Catalytic Hydrogenation of Nitrate over Immobilized Nanocatalysts in a Multi-Phase Continuous Reaction System: System Performance, Characterization and Optimization

Abstract

Nitrate catalytic reduction in a continuous system was studied in the presence of Pd-Cu macrostructured catalysts synthesized through a novel washcoating methodology of the pre-formed bimetallic powder catalyst. The present work aims to understand the behavior of the macrostructured bimetallic catalyst in the presence of different reaction conditions in order to achieve the design of an optimized facility that can produce the best catalytic results: maximum NO₃⁻ conversion with enhanced N₂ selectivity. The residence time of the inlet solution and the catalyst concentration in the reactor proved to be the parameters that most influenced the conversion and selectivity due to the important role that these parameters play in the hydrodynamic conditions of the reactor. A higher loading of catalyst and lower inlet flow rates allow promoting a higher contact time between the three phases that participate in the reaction (G-L-S). The most efficient reaction conditions (three pieces of the macrostructured catalyst, liquid flow rate of 10 mL min⁻¹, and a total gas flow rate of 200 Ncm³ min⁻¹ (1:1 H₂:CO₂)) allowed obtaining an NO₃⁻ conversion of 51% with a corresponding N₂ selectivity of 23%. Also, the conversion results strongly depended on the total gas flow rate used during the reaction since this assists the mixing between the three phases and promotes a greater contact that will contribute to enhanced catalytic results.

1. Introduction

Nitrate (NO₃⁻) is an environmental pollutant that can be commonly found in surface and groundwater. Water contamination with NO₃⁻ presents environmental and human health concerns related to methemoglobinemia development; several medical papers have described NO₃⁻ and NO₂⁻ as precursors for cancerous processes and thyroid malfunctions in babies and pregnant women [1,2,3,4]. Also, the carcinogenic nitrosamines may also be formed in adult stomachs upon the ingestion of nitrate-contaminated water [5,6]. The European and US legislation have established drinking water maximum concentrations of 50, 0.1, and 0.5 mg L⁻¹ for NO₃⁻, NO₂⁻, and NH₄⁺, respectively [7]. Hence, there is an increasing urgency to develop removal and control technologies for NO₃⁻ abatement.

Conventional water treatment systems include several methodologies to remove inorganic contaminants. However, these techniques still present some difficulties in their application [8,9]. Many procedures, such as reverse osmosis, ion exchange, and biological denitrification, have been studied over the past few years to perform NO₃⁻ abatement and showed efficient results for the treatment of water contaminated with NO₃⁻, within the legal limits stipulated, being techniques that are already implemented at full industrial scale [10,11,12,13]; however, disadvantages associated with high implementation costs or promotion of ion concentration without further transformation are the main disadvantages of these techniques, being imperative the study of solutions that can promote conversion into less toxic species [14].

Heterogeneous catalytic reduction allows the direct conversion of NO₃⁻ into less toxic species that could safely be released into the environment [15]. NO₃⁻ can efficiently be converted into N₂ in the presence of adequate bimetallic catalysts and a reductive agent; the main drawback is associated with the formation of NO₂⁻ as an intermediate by-product and excessive NH₄⁺ production, which is an undesirable by-product [1,16,17]. Generally, the most accepted reaction mechanism for NO₃⁻ catalytic reduction was proposed by Epron et al. [18], where it is suggested that NO₃⁻ adsorbs on the transition metal, which is responsible for its reduction to NO₂⁻ through a redox reaction. Great efforts have been made, over the past few years, concerning the development of effective catalysts for NO₃⁻ reduction; however, their application is still limited due to NH₄⁺ formation during the process [19]. The bimetallic pair Pd-Cu was demonstrated as one of the most promising for NO₃⁻ conversion with interesting selectivities to N₂ formation [16,19,20,21,22,23,24]. However, the efficiency of the process also depends on the reaction conditions applied during the conversion process, namely the catalyst concentration and durability (high catalytic activity without deactivation), gas flow rates, the reaction pH, and liquid flow rate (in the case of NO₃⁻ catalytic reduction in a continuous reactor), among other factors [1,25,26,27].

Several studies are reported focusing on the design and study of catalytic NO₃⁻ reduction in continuous systems [28,29,30,31,32]. One of the first studies was conducted by Pintar et al. [33], who studied the catalytic hydrogenation of NO₃⁻ in fixed-bed reactors, showing that NO₃⁻ conversion and selectivity were strongly influenced by liquid and gas flow maldistribution associated with the use of compact material beds. To avoid this type of problem, monolithic reactors emerged as interesting supports for the powder catalysts allowing immobilization of the powdered materials on structured supports and their application in heterogeneous catalytic continuous systems [34,35,36,37]. These types of structures have been widely studied over the past few years as efficient solutions to perform the catalyst immobilization for their application at the industrial level [38,39]. Also, the possibility of immobilizing powder catalysts on a structured support consists in a more sustainable approach for applying such catalysts in a catalytic water treatment system, contributing to the stability and security of the system [40,41]. However, to create an efficient system for NO₃⁻ conversion using a monolithic reactor, it is still required to optimize all the parameters that influence the catalytic reaction. This approach has not been adequately studied yet, since most of the studies in continuous systems are focused on fixed-bed reactors and there are very few studies on the application of monolithic reactors as support for the catalyst (macrostructured catalysts); also, most of these studies are more related to the synthesis of stable and active macrostructured catalysts and not to the optimization of reaction parameters that drive the reaction [37]. Thus, as important as the catalyst design, there is a need to understand what variables of the continuous system operation mostly contribute to the efficiency of the process using macrostructured catalysts.

The aim of the present work is the study of the catalytic liquid-phase reduction of aqueous NO₃⁻ solution in a continuous-flow three-phase reactor, at atmospheric pressure and room temperature, using H₂ as a reducing agent and CO₂ as a pH buffer over nanostructured catalysts immobilized on macro-structured supports. For this purpose, different inlet liquid and gas flows will be considered, as well as different H₂ and CO₂ ratios, and varying catalyst loading. It is intended to understand which would be the most important parameters to control, in order to obtain high NO₃⁻ conversion and selectivity to N₂, towards the up-scaling of the hydrogenation reaction of NO₃⁻ using a continuous reactor.

2. Materials and Methods

2.1. Bimetallic Macrostructured Catalyst Preparation

Bimetallic macrostructured catalysts were synthesized through a washcoating procedure using a pre-formed 5 wt.% Pd and 2.5 wt.% Cu powder catalyst supported on ball-milled carbon nanotubes (5%Pd–2.5%Cu/CNT(BM 2h)) [42]. The powder catalyst was synthesized by co-incipient wetness impregnation, and the synthesis procedure has been extensively described in previous works [22]. This consists of a novel methodology that allows the direct incorporation of pre-formed powder catalysts, being an interesting methodology that allows taking advantage of the mechanical modifications that are only compatible with powder material synthesis and are responsible for increasing the catalytic activity of the catalysts. This catalyst was selected for these studies considering the novelty of the process and the practicality of the synthesis methodology when compared with more conventional synthesis methods.

In the first stage, cordierite honeycomb monoliths (1 cm diameter, 6 cm length, and with 100 channels per square inch, from Corning) were thermally treated at 900 °C for 1 h in air flow to remove any impurities. A coating solution was prepared using the 5%Pd–2.5%Cu/CNT(BM 2h) powder catalyst properly dispersed in ultra-pure water by using an organic binder (Triton X-100^TM (TX100) from Sigma-Aldrich (St. Louis, MI, USA)). The amounts of binder and dispersant used were adjusted to maintain a 1:1 weight ratio with the powder catalyst in the slurries [43]. The monolith coating was performed 5 times, by dip-coating (30 mm min⁻¹ of dipping speed and 1 min of immersion) using a Bungard RDC21 K coater, with a drying step (at 100 °C in an oven) between each coating. To perform full TX100 removal from the macrostructured catalysts structure, without compromising the catalysts activity and avoiding metal nanoparticles sintering, a washing procedure was applied to the catalyst by successive dip-coatings in a solution of isopropanol and acetone (IPA:ACN) (1:1 v/v) [44,45]. The resulting macrostructured catalyst was identified as a Pd-Cu/WCP_wash catalyst. An image of the macrostructured catalysts before and after coating has been added in Supplementary Information (Figure S1).

2.2. Experimental Set-Up and Procedure

Continuous catalytic experiments were performed in a custom-made experimental set-up, working in continuous mode, consisting of a bubble column with an internal recirculation loop. Figure 1 presents a schematic representation of the continuous system used during the catalytic experiments.

For the catalytic reduction experiments, an initial solution of NO₃⁻ (prepared from NaNO₃ from Sigma-Aldrich) was introduced, using a peristaltic pump, at the top of the column, at different initial NO₃⁻ concentrations (30, 60, and 90 mg L⁻¹). Different liquid flow rates (5, 7.5, and 10 mL min⁻¹) were studied in order to promote different residence times inside the reactor and understand how this can influence NO₃⁻ conversion and selectivity. The liquid flow rate of the recirculation loop is around 10 times the inlet flow rate of the system. The residence time distribution (RTD) was calculated for each of the studied liquid flow rates using a step-injection tracing method using a dye tracer. The concentration of the dye was measured at regular intervals using an UV-Vis spectrophotometer; RTD and mean residence time for each liquid flow rate were obtained by modelling the evolution of the outlet dye concentration using the Dankwertz equation applied to a continuous flow reactor [43,46]. The system presented an RTD of 20 min and took about 60 min to reach the steady state.

H₂ was used as a reducing agent and CO₂ as a pH buffer in order to maintain the solution pH value close to 5.5 (to help avoiding NH₄⁺ formation). The gas was introduced at the bottom of the column, allowing the formation of a counter-current system, by using a gas diffuser. Different gas flow rates and H₂:CO₂ combinations were tested in order to optimize the conversion and selectivity. In the present work, a pre-reduction step of the catalyst was not performed; however, by the time the system reached the stationary state, it was considered that the used catalyst was pre-reduced.

Bubble size distribution was obtained in order to ascertain how the change in the total gas flow rate (H₂ + CO₂) could influence the bubble size distribution throughout the reaction. These experiments were performed using a focusable VLMTM2 Laser line (1.1 Mw, 635 nm 30°). The residence times were also calculated for these experiments to ascertain whether differences in the total gas flow rate could influence the liquid residence time in the column reactor.

To monitor the progress of the reaction, liquid samples were collected at the outlet of the reactor, at specific times, to quantify NO₃⁻, NH₄⁺, and NO₂⁻ concentrations. This quantification was performed by ionic chromatography in a Metrohm 881 Compact IC Pro apparatus, equipped with a Metrosep A Supp 7 Anionic Exchange Column (250 mm × 4.0 mm) for anionic species quantification and Metrosep C4 Cationic Exchange Column (250 mm × 4.0 mm) for cationic species analysis. The obtained results correspond to an average of five points taken at fixed intervals (30 min) along the reaction after the stationary state was reached. Solution pH was controlled by taking samples along the reaction time after the steady state was reached.

NO₃⁻ conversions and the corresponding NO₂⁻, NH₄⁺, and N₂ selectivities were calculated as described in [47]. The catalytic experiments performed during the present work were performed in triplicate. The maximum error registered between experiments was of 4%.

The first-order reaction rate of each experiment was calculated by application of the derived balance equation (Equation (1)). A continuous stirred-tank reactor behavior was assumed, as previously reported, due to the high ratio between the internal circulation loop and inlet flows (~10) [43]. In this equation, $C_{Steadystate}$ corresponds to NO₃⁻ concentration measured at steady-state (in mg L⁻¹), and $C_{Feed}$ represents the inlet concentration of NO₃⁻ that is fed to the reactor (in mg L⁻¹). $k_{het}$ is the heterogeneous first-order reaction rate constant, and $t_r$ represents the mean residence time in minutes (min).

$$\left[C_{Steadystate}\right]=\frac{C_{Feed}}{1+k_{het}·t_r}$$

(1)

The previous equation was modified to allow a comparison between the catalytic performance of the macrostructured catalysts (Equation (2)). A deviation coefficient, $d$, was added to the equation representing the apparent deviation from the powder performance (which is considered as the optimal performance of the catalyst) that includes factors such as the mass transfer limitations associated with the immobilization of the catalyst on a macro-structured support and all the factors that condition the reaction kinetics in a continuous reactor [43].

$$C_{{{NO}_3^{-}}_f}=\frac{C_{{{NO}_3^{-}}_i}}{1+(k_{{NO}_3^{-}}·d_{{NO}_3^{-}})·t_r}$$

(2)

For the previous equations, $C_{{{NO}_3^{-}}_i}$ represents the initial NO₃⁻ concentration feed to the reactor, and $C_{{{NO}_3^{-}}_f}$ stands for NO₃⁻ concentration measured on steady-state (both in mg L⁻¹); $k_{{NO}_3^{-}}$ corresponds to the first-order reaction rate constant obtained for the corresponding powder catalyst tested in a semi-batch reactor, and $t_r$ represents the mean residence time on the reactor (in min). An arrangement of Equation (2) allows to calculate the deviation coefficient ($d_{{NO}_3^{-}}$) associated with NO₃⁻ conversion, which represents the deviation associated with the reaction conditions and how these parameters influence NO₃⁻ conversion.

$$d_{{NO}_3^{-}}=\frac{\frac{C_{{{NO}_3^{-}}_i}}{C_{{{NO}_3^{-}}_f}}-1}{k_{{NO}_3^{-}}·t_r}$$

(3)

Another approach to this equation consisted of its arrangement considering the formation of N₂ during reaction (Equation (4)). For this specific case, $C_{{N_2}_f}$ corresponds to N₂ concentration after the reaction reaches the steady state, and $C_{{{NO}_3^{-}}_i}$ corresponds to the initial concentration of NO₃ (both in mg L⁻¹); $k_{{NO}_3^{-}//N_2}$ represents the heterogeneous first-order reaction rate constant for N₂ formation obtained for the powder catalyst during testing in a semi-batch reactor; $t_r$ corresponds to the residence time (in min); and $d_{{NO}_3^{-}//N_2}$ represents the deviation coefficient associated with N₂ formation. This coefficient is representative of the contribution of factors inherent to the continuous reaction process that led to the formation of N₂ during NO₃⁻ conversion.

$$C_{{N_2}_f}=C_{{{NO}_3^{-}}_i}-\frac{C_{{{NO}_3^{-}}_i}}{1+[k_{{NO}_3^{-}//N_2}\times d_{{NO}_3^{-}//N_2}]·t_r}$$

(4)

The arrangement of Equation (4) to obtain the deviation coefficient of N₂, $d_{{NO}_3^{-}//N_2}$, is present in Equation (5).

$$d_{{NO}_3^{-}//N_2}=\frac{\frac{C_{{{NO}_3^{-}}_i}}{C_{{{NO}_3^{-}}_i}-C_{{N2}_f}}-1}{k_{{NO}_3^{-}//N_2}·t_r}$$

(5)

Thus, $d_{{NO}_3^{-}//N_2}$ is a representative coefficient of the added interfering factors that will be in the origin of an increase or decrease of N₂ selectivity of the process.

3. Results and Discussion

3.1. Bimetallic Macrostructured Catalyst Characterization

For the present work, only one methodology was selected for the synthesis of the macro-structured catalyst, which was used throughout the study so that all the differences in performance corresponded to the influence of the operating conditions of the system.

Briefly, the catalyst presented a well-adhered carbon coating on a macro-structured cordierite support, with a carbon content of about 2% to the total mass of the macro-structured catalyst and a metal percentage of 1.6% of Pd and 0.4% of Cu per mass of carbon present in the structure (results obtained through ICP-OES analyses). It is important to note that these percentage values obtained through ICP-OES correspond to the mass of carbon present in the catalyst, which in this case is around 200 mg for each macrostructured catalyst. TEM micrographs of this catalyst allowed us to conclude that some of the metal nanoparticles were found inside the carbon nanotubes due to a possible migration during the synthesis of the powder catalysts (before incorporation on the cordierite structure) and/or the washing methodology with an organic solvent solution applied for the removal of the surfactant from the catalyst structure, which can promote metal nanoparticles migration [44,48,49].

X-ray diffraction (XRD) results of 5%Pd–2.5%Cu/CNT(BM 2h) powder catalyst allowed to identify the formation of Pd-Cu alloys. Since the application of a washcoating procedure allowed the incorporation of this powder catalyst directly on the cordierite, these structures will also be present on the macrostructured catalyst playing a predominant role in the reaction scheme. The proximity of both metals facilitates the migration of chemisorbed H⁺ and the reactants through the active centers of the catalyst, promoting a higher NO₃⁻ conversion degree.

SEM-EDS analysis of the structured catalyst allowed us to evaluate the metal phase distribution along the coating layer (in the cross-section), showing that the washcoating methodology results in catalysts with a homogeneous distribution of the metallic phase through the different coating layers. This distribution proved to represent a key role in the catalyst activity for NO₃⁻ conversion since the depth of the coating layers limits the diffusion of reactants leading to a greater difficulty for these species to contact with the catalyst active centers that are more deeply deposited. Pd-Cu/WCP_wash bimetallic catalyst was characterized in detail in previous works [42]. The used catalysts proved to be active and stable for at least 30 h of reaction.

3.2. Continuous Catalytic Experiments Results

3.2.1. Parameter Optimization Experiments

Several experiments were performed to study the influence of each parameter on the process conversion and selectivity. These studies aim not only to select the best values for each of the intervening parameters in the reaction, in order to build an active catalytic system with enhanced conversion and selectivity, but also to understand the behavior of the macrostructured bimetallic catalyst with different reaction conditions to inform the design of an optimized facility that can produce the best catalytic results. All the experiments were always performed for the same catalyst, so the obtained results correspond to the influence of the operating conditions during the reaction.

In a first stage, a blank test without the addition of a catalyst to the reaction system was performed. This experiment showed that in the absence of a catalyst, no NO₃⁻ conversion was registered.

Inlet NO₃⁻ Concentration

At this stage, different inlet NO₃⁻ concentrations were studied. The results obtained are presented in Figure 2.

Generally, and as it was expected, NO₃⁻ conversion slowly decreased with the increase of the inlet NO₃⁻ concentration in the feed solution of the system [50,51]. However, it is important to highlight that only a 5% decrease in activity was noticed from inlet NO₃⁻ concentrations of 90 mg L⁻¹ to 30 mg L⁻¹, meaning that the catalyst retains its activity at much higher NO₃⁻ concentrations. In what concerns the process selectivities, it was possible to verify a slight increase in N₂ selectivity; however, this increase may not be directly related to the performance of the system to obtain higher N₂ conversions for higher initial NO₃⁻ concentrations, but with the slight loss of activity registered.

Previous studies reported that NO₃⁻ initial concentrations have a significant effect on the catalytic conversion and corresponding selectivity [50,51,52,53]. No significant differences in conversion were reported for the inlet NO₃⁻ concentration ranges reported in the present work. In general, the largest conversion differences were reported for differences in initial NO₃⁻ concentrations four times higher than those tested. Such results depend on other factors, such as the properties of the reaction medium and the catalyst used. Nevertheless, it is possible to conclude that an increase in NO₃⁻ inlet concentration, for the same number of active centers, leads to, on the one hand, the accumulation of NO₃⁻ that remains unreacted due to the unavailability of Cu active centers and, on the other hand, to the accumulation of NO₂⁻ that is not converted due to the lack Pd active centers. This way, for this specific case, the design of the catalyst plays an important role in the conversion efficiency since the presence of more active sites will lead to a higher conversion. Also, in the case of the studied catalyst, the presence of metal alloys promotes a higher selectivity to N₂ [42].

Considering these results, and since higher conversion results were obtained for a lower NO₃⁻ initial concentration, 30 mg L⁻¹ was selected as inlet concentration for further experiments in order to be able to further differentiate the results obtained by changing the different reaction parameters.

Liquid Flow Rates Optimization

Figure 3 presents the results obtained for NO₃⁻ conversion and corresponding N₂, NH₄⁺, and NO₂⁻ selectivities, using different liquid flow rates during continuous catalytic experiments conducted with Pd-Cu/WCP_wash macrostructured catalyst.

Through the obtained results, it is possible to conclude that NO₃⁻ conversion substantially increases with the increase of the residence time in the reactor. This result is evident since higher residence times in the reactor allow more chances to promote the contact between NO₃⁻ and the catalyst [8,9,54]; however, selectivity for NH₄⁺ formation was favored for lower liquid flow rates. This increase may be associated with raising the contact time between H₂ present in the liquid phase and the catalyst, which could lead to an overreduction phenomenon, which promotes a higher NH₄⁺ formation. Generally, the reaction mechanism on a Pd-Cu supported catalyst starts with the adsorption of NO₃⁻ on Cu centers where it is going to be reduced to NO₂⁻ (2Cu⁰ + 2NO₃⁻ → 2CuO + 2NO₂⁻) and, then, NO₂⁻ is desorbed from the oxidized promoter metal into the aqueous solution or directly to the reduced Pd-H_ads complex, where NO₂⁻ can be reduced to N₂ (2NO₂⁻ + Pd-4H_ads → Pd⁰ + N₂ + 4OH⁻) or NH₄⁺ (NO₂⁻ + Pd-6H_ads → Pd⁰ + NH₄⁺ + 2OH⁻). The noble metal is responsible for regenerating the previously oxidized Cu center through chemisorbed hydrogen (CuO + 2H_ads → Cu⁰ + H₂O). Usually, N₂ formation is favored at an acidic pH, while NH₄⁺ formation is favored at alkaline pH. Through the equations deduced for the reaction, it is possible to verify that the presence of H₂ in large quantities can lead to an overreduction phenomenon, characterized by the presence of large amounts of reducing agent, contributing to NH₄⁺ formation instead of N₂ (NO₂⁻ + Pd-6H_ads → Pd⁰ + NH₄⁺ + 2OH⁻ or 2NO₂⁻ + Pd-4H_ads → Pd⁰ + N₂ + 4OH⁻). Also, it is important to highlight that changes in liquid flow will have consequences regarding the mass transfer inside macrostructured catalysts channels. This can lead to significant differences in the mixing of the three phases, which will be reflected in the conversion and selectivity values achieved.

For a flow rate of 10 mL min⁻¹ (corresponding to a residence time of around 20 min), it was possible to achieve an NO₃⁻ conversion of around 20% with a corresponding N₂ selectivity of 51%. Approximately the double of this conversion was achieved for a liquid flow rate of 5 mL min⁻¹ (around 47% of NO₃⁻ converted); however, NH₄⁺ selectivity also increased with the increasing conversion (registering selectivities of around 68% to NH₄⁺ and 27% to N₂). For a liquid flow rate of 7.5 mL min⁻¹ (residence time of around 28 min), an NO₃⁻ conversion of around 35% was registered with a corresponding N₂ selectivity of 30%.

Previous works point out the formation of higher NO₂⁻ concentrations for higher residence times in the reactor [27]. As reported in several papers, NO₃⁻ is reduced to NO₂⁻ on the transition metal centers of the catalyst to be further reduced to NH₄⁺ or N₂ on the noble metal centers [19,25,55,56]. Thus, higher liquid flow rates, and consequently shorter residence times of the reactants inside the reactor, lead to an accumulation of by-products that did not have time to contact with the catalyst during their passage through the reactor. However, during the present work, no significant increase in NO₂⁻ concentration was detected when higher inlet liquid flow rates were implemented, with the large capacity of the catalyst to reduce NO₂⁻ being significant.

Taking into account the obtained results and the aim of the present work, it is necessary to reach a compromise between high NO₃⁻ conversion efficiencies and selectivity to N₂. Thus, fixing the feed liquid flow rate at 10 mL min⁻¹ appears the preferred option for further optimization experiments.

Gas Flow Rates Optimization

Several experiments were performed varying H₂ and CO₂ flow rates in order to assess how the variation of these parameters would influence the conversion process and the respective selectivities for N₂ and NH₄⁺ formation. Figure 4 presents the obtained results regarding NO₃⁻ conversion and N₂ selectivity for the different gas flow rate combinations studied. The obtained results regarding NO₂⁻ and NH₄⁺ selectivity were included in the Supplementary Information (Figure S2).

One of the most important factors to control during catalytic reduction using H₂ as a reducing agent is the flow rate at which it is fed to the reactor, which strongly influences the reduction kinetics, the conversion efficiency, and the selectivity results [25]. Generally, higher H₂ flow rates are associated with the improvement of NO₃⁻ conversion, and the obtained results show that NO₃⁻ conversion only significantly increases for an H₂ flow rate of 100 Ncm³ min⁻¹. For lower H₂ flow rates (<100 Ncm³ min⁻¹), the conversion results vary from 11 to 13% for all H₂:CO₂ ratios, which are not significant variations. N₂ selectivity values varied more remarkably for different H₂:CO₂ combinations. Typically, in the studies performed in the absence of the CO₂ buffer, a higher H₂ flow rate is associated with a higher NO₃⁻ conversion and higher NO₂⁻ and NH₄⁺ selectivity. An increase in the H₂ flow rate provides an excess of adsorption of H₂ on the monometallic Pd sites, which may help the reaction of some of the intermediate by-products, such as NO₂⁻ and NO_x, with an atomic hydrogen to form N-H bonds and, this way, promote a higher NH₄⁺ formation [25,27,47,56,57,58,59].

As previously mentioned, CO₂ was used as a buffer to help control the selectivity of the process by maintaining the pH of the reaction medium, which will strongly influence the selectivity of the catalytic reduction process [25]. For all the experiments using CO₂ as a buffer, the pH value was kept at values close to 5, even when using lower CO₂ flow rates. The obtained results show that N₂ selectivity increased for higher CO₂ flow rates for each of the studied H₂ flow rates. Previous studies focused on the evaluation of NO₃⁻ abatement in a fixed-bed reactor showed that both N₂ selectivity and the catalyst activity significantly improved when CO₂ was used to assist in pH adjustment of the reaction medium [17,25,60]. The most efficient results were obtained for a total flow rate of 200 Ncm³ min⁻¹ with a ratio of 1:1 for each gas (100 Ncm³ min⁻¹ of each gas). For this combination, an NO₃⁻ conversion of 20% was obtained with a corresponding N₂ selectivity of 50%.

Additional experiments were conducted without the use of CO₂ as a pH buffer, at 100 and 200 Ncm³ min⁻¹ H₂ flow rate. The latter allows direct comparison with the best performing conditions (200 Ncm³ min⁻¹ 1:1 H₂:CO₂) as the same hydrodynamic conditions are maintained. The obtained results (present in Supplementary Information) verified that the absence of CO₂ leads to a slight increase in the solution pH from 5 to 8. Generally, it was possible to verify a large NO₂⁻ accumulation during reaction for both experiments, although this accumulation was more evident for the experiments with higher H₂ flow rate (200 Ncm³ min⁻¹). Also, NO₃⁻ conversion results were not as efficient as in the case where CO₂ was used, showing a decrease to around 8% and 12% for the experiments with 100 Ncm³ min⁻¹ and 200 Ncm³ min⁻¹ of H₂, respectively. This was previously reported by Mendow et al. [17], who related the increase in the pH of the reaction medium with a higher conversion of NO₃⁻ into NO₂. The reduction of NO₃⁻ by itself leads to an increase in the solution pH, due to their replacement by hydroxide ions (OH⁻), to keep the electroneutrality of the aqueous phase. In the absence of a pH buffer, the solution pH will increase, being responsible for negatively charging the surface of the used catalyst and, consequently, working against NO₃⁻ and NO₂⁻ adsorption due to electrostatic repulsions [15,16,17,18,19,20,21,22]. This will lead to a decrease in NO₃⁻ removal and an accumulation of NO₂⁻ in solution. Thus, it is evident, through the obtained results, that using CO₂ as a buffer plays a key role in the catalyst selectivity, allowing a good compromise between higher NO₃⁻ conversions (mainly achieved with high H₂ flow) while achieving higher selectivities for N₂ formation. Considering these results, the ideal ratio of 1:1 was tested for higher H₂ and CO₂ flow rates (125:125 and 150:150 Ncm³ min⁻¹) to understand whether it would be advantageous to increase the gas flow rates. The obtained results are presented in Figure 5.

The obtained results show that there is not a significant advantage in increasing the total gas flow rate beyond 200 Ncm³ min⁻¹, since the conversion and selectivity values remained practically unchanged, excluding the NO₂⁻ selectivities registered. As previously mentioned, an increase in the H₂ flow rate will result in an over-reduction phenomenon that will promote a higher NH₄⁺ formation; however, the obtained results show that for higher H₂ flow rates, there is a higher accumulation of NO₂⁻.

Previous studies focused on the study of the influence of gas flow rates on the residence time distribution of the reactants in the reactor show that an abrupt increase in the gas flow rate fed to the reactor could lead to significant changes in the conversion values and selectivity of the process and even lead to the accumulation of reaction by-products, as proven by the results present in Figure 5. This can happen due to a promotion of a higher recirculation of the formed vortices inside the catalyst channels that are generated by the feed of higher gas flow rates to the reactor, which can lead to an increase in the mass transfer limitations in the three-phase system and promoting the release of NO₂⁻ before it can react at the Pd centers to be converted to NH₄⁺ or N₂ [17].

The influence of the total gas flow rate (H₂ + CO₂) fed to the reactor in the residence time of the reactants was also studied under the scope of the present work at selected total gas flow rates: 100, 200, and 300 Ncm³ min⁻¹. The obtained results verified that there are no significant changes recorded in the residence time distributions, meaning that a sharp increase in the total gas flow rate should not cause changes in the contact time of the liquid phase reactants with the catalyst. Therefore, all changes recorded throughout these experiments are directly related to the variation that each parameter has at the reaction level.

Coupled with these RTD tests, bubble size measurement experiments were also performed to verify if higher or smaller gas flow rates could influence the bubble size that passes through the system. Figure 6 presents the images collected during these experiments.

Generally, for blank experiments (without catalyst present in the reactor), the average bubble size remained constant (between 0.2 and 0.6 cm) for different total gas flow rates. When a catalyst is added to the column, the average bubble size, after contact with the macrostructured catalyst, increases to an average of about 1.0 to 1.5 cm. This naturally happens after gas diffusion through the channels of the macrostructured catalysts, meaning that the gas bubbles fully contact with the catalyst channels promoting the interaction between the three phases of the system (G-L-S). For this type of system, the hydrodynamic regime of interest, to promote an enhanced mass transfer, is the Taylor flow regime characterized by the formation of a segmented flow, with well-defined gas bubbles and liquid slugs crossing the channels [61,62,63,64].

Influence of the Catalyst Amount

The last step of the optimization experiments consisted of studying the influence of the catalyst amount during NO₃⁻ catalytic reduction. For this purpose, catalytic reduction experiments with one, two, and three pieces of catalyst in the monolithic form (all synthesized through the same methodology described in the Section 2) arranged in series in the column were performed. The obtained results are presented in Figure 7.

Typically, it is known that a higher content of catalyst results in enhanced NO₃⁻ conversion [25,27]. The conversion efficiency obtained with two macrostructured catalysts doubled when compared with the single catalyst reaction (20% vs. 41%). However, the process selectivity was strongly compromised with the addition of catalysts to the reaction system. N₂ selectivity decreased by half for the reaction performed with two macrostructured catalysts (from 50% registered with one catalyst to 23% registered with two catalysts). When a third catalyst was added, N₂ selectivity remained constant despite an increase in the conversion value (conversion efficiency of 51%). Previous studies reported an increase in NH₄⁺ formation during reactions with higher catalyst loading [50,52,54,55]. This increase in NH₄⁺ concentration may be mainly associated with the increase in the number of active sites on the catalytic system and the consequent availability of more centers for H₂ chemisorption being responsible for an over-reduction phenomenon that will promote a higher NH₄⁺ formation [25].

An important detail to note with the increase of the catalyst amount during the reaction is the decrease in NO₂⁻ selectivity for higher catalysts loading (from around 8% with two structured catalysts to 3% with three catalysts). This phenomenon was previously reported in other studies focused on the study of the influence of catalyst quantity, and this decrease was attributed to the existence of a greater Pd availability to perform NO₂⁻ reduction and, thus, resulting in a less accumulation of NO₂⁻ [25,27,54].

After reaction, catalysts were characterized, mainly in terms of metal leaching, which would be one of the main reasons related to the loss of activity/deactivation of the catalyst. ICP-OES analyses were performed for the liquid solutions collected after reaction. The obtained results did not show any signs of leaching of any of the metals present in the catalyst structure. In previous studies, TEM images of used bimetallic catalysts were obtained, and no significant changes in the catalyst structure were recorded, including no signs of nanoparticles sintering [22].

3.3. Assessment of Factors Affecting NO₃⁻ Conversion and N₂ Formation

To assess the conditioning factors that affect NO₃⁻ conversion and selectivities, a deviation coefficient d was calculated for each of the previously presented experiments. This coefficient corresponds to an apparent deviation of the results from the expected corresponding powder catalyst performance due to the characteristics of the continuous reaction system. Two different coefficients were calculated: one considering NO₃⁻ conversion ($d_{{NO}_3^{-}}$), in each of the previous experiments, to understand how the induced changes in the reaction system affected the conversion, and other coefficient focused on N₂ formation ($d_{N_2}$), which will allow evaluating how the changes in the reaction system will contribute to increase or decrease the formation of N₂, since the major goal of catalytic reaction processes is the conversion of NO₃⁻ into less toxic compounds (namely N₂).

These results allow performing a systematic study where it is possible to identify which of the variants has the most significant impact on the results achieved. The closer the deviation coefficient ($d_{{NO}_3^{-}}$ or $d_{N_2}$) value is to 1, the closer the reaction system will be to what are the ideal reaction conditions (identical to what was observed in the reactions with the powder catalyst in a stirred tank system assuming no mass transfer limitations). The obtained results are organized in different graphs grouped in Figure 8. Figure 8A–D present the results related with the deviation coefficient, considering NO₃⁻ conversion, as a function of each of the studied variables (inlet liquid flow rate, H₂ flow rate, CO₂ flow rate, and number of macrostructured catalysts, respectively). Figure 8E–H present the results related with the deviation coefficient, considering the selectivity to N₂, as a function of the studied variables. Some of the results achieved with the powder catalyst tested in the batch reactor can be found in the Supplementary Information.

An overall analysis of the obtained results allows us to verify that the different variations in each of the parameters lead to a calculated coefficient value lower than one meaning the existence of mass transfer limitations when applying the structured catalysts in the continuous catalytic system. It is important to highlight that this coefficient, when calculated taking into account the results obtained in batch tests using the homologous catalyst in powder form, will indicate deviations from the ideal mass transfer regime (assuming no limitations) referring to external factors (such as the change of parameters in the reaction) and internal factors (related to the diffusion of reactants through the coating layer). The comparison of the deviation coefficients between each other may allow to take some conclusions regarding the system performance in different conditions (by changing the reaction parameters).

Generally, the obtained results prove that changing the flow rate of the liquid fed to the reactor and increasing the catalyst amount in the system allowed to obtain the most significant differences and the more linear correlations in terms of conversion and selectivity.

The increase of the liquid flow rate corresponds to a decrease in the residence time in the reactor and in the contact time of reactants with the solid phase catalyst, and the latter may be one of the factors that most influences the conversion and selectivity results. However, changes on the residence time of the liquid phase will have consequences on the mixing of the three phases inside the catalyst channels, causing variations that can be quite significant for the reaction system. For instance, a lower liquid flow rate will promote a higher residence time, which in turn may lead to further saturation of the liquid phase with H₂ and affect the mass transfer from the liquid and gaseous phases to the solid catalyst. This, in turn, promotes the highest conversion of NO₃⁻ and greater formation of NH₄⁺, due to the higher availability of H₂ in the system (results presented in Figure 8A,E).

The results corresponding to the different gas flow rates (present in Figure 8B,C,F,G) show poor linear correlations for all the combinations studied, meaning that these changes have not shown the most significant contribution to the conversion and selectivity of the process. However, the results previously presented and discussed in Figure 4 allowed to highlight an optimal combination of the two gases to obtain the maximum conversion in the studied system and the best selectivity to N₂ (100 H₂:100 CO₂ cm³ min⁻¹). In terms of NO₃⁻ conversion, the variation in H₂ flow rate proved to be the most relevant factor; however, it is important to notice that, for lower H₂ flow rates (50 and 75 cm³ min⁻¹), the deviation coefficient d_NO3⁻ slightly increases for higher CO₂ flow rate values. These results may be related to the influence of the total gas flow rate on the hydrodynamic operation of the reactor, in particular inside the macro-structured catalyst. As previously mentioned in Section Gas Flow Rates Optimization, higher gas flow rates do not significantly impact the residence time distribution of the liquid in the reactor; however, they may significantly impact the mass transfer within the macrostructured catalyst channels. Several aspects of the Taylor flow regime inside the macrostructured catalyst channels are significantly affected by the variations in the operating conditions of the system, namely the liquid and gas flow rates. There are different parameters that are responsible for the mass transfer rate inside the structured catalyst channels, such as liquid slug length, shape, and velocity, the gas bubble length, shape, and velocity (which is conditioned by the liquid velocity), the gas hold-up, pressure drop, recirculation of the liquid and gas inside the slugs, and the film thickness between the gas bubbles and the catalyst wall [65,66,67,68]. For the specific case of the present work, the mass transfer inside the monolith channels may be promoted by a higher total gas flow rate, which is why, even for lower H₂ flow rates, there is a slight increase in NO₃⁻ conversion when the CO₂ flow rate increases. Higher total gas flow rates will promote a greater formation of recirculation vortexes inside the liquid slugs, promoting a higher mixture of the gas and liquid phase that will be in close contact with the catalyst channels. A similar effect is seen with changing liquid flow rates, affecting the formation of the liquid slug vortices inside the catalyst, and hence the changes in the corresponding d_NO3⁻. Previous studies have reported that the decrease on the liquid velocity inside the monolith channels would enhance the mass transference through the thin film of liquid joining the three phases participating in the reaction (G-L-S) since this film becomes thinner [69]. In what concerns to the gas to liquid mass transfer (G-L), previous studies reported that this is not necessarily a function of the bubble length but will strongly depend on the liquid slug length, and typically the thinner the slug length, the higher the mass transference between both phases and, consequently, between the third phase involved in the reaction (as previously mentioned) [69].

Higher CO₂ flow rates proved to have a more significant impact on the decrease of undesirable by-product formation (Figure 8G).

The amount of catalyst present in the reactor seems to be one of the most relevant factors to improve NO₃⁻ catalytic conversion (Figure 8D); however, it was revealed to be quite detrimental regarding the formation of sub-products, mainly NH₄⁺ formation (Figure 7 and Figure 8H). It is evident that a higher amount of catalyst will positively contribute to the reaction extent since it adds active centers to the system where the reaction can occur. On the other hand, as more catalysts are added to the system, the contact time of the reactants with the catalyst increases. Figure 9 presents the results obtained for the deviation coefficient considering the contact time with the catalyst, instead of the residence time in the reactor in Equation (3) (t_r).

As previously discussed, the main factor influencing the increase of NO₃⁻ conversion in a continuous process using macrostructured catalysts is the intimate contact between the three phases involved in the reaction. Thus, increasing the contact time between reactants and solid catalysts by adding three pieces of the catalyst in the monolithic form to the system (maximum held by the reactor) should result in a linear increase in d_NO3⁻. However, the deviation coefficient only significantly increased for the experiment with two monoliths in the reactor, whereas when the third monolith was added, the coefficient remained constant. On one hand, it would be possible that the increase in the number of active centers was not contributing to the reaction kinetics after a certain amount. However, the linear increase in Figure 8D suggests otherwise. Thus, this is likely related to the formation of larger gas bubbles after the first monolith in series, since the process of formation of these bubbles is then not present between the second and third monoliths, which in turn affects the mass transfer between phases inside the monoliths by altering the hydrodynamic regime.

3.4. Final Remarks

The results previously presented allow to conclude that there is a significant influence of the reaction conditions on the hydrodynamic behavior of the reactor. Generally, there is an ideal liquid and total gas flow rate, which results in the most promising NO₃⁻ conversion and N₂ selectivity results. The reaction conditions that have a higher influence on the catalytic results are those with a more significant impact on the residence time of the pollutants inside the reactor and that promote an extensive contact between the three phases (three monoliths, liquid flow rate of 10 mL min⁻¹, and a total gas flow rate of 200 Ncm³ min⁻¹ (1:1 H₂:CO₂)). The obtained results indicate that there are no significant changes in the selectivity when varying the reaction conditions. Thus, the system is optimized to obtain results that allow the best compromise between obtaining an efficient NO₃⁻ conversion and good selectivity to N₂ formation. Therefore, the current system would only be efficient (and optimizable) in a specific range of conditions that would allow conversion of nitrate below the 50 mg L⁻¹ limit while maintaining NH₄⁺ concentration below the legal limit. Nevertheless, it is also evident that there are still further opportunities to improve selectivity to N₂ by optimizing the catalysts’ formulation by controlling the formation of structures that would promote its formation during reaction. As it was previously shown with the obtained results, one of the parameters that most influenced the catalytic results obtained with the designed continuous catalytic system was the catalyst concentration inside the reactor; this is an additional parameter where there must be additional optimization either in the amount of powder catalyst possible to incorporate in the macroscopic structure, or by the availability of active centers to participate in reaction. At this point, in the present work, an optimized continuous catalytic system to be implemented for NO₃⁻ conversion is presented; however, additional studies need to be performed (mainly concerning catalyst design) in order to implement this type of system at a full scale.

4. Conclusions

The present work aims to study and optimize the reaction conditions applied in a continuous catalytic system for the selective conversion of NO₃⁻ to N₂. Several factors, such as the liquid flow rate fed to the system, the gas flow rate, the combination of both gases (H₂ and CO₂), and the amount of catalyst present in the reaction, strongly condition the conversion process and the respective selectivity. Therefore, this work allowed identifying how these factors interfere in the reaction system and, in this way, perform its optimization towards the best compromise between an efficient NO₃⁻ conversion and good selectivity for the formation of less toxic by-products (N₂).

Generally, the inlet liquid flow rate and the catalyst amount inside the reactor were the parameters that most influenced the conversion and selectivities due to the important role these parameters play in defining the hydrodynamic conditions of the reactor. Higher loading of catalyst (higher number of macrostructured catalysts in the reactor) promoted higher contact time between the three phases (G-L-S), leading to more efficient conversion results; however, N₂ selectivity was somewhat compromised during these experiments. In this case, the optimization of gas flow rate and the ratio of both gases play a major role in the selectivity control. In addition, the optimized gas flow rate and ratio also correspond to the best conditions that promote a higher mixture of gas and liquid phases inside the macrostructured catalyst channels, thus promoting greater contact between the three phases. The most efficient gas combination was a total gas flow rate of 200 Ncm³ min⁻¹ (1:1 of each gas), which proved to be preponderant in promoting a greater mixture of the various phases involved in the reaction.

This way, the developed work allowed us to study, in detail, the factors that most contribute to the efficiency of the reaction conditions, allowing its optimization to achieve a more efficient catalytic system for selective NO₃⁻ conversion.