Induced biodenitrification of nitrate‐polluted groundwater: engineering strategies and assessment of chemical, microbial and isotope effects

Nitrate pollution is a widespread problem that affects water bodies in many regions of the world, undermining water quality and therefore its safe use. Despite the application of improved management practices, nitrate pollution seems to increase, particularly in groundwater. The Nitrate Vulnerable Zone (NVZ) designation in Europe, for instance, has increased from 35.5% of the EU-15 territory at the end of 1999 to 44% at the end of 2003, and the Commission’s report for the period 2004-2007 revealed that 15% of groundwater monitoring stations in the EU-27 territory showed nitrate levels above the limit of 50 mg of nitrates per liter. Some trends towards nitrate attenuation are observed, but at least 33% of water bodies will clearly fail in achieving the 2015 goals set by the Water Framework Directive. Several efforts have been addressed to either reduce nitrogen inputs or to decrease its already accumulated levels, particularly by designing nitrate-removal technologies aimed at recovering drinking-water standards. This PhD thesis, hence, focuses on the optimization of an already existing technology for nitrate­removal: enhanced in situ biodenitrification (EISB), which is now regaining attention due to its economic and environmental benefits and its potential for scale-up and design of case­specific solutions. EISB is an engineered application of microbial heterotrophic denitrification aimed at in situ nitrate removal from groundwater. Aimed at stimulating facultative denitrifiers, EISB is based on the injection of a C source into the aquifer. Microbial denitrification is then enhanced in a designated area of the aquifer, creating a biologically active zone (often referred as biowall) which removes nitrate from the naturally-flowing groundwater. Among the different factors that affect the technical feasibility of EISB, the type and quantity of the injected C source is a key issue, particularly due to its influence upon the microbial processes that determine the treatment performance. The understanding of the subsurface geology and hydrogeology is also an issue of concern, particularly if highly heterogeneous media, such as fractured aquifers, are meant to be remediated. Aimed at achieving our research goal, several EISB experiments were developed at different scales -batch, flow-through column and pilot scale-and involving different geological media -granular and fractured-. Combined chemical, microbial and isotope monitoring tools where applied to gain a better insight on the denitrification process and thus improve technology design and optimization. The first set of batch-scale experiments focused on testing the viability of in situ heterotrophic denitrification and determining the most suitable biostimulants for a case­specific scenario in the Osona region, a Catalan NVZ showing historic nitrate pollution up to 200 mg/L. Native microbiota was stimulated and nitrate reduction was effectively achieved by addition of a carbon source (ethanol or glucose) as well as a phosphorous source (disodium hydrogen phosphate). Transient nitrite accumulation was observed, especially when using glucose as the C source. The N and O isotope fractionation was determined to be -13.0‰ and -17.1‰ for eN and -8.9‰ and -15.1‰ for eO in ethanol and glucose-amended experiments respectively, resulting in eN/eO values of 1.46 (ethanol-amended experiment), and 1.13 (glucose-amended). Organic carbon (OC) consumption in batch­scale experiments, expressed as .C/.NO3 -, varied slightly depending on the type of C source used: 1.6 mmolOC/mmolNO3 -for ethanol and 2.2 for the glucose, similarly to stoichiometric values associated with nitrate respiration (0.83 and 1.25 mmolOC/mmolNO3 ­respectively). When deriving stoichiometric reactions that accounted not only for the amount of electron donor used for nitrate respiration but also for cell synthesis, the following values were determined: 1.9 and 2.0 mmolOC/mmolNO3 -for ethanol and glucose­induced biodenitrification respectively. These values were used for the numerical modeling of batch-scale experiments, aimed at quantifying microbial kinetics by applying the modified Monod expression. The (geochemical) numerical model also indicated a different effect of mineral precipitation on ethanol or glucose-induced denitrification, an effect that is linked to a different alkalinity production. Such effect could be taken into account when designing and/or optimizing EISB systems, particularly as a way to control geochemical clogging. A pilot-scale application was then performed at the site, aimed at assessing the viability of EISB in a fractured aquifer. Ethanol was now used as the main C source, and based on lab­scale results, P was also added. Again, transient nitrite accumulation was detected, and evidences for incomplete denitrification and coexistence of other respiration processes (such as iron or sulfate reduction) and autotrophic denitrification were observed. Sulfate isotope characterization proved that autotrophic denitrification linked to sulfide oxidation could be occurring along with heterotrophic denitrification, while sulfate­reduction couldn’t be verified. Overall, results suggested that stimulated heterotrophic denitrification could be applied as a remedial alternative in a fractured media and despite the complexity of the formation. However, a deep understanding of the system is required and efforts must be addressed to control microbial population and stability as a key issue to avoid the decrease of groundwater quality due to incomplete denitrification or secondary respiratory processes. Different engineering approaches such as feeding or pumping strategies could help improving the system performance. Aimed at testing the impact of such engineering approaches upon resulting water quality, a second study-case was studied, now in an alluvial media. . A flow-through experiment was built to simulate an EISB system and assess the influence of different C addition strategies upon the denitrification process. Heterotrophic denitrification was stimulated by the periodic addition of a C source (ethanol), and 4 different addition strategies were evaluated, being the first-one a weekly injection, and the others a daily injection with decreasing amounts of C. Enhanced denitrification was stimulated following the first C addition, easily achieving drinking water standards for both nitrate and nitrite. Water quality in terms of remaining C, denitrification intermediates and other anaerobic respiration products varied during the experimental time. Ethanol, for instance, showed a cyclic behavior during the weekly feeding strategy while it was completely depleted when injected daily. A quasi steady­state nitrate outflow, similar to ethanol’s, was obtained in daily injection scenarios, with nitrate levels ranging from non-detected values and up to 10 mg/L, and nitrite’s remaining undetected. No dissimilatory nitrate reduction to ammonium was ever detected and some secondary microbial respiration processes, mainly manganese reduction, were suspected to occur temporarily. Overall, results showed that biodenitrification could be successfully achieved by a daily addition of a C source slightly higher than the stoichiometric value, diminishing the accumulation of non-desired products and the biofilm growth and still obtaining the required denitrification results. Reducing the C/N ratio enables us to reduce treatment costs while achieving a better water quality in terms of remaining C and residual microflora, and potentially reducing the biofouling effect due to the increase of endogenous respiration. Endogenous activity –that provides internal C for denitrification-may become important when low C/N values are used, keep denitrification temporarily ongoing and reducing the biofilm growth, but may affect the biodenitrification performance at longer operation times. Such aspects should be further evaluated using modeling and/or experimental tools. Furthermore, results suggested that not only the feeding strategy but also the biofilm life-time have a direct effect on microbial population structure and hence on the biodenitrification performance, reducing the accumulation of nitrite over time. The obtained eN/eO fractionation values for the flow-through experiment (1.01) fell within the low-end of previously reported data (varying from 0.9 to 2.3), an effect that may be linked to faster microbial kinetics in enhanced vs. natural biodenitrification. Similar low values were observed in our previous batch-scale experiments as well as in other work conducted in our lab. Concerning ethanol’s fractionation, on the other side, a two-trend behavior was observed, probably indicating a change in the dominating C­consuming population. Interestingly, the second trend suggests an inverse fractionation of the C source that got depleted while being consumed.