/l) 120 0.025

100

80

60

40

0.075
20

0
0 5 10 15
Time(h)

200

180

160
Ks
140
Elemental Sulfur (mg/l)

120

60.22
100

80

60

40

0.22
20

0
0 1 2 3 4 5 6 7 8 9 10
Time (h)

200

180

160
Y
140
Elemental Sulfur (mg/l)

120

100 0.911

80

60

40

0.411
20

0
0 1 2 3 4 5 6 7 8 9 10
Time (h)

Figure 5. Output sensitivity of parameters to elemental sulfur bioreduction: (A) µeSRE ;

eSRE
(B) K S 0 ; (C) YeSRE .

27 
 
>The sulfate-reducing (SR) and sulfide-oxidizing (SO) process was modeled.

>The Monod-type model was used to get best-fit kinetic parameters.

>The molar ratio of oxygen to sulfide significantly affects the SR+SO process.

>Overlooking S0 bioreduction step would overestimate the yield of S0.

28