Enhancement of Pd-based catalysts for the removal of nitrite and nitrate from water

Nitrate pollution is one of the most important issues in water quality (Gupta & Ali 2012). The European Commission highlighted the situation of some hypertrophic rivers, lakes and ground waters where NO₃⁻ exceeds legal limits in more than 10% of the sampling points in the European Union (EU) (Eurostat 2012). Supranational organizations such as the European Commission and the World Health Organization (WHO) have developed some regulations and guidelines regarding NO₃⁻ accumulation in ground waters. The guideline values for NO₃⁻ and NO₂⁻ in drinking water dictated by the WHO are 50 and 3 mg/L, respectively. Several nations have also limited the maximum NH₄⁺ levels for drinking water to 0.5 mg/L (Della Rocca et al. 2007).

The removal of NO₃⁻ from water has been traditionally carried out by biological nitrification-denitrification which is a well-known, effective and cost-effective method. However, this method is not recommended for the treatment of drinking water because of the risk of biological contamination. The application of other economically feasible, effective and environmentally friendly solutions is required. Other systems such as electroreduction can have drawbacks, such as the production of chlorates and perchlorates (Sánchez-Carretero et al. 2011). Some other technologies based on the reduction of NO₃⁻ and perchlorate such as advanced reduction processes have also been tested recently (Vellanki et al. 2013).

The catalytic reduction of NO₃⁻ is one of the more suitable technologies for the potabilization of ground water polluted with NO₃⁻. The catalytic reduction of NO₃⁻ uses H₂ as reactant in the presence of a bimetallic catalyst containing a noble metal from group VIII, usually palladium, and a transition metal, usually copper, as catalytic promoter for the conversion of NO₃⁻ to N₂ (Hörold et al. 1993; Barrabés & Sá 2011; Pizarro et al. 2015). The reported costs of NO₃⁻ catalytic reduction are in the same range as other technologies such as biological denitrification, reverse osmosis and ionic exchange, and there are other advantages such as no brine production, innocuous end products, no biohazard risks, stability, and ease of fine-tuning and start-up, which have to be considered in the life cycle analysis comparison (Centi & Perathoner 2003). The main cost is the use of hydrogen that is used in very low concentration when the H₂ is limited in the reaction medium. The cost of this process decreases with the life of the catalyst; therefore the main objective is to obtain catalysts with high stability. Hence, NO₃⁻ can be converted selectively to N₂ using Pd and other metals as promoters under adequate operation conditions. Several works have studied the denitration process using monometallic Pd supported on SnO₂ (Gavagnin et al. 2002) and ZrO₂ (Marella et al. 2000), showing some advantages over traditional supports such as Al₂O₃ or SiO₂. These conventional supports present narrow pores that allow the formation of an internal OH⁻ gradient that diminishes their activity and selectivity to N₂, increasing the undesirable NH₄⁺ production. The development of catalysts with a higher pore width (i.e. Pd/SnO_2, Pd/ZrO₂) decreases the effect of alkalinity inside the pore system, causing a noticeable increase of the selectivity to N₂ compared with catalysts with narrow pores.

Previous works have reported low NH₄⁺ selectivity using Al₂O₃, activated carbon, TiO₂, SiO₂ and SnO₂ as catalytic supports. The selectivity and activity greatly depends on the catalyst properties (active phase and surface charge) and the operation conditions, mainly pH, N:H ratio (liquid phase) and the competition between different species present in the feed stream. The catalytic supports also play an important role since they present significant differences in their surface charge, area, particle size, pore width and hydrophobicity. NH₄⁺ selectivity is also reported as 5% to 10% using these supports (Hörold et al. 1993; Barrabés & Sá 2011; Pizarro et al. 2015). The interaction between Pd and the promoting metal such as Cu, In or Sn in bimetallic catalysts supported on inert materials is also more improbable than that present in reducible metal oxides where Pd nanoparticles are in full contact with the active supports such as SnO₂ (Gavagnin et al. 2002; Sá et al. 2007; Bošković et al. 2008; Guo et al. 2012), CeO₂ (Epron et al. 2002; Devadas et al. 2011; Kim et al. 2014; Lee et al. 2015) or TiO₂ (Gao et al. 2003; Sá & Anderson 2008; Zhang et al. 2008; Krawczyk et al. 2011; Chen et al. 2013; Kim et al. 2013, 2014).

The majority of studies (close to 75%) focused on the catalytic reduction of NO₃⁻ concluded that Pd is the most active phase to reduce NO₂⁻. In addition, Pd is more selective to N₂ production than Rh, Ru or Pt catalysts (Hörold et al. 1993; Barrabés & Sá 2011). Studies on the use of other less toxic metals such as zero valent iron for NO₃⁻ reduction reported an undesirable high selectivity to NH₄⁺ and iron leaching caused by its corrosion. Pd has demonstrated the highest stability so far and scarce leaching in the pH range for water potabilization. The aim of this work is to study and compare the catalytic reduction of NO₂⁻ and NO₃⁻ with H₂ as reducing agent using Pd catalysts supported on Al₂O₃, SiO₂ and SnO₂. The effect of the H₂ flow and the supplementation of CO₂ have also been evaluated. A novel Pd-In/γ-Al₂O₃ catalyst has been also prepared and evaluated in terms of activity, selectivity to N₂ and stability treating NO₃⁻ bearing water.

Al₂O₃ Brockmann (Sigma Aldrich), SiO₂ (Scharlab), and SnO₂ (Sigma) were used as supports to prepare the monometallic catalysts tested in the batch experiments. Pd was incorporated in the oxides by wet impregnation of PdCl₂ (Sigma, 99.99%) dissolved in an acidic solution (0.1 M HCl) or Na₂PdCl₄ (Sigma, 99.99%) dissolved in water (1 mL/g of support). The bimetallic catalysts supported on Al₂O₃ or SiO₂ were co-impregnated using PdCl₂ (Sigma, 99.99%) (0.1 M HCl) with InCl₃ or SnCl₄ (liquid). The impregnated materials were dried overnight at 100 °C and calcined at 500 °C for 2 h. The resulting catalysts were reduced in H₂ atmosphere (35 NmL/min) at 100 °C for 1 h before the catalytic experiments. The Pd content in the monometallic catalyst was 5 wt%. Bimetallic catalysts were prepared with 5 wt% of Pd with a ratio Pd:Me 4:1 (w:w), where Me is Sn or In.

Room temperature powder X-ray diffraction (XRD) patterns were obtained with a Bruker D8 Advance A25 diffractometer with a Bragg Brentano configuration and a lineal detector Lynxeye (Bruker). The X-ray source was a copper long fine focus X-ray diffraction tube operating at 40 kV and 30 mA, with an angular step of 0.015° and a time per step of 0.1 s. Crystalline phases were identified by referencing the PDF 2014 database.

The porous structure of the catalyst was assessed from the −196 °C N₂ adsorption-desorption isotherms, obtained in a Micromeritics ASAP 2020. The samples were previously outgassed at 160 °C and 0.66 Pa for 16 h. The specific surface area was calculated from the BET method (S_BET). The particle size distribution of the metallic phase was determined by transmission electron microscopy (TEM) using a Phillips CM-200 microscope with a point of resolution of 2.8 Å coupled with an energy dispersive EDAX X-ray spectrometer.

The reduction experiments in batch mode were performed in batch glass reactors (0.6 L capacity) at 25 °C and atmospheric pressure using 1.61 mM initial concentration of NO₂⁻ (NaNO₂, Scharlau, 98%) or NO₃⁻ (NaNO₃, Scharlau, 99%) with 1 g/L of catalyst. Internal and external mass transfer limitations were discarded after preliminary experiments where the effects of agitation (700 rpm), bubbling and particle size were checked. The reactor was placed in a thermostatic bath and the experiments were conducted under continuous H₂ or H₂ + CO₂ flow.

The catalyst was previously reduced at 100 °C under H₂ flow (35 N mL/min) for 1 h and outgassed with N₂ for 30 min at ambient temperature. The temperature of the reactor was maintained at 25 °C before and during the catalytic tests. The catalyst was placed inside the reactor with 450 mL of distilled water. The catalyst was maintained in this aqueous suspension under continuous stirring and bubbling H₂ or H₂ + CO₂ for 20 min prior to the catalytic test in order to favour the catalyst wetting. A H₂ gas flow of 15 N mL/min was used, which was reduced to 8 N mL/min when CO₂ was supplemented (8 N mL/min). Then, a concentrated solution containing (16.1 mM of NO₂⁻ (NaNO₂) or NO₃⁻ (NaNO₃) was fed to the catalytic reactor, reaching an initial concentration of 1.61 mM. Catalytic tests of 1 and 3 h were performed for NO₂⁻ and NO₃⁻ reduction, respectively. Samples were withdrawn from the reactor and filtered through regenerated cellulose filters with a pore diameter of 0.45 μm.

The stability of the catalysts was checked upon long-term continuous experiments (up to 500 h) carried out in a fixed-bed reactor (Pyrex glass, 30 cm length, 9 mm inner diameter). The catalysts were deposited on a porous glass support placed in the middle part of the reactor. The catalyst used consisted of 0.3 g of Pd-In/Al₂O₃ synthetized according to the aforementioned method. An aqueous solution of 100 mg/L of NO₃⁻ was fed (0.5 mL/min) with H₂ and CO₂ at different flow rates (0.3–1.5 N mL/min) and fractions (3.75–100% H₂/CO₂, v/v). Liquid samples were taken from the reactor at different times on stream.

NO₃⁻, NO₂⁻, NH₄⁺ and total nitrogen (N_T) were analyzed by colorimetric titration using Hach Lange reagents and a benchtop VIS spectrophotometer DR 3900™ with RFID (Hach Lange). The pH, electrical conductivity (EC) and temperature were analyzed using a Crison MM 40 equipment.

The XRD diffractograms of the Pd catalysts prepared by impregnation of PdCl₂ on SiO₂, Al₂O₃ and SnO₂ are depicted in Figure 1(a)–1(c), respectively. The mean Pd particle size was measured using the Scherrer equation giving values of 14, 9 and 11 nm for Pd/Al₂O₃, Pd/SiO₂ and Pd/SnO₂, respectively. The crystallinity of these catalysts samples resulted in 40%, 73% and 77%, respectively. Table 1 shows the BET area and the average pore diameter of these monometallic catalysts. A high increase of the pore width upon the impregnation stage was observed for Pd supported on Al₂O₃ probably due to the acidic nature of the PdCl₂ solution. Contrarily, pore size diminished when Pd was incorporated to SiO₂, which seems to be related to pore system blockage. The effect of Pd incorporation on SnO₂ was negligible. A similar response was observed with the bimetallic catalysts (Table 2), but the effect of the impregnation was more severe in this case. The main difference with respect to the monometallic catalysts is the significant decrease in the BET surface area values obtained; this is because a high percentage of the metals are co-impregnated. The metal loads obtained are close to the expected nominal values, which indicates that the impregnation method proposed is highly effective.

TEM images of the monometallic catalysts are depicted in Figure 2. As can be observed in Figure 2(a) and 2(b), the high load of Pd achieved in the Al₂O₃ catalysts favoured the agglomeration of the Pd nanoparticles. The micrographs of Pd/SiO₂ (Figure 2(c) and 2(d)) and of Pd/SnO₂ (Figure 2(e) and 2(f)) show a high dispersion of the metallic active phase. The average nanoparticle diameters of the Pd nanoparticles calculated were 62, 20 and 5 nm for Pd-Al₂O₃, Pd/SiO₂ and Pd/SnO₂, respectively, following the procedure described elsewhere (Pizarro et al. 2014).

Figure 3 shows the NH₄⁺ concentration observed at different NO₂⁻ conversion with monometallic catalysts impregnated with PdCl₂. The analysis of the selectivity to NH₄⁺ formation revealed a broad response of the catalysts tested. The Pd/SiO₂ catalyst was especially attractive since a concentration of NH₄⁺ lower than 3.5 mg/L was obtained for a NO₂⁻ conversion close to 50%. This has been related to its small nanoparticle size, which favours the selectivity to N₂ (Matatov-Meytal et al. 2003). The use of Al₂O₃ and SnO₂ as supports led to NH₄⁺ concentrations of 7.0 and 8.5 mg/L, respectively, for a similar NO₂⁻ conversion value (50%).

The Pd/SnO₂ catalyst showed the highest reaction rates and conversion (70%), which reveals the positive impact of highly dispersed Pd nanoparticles (Figure 2). However, this catalyst led to an NH₄⁺ concentration higher than that obtained using Pd/Al₂O₃ and Pd/SiO₂ catalysts. This fact could be related to the synergistic interaction between Pd and the SnO₂ reducible support.

The catalytic reduction of NO₂⁻ was also evaluated using bimetallic Pd-Sn and Pd-In catalysts supported on Al₂O₃ and SiO₂. As can be observed in Figure 4(a), the addition of co-metals, especially In, increased the conversion extension up to 0.82. Figure 4(a) shows the NH₄⁺ concentration produced with Pd-Sn/Al₂O₃ and Pd-Sn/SiO₂ catalysts. The conversion obtained with these catalysts exceeds the values obtained with the monometallic catalysts supported on Al₂O₃ or SiO₂. Previous works have reported that bimetallic catalysts favour the dispersion of the Pd metallic phase due to the strong interaction between Pd and the SnO₂ phase, expanding the active surface of Pd, especially when a co-impregnation method is used (Bošković et al. 2008; Pizarro et al. 2015).

The addition of Sn (Pd-Sn/Al₂O₃) as co-metal also enhanced the selectivity to N₂, reducing the NH₄⁺ concentration from 6.7 to 3.6 mg/L at a NO₂⁻ conversion of 50%. Unlike the results observed with the monometallic catalysts, the use of SiO₂ in bimetallic catalysts produced more NH₄⁺ than those supported in Al₂O₃.

The results obtained in the catalytic reduction of NO₂⁻ with bimetallic Pd-In catalysts supported on Al₂O₃ and SiO₂ are shown in Figure 4(b). The presence of In led to a higher activity than that obtained with Sn doped catalysts. This enhancement of the catalytic activity seems to be related to a strong interaction between Pd and In₂O₃ phases that results in a close contact between the two active phases. The production of NH₄⁺ with Pd-In catalysts yielded concentrations of 5.0 and 12.5 mg/L at a NO₂⁻ conversion of 50% with Pd-In/Al₂O₃ and Pd-In/SiO₂, respectively.

The catalytic reduction of NO₃⁻ using a monometallic Pd-SnO₂ catalyst was tested using two different salts as precursors: PdCl₂ and Na₂PdCl₄ (Figure 5). First, the NO₃⁻ reduction was performed with H₂ as the only gas fed (Figure 5(a)). The conversion reached with the catalyst synthesized with PdCl₂ (90%) was higher than that synthesized with Na₂PdCl₄ (60%) while a slight difference was observed on the selectivity to NH₄⁺ at a similar NO₃⁻ conversion. Nitrite production was favoured with the catalyst impregnated with Na₂PdCl_4.Figure 5(b) shows the conversion of NO₃⁻ and the concentration of NH₄⁺ reached when feeding H₂:CO₂ (50%, v/v), where a significant enhancement in the catalytic performance was observed. The addition of CO₂ in the inlet gas led to a total NO₃⁻ conversion and decreased the selectivity to NH₄⁺ as a consequence of the OH⁻ neutralization in the pore system. The use of PdCl₂ as salt precursor resulted in a higher catalytic activity and a high NO₃⁻ reduction rate due to the close interaction between Pd and the reducible support. In addition, NO₂⁻ accumulation in the reaction medium and the increase of the pH medium, were mitigated in the presence of CO₂.

The catalytic reduction of NO₃⁻ was also tested with Pd-Sn and Pd-In catalysts impregnated with PdCl₂ supported on Al₂O₃ and SiO₂ in absence and presence of CO₂. Figure 6(a) shows the NO₃⁻ conversion and the concentration of NO₂⁻ and NH₄⁺, corresponding to the catalytic reduction of NO₃⁻ with Pd-Sn/SiO₂ and Pd-In/SiO₂ catalysts without pH buffering. The Pd-In catalyst yielded total conversion of NO₃⁻ in less than 3 h while the Pd-Sn catalyst reached less than 25% conversion. The latter was deactivated by the alkaline medium generated in the catalytic reaction studied.

The performance of these bimetallic Pd-Sn and Pd-In catalysts supported on SiO₂ was improved significantly when adding CO₂ to control the pH value between 5.0 and 5.5 in the reaction medium (Figure 6(b)). The NO₃⁻ conversion was significantly increased by supplementing CO₂, reaching a conversion of 65% with the Pd-Sn/SiO₂, and total conversion was achieved with the Pd-In/SiO₂ catalyst. The selectivity to NH₄⁺ obtained with the latter clearly diminished, with concentrations from 21 to 14 mg/L and from 5 to less than 3 mg/L at a NO₃⁻ conversion of 100% and 50%, respectively. The initial rate of NO₃⁻ reduction reached with the Pd-In/SiO₂ catalyst was fivefold higher than that of Pd-Sn/SiO₂ (Table 3).

NO₃⁻ conversion and selectivity to NH₄⁺ was significantly improved when CO₂ was added using the Pd-Me bimetallic catalysts supported on Al₂O₃. Figure 7(a) shows the results obtained during the catalytic reduction of NO₃⁻ experiments with Pd-Sn/Al₂O₃ and Pd-In/Al₂O₃ in absence of CO₂. Bimetallic Pd-Sn/Al₂O₃ catalyst led to a higher conversion level (50%) than the Pd-Sn/SiO₂ catalyst (25%) showed in Figure 6(a). Pd-In/Al₂O₃ catalyst reached total conversion and a similar selectivity to NH₄⁺ to that obtained with the Pd-In/SiO₂ catalyst (Figure 6(a)). NO₂⁻ is accumulated in the absence of a pH buffering source (Figure 6(a)), which is effectively transformed by the addition of CO₂ (Figure 6(b)).

The results obtained with the Pd-Sn/Al₂O₃ and Pd-In/Al₂O₃ catalysts without and with CO₂ in the gas feed are depicted in Figure 7(a) and 7(b), respectively. A total NO₃⁻ conversion was achieved with both catalysts when CO₂ was added, the Pd-In catalyst being less selective to NH₄⁺, reaching a concentration of 7.5 mg/L in the final effluent (Figure 7(b)). The initial rate of NO₃⁻ reduction reached by the Pd-In/Al₂O₃ catalyst was twice the observed with the Pd-Sn/Al₂O₃ catalyst (Table 3). Again, the pH control by adding CO₂ was an effective strategy to diminish the production of NO₂⁻ and NH₄⁺. These results indicate that the bimetallic catalysts supported on Al₂O₃ are the most appropriate for the reduction of NO₃⁻; the Pd-In/Al₂O₃ catalyst was selected for the next experiments. The wider pores of Al₂O₃ than that observed for SiO₂ could present a positive effect on the neutralization of OH⁻ when CO₂ is supplemented.

It has been reported that the H:N ratio in the gas feed can have a decisive influence on the selectivity to N₂ obtained during the catalytic reduction of NO₃⁻ (Hörold et al. 1993). With the aim of analyzing the influence of the amount of H₂ present in the reaction medium, the H₂ flow was varied from 0.3 to 8.0 N mL/min maintaining a CO₂ gas flow of 8.0 N mL/min with the Pd-In/Al₂O₃ catalyst. Figure 8(a) shows the concentration of NH₄⁺ obtained at different NO₃⁻ conversions. The amount of NH₄⁺ produced with the lowest H₂ flow rate (0.3 N mL/min) was 0.33 mg/L of NH₄⁺ (selectivity to NH₄⁺ of 2.7%) for a NO₃⁻ conversion of 55%, reaching the legal limits established by the environmental legislation. This result improves those obtained under H₂ and CO₂ saturation as reflected in Table 3. This shows that selectivity towards NH₄⁺ can be effectively diminished by controlling the H:N ratio in the gas feed. High values of NO₃⁻ conversion (around 80%) were reached at increasing H₂ flows (4.0 and 8.0 N mL/min).

The availability of chemisorbed hydrogen promotes the formation of NH₄⁺ from NO₂⁻ (Ebbesen et al. 2008). This fact determines the selectivity when different relative concentrations of NO₃⁻/NO₂⁻ are treated, different sizes of Pd nanoparticles are used or the solubility of H₂ is modified by changing the operation conditions.

The stability of the Pd-based catalysts has been scarcely studied, showing high and constant activities treating synthetic NO₃⁻-polluted waters (Wang et al. 2009). The influence of the amount or relative fraction of the gases fed has been also analyzed during long-term stability tests. Other long-term experiments with Pd-In catalysts using different experimental conditions such as powdered Pd/Al₂O₃ catalysts, H₂ or formic acid as reducing agents and lower temperatures have been conducted in previous studies (Prüsse et al. 2000; Chaplin et al. 2007, 2009).

The Pd-In/Al₂O₃ was used in a fixed-bed reactor varying the gas flow and the proportion of H₂ to CO₂. Four different gas mixtures were evaluated, stages I–IV (Figure 8(b)). A gas composition of 0.3 N mL/min of H₂ diluted in 8 N mL/min of CO₂ (3.6% H₂) was fed during stage I. This experiment lasted for more than 50 h, detecting a very low amount of NH₄⁺ in the resulting effluent (0.2 mg/L) but reaching a low NO₃⁻ conversion of around 11%. Higher NO₃⁻ conversions (28%) were obtained when the CO₂ flow was reduced to 1.2 N mL (25% H₂) where no more than 0.5 mg/L of NH₄⁺ was detected. The addition of a higher amount of H₂ (0.75 N mL/min), keeping the relative H₂:CO₂ fraction constant (50%), also led to an increase of NO₃⁻ conversion up to 32% and an NH₄⁺ concentration of around 1.0 mg/L. During the last stage H₂ was fed solely to the reactor at a flow rate of 1.5 N mL/min. The catalytic activity showed a significant increase of the NO₃⁻ conversion by 31–37% with respect to the previous experimental stages. Nevertheless, the selectivity to NH₄⁺ increased, detecting 2.0 mg/L in the discharge due to the increase of the pH values from 4–5 to around 9 (Figure 8(b)). Decreasing H₂ feed flow values reduces the production of NH₄⁺ dramatically, and the pH of the resulting effluents are leveled to neutral values.

Finally, a catalytic test was performed under the same conditions evaluated in stage III to analyze the stability of the catalyst (Figure 8(c)). The space-time used was 6.21 kg_cat·h/mol. The results showed herein demonstrate that the Pd-In/Al₂O₃ catalyst was stable during 500 h on stream. The conversion values were maintained between 30% and 42% with a production of NH₄⁺ between 1.6 and 2.2 mg/L during the whole experiment. The pH values directly depend on the amount of CO₂ added, concluding that values lower than 5.5 are optimal for the catalytic reaction. The concentration of H₂ is one of the main limiting steps that determines the N:H ratio in this process which can be fine-tuned in future applications.

The activity and selectivity to N₂ of monometallic Pd catalysts used for NO₂⁻ removal in water have been evaluated with SnO₂, Al₂O₃ and SiO₂ as supports, respectively. The use of bimetallic Pd-Sn and Pd-In catalysts improved the NO₂⁻ conversion, especially with the Pd-In catalysts where the metallic phases showed a strong interaction. Pd-In catalysts showed high activity in the NO₃⁻ reduction, leading to low NH₄⁺ concentration effluents. The Pd-In/Al₂O₃ prepared in this work has shown promising stability (more than 20 days in continuous operation) and a high selectivity to N₂. The addition of CO₂ in the gas feed as buffering agent and the limitation of H₂ dose are effective to control NH₄⁺ formation, and offer an efficient strategy to ensure a long-term operation satisfying drinking water quality standards.

The authors would like to express their gratitude to A. Buenaventura, Director of Abengoa Research, for his help. The authors also thank Santiago Medina Carrasco and the XRD laboratory team from the University of Seville for their help in the development of the work. The authors would also like to express their gratitude to Maria Elena Guillén, Manuel Antonio Díaz and Javier Echave for their help in this work.