Denitrification of groundwater with pyrite and Thiobacillus denitrificans

Clara Torrentó (1,*), Jordi Cama (1), Jordi Urmeneta (2), Neus Otero (3) and Albert Soler (3). (1) Hydrogeochemistry Group, Institute of Environmental Assessment and Water Research IDAEA, CSIC, C/Jordi Girona, 18-26, 08034 Barcelona, Spain. clara.torrento@idaea.csic.es, jordi.cama@idaea.csic.es (2) Department of Microbiology and Biodiversity Research Institute (IRBio), University of Barcelona, Av. Diagonal 645, 08028 Barcelona, Spain. jurmeneta@ub.edu (3) Mineralogia Aplicada i Medi Ambient Group, Department of Crystallography, Mineralogy and Ore Deposits, University of Barcelona, Martí i Franquès s/n, 08028 Barcelona, Spain. notero@ub.edu, albertsolergil@ub.edu (*) Corresponding author: Clara Torrentó e-mail: clara.torrento@idaea.csic.es Fax: +34 93 411 00 12.


INTRODUCTION
Groundwater contamination by nitrate usually originates from anthropogenic sources, mainly as a oxidation of pyrite based on geochemical and/or isotopic data (Aravena and Robertson, 1998 (3)

45
Our interest is to characterize this pyrite-driven denitrification reaction and assess its feasibility.

46
Although much work has been devoted to enhancing autotrophic denitrification by adding several 47 inorganic electron donors, such as zero-valent iron, ferrous ions, elemental sulfur, and iron bearing

88
Pyrite samples were crushed and sieved to obtain two particle sizes, one ranging from 25 to 50 µm 89 and the other from 50 to 100 µm. The samples used in two blank (TD-blank-21 and TD-blank-22, 90 Table 1) and in two pyrite-amended (TD-13 and TD-14,   Table 2).

155
In the experiments designed to characterize nitrogen and oxygen isotope fractionation associated 156 with the process (Table 3), the procedure was the same but using 250 mL glass Witeg bottles with 100 Flow-through experiments were performed to investigate pyrite-dependent denitrification under 162 similar conditions to the natural environment and to evaluate the long-term performance of the 163 process. Three types of flow-through experiments were performed: inoculated, blank and non-164 inoculated (Table 4). By means of a peristaltic pump, input solutions were circulated through 50 mL 165 polyethylene reactors in which 50-100 µm powdered Cerdanya pyrite (approximately 1 g in the blank 166 and non-inoculated experiments and 10 g in the inoculated experiment) was placed.

167
The T. denitrificans-inoculated experiment was carried out to evaluate the response and the 168 denitrification capability of the pure culture over long term (several months). After 15 d of inoculation 169 (6.6×10 7 cells mL -1 ), solution was circulated through the reactor with a flow rate of 0.003 mL min -1 , 170 yielding a hydraulic retention time (HRT) of 11.6 d. Reactors, tubing, pyrite powder and solutions 171 were sterilized before use in the inoculated experiment and also in the blank experiment.

172
The non-inoculated experiments, with non-sterilized pyrite powder, were performed to stimulate 173 activity of indigenous bacteria. The flow rate ranged between 0.009 and 0.014 mL min -1 , yielding HRT 174 of 2.3-3.9 d. These non-inoculated experiments were replicated to ensure the reproducibility of the 175 results (Table 4).

176
Input solution in the inoculated experiment was the modified T. denitrificans medium solution with 177 2.5 mM KNO3 (nitrate loading rate of 0.21 mmol NO3 -L -1 d -1 ). Input solutions in the blank and non-178 inoculated experiments consisted of NaNO3 solutions with nitrate concentration between 0.4 and 2.5 179 mM, yielding nitrate loading rates from 0.11 to 0.50 mmol NO3 -L -1 d -1 . In the two solutions, no other 180 electron donor was added to ensure that pyrite was the only electron donor available for cells. In 181 order to ensure an optimal pH, pH of influent solutions was between 6. 5   In the control batch experiments, nitrate concentrations remained unchanged up to 60 d (Table 1).

210
Consumption of nitrate over time was only observed in the pyrite-amended, T. denitrificans-inoculated 211 batch experiments (Fig. 1). The time needed to consume nitrate was dependent on pyrite grain size 212 and initial nitrate concentration. In most of the experiments with 25-50 µm pyrite, nitrate content was 213 mostly consumed within 14 d (Fig. 1A). In cultures amended with 50-100 µm pyrite, the time needed 214 to consume most nitrate was longer and decreased by lowering the initial nitrate concentration. With 215 an initial concentration of approx. 4 mM NO3 -, 35 to 80% of the nitrate content was consumed after 60 216 d; with approx. 2.5 mM NO3 -, nitrate was completely consumed within 60 d; and with approx. 1 mM 217 of NO3 -, complete consumption of nitrate occurred within 14 d (Fig. 1B).

218
An initial stage of 7 d during which nitrate concentration barely decreased was observed with the 219 lowest initial cell density (∼10 5 cells mL -1 ) (data not shown). This occurred because a longer adaptation 220 time was necessary for bacteria to grow into a population large enough to bring about a detectable 221 change in nitrate concentration. Nevertheless, the final percentages of reduced nitrate tended to 222 resemble those of experiments with higher initial cell density (Table 2).

223
As regards the flow-through experiments, nitrate reduction occurred in all the non-inoculated and 225 inoculated experiments, but not in the blank experiment. In the T. denitrificans-inoculated experiment, 226 partial nitrate removal occurred for 70 d ( Fig. 2A). Subsequently, complete nitrate removal was 227 achieved and lasted until the end of the experiment (200 d), indicating a high long-term efficiency of 228 T. denitrificans in nitrate removal using pyrite as the electron donor under the study conditions. Figure   229 2B shows the consumption of nitrate in one representative non-inoculated flow-through experiment.

230
In these experiments, a maximum nitrate reduction was achieved after 50-200 d (   probably as a result of changes in the experimental conditions that control the activity and growth of 240 bacteria (such as oxygen concentration or nutrient availability). At pH 4.5 (NON-1), nitrate reduction 241 was less effective than that observed in experiments carried out at pH 6.5-8, confirming the marked 242 decrease in microbial activity due to acid pH (Table 4). Nitrate reduction efficiency was dependent on 243 the nitrate loading rate. As is shown in the Table 4, when the nitrate loading rate ranged between 0.11 244 and 0.25 mmol NO3 -L -1 d -1 , nitrate reduction was effective (overall nitrate removal of 40-80%), lasting up to 150-350 d. By contrast, with high nitrate loading rates (0.33-0.50 mmol NO3 -L -1 d -1 ), nitrate 246 reduction efficiency was lower (overall nitrate removal lower than 30%), lasting only 20-70 d (e.g.

247
NON-6b, Fig. 3D). It should be noted that, although efficiency in nitrate removal was different, the 248 maximum amount of nitrate removed was similar in the two cases (between 0.12 and 0.48 mM for 249 lower nitrate loading rates and 0.31-0.38 mM for higher ones). Therefore, a maximum nitrate removal    was coupled with pyrite dissolution and was mediated by bacteria. Iron concentration was below 304 detection limit, suggesting that most of the Fe 2+ resulting from pyrite oxidation was oxidized to Fe 3+ 305 and precipitated. As stated in section 3.1, ammonium production could be excluded. Accordingly, the 306 overall reaction can be expressed as eq.(3).

307
If nitrate reduction was coupled to pyrite dissolution via eq. (3), the measured molar ratio of nitrate 308 consumed to sulfate produced should be close to the stoichiometric ratio of this reaction, which is 1.5.

309
However, in some experiments transient nitrite accumulation occurred, and therefore, the expected nitrate/sulfate ratio was calculated based on the amount of nitrite accumulated according to the 311 following reaction: where nitrate/sulfate ratio is 3.75.

314
In most of the experiments, the final products of the overall reaction were gaseous N-compounds 315 (i.e. NO, N2O or N2). If the product was NO or N2O, the nitrate/sulfate ratio should be 2.5 (eq. 5) and 316 1.9 (eq. 6), respectively:

319
In the inoculated pyrite-amended batch experiments, the nitrate/sulfate ratio was calculated using 320 sulfate released after time 0 given that nitrate reduction started after this time. The ratio ranged from 321 0.4 to 2.0, being lower than the possible stoichiometric ratios in most experiments (Table 2).

322
Nevertheless, the ratio was 1.5 within a 15% error in seven experiments. The low nitrate/sulfate ratio indicates excess of sulfate, which, as stated above, could be explained by additional oxidation of pyrite 324 by traces of dissolved oxygen as observed in the blank experiments. In fact, the excess of sulfate 325 produced in the inoculated experiments (assuming that the reaction occurs via eq. 3) ranged from 0.2 326 to 5.0 mM in agreement with sulfate produced in the blank experiments (between 0.2 and 4.9 mM). It 327 is important to note that in the experiments in which pyrite was previously washed with HCl, the 328 molar nitrate/sulfate ratio was similar to that of the rest of the experiments, as occurred with the efficiency and rate of nitrate removal (Table 2). This suggests that the presence of possible In the non-inoculated flow-through experiments, the measured nitrate/sulfate ratio at the time of 332 maximum nitrate removal was significantly higher than the possible stoichiometric ratios (values 333 higher than 10, Table 4). In fact, the percentage of nitrate reduction due to pyrite dissolution was 334 calculated to be 1-30%. Moreover, this percentage could be lower since an amount of sulfate was

346
On the other hand, some deficit in sulfate, considering the expected sulfate production, could be 347 partially attributed to passivation of the pyrite surface owing to precipitation of iron (hydr)oxide solid 348 phases. XPS examination showed an enrichment of Fe onto the pyrite surface since surface Fe/S ratios 349 increased from 0.50 to up to 0.77 (Table 5), which is consistent with the absence of iron in solution.

358
In the inoculated flow-through experiment, the measured nitrate/sulfate ratio was also high (IN-1, 359 Table 4). An iron coating may account for one part of the one part of the deficit in sulfate with respect 360 to the expected sulfate production. XPS confirmed iron enrichment on the surfaces (Table 5)   In pyrite-amended batch experiments, nitrate reduction rates were computed assuming zero-order 368 kinetics and using linear regression to fit the remaining nitrate concentrations vs. time (Fig. 1).

371
Nitrate reduction rates were higher in the experiments with 25-50 µm pyrite (2.12±0.83 mmol NO3 -372 kgpy -1 d -1 ) than in the 50-100 µm ones (0.39±0.31 mmol NO3 -kgpy -1 d -1 ). With initial nitrate concentration 373 of approx. 1 mM, the nitrate reduction rate was higher than the rates with approx. 2.5 and 4 mM NO3 -(0.62±0.34, 0.19±0.01 and 0.28±0.23 mmol NO3 -kgpy -1 d -1 , respectively). The variability in average rates 375 of the experiments with similar initial conditions (Table 2) could be attributed to different microbial 376 activity (especially in those experiments with low cell density) and/or certain degree of heterogeneity 377 in the range of grain size of the pyrite powders, which has been demonstrated to significantly modify 378 nitrate reduction rates and nitrate removal efficiency.

379
Rate dependence on pyrite grain size implies that the reduction rate depends on exposed pyrite 380 surface area. The larger the surface area, the higher the rate. A large surface area could enhance mass 381 transfer from solid surfaces to solution and/or bacterial attachment to the surface of pyrite grains.

382
Further experiments are necessary to ascertain whether the rate-limiting factor in the overall process is 383 mass transfer or bacterial adhesion.

384
In the flow-through experiments, the pyrite-mass normalized nitrate reduction rate, RNO3 (mol g -1 s -385 1 ) was calculated from the maximum consumption of nitrate according to the expression: where q is the flow rate (L s -1 ) of the solution through the reactor, CNO3 and C 0 NO3 are the 388 concentrations (mol L -1 ) of nitrate in the output and input solutions, respectively, and m is the pyrite 389 mass (g).

390
In the non-inoculated experiments, computed nitrate reduction rates ranged between 1.62 and 5.42 391 mmol NO3 -kgpy -1 d -1 (Table 4). Lower nitrate reduction rate was computed in the experiment 392 performed at pH 4.5 (1.31 mmol NO3 -kgpy -1 d -1 , Table 4). The nitrate loading rate faintly affected nitrate 393 reduction rates, although, as discussed above, nitrate reduction efficiency was higher in experiments 394 with low nitrate loading rates (0.11-0.25 mmol NO3 -L -1 d -1 ). The nitrate reduction rate obtained in the Hence, the results indicate that nitrate reduction rates increased by decreasing grain size and initial 398 nitrate concentration. The nitrate reduction rates were lower in the inoculated flow-through 399 experiment than in the non-inoculated ones, although efficiency in nitrate removal was higher in the 400 former.   (Table 3). The initial the experimental runs. In the 50-100 µm experiment, after 60 d, δ 15 NNO3 and δ 18 ONO3 increased to +8.4‰ 417 and +34.9‰, respectively, with 52% reduction of initial nitrate. In the experiment with 25-50 µm 418 pyrite, after 16 d, δ 15 NNO3 and δ 18 ONO3 increased to +2.6‰ and +29.2‰, respectively, with 18% 419 reduction of initial nitrate. Figure 6A

425
To our knowledge, isotope fractionation during autotrophic denitrification in laboratory cultures 426 has not been reported to date. Therefore, the ε ranges obtained in this study using T. denitrificans 427 culture were compared with those reported in experiments with heterotrophic denitrifying strains 428 under different growth conditions (Table 6). However, it should be noted that the NO3/SO4 ratio of the 429 TD-20 experiment (2.8) was significantly higher than the stoichiometric ones, suggesting the possible 430 occurrence of heterotrophic contamination (Table 3). In this case, εN and εO could be associated with

666
Nitrate reduction commenced at the start of the experiment and lasted 120 d. In the first 70 d, nitrate 667 was reduced to nitrite, which in turn, reduced to a N-gaseous compound. Thereafter, between 70 and 668 120 d, nitrate reduced to nitrite, and nitrite was not reduced. After 120 d, nitrate reduction ceased.