Potential of endogenous PHA as electron donor for denitrification

Potential of endogenous PHA as electron donor for denitrification S. Santorio, A. Fra-Vázquez, A. Val del Rio, A. Mosquera-Corral Accepted Manuscript How to cite: Science of The Total Environment, 695 (2019), 133747. https://doi.org/10.1016/j.scitotenv.2019.133747 Copyright information: © 2019 Elsevier B.V. This manuscript version is made available under the CC-BY-NC-ND 4.0 license (http://creativecommons.org/licenses/by-nc-nd/4.0/) PHA PHA PHA N2 NO385 mL 120 mL Head Space MANOMETRIC TRACKING Nitrogen Uptake Rate (NUR) test Time PHB PHV NO2NO3Time N2 Validation Validation TEST PROCEDURE SEDA: Specific Endogenous Denitrifying Activity Gas Chromatography •No preference between HB and HV •PHA < 5 % slight SEDA •Absence of N2O in the gas phase Denitrification by cell stored polyhidroxyalcanoates (PHA) 0.39 g N2-N/(g VSSact·d) Maximum SEDA value *Graphical Abstract HIGHLIGHTS A denitrifying activity method using stored PHA as carbon source was optimized. Specific endogenous denitrifying activity (SEDA) of 0.39 g N/(g VSS d) was measured. No exists a preference in the consumption of hydroxybutyrate over hydroxyvalerate. PHA concentrations lower than 5% do not allow the obtainment of maximal SEDA value. No N2O was detected in the gas phase during the SEDA test with stored PHA. 1 Potential of endogenous PHA as electron donor for denitrification 1 S. Santorioa*, A. Fra-Vázqueza, A. Val del Ríoa and A. Mosquera-Corrala. 2 a Department of Chemical Engineering, School of Engineering, Universidade de Santiago de 3 Compostela. E15705. Santiago de Compostela, Spain. 4 5 * Corresponding author at: Department of Chemical Engineering, School of Engineering,6 Universidade de Santiago de Compostela. E15705. Santiago de Compostela, Spain. Tel.: 7 +34 8818 16783. 8 9 E-mail address: [email protected] 10 11 ABSTRACT 12 The use of wastewater streams to obtain polyhydroxyalkanoates (PHA) as high added-value 13 products is widely studied. However, nitrogen removal is not well integrated into this 14 process. In this study, the optimal conditions to track the specific endogenous denitrifying 15 activity (SEDA) driven by PHA as carbon source were selected as: sludge concentration of 16 0.5 - 2 g VSS/L, CODPHA/N ratio higher than 5.4 g/g and between 40 - 60 mg NO3--N/L.17 The seeding biomass used to perform the activity tests was collected from two sequencing 18 batch reactors and was able to store up to 69 % wt/wt of PHA. SEDA values of 0.26 - 0.39 g 19 N2-N/(g VSS d) were achieved, which proved the potential of PHA-accumulating mixed 20 microbial cultures to be used in nitrogen removal processes. The results indicated that there 21 is not a preference in the consumption of hydroxybutyrate over hydroxyvalerate and that 22 PHA concentrations lower than 5 % wt/wt do not allow the obtainment of the maximum 23 SEDA value. Finally, N2O gas production was not detected in the SEDA experiments. 24 25 Keywords: Activity test; Denitrification; Polyhydroxyalkanoates; Specific endogenous 26 denitrifying activity. 27 28 2 1. INTRODUCTION29 The use of residues to obtain valuable products, promoted by the circular economy concept, 30 is the future. In this context, the production of biopolymers using wastewater as raw material 31 is one of the promising alternatives for the development of future wastewater recovery 32 facilities (WWRF). Polyhydroxyalkanoates (PHA) are an environmentally friendly 33 alternative to petrochemical plastics due to their polymeric nature but with the advantage of 34 being biodegradable. Their production from wastewater streams has been deeply studied and 35 it comprises at least two steps: enrichment and accumulation ones (Kumar and Kim, 2018). 36 The enrichment step consists of the selection of the microorganisms in a mixed microbial 37 culture able to intracellularly store PHA as carbon source, by applying certain stress 38 conditions such as the aerobic dynamic feeding (ADF) (Serafim et al., 2004). Then, during 39 the accumulation step the objective is to maximize the amount of stored PHA previous to 40 their subsequent extraction. 41 Unlike acetate or other external carbon sources, PHA are endogenous organic compounds 42 used as electron donor by bacteria in the absence of an exogenous source (Third et al., 43 2003). This condition is due to their related slow metabolic activity, which couples with the 44 denitrification process accurately (Beun et al., 2002, 2000). According to the applied ADF 45 strategy a feast-famine regime is established where PHA are stored inside the biomass 46 during the feast phase while they are consumed in the famine one. If the wastewater stream 47 used in the system contains nitrogen the PHA could be used during this famine phase to 48 denitrify. Therefore, when organic matter present in the wastewater is not enough to achieve 49 complete denitrification, the use of PHA as electron donor is an interesting option since 50 endogenous oxidation of PHA is 6 times slower than the traditional heterotrophic oxidation 51 of organic matter, and thus a more efficient use of it can be done. 52 3 Thus, biomass with PHA-storage capacity can be used to denitrify without the need of 53 external carbon source addition, which in many cases limits the efficiency of the nitrogen 54 removal. So, the availability of this internal carbon source contributes to accomplish the 55 stringent discharge limits for nitrogen. Therefore, to evaluate the PHA-accumulating 56 capacity and capability for denitrification of enriched mixed cultures is of interest. To 57 develop these processes to determine the denitrifying potential of an sludge using PHA is 58 important. 59 The denitrification activity is commonly determined by two types of tests: liquid phase and 60 manometric measurements. The denitrification test following the composition of the liquid 61 phase is the most applied alternative, where the evolution of nitrate concentrations is 62 followed by chemical measurements, and it is defined as nitrate uptake rate (NUR) test. In 63 addition, nitrite and organic compounds consumption can be monitored too (Kristensen et 64 al., 1992; Kujawa and Klapwijk, 1999; Sözen et al., 2002). On the other hand, the 65 manometric assays are based on the measurement of the dinitrogen gas production by 66 following the gas phase. In these assays, to maintain the temperature constant and enough 67 mixing to favour the mass transfer are crucial. The applicability of manometric 68 measurements to determine denitrifying activities has been well demonstrated (Sánchez et 69 al., 2000). While the NUR test require a significant amount of chemical analysis that are 70 time-consuming, biomass activities determined by manometric assays are faster and require 71 less chemical (Buys et al., 2000). 72 In this context, this is the first study focused on the evaluation of the PHA potential use as 73 endogenous carbon source for denitrification by using a manometric method to measure the 74 specific endogenous denitrifying activity (SEDA). A deep discussion is provided about the 75 influence of parameters like CODPHA/N ratio and biomass properties that limit the values of 76 4 SEDA: origin and concentration of biomass, and concentration and composition of PHA. 77 78 2. MATERIALS AND METHODS79 2.1 Source of seeding sludge with PHA-storage capacity 80 The biomass used for the batch denitrifying activity assays was collected from two different 81 lab-scale sequencing batch reactors (SBR) already enriched in PHA-accumulating 82 microorganisms. Moreover, biomass samples were also collected from accumulation fed-83 batch reactors where the PHA accumulation of the biomass from the previous SBR was 84 maximized. In this way, a range of accumulated percentages of PHA inside the cells, low in 85 the enrichment step and high in the accumulation step, were available for the experiments. 86 87 2.1.1 Enrichment step 88 The two enrichment SBR were operated under the ADF strategy to impose the feast/famine 89 regime. These SBR had a useful volume of 2 L, operated at a hydraulic retention time (HRT) 90 of 24 h, at constant temperature of 30 ºC. Allylthiourea, 4.95 mg/L, was added to avoid the 91 growth of ammonium oxidizing bacteria. The first reactor (SBR-A) was fed with a mixture 92 of volatile fatty acids (VFA) obtained after the acidogenic digestion of fish-canning 93 effluents, with high concentrations of ammonium, proteins and NaCl. The second reactor 94 (SBR-B) was operated under the same conditions as SBR-A, but fed with a synthetic media 95 containing a mixture of VFA and concentrations of nitrogen compounds and NaCl lower 96 than in SBR-A. The characteristics of the SBR feedings can be looked up in Supplementary 97 Material. The operational cycles of both SBR-A and SBR-B were completely aerated and 98 the feeding media did not contain nitrate. Therefore, the operational configuration did not 99 allow the denitrifying process to take place. 100 5 101 2.1.2 Accumulation step 102 After the enrichment step, accumulation fed-batch assays were carried out to maximize the 103 PHA storage inside the cells. These assays were conducted in aerobic conditions adding 104 pulses of a synthetic VFA mixture, mimicking the VFA composition of the feeding added to 105 the respective enrichment SBR. The VFA pulses contained 28 Cmmol and were added when 106 the dissolved oxygen (DO) concentration suffered a sharp increase, which indicated that the 107 VFA added in the previous pulse were consumed. In these assays, nitrogen compounds were 108 not added to avoid the biomass growth and promote the maximum PHA accumulation. 109 110 2.2 Key aspects for the denitrifying test with endogenous carbon source 111 2.2.1 Biomass collection procedure 112 Biomass from the enrichment SBR-A and SBR-B was collected immediately after the end of 113 the feast phase to avoid the consumption of the stored PHA in the subsequent famine phase. 114 Biomass from the accumulation fed-batch assays was collected after the last feeding pulse 115 was consumed to avoid the presence of VFA in the liquid media. 116 117 2.2.2 Preparation of the biomass 118 In order to avoid the consumption of the stored PHA, the biomass samples were 119 immediately centrifuged at 3000 rpm after collection (Loosdrecht et al., 2016). Then, they 120 were washed four times and re-suspended in a buffer solution (0.74 g/L K2HPO4 and 0.14 121 g/L KH2PO4), which provided an initial pH of 7.7 ± 0.1, and introduced into the vials for the 122 batch assays. The number of washes (4) was optimized in preliminary experiments by 123 measuring the total organic carbon (TOC) concentration in the liquid phase to verify its 124 complete absence and guarantee that exogenous organic matter was not present. 125 6 126 2.2.3 Estimation of the stored PHA 127 Since the electron donor for denitrification is endogenous, it would be ideal to determine the 128 content of PHA and the volatile suspended solid (VSS) concentrations of the collected 129 biomass samples. In this way, the exact mass of available carbon source would be quantified 130 and, consequently, the corresponding mass of nitrogen, necessary to have an adequate 131 CODPHA/N ratio for denitrification, could be added to the test. However, to determine PHA 132 and VSS concentrations before running the denitrifying test is no posible because it would 133 take too long and the carbon source would be depleted. Fortunately, although it is no 134 possible to know accurately the amount of carbon source available in the collected sample, it 135 is possible to estimate its value. In the case of biomass from the enrichment reactors, 136 considering operational steady state conditions, this estimation was based on the measured 137 amount of PHA accumulated in previous cycles because these values are normally stable in 138 consecutive cycles. In the case of biomass from the accumulation step, the amount of PHA 139 accumulated correlated with the pulses fed following the results obtained from previous fed-140 batch assays. Therefore, with the pulsed feeding method is possible to estimate the PHA 141 content of the biomass, taking into account past experiences and the number/concentration 142 of pulses added. 143 144 2.2.4 Selection of the minimum nitrate concentration added 145 The concentration of nitrate used in each experiment (between 25 - 100 mg NO3--N/L in the146 vials) was selected based on the estimated amount of PHA accumulated inside biomass, to 147 fix an adequate CODPHA/N ratio for the denitrification process. Therefore, the CODPHA/N 148 ratio used to perform the maximum SEDA tests must be higher than the stoichiometric ratio 149 required by microorganism to conduct the denitrification catabolic reaction from PHA as 150 13 [Figure 2] 286 Regarding the accuracy of the manometric method, the calculated relative errors from the 287 nitrogen balance (ε2) were in the range of 4.2 - 7.4 % (assays A6, B2, B3 and B4), showing 288 the accuracy of the manometric method to obtain SEDA values based only on gas 289 production. In all cases SEDAM values were a little bit higher than SEDAL values. This 290 behaviour can be attributed to a slight negative effect during sampling in the liquid phase 291 measurement, where a fraction of the liquid is removed from the vial to measure the nitrate, 292 nitrite and PHA concentrations, while the manometric measurements are less invasive. 293 In addition, the pH was monitored in the assay A6 where the concentration of 75 mg NO3--294 N/L was applied (Table 1). The pH values remained between 7.6 and 8.0 in the triplicate 295 experiments. Thus, these results showed that phosphate buffer media maintained the pH 296 value below the inhibitory values for denitrification (Glass and Silverstein, 1998), even in 297 the assay with the higher nitrate concentration (assay A5.2). 298 Summarizing, the monitoring of the liquid phase during the denitrification test driven by 299 PHA provides more information about the consumption of HB, HV and nitrate 300 concentrations, while the assays based on gas production are faster, easier and cheaper in 301 terms of chemical reagent usage with the same reliability in terms of SEDA values. For this 302 reason, the subsequent SEDA determinations in this research work were performed by the 303 manometric test. 304 305 3.2 Definition of the initial conditions to determine the SEDA 306 A total of 17 assays (made in triplicate) were performed (by the manometric test) to obtain 307 the optimal conditions required to achieve the SEDA (Table 1). In these assays the influence 308 of parameters, like the concentration of seeded biomass, PHA concentration and HB:HV 309 ratio, nitrate, and the CODPHA/N over the denitrifying activity were evaluated. 310 14 [Table 1] 311 3.2.1 Source of the seeding sludge 312 The SEDA values obtained using both types of biomass (from SBR-A and SBR-B), showed 313 that the specific denitrification activity measured was different for each type of biomass. 314 This value ranged from 0.1 to 0.2 g N2-N/(g VSSact d) and from 0.3 to 0.4 g N2-N/(g VSSact 315 d) for the biomass from SBR-A and SBR-B, respectively (Table 1 and Figure 3.a). 316 The results of the tests “A” performed with the biomass from the reactor SBR-A, fed with 317 acidified fish caning wastewater, revealed the direct dependence of the measured specific 318 denitrification activity with the increase in the percentage of PHA accumulated inside the 319 cells (Figure 3.b). However, the comparison of assays A4 and A6 do not corroborate this 320 relationship and different values of 0.127 ± 0.004 and 0.218 ± 0.009 g N2-N/(g VSSact d), 321 respectively, were measured despite the similar PHA content inside the biomass (35.0 and 322 39.1 % PHA, respectively) (Table 1). The lower activity in assay A4 could be attributed to 323 an episode of biomass inhibition during the operation of the SBR-A, caused by an 324 unexpected increase of the NaCl concentration in the feeding up to 5 g NaCl/L. 325 The biomass from SBR-B, enriched with a synthetic media containing a mixture of VFA, 326 was able to accumulate up to 69 % of PHA (wt/wt). In all the assays, the biomass from 327 SBR-B exhibited specific activities higher than those measured for the biomass from SBR-328 A. More studies should be performed to better determine the influence of the origin of the 329 biomass over the SEDA results. For example, using biomass capable of accumulating HB 330 and HV at different ratios the obtained SEDA could change. In addition, biomass from SBR-331 A was adapted to high salt concentrations, but during the activity assays the buffer liquid 332 media did not contain NaCl, which could affect the SEDA value. Previous studies report 333 already on the effect of the salt concentration of the culture media on the denitrifying 334 activity of the biomass (Dinçer and Kargi, 1999; Jafari et al., 2015). 335 15 336 3.2.2 Effect of PHA concentration and composition used as electron donor 337 As a general response, in all experiments the SEDA measured increased with the amount of 338 carbon source present as accumulated PHA (Figure 3.a) demonstrating the direct 339 relationship between both parameters. Experiment B5 illustrates this correlation (Table 1). 340 The three assays, B5.1, B5.2 and B5.3, were performed in parallel with three different PHA 341 concentrations (1670, 2749 and 4657 mg CODPHA/L) but with the seeding biomass from the 342 same origin, and equal percentage of accumulated PHA and nitrate concentration. The 343 values of SEDA increased from 0.337 g N2-N/(g VSSact d) at 1670 mg CODPHA/L (B5.1) to 344 0.377 g N2-N/(g VSSact d) at 4657 mg CODPHA/L (B.5.3). 345 A similar relationship is exhibited between the increase of SEDA and the increase of the 346 percentage of PHA accumulated inside the biomass (Figure 3.b). PHA storage percentages 347 lower than 5% (wt/wt), in experiments A1.1, A1.2 and A2 led to poor denitrifying activities. 348 This behaviour could be attributed to the preference of the microorganism to keep always a 349 certain amount of PHA as storage energy instead of using them as electron donor in the 350 denitrification process, even if in these experiments a CODPHA/N ratio below the 351 stoichometric value was used (Table 1). These low values could be the cause also for these 352 low activities. Since the determination of the percentage of PHA inside the biomass is a time 353 consuming procedure, it is not possible to know in advance its precise value before carrying 354 out the denitrifying assay. For this reason, experiments with low PHA percentage in the 355 biomass, estimated from previous experiences, but with an adequate CODPHA/N ratio need 356 to be carried out to verify this hypothesis. Bengtsson et al. (2017) reported a complete post-357 anoxic denitrification with 1.8 % of PHA accumulated. Nevertheless, the denitrifying 358 activity was poor (0.03 g N2-N/(g VSSact d)) and comparable with the value obtained in the 359 present study of 0.025 ± 0.003 g N2-N/(g VSSact d) with 1.3 % of PHA (assay A2). 360 16 [Figure 3. a] [Figure 3. b] 361 No correlation was found between the HB:HV ratio at the beginning of the assay and the 362 obtained values of the SEDA. Assays A3 and A5.1 achieved similar SEDA values with 363 different HB:HV ratios, 74:26 and 92:8, respectively. However, since all the assays carried 364 out with the biomass from SBR-B had similar HB:HV ratio and the SEDA obtained were 365 not very different, further studies are needed to confirm this result maintaining all the 366 parameters constant and different HB:HV ratios. This ratio can be changed by performing 367 the accumulation batch assays using feeding media containing different mixtures of VFA 368 (Albuquerque et al., 2011). Furthermore, as it was explained before, the HB:HV ratio was 369 similar at the beginning and end of the experiment (Table 1), which could indicate the 370 absence of preference of the microorganisms for HB or HV probably due to their similar 371 chemical structure or the presence of both as a copolymer. 372 373 3.2.3 Optimal nitrate concentration 374 A wide range of nitrate concentrations was tested (25 - 100 mg NO3--N/L), depending on the 375 content of PHA inside the biomass. Experiments with the similar nitrate concentration as B4 376 and B5.3 (50 - 58 mg NO3--N/L) or B6 and B7 (50 mg NO3--N/L) showed different SEDA 377 values (Table 1). Moreover, experiments carried out simultaneously with biomass from the 378 same origin but different nitrate concentrations resulted in similar SEDA. For example, 379 assay A5.1 (50 mg N/L) showed a similar activity as A5.2 (100 mg N/L), of 0.121 ± 0.029 380 and 0.108 ± 0.007 g N2-N (g VSSact d), respectively. Therefore, the initial nitrate 381 concentration in the range of values tested, without considering other parameters, did not 382 affect the SEDA, as long as an adequate CODPHA/N for the denitrification process was 383 provided. 384 However, from the practical point of view, an adequate nitrate concentration must be 385 17 considered when determining the SEDA values. For example, experiments with 100 mg N/L 386 lasted too long, requiring more than 15 hours to reduce all the nitrate to nitrogen gas, while 387 an appropriated duration ranges at approximately 5 - 6 hours (Figure 1). On the other hand, 388 in the experiments performed with nitrate concentrations lower than 25 mg N/L, such as 389 assay A1.2, the pressure increments measured in the headspaces of the vials were low due to 390 the slight production of N2 gas and presented the highest relative error (ε2 = 42%). Thus, the 391 nitrate concentration range to perform the SEDA determination is recommended to be 392 between 40 - 60 mg NO3--N/L. 393 394 3.2.4 Optimal CODPHA/N ratio 395 The COD/N ratio is a key factor for the denitrification process (Henze et al., 2008). A ratio 396 of 2.85 g CODPHA/g N is necessary to avoid organic matter limitation, according to the 397 stoichiometric expressions without considering the cellular growth (Eq. 1 - 4), to obtain the 398 SEDA. When the COD/N ratio is low the amount of electron donor is deficient and 399 consequently the microbial activity decreases (Figure 4). 400 [Figure 4] 401 Due to an error in the estimation of the PHA content inside the biomass previous to the 402 performance of the assays A1.1, A1.2 and A2, the applied CODPHA/N ratios were lower than 403 1.5 g/g, below the stoichiometric relationship, and thus the measured SEDA was barely 404 detectable. For CODPHA/N ratios higher than 5.4 g/g significant denitrification activities 405 were measured. The assays where the biomass from SBR-A was used, A3 and A4 presented 406 similar SEDA, although the applied CODPHA/N ratios were of 5.5 and 21.5 g COD/g N, 407 respectively. This behaviour is repeated in the assays B5.1 and B5.3 where the ratio was 408 28.8 and 80.3 g CODPHA/g N, respectively. Obtained results seem to indicate that the 409 18 COD/N ratio does not affect the SEDA, if it is higher than 5.4 g CODPHA/g N (Figure 4). 410 This value is close to that determined by Beun et al. (2002), who reported a maximum 411 theoretical stoichiometric ratio for the denitrification using PHA as carbon source, and 412 considering no growth, of 6.88 g CODPHA/g N. Further studies are needed in the range of 413 2.85 to 5.4 g CODPHA/g N to determine the minimum value of this ratio to achieve the 414 SEDA. For this purpose, the evolution of the PHA concentration during the operational 415 cycles in the accumulation experiments needs to be determined to identify the best moment 416 for the biomass collection to perform the denitrifying assay, as it contains the desired PHA-417 accumulated percentage. 418 419 3.2.5 Optimal biomass concentration 420 No relationship was found between the SEDA values and the Xact concentration in the 421 experiments (Table 1). Similar activities were achieved in the range of 0.45 - 1.26 g 422 VSSact/L in the case of biomass B and 0.80 g VSS/L - 2.76 g VSS/L in the case of biomass 423 A (Table 1). The corresponding VSS range (considering also the PHA mass) was 0.8 - 2.9 g 424 VSS/L. 425 To elucidate the influence of the VSS concentration on the denitrification activity two assays 426 (B6 and B7) were carried out, in parallel, inoculated with biomass from SBR-B and in the 427 same operational conditions: 50 mg NO3--N/L, 25 % PHA and 70:30 HB:HV ratio. 428 However, the seeding biomass concentration was half in assay B7 (0.7 g VSS/L) in 429 comparison with assay B6 (1.4 g VSS/L). Consequently, the CODPHA concentration and 430 CODPHA/N ratio were also half in B7 (Table 1), since the concentration of biomass 431 determines the amount of PHA in the test when the carbon source is endogenous. In addition 432 the slope of the curve describing the gas production of the assay B6 was much higher than 433 that of the assay B7 (Figure 5). The SEDA determined for B6 and B7 was of 0.390 ± 0.021 434 19 and 0.260 ± 0.008 g N2-N/(g VSSact d), respectively. 435 [Figure 5] 436 In these assays B6 and B7 the biomass contained only 25 % PHA, lower than in all the 437 previous assays carried out with biomass from SBR-B (between 40 - 69 % PHA). However, 438 the assay B6 presented a significantly higher SEDA. Therefore, when the storage of PHA is 439 deficient, a high SEDA value can be achieved by increasing the VSS concentration, which 440 means increasing the COD concentration added to the assay. 441 When the COD/N ratio was fixed at 11.7 g CODPHA /g NO3--N in assay B6 the nitrogen442 production profile (Figure 5) coincided with that from other studies where the denitrifying 443 activity was determined using external carbon sources (Buys et al., 2000; Sánchez et al., 444 2000). Nevertheless, the nitrogen production profile in the assay B7, where this ratio was of 445 6.3 g CODPHA /g NO3--N, presented two slopes well-differentiated, corresponding to446 productions of 11.3 mg N2-N/d at the beginning and to 6.0 mg N2-N/d after minute 250 of 447 the experiment (Figure 5). This profile resembles those obtained by previous mentioned 448 authors when the used COD/N ratio was not high enough. 449 However, this ratio was, in both assays B6 and B7 higher than 2.85 g CODPHA/g N, enough 450 to obtain a similar SEDA, according to the stoichiometry of the process and the previous 451 obtained results. Thus, the fact that both activities were different should be attributed to 452 another factor, which might be in this case the lower biomass concentration at the beginning 453 of assay B7 in comparison with assay B6. Both experiments started with 25 % of PHA 454 stored inside the cells, nevertheless, B6 contained 18 % of PHA stored after complete nitrate 455 denitrification while B7 contained 6.8 % of PHA, due to the different biomass concentration. 456 This fact indicates that PHA consumption was higher, by cell, in B7. Therefore, these results 457 seem to support the hypothesis that the lower PHA concentration inside the cells the lower 458 20 its use for denitrification, as the biomass prefers to store it as carbon source. 459 460 3.3 Absence of nitrous oxide in the gas phase 461 Some authors reported a relationship between the endogenous denitrification driven by PHA 462 with the production of N2O gas (Wang et al., 2015; Wei et al., 2014; Zhou et al., 2012). 463 They attributed this behaviour to two causes. The lack of electron donors during 464 denitrification, where the nitrate reductase enzyme (Nar) has a competitive advantage over 465 nitrite and nitrous oxide reductase (Nir and Nos) ones, causes NO2accumulation and N2O 466 gas generation. Or the fact that endogenous PHA degradation kinetics are at least 6 times 467 lower than soluble COD degradation kinetics (Third et al., 2003). 468 The composition of the gas phase was analysed at the end of all assays to determine the 469 biogas composition and calculate the mass balances. N2 and CO2 were detected. However, in 470 this research work, no N2O was detected in any assay, not even in the cases where the gas 471 phase was measured during all test performance. This results seem to confirm that if the 472 available PHA provides enough electron donor to maintain a high CODPHA/N ratio the 473 denitrification process will no produce N2O gas. This behaviour coincides with Zhou et al. 474 (2012) who reported that at high COD ratios the nature of the carbon source does not seem 475 to affect nitrous oxide accumulation. 476 477 3.4 Potential of PHA driven denitrification 478 This research study reports on remarkable denitrifying activities using PHA as a sole 479 electron donor. SEDA values close to 0.4 g N2-N/(g VSSact d) have been measured in the 480 assays B6 and B5.3 (Table 1). These results are comparable with the values obtained by 481 other research works where the specific denitrifying activity was determined using 482 21 endogenous and/or exogenous organic carbon source (Table 2). 483 [Table 2] 484 Denitrification using endogenous PHA is not as widely studied as heterotrophic 485 denitrification for wastewater treatment. However, there are many research studies on 486 denitrification driven by PHA. Experiences with granular sludge able to store PHA provided 487 denitrifying activities between 0.07 and 0.11 g N/(g VSS d) without external carbon source 488 supply (Qin et al., 2005; Val del Río et al., 2013). Bengtsson et al. (2017) reported an 489 activity of 0.03 g N/(g VSS d),in a post-denitrification unit with PHA as carbon source. This 490 low value could be attributed to an insufficient concentration of PHA inside the cells (1.8 ± 491 0.9 % PHA. With the presence of 3 % of PHA, Basset et al. (2016) achieved 0.2 g N/(g VSS 492 d) in a denitritation system. 493 The value reported by Khan et al. (2002) was the most similar one to that obtained in the 494 present study. However, in their case the PHA was fed as external carbon source (synthetic 495 powder), which suggests that the specific denitrifying activity driven by PHA is more 496 related to their chemical structure than to the endogenous nature of the carbon source. 497 In the case of heterotrophic denitrification higher activities of 1.91 and 0.92 g N2-N/(g VSS 498 d) were obtained by Buys et al. (2000) and Courtens et al. (2014), respectively. This high499 values could be attributed to an adaptation of the seeding sludge to denitrify high nitrate 500 concentrations (1000 - 400 mg N/L). Nevertheless, Ficara and Canziani (2007) and Val Del 501 Río et al. (2015) using acetate as carbon source reported values close to those from the 502 present research study of 0.46 and 0.22 g N/(g VSS d). 503 For all that, the results of this study highlight the potential of the denitrification driven by 504 PHA as carbon source, showing that it might be the basis of a competitive technology for 505 nitrogen removal. It is also important to consider that the seeding biomass was collected 506 from a SBR continuously aerated, nitrate was not added in the feeding, but allylthiourea was 507 22 to inhibit ammonium oxidizing bacteria activity. Thus, no denitrifying microorganisms were 508 specifically enriched in the biomass. Therefore, future research is needed to evaluate the 509 SEDA from denitrifying PHA-accumulating biomass. 510 511 4. CONCLUSIONS512 The batch assay procedure has been optimized to determine the specific denitrifying activity 513 of sludge using stored PHA as carbon source. The optimal conditions to determine the 514 SEDA were defined as follows: 0.5 – 2.0 g VSS/L, CODPHA/N ratio higher than 5.4 g/g and 515 40 - 60 mg NO3--N/L.516 The results indicated that a correlation between the HB:HV ratio in the sludge and the 517 obtained values of the SEDA does not exist, and there was no preference of the 518 microorganisms for HB or HV. 519 PHA concentrations under 5% inside the biomass do not allow for the obtainment of the 520 maximal SEDA values, which support the hypothesis that the lower the PHA concentration 521 inside the cells the lower its use for denitrification, as the biomass prefers to store it as 522 carbon source. 523 The denitrification with stored PHA using large CODPHA/N ratios does not produce N2O 524 gas. 525 A SEDA of 0.39 g N2-N/(g VSSact d) was achieved enhancing the potential of PHA as 526 carbon source for denitrification. 527 528 "E-supplementary data of this work can be found in online version of the paper" 529 530 ACKNOWLEDGEMENTS 531 Figure 1. Denitrifying activity test using the same biomass source and monitoring (a) the liquid phase composition: NO3--N (■) and NO2--N (●) concentrations, and HB (▲) and HV (˟) percentages of consumption; and (b) N2 gas production inside the vial (〇) in the manometric test throughout time. The results are average values of the performed triplicates with the corresponding standard deviation for assay B2 in Table 1. 0 3 6 9 12 15 18 0 10 20 30 40 50 60 0 50 100 150 200 250 300 350 400 450 HB, HV (%) NO2--N, NO3--N (mg/L) Time (min) (a) 0 1 2 3 4 5 0 50 100 150 200 250 300 350 400 450 N2-N (mg) Time (min) (b) Figure 1 Figure 2. Comparison between SEDA values determined by both methods: Liquid phase measurements (■); Gas phase measurements (■).The results are average values of the performed triplicates with the corresponding standard deviation. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Assay 1 Assay 2 Assay 3 Assay 4 SEDA (g N2-N/(g VSSact d) Figure 2 (a) (b) Figure 3. Measured SEDA with respect to the PHA content: (a) PHA as COD concentration in the liquid media; and (b) PHA as percentage inside the biomass cells. Seeding sludge collected from SBR-A (●), and from SBR-B (▲). The results are average values of the performed triplicates with the corresponding standard deviation. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0 1000 2000 3000 4000 5000 SEDA (g N2-N/(g VSSact d) PHA (mg CODPHA/L) 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 SEDA (g N2-N/(g VSSact d) PHA (%wt/wt) Figure 3 Figure 4. Correlation between SEDA with respect to CODPHA/N ratio: seeded with SBR-A biomass (●), and seeded with SBR-B biomass (▲). The results are average values of the performed triplicates with the corresponding standard deviation. 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 SEDA (g N2-N/(g VSSact d) g CODPHA/g NO3--N Figure 4 Figure 5. Profile of N2 gas production: B6, assay with 1.4 g VSS/L (□); B7, assay with 0.7 g VSS/L (x). The results are average values of the performed triplicates with the corresponding standard deviation. 0 1 2 3 4 5 6 0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 N2-N (mg) Time (min) Figure 5