Potential of low-enthalpy geothermal energy to degrade organic contaminants of emerging concern in urban groundwater

Low-enthalpy geothermal energy (LEGE) is a carbon-free and renewable source to provide cooling and heating to infrastructures (e.g. buildings) by exchanging their temperature with that of the ground. The exchange of temperature modifies the groundwater temperature around LEGE installations, which may contribute to enhancing the capacity of aquifers to degrade organic contaminants of emerging concern (OCECs), whose presence is significantly increasing in urban aquifers. Here, we investigate the impact of LEGE on OCECs and their bioremediation potential through numerical modelling of synthetic and real-based cases. Simulation results demonstrate that: (i) LEGE facilities have the potential to noticeably modify the concentrations of OCECs; and (ii) the final impact depends on the design of the facility. This study suggests that optimized LEGE facility designs could contribute to the degradation of OCECs present in urban aquifers, thus improving groundwater quality and increasing its availability in urban areas.

the facility and the energy obtained, but all of them satisfy the minimum energetic requirements of the NLB. The concentration of diclofenac (C DCF ) and carbamazepine (C CBZ ) are normalized, respectively, by the concentration of diclofenac (C DCF,0 ) and carbamazepine (C CBZ,0 ) under unperturbed (i.e., initial) conditions. The first simulated scenario (Sce1) barely modifies the concentrations of the assessed OCECs in the aquifer. Only small variations occur around the injection well. The limited changes occur because variations of groundwater temperature induced by GWHP are low and restricted to the surrounding of the injection well (Fig. 3). In addition, the pumped and injected flow-rates are relatively low, and thus, the volume of mobilised water is low in comparison with that of the aquifer. In the scenario Sce2, small variations with respect to the unperturbed conditions are observed downgradient of the injection wells. The concentration of diclofenac and carbamazepine decreases downgradient of the injection well that introduces hot water and increases downgradient of the well injecting cold water (Fig. 3), which agrees with Eqs. (1) and (3). The normalized concentration of diclofenac at the end of the simulated period decreases up to 0.8 around the hot injection well and increases up to 1.15 around the cold injection well. Similarly, the normalized concentration of carbamazepine decreases down to 0.7 and increases up to 1.3 around the hot and cold injection wells, respectively. The volume of mobilised water seems to be an important factor controlling the impact of GWHP, which is supported by the results of Sce3, Sce4, and Sce5 scenarios, where the pumped and injected flow-rates are doubled. When the flow rate is increased, the volume of groundwater affected by the GWHP facility, and thus, the variations of temperature (Fig. 3) increase. Consequently, the influence of GWHP on the selected OCECs also increases.
The simulated GWHP in the scenario Sce3 reduces the concentration of the chosen OCECs downgradient and around it. This greater reduction of the concentration compared to that observed in Sce1 and Sce2 scenarios occurs because during summer the potential of the groundwater is exploited to the maximum obtaining more energy than that required by the NLB. The injected water during hot months is at 37.5 °C, which produces a plume of hot groundwater (Fig. 3) where the degradation rates of diclofenac and carbamazepine increase.
The largest reductions in the concentrations of the OCECs occur in the scenario Sce4 because the proposed GWHP scheme exploits the cooling potential of the aquifer (i) to the maximum, and (ii) during the whole year. Thus, the temperature of the hot water plume is higher than in the other scenarios (Fig. 3). Despite in the scenario Sce5 the cooling potential of the aquifer is also exploited during the whole year and the hot water is injected through a separate well, the decrease in concentrations is lower than that observed in the scenario Sce4. This fact occurs because the energy obtained for cooling purposes during cold months in the scenario Sce5 is lower than that obtained in the scenario Sce4, and then the size of the hot water plume is smaller than in the scenario Sce4 (Fig. 3). Note that, in the Sce2 and Sce5 scenarios, the normalized concentration of diclofenac  www.nature.com/scientificreports/ and carbamazepine is higher than 1 around and downgradient the cold injection well (i.e., is higher than the concentration under unperturbed conditions). This fact occurs because the groundwater temperature in this area decrease in comparison to unperturbed conditions. Then, the removal rate in this area is lower and the final OCECs concentration is higher than under natural conditions. The adopted values for the hydraulic conductivity (K), the effective porosity (θ eff ), the longitudinal and transversal dispersivities (D L and D T ) and the thermal diffusivity (D m ) can play a significant role in the behavior of OCECs under the influence of GWHP. Although the parameters used in our model have been derived from field investigations, the sensitivity of the model to them has been assessed to investigate their effect on the results. Thus, some additional simulations have been developed by varying the values of K, θ eff , D L , D T and D m . Simulations are developed considering scenario Sce4. This scenario is chosen because it is the most favourable to enhance the degradation capacity against the selected OCECs. The results of these simulations and their discussion are shown in the supplementary material (Appendix B).
Evolution of concentration and groundwater temperature at the downgradient boundary. Similar conclusions can be drawn when observing the concentration of the selected OCECs (carbamazepine and diclofenac) in the water that flows out the aquifer along the downgradient boundary. Figure 4 shows the concentration of diclofenac (Fig. 4a) and carbamazepine (Fig. 4b) through the downgradient boundary normalized by the initial concentration (i.e., under unperturbed conditions). Figure 4c displays the evolution of the groundwater temperature through the downgradient boundary. Results are obtained considering the whole volume of water crossing the boundary. The Sce1 and Sce2 scenarios barely modify the concentration of the studied OCECs. Despite the concentrations of diclofenac and carbamazepine vary in the scenario Sce2, their decrease around and downgradient of the hot well is compensated by their increase (i.e., reduction of the degradation rate) around the cold well. As a result, the concentrations of both OCECs in the scenario Sce2 slightly increase in the downgradient boundary in comparison with the unperturbed conditions. Concentrations significantly decrease for the Sce3, Sce4 and Sce5 scenarios. After 10 years of operation, the normalized concentration of diclofenac decays down to 0.9, 0.76 and 0.84 for the Sce3, Sce4 and Sce5 scenarios, respectively, while that of carbamazepine decreases down to 0.77, 0.54 and 0.67 for the Sce3, Sce4 and Sce5 cases, respectively. It is important to highlight that after 10 years of operation, the concentration of diclofenac and carbamazepine through the downgradient boundary continues to decrease and a steady state is not reached. Thus, the concentration of these OCECs will continue decreasing and the final impact of the GWHP facility will be larger for longer operational for the Sce4, Sce5 and Sce3 scenarios, respectively, when considering the whole volume of water crossing the downgradient boundary. As in the synthetic case (see Appendix A of the supplementary material), the observed changes in the concentration of the OCECs in the downgradient boundary start earlier than the groundwater temperature variations. This fact is also related to the difference between the considered retardation factors for the modelled OCECs and the heat. The main outcome that can be drawn from the simulation results is that GWHP facilities have the potential to modify the concentration of OCECs, as well as organic compounds across the aquifers, and that their impact depends on the facility design. Theoretically, GWHP scenarios that produce large plumes of high temperature reduce in a higher degree the concentration of OCECs. The impact of GWHP is higher for the carbamazepine than for the diclofenac, which is a consequence of the temperature dependency of both compounds. From 20 to 35 °C, the degradation velocity of diclofenac increases by a factor of 2 while it increases by a factor of 5 for carbamazepine.
Unfortunately, numerical results could not be compared with real measurements because the site is still under construction and the geothermal facility has not been built yet. Once the construction will be finished, it is planned to periodically take groundwater samples upgradient and downgradient the study site to analyse the behaviour of OCECs under the influence of the geothermal facility.

Discussion
This research represents a step forward in the field of urban water resources and groundwater remediation, as it shows that LEGE facilities can significantly modify the concentration and distribution of OCECs. The comparison between different GWHP scenarios indicates that the facility design (given uses, energy production or position of wells) plays a critical role in the behaviour of OCECs, increasing their attenuation as the thermal plume becomes larger. This fact suggests that properly designed GWHP facilities have the potential to improve the quality of groundwater by degrading OCECs, and thus increasing the quality and amount of available freshwater resources.
The variations observed in the concentration of OCECs as a result of the simulated GWHP facilities can be considered as relatively low, in the order of ng L −1 . However, it should be borne in mind that hundreds, or even thousands, of different OCECs can be found in urban aquifers 44 . Therefore, the global impact of GWHP facilities towards OCECs attenuation will be much larger, substantially improving the quality of groundwater resources. www.nature.com/scientificreports/ In addition, the two considered OCECs (carbamazepine and diclofenac) have low retardation factors. Thus, it is expected that GWHP impact will increase in other OCECs with high retardation factors, as the residence time within the thermal plume will be longer. However, the increase of groundwater temperature induced by GWHP could have counter-productive effects under some circumstances, especially when transformation products are more persistent, mobile, and harmful (enhanced toxicity) than their parent compound 45 . Therefore, it will be needed to analyse the nature of the potential transformation products at any specific case to evaluate the benefits and disadvantages, in terms of groundwater quality, of installing a LEGE facility with remediation purposes. The GWHP scenarios presented in this investigation are designed to take account of their utilization in a maritime Mediterranean climate (Köppen-Geiger classification: Csa 46 ), like in Barcelona (Spain), where cooling requirements are usually higher than heating ones. Thus, the volume of hot water introduced in the aquifer is expected to be higher than that of cold water. Instead, if GWHP facilities are used mainly for heating purposes, the volume of injected cold water will increase, having negative consequences regarding the groundwater quality because the degradation rate of OCECs will decrease.
A factor that may improve bioremediation in urban aquifers is subsurface urban heat islands (SUHI). The temperature increase of a few degrees associated with SUHI will increase the degradation rates of OCECs improving the groundwater quality. The contribution of SUHI to improve bioremediation of urban aquifers deserves to be deeply investigated since many anthropogenic OCECs reach aquifers in urban areas, where SUHI occur. www.nature.com/scientificreports/ Finally, it is needed to highlight that although the impact on groundwater temperature of GWHP seems to be beneficial in terms of groundwater quality, there are some issues that deserve further investigation. For example, it is necessary to reach an agreement between the reduction of OCECs and the potential negative impacts related to the creation of a large thermal plume that could affect the efficiency of LEGE facilities located downgradient. LEGE design at the city scale should take into consideration upgradient LEGE facilities. For example, a LEGE downgradient of the LEGEs considered in scenarios Sce 2 and 5 could improve its performance by drilling two pumping wells, taking water from the cold plume for cooling and water from the hot plume for heating. Consequently, the design of LEGE facilities should include the influence of aquifer properties on temperature variations and the interactions between adjacent LEGEs. In addition, it is needed to consider biodegradation potential in the presence of a wide range of OCECs and under variable redox conditions. In this regard, changes of aquifer temperature can have a critical effect on various (a)biotic processes like microbial activity, redox (electron transfer reactions), pH, as well as contaminant transport and fate (e.g. (co)precipitation-dissolution, adsorption-desorption, (bio)transformation). For instance, an increase in aquifer temperature could foster consumption of dissolved oxygen and organic carbon concentrations by microbial activity, resulting in suboxic/anoxic conditions in the aquifer. Additionally, these low redox conditions could dissolve manganese and iron oxides, contributing to the mobility of arsenic (As) turbidity or clogging 47 . In any case, attaining sulphate-reducing conditions could release sulphide ion which can be toxic and corrosive 48 .

Methods
Geothermal scheme. There are two types of ground source heat pumps (GSHPs), closed-loop systems, where the heat is exchanged with the ground through the circulation of a carrier fluid in borehole heat exchanger (BHE) buried into the ground 49,50 , and open-loop systems, also named groundwater heat pumps (GWHPs) 51,52 . In GWHP schemes, groundwater is pumped from aquifers and carried to surface heat exchangers, where heat is exchanged with the working fluid of a heat pump. Subsequently, water is commonly returned to the aquifer through a discharge well 53 . In this study, we consider GWHP schemes since these systems are less costly and more efficient 50 than GSHP schemes when clean water is available 54 . Clean groundwater refers to main quality standards, such as pH, hardness, iron content, dissolved oxygen and turbidity, because if these parameters are not acceptable, as previously noted, corrosion, incrustation, erosion or clogging may occur 55 . Thus, despite deterioration of urban groundwater quality due to the presence of OCECs, for example for potable water, this is not significant for water quality standards for GWHP.
General description of the modelled site. The influence of GWHP on selected OCECs is investigated using a model based on a real site. The site, which is located in Barcelona (Spain), is an industrial complex where an important textile factory, named "Can Batlló", which was operative from 1880 to 1964. Currently, it is planned to construct green areas and public facilities in the area occupied by the main factory and the adjacent industrial units. One of the planned actions is to transform the old main existing building of the factory to build a new modern library building (NLB). In accordance with the policies and commitment of the Barcelona City Council to climate change mitigation, it is planned to use LEGE to provide heating and cooling to the future NLB. In this context, some previous investigations, which only have addressed the problem from an energy point of view, have been developed to establish the viability of different LEGE scenarios 56 . As a result of these investigations, it has been decided to use a closed-loop type geothermal facility to cover only the heating and cooling requirements of the NLB. Here, we want to go further and give a vision about the potential of a hypothetical LEGE facility of the open-loop type (i.e., GWHP) to cover the energetic demand and to improve the quality of groundwater by enhancing the removal of OCECs.

Geographical, geological and hydrogeological context. The study site (Can Batlló) is located in
Barcelona (North-East of Spain). The site is placed in the Barcelona plain between the deltas of the Besòs (North-East) and Llobregat (South-West) rivers (Fig. 5a). The Barcelona plain is also surrounded by the Collserola mountain range (West and North-West) and the Mediterranean Sea (East and South-East) (Fig. 5a).
Geologically, Can Batlló is located at the intersection of three sedimentary units (Fig. 5b). These units are: (1) the Barcelona plain; (2) the Montjuïc deposits and (3) the Llobregat delta river unit. Specifically, Can Batlló is found between 3 paleochannels of Quaternary age (Fig. 5c). From borehole information and collected data, Miocene, Pliocene and Quaternary materials can be distinguished below the study site (Fig. 5b). Quaternary materials are between 10 and 30-m thick and largely correspond to the filling deposits of the Barcelona plain and Montjuïc streams. These Quaternary materials consist in clay and silt deposits intercalated with layers of sand and gravel. The Miocene materials are deposits of the Serravallian (Middle Miocene) and they reach up to 40 m of thickness. They consist of clay and marl deposits intercalated with layers of sand and gravel. Finally, two types of Pliocene materials can be distinguished in Can Batlló: (1) yellowish silt and (2) bluish marl. Both are made of compacted, slightly permeable and slightly cemented materials, which form part of the Llobregat river delta basement. Hydrogeologically, according to the piezometric map (Fig. 5c), the flow direction depends on the three Quaternary paleochannels. The water table is located in the Quaternary materials between 6 and 8 m.a.s.l. The average hydraulic gradient ranges from 0.005 to 0.006 and the groundwater flows towards the South-East. Hydraulic tests carried out in the surroundings reveal an average effective hydraulic conductivity ( K ) of 3.3 m d −1 for the Quaternary and Miocene materials that can be considered as a single multilayer aquifer. Heating and cooling requirements. The minimum energetic requirements for the NLB has been calculated considering the dimensions and uses of the conditioning system (Fig. 6). The maximum net monthly energy required for heating is 140 MWh (in winter), and for cooling is 165 MWh (in summer).

Presence of OCECs
Numerical approach. PHT3D 58 code was used to build the numerical model. This code solves advectivedispersive-reactive transport processes by coupling MT3DMS with PHREEQC 59 . The numerical model simulates the Quaternary and Miocene materials because (i) the Pliocene formation has a low hydraulic conductiv-  (Figs. 5c and 7). The production and injection wells are modelled by using Neumann BCs. Concerning the transport BCs, the concentration of OCECs is prescribed on the upgradient boundaries according to the measured concentrations at the field. Additionally, a constant input of mass of 6.1·10 −12 and 1.1·10 −12 mol d −1 per square Figure 6. Net monthly energy required for temperature regulation (i.e., heating and cooling) of the future NLB. Positive values (red bars) refer to energy needed for heating while negative ones (blue bars) refer to energy required for cooling. Table 1. Aquifer properties at the study site (Can Batlló). Values are derived from pumping tests and thermal response tests performed at the site. The value of θ eff is calculated by considering the measured porosity for the more permeable layers (0.1) and their total thickness (20 m) in comparison with the whole thickness of the simulated materials (60 m). The thermal diffusivity coefficient is selected in agreement with the reference values specified in the guidelines for thermal use of the underground of the German Engineer Association 60 . www.nature.com/scientificreports/ meter is imposed in the whole domain for diclofenac and carbamazepine, respectively, to mimic the observed concentrations in the aquifer. Diclofenac and carbamazepine are modelled using the Monod kinetics and considering the redox conditions that enhance their degradation. The degradation of carbamazepine is modelled as the 1st order Monod kinetics including an inhibition term to account for the dissolved oxygen concentration (O 2 ). The carbamazepine degradation rate ( r CBZ ) is modelled as where MX CBZ is the maximum degradation rate constant of carbamazepine, C CBZ is the concentration of carbamazepine, K inhO 2 is the inhibition coefficient for O 2 , C O 2 is the concentration of O 2 , and f T is a function that depends on the temperature. f T is defined with the Arrhenius equation adding a normalization factor ( β ). β allows normalizing the result to 1 when the temperature of groundwater is between 35 and 40 °C, which is when the highest microorganism activity occurs 38,61-63 , as where A is a pre-exponential factor, E A is the activation energy, R is the gas constant and T the temperature in Kelvin. Similarly, the degradation of diclofenac was approximated as a 1st order degradation. In this case, a Monod term to incorporate the influence of O 2 39 is included. The diclofenac degradation rate ( r DCF ) is modelled as: where K O 2 is the Monod half-saturation constant of diclofenac, MX DCF is the maximum degradation rate constant of diclofenac, and C DCF is the concentration of diclofenac. Parameters used in Eqs. (1) to (3) are obtained from bibliographical data ( Table 2). Parameters for computing f T (A and E A ) are derived from laboratory data provided for carbamazepine by Burke et al. 14 . MX CBZ and MX DCF are obtained from Barkow et al. 39 . K inhO 2 is obtained by fitting Eq. (1) to ensure that r CBZ is equal to MX CBZ in the absence of oxygen and, according to Regnery et al. 64 , very low (0.001 d −1 ) under oxic conditions (O 2 ≥ 1 mg L −1 ). Low removal rate under aerobic conditions has been corroborated by several authors 15,[65][66][67] . K O 2 is calculated by fitting Eq. (3) to ensure that r DCF matches MX DCF under aerobic conditions and is equal to 0.03 d -1 under sub-oxic/anoxic conditions. This value has been calculated by averaging data from Banzhaf et al. 68 and Heberer et al. 69,70 summarized in Henzler et al. 65 . Despite the retardation factors of carbamazepine and diclofenac are low 57,65 , we consider them to increase the accuracy of the results with a value of 1.9 and 1.41 for the carbamazepine and diclofenac, respectively 71 .
(1)  www.nature.com/scientificreports/ Simulated exploitation scenarios. Five different hypothetical scenarios that satisfy the energetic requirements of the NLB are simulated to compute their impact on OCECs distribution. They differ in the pumped and injected flow rates, the number and location of the wells and the uses given to the facility. The distance between wells (production and injection) is calculated to avoid thermal breakthrough (t BR ) during the simulated period according to 72 where S VCaq is the volumetric heat capacity of the aquifer (2800 J kg −1 K −1 ; Ref. 73 , S VCwat is the volumetric heat capacity of the water at 20 °C (4180 J kg −1 K −1 ), L is the distance between wells and α is defined as where b is the saturated thickness and Q is the pumping/injection rate. The temperature of injected water in the models is computed according to the thermal potential of the system (P GW ) as follows: where T is the temperature difference between the production and injection wells. The groundwater temperature under unperturbed conditions is assumed constant and equal to 20 °C 74 . Scenarios 3, 4 and 5 consider the possibility of obtaining more energy than needed to maximize the usefulness of the GWHP facility. The considered scenarios are as follows: • Scenario 1 (Sce1): This scheme provides only the required energy for the climatization of the NLB. The GWHP system is made up by one production and one injection well (Fig. 8). The pumped and injected flow rates are constant and equal to 432 m 3 d −1 . • Scenario 2 (Sce2): This scheme provides only the required energy for the climatization of the NLB. The GWHP system is made up by one production and two injection wells (Fig. 8). The pumped and total injected flow rates are constant and equal to 432 m 3 d −1 . The two injection wells do not inject water at the same time.
One is activated when the facility is used for heating (i.e., cold water is injected) while the other is activated when the facility is used for cooling (i.e., hot water is injected). • Scenario 3 (Sce3): This scheme provides the required heating energy for the climatization of the NLB during cold periods, while during hot periods, more cooling energy than that needed by the NLB is extracted with the objective of providing cooling energy to neighbourhood buildings, factories or other nearby infrastructures. The obtained cooling energy during hot periods is the one that yields an injection temperature of 37.5 °C. This scenario consists in two wells (one for production and one for injection) and the pumped and injected flow rates are of 864 m 3 d −1 (Fig. 8). • Scenario 4 (Sce4): This scheme considers the possibility of providing, in addition to the heating and cooling energy needed by the NLB, continuous cooling energy to neighbourhood factories and infrastructures that need it, such as data centres containing high-performance computing systems or information technology equipment 75 . This scenario consists in two wells (one for pumping and one for injection) (Fig. 8) that pump and inject 864 m 3 d −1 and two heat exchangers. During cold months, the needed heating energy is extracted from 30% of the pumped water using one of the heat exchangers, while the rest of the pumped water (70%) is used to produce cooling energy in a second heat exchanger. The extracted energy for cooling is equal to that obtained by the system during hot periods (i.e., 527 MWh). Outflow water from both heat exchangers is mixed and injected into the aquifer through the same well. The maximum temperature of injected water reached during cold months is 36.5 °C. During hot periods, energy for cooling is extracted from the whole pumped water using only one heat exchanger, and the extracted energy is that for which the temperature of the injected water is 37.5 °C. • Scenario 5 (Sce5): This scheme also considers the possibility of providing, in addition to the heating and cooling energy needed by the NLB, continuous cooling energy to neighbourhood factories and infrastructures 75 . Differently from scenario Sce4, Sce5 consists in three wells (one for pumping and two for injection) and two heat exchangers (Fig. 8). The production well pumps 864 m 3 d −1 . During hot periods, all pumped water is used for cooling, the extracted energy is that for which the temperature of the injected groundwater is 37.5 °C (i.e., 527 MWh, higher than the demand of the NLB) and the hot water is returned to the aquifer through the hot injection well. During cold months, required heating energy by the NLB is obtained from the half of the pumped water (i.e., 432 m 3 d −1 ) and the resulting cold water from the heating process is returned to the aquifer through the cold injection well. The other half of pumped groundwater (432 m 3 d −1 ) is used to obtain cooling energy. The extracted energy for cooling during these cold periods is that for which the variation induced in the water temperature is 17.5 °C (i.e., 264 MWh) and the resulting hot water is injected through the hot well (Fig. 8).
The excess of energy obtained at Sce2, Sce4 and Sce5 scenarios could be shared (i.e., commercialized) with nearby buildings, factories or other infrastructures like data centres containing information technology equipment, which will contribute to maximize the efficiency of the installation. Considering that Barcelona is located in a maritime Mediterranean climate region (Köppen-Geiger classification: Csa 46 ), where atmospheric temperature can reach up to 39 °C in summer 76 , and the demand for cooling is high, this sharing adds value to the installation. www.nature.com/scientificreports/ Table 3 summarizes the energy obtained from the GWHP facility for each scenario and Fig. 8 displays the location of the wells and the temperature of the injected water at the considered scenarios.

Data availability
Data generated or analysed during this study are included in the article/supplementary material, further inquiries can be addressed to the corresponding author.