Hydroxylamine intermediate governs selectivity in nitrite hydrogenation on Pd-based catalysts for sustainable water treatment

Abstract

Catalytic hydrogenation of nitrate (NO₃⁻) and nitrite (NO₂⁻) is a promising route for drinking water purification and rebalancing the global nitrogen cycle. Recently, hydroxylamine (NH₂OH) was detected as a persistent reaction intermediate although ammonium (NH₄⁺) and dinitrogen (N₂) were assumed to be the only significant reaction products for several decades. In this work, we systematically investigate NO₂⁻ hydrogenation over Pd/Al₂O₃ and SnPd/Al₂O₃ while explicitly quantifying NH₂OH under various conditions, including changes in H₂ partial pressure, the initial NO₂⁻ concentration, and reaction temperature, and through co-feeding of NH₂OH. We reveal that NH₄⁺ selectivity depends strongly on the NO₂⁻ conversion level, reflecting shifts in surface coverages as the reaction progresses. Suppression of both NH₂OH and NH₄⁺ formation is only achievable under H₂-deficient conditions, though this comes at the expense of lower overall hydrogenation activity. Elevated temperatures enhance NH₂OH decomposition and thereby promote NH₄⁺ formation, while leaving N₂ selectivity largely unaffected. Co-feeding experiments further show that externally introduced NH₂OH does not influence the NO₂⁻ hydrogenation rate. We critically reviewed prior mechanistic studies on NO₂⁻ hydrogenation and propose a refined Langmuir–Hinshelwood scheme that explicitly incorporates NH₂OH as a desorbed intermediate. This work highlights the importance of NH₂OH in the reaction network and underscores the need to include it in the assessment of the reaction selectivity.

---

1 Introduction

Nitrate (NO₃⁻) and nitrite (NO₂⁻) are water contaminants that pose a serious threat to both human health and aquatic ecosystems if not properly controlled. Although these species are natural components of the biogeochemical nitrogen cycle, the excessive use of nitrogen-based fertilizers has severely disrupted this balance in recent decades.^1–4 As a consequence, nitrogen nutrient levels have risen in inland, coastal, and marine waters, driving eutrophication.⁵ Eutrophication negatively impacts entire ecosystems by reducing biodiversity of animals and plants. In addition, elevated nitrate and nitrite concentrations can directly affect human health. When present in drinking water, these species may cause methemoglobinemia—commonly known as “blue baby syndrome”⁶ and have, therefore, strict limits (50 and 5 mg L⁻¹, respectively).⁷

Catalytic nitrate and nitrite reduction has emerged as a promising technology to convert these species into harmless dinitrogen (N₂).⁸ In contrast to alternative remediation methods such as reverse osmosis, ion exchange, or electrodialysis, catalytic reduction degrades the contaminants rather than concentrating them into waste brines.⁸ Nitrate is first reduced to nitrite, which is then further hydrogenated to N₂. In this reaction network, the hydrogenation of nitrate to nitrite is the rate determining step (RDS) and requires a bimetallic catalyst (e.g. Sn–Pd, Cu–Pd, or In–Pd), while the subsequent nitrite hydrogenation can be catalyzed by Pd alone.^9–12

NO₃⁻ + H₂ → NO₂⁻ + H₂O (1)

2 NO₂⁻ + 3H₂ + 2H⁺ → N₂ + 4H₂O (2)

The main challenge for application remains the formation of ammonium (NH₄⁺), which follows even stricter regulations (0.5 mg L⁻¹).¹³

NO₂⁻ + 3H₂ + 2H⁺ → NH₄⁺ + 2H₂O (3)

While the formation of NH₂OH in electro- and photocatalytic NO₃⁻ and NO₂⁻ reduction has been reported in the past,^14–16 its presence in thermo-catalytic conversion was only reported very recently. The results showed that hydroxylamine (NH₂OH) is an omnipresent and long-lasting reaction intermediate under various reaction conditions on several different catalysts (eqn (4)).¹⁷ Until then, a pervasive assumption in the field was that N₂ and ammonium (NH₄⁺) were the only relevant reaction products of thermocatalytic NO₃⁻ and NO₂⁻ hydrogenation under relevant reaction conditions for drinking water purification.^{8–12,18–22}

NO₂⁻ + 2H₂ + H⁺ → NH₂OH + H₂O (4)

Hydroxylamine is highly toxic and harmful for water bodies²³ and therefore represents a major challenge for the application of NO₃⁻ and NO₂⁻ as water purification technology. Upon extension of the reaction time, NH₂OH can be hydrogenated to NH₄⁺ (eqn (5)) or disproportionate into NH₄⁺ and nitrous oxide (N₂O, eqn (6)). The latter is quickly further hydrogenated into N₂ (eqn (7)).²⁴

NH₂OH + H₂ + H⁺ → NH₄⁺ + H₂O (5)

4 NH₂OH + 2H⁺ → 2 NH₄⁺ + N₂O + 3H₂O (6)

N₂O + H₂ → N₂ + H₂O (7)

Thus, NH₂OH is at least 50% decomposed into undesired NH₄⁺. In previous research it was shown that NH₂OH decomposition proceeds exclusively via the hydrogenation pathway in the absence of NO₃⁻ and NO₂⁻ and in the presence of H₂ over a Pd/Al₂O₃ catalyst. In contrast, in the presence of residual NO₃⁻, NO₂⁻ or their remaining surface species adsorbed on the catalyst, the hydrogenation and the disproportionation pathway contribute to the NH₂OH decomposition.¹⁷

In this study, NH₂OH is used as a reaction marker to unravel the interplay between the reaction rate of NO₂⁻ hydrogenation and the selectivity to NH₂OH, NH₄⁺ and N₂. To do this we investigated detailed reaction kinetics for the NO₂⁻ hydrogenation on Pd/Al₂O₃ and SnPd/Al₂O₃ catalysts at different H₂ partial pressures, initial NO₂⁻ concentrations, and temperatures while monitoring the NH₂OH and NH₄⁺ concentrations to deliver a complete picture of the selectivity patterns and reaction order dependencies. The results drastically change the prevailing view that ammonia and nitrogen are the only two reaction products while showcasing the complex interdependence between hydrogen, nitrite, and hydroxylamine ratios in the observed kinetics and product distribution.

2 Materials and methods

2.1 Materials

Pd/Al₂O₃ and Tin(II) chloride (SnCl₂) were purchased from Fisher Scientific and potassium nitrate (KNO₃), potassium nitrite (KNO₂), benzaldehyde (≥99.5%), benzaldehyde oxime (≥96.5%) and hydroxylamine (50 wt% in H₂O) were purchased from Sigma Aldrich.

2.2 Catalyst preparation

Pd/Al₂O₃ was dried before use and sieved to a support particle size <25 µm. SnPd/Al₂O₃ was prepared from the same Pd/Al₂O₃ (<25 µm) via controlled surface deposition.^25,26 Therefore, Pd/Al₂O₃ was suspended in water and exposed to a continuous gas flow of 80 : 10 : 10 mL min⁻¹ H₂ : He : CO₂ to reduce the catalyst for 30 min at RT. SnCl₂ was dissolved in degassed Milli-Q water and a volume containing Sn in amounts corresponding to 40% of the monolayer capacity of Pd based on CO-chemisorption was added to the catalyst suspension. The suspension is stirred for another 30 min and subsequently filtered to recover the catalyst powder. The filter cake was dried in a vacuum at 70 °C followed by calcination (500 °C in 50 mL min⁻¹ air, 5 °C min⁻¹, 5 h) and reduction (500 °C in 30 mL min⁻¹ H₂ and N₂ each, 5 °C min⁻¹, 5 h) in a tube oven. The catalyst is stored in air at ambient temperature.

2.3 Catalyst characterization

The metal loadings were determined with X-ray fluorescence (XRF). For the measurements a Bruker S8 Tiger Series 1 4 kW X-ray spectrometer was used. The samples were measured as powder, in a sample cup with a Mylar film, in a helium environment. The concentrations of each element were calculated from the Kα or Lα lines of all elements, using the internal database. The metal dispersion and metal particle sizes were determined by CO-chemisorption. Therefore, the samples were reduced in H₂ for 1 h and purged with He for 20 min followed by CO-pulsing (all at RT, Micromeritics Chemisorb 2750). A Pd : CO 1 : 1 ratio and the geometry factor for a hemisphere (6) are assumed. The Brunauer–Emmet–Teller (BET) surface area was determined via N₂ physisorption at −196 °C after outgassing for 24 h at 300 °C (Micromeritics Tristar 3000). The characterization results are summarized in Table 1.

Table 1 Catalyst material properties based on N₂ physisorption, XRF and CO-chemisorption

Catalyst	BET surface area	Pd loading	Sn loading	Average Pd particle size
Pd/Al2O3	110 m^2 g^−1	8.1 wt%	—	10.3 nm
SnPd/Al2O3	—	8.4 wt%	0.7 wt%	—

---

2.4 Catalytic testing

In a typical experiment, Pd/Al₂O₃ or SnPd/Al₂O₃ was suspended in 297 mL of deionized water in a 1 L baffled glass reactor equipped with a four-blade magnetic stirrer. The catalyst was reduced in situ under continuous gas flow (80 : 10 : 10 mL min⁻¹ H₂ : CO₂ : He) for 1 h at 600 rpm. After the activation period, 3 mL of a KNO₂ stock solution was injected to reach the desired initial concentration and to initiate the reaction. Liquid samples were withdrawn periodically, passed through a 0.2 µm syringe filters to remove catalyst particles to terminate the reaction in the sample, and subsequently analyzed. NO₃⁻, NO₂⁻ and NH₄⁺ concentrations were analyzed via ion chromatography (IC, a Dionex ICS-3000 with electronic suppression, Thermo Fisher AS19 and CS12 columns with 20 mmol L⁻¹ potassium hydroxide (KOH) and 20 mmol L⁻¹ methane sulfonic acid (MSA) as eluents and an Automate 2000 autosampler). For NH₂OH determination, 0.5 µL benzaldehyde was added right after sampling to derivatize NH₂OH to benzaldehyde oxime.¹⁷ Benzaldehyde oxime was quantified via HPLC (Shimadzu HPLC10AVP with an autosampler, a C18 hypersil gold column, a MeOH : H₂O (30 : 70 V% eluent, and a UV-vis detector at 248 nm). IC and HPLC analysis were conducted from the same sample vial and RT refers to 22 °C. In some experiments the analysis of the gas phase was attempted but ambient N₂ could not sufficiently be suppressed to ensure satisfactory sensitivity. In these approaches NO, NO₂ or N₂O could be detected. The selectivity was calculated as or and the remaining fraction to 1 considered as N₂.

3 Results and discussion

3.1 NO₂⁻ hydrogenation on Pd/Al₂O₃ at RT

The NO₂⁻ hydrogenation was conducted in a concentration window of 0.26–6.5 mmol L⁻¹ at 0.8 bar H₂ and at H₂ partial pressures between 0.05 and 0.8 bar for 0.8 mmol L⁻¹ NO₂⁻ in the absence of internal (SI, Fig. 1) and external mass transport limitations.^27,28 Initial rate constants were obtained from concentration–time profiles (SI, Fig. 2) and plotted in log–log form (Fig. 1) to determine the reaction orders. At RT, a reaction order close to zero with respect to NO₂⁻ and ∼0.2 with respect to H₂ was observed, confirming previous results from our group.^27,28

Fig. 1 Effect of NO₂⁻ (A) and H₂ concentration (B) on the NO₂⁻ hydrogenation rate on Pd/Al₂O₃. Reaction conditions: 300 mL H₂O, 0.3–6.5 mmol L⁻¹ KNO₂, 10 mg Pd/Al₂O₃, 10 mL min⁻¹ CO₂, 5–80 mL min⁻¹ H₂ and He to balance to 100 mL min⁻¹ total flow, RT, 600 rpm.

Thus, the NO₂⁻ hydrogenation activity at room temperature is independent of the NO₂⁻ concentration, while slightly accelerated by increasing H₂ partial pressure. The low sensitivity towards both NO₂⁻ and H₂ concentrations suggests that the catalyst surface is predominantly covered by intermediates prior to the rate-determining step, leaving few sites available for additional NO₂⁻ or H₂ adsorption. This interpretation is consistent with previous Attenuated Total Reflection Infrared Spectroscopy (ATR-IR) studies as discussed later.^29–31

To evaluate the selectivity of the reaction, NH₄⁺ and NH₂OH concentrations were measured delivering a typical reaction profile as shown in Fig. 2A. Initially both NH₄⁺ and NH₂OH concentrations increase with decreasing NO₂⁻ concentration. While the NH₄⁺ concentration consistently increases, the NH₂OH concentration passes through a maximum and subsequently decreases illustrating its intermediary nature. The remaining part of the converted NO₂⁻ is assumed to be N_2. The NH₄⁺ selectivity ranges from <1% to 10%, depending on the H₂ partial pressure and the NO₂⁻ conversion level. The NH₄⁺ concentration increases at a faster rate when NO₂⁻ is near full conversion, reflecting shifts in surface coverages during the reaction. When plotted as a function of NO₂⁻ conversion, the NH₄⁺ selectivity follows a U-shaped dependence (Fig. 2B). Before NO₂⁻ injection (i.e., at the start of the reaction), the catalyst is activated in H₂-saturated water, giving high surface H-coverage and a high H : N ratio, which promotes NH₄⁺ formation. As the reaction proceeds, H-coverage decreases while N-coverage increases, favoring N–N coupling and thereby enhancing N₂ selectivity. At high NO₂⁻ conversion, N-species are depleted, the H : N ratio shifts back toward higher H-coverage, and NH₄⁺ selectivity increases again. Thus, the time required to achieve a pseudo steady state is an intrinsic disadvantage of batch experiments due to their transient nature. The competitive adsorption of NO₂⁻ and H₂ was demonstrated by Postma et al.³² and Huang et al.²⁸ but is often not considered in batch experiments. This highlights the importance of monitoring NH₄⁺ selectivity over the entire conversion range, since relying on initial selectivity is likely biased by the initial high H-coverage. While NH₄⁺ formation generally decreases with decreasing H₂ partial pressure, substantial NH₄⁺ suppression occurs only under strongly H₂-deficient conditions (0.05 bar, Fig. 2B), which also reduces the NO₂⁻ hydrogenation rate. This dependence reflects the higher H₂ demand for NH₄⁺ formation relative to NO₂⁻ hydrogenation (3 vs. 1.5 H₂ per NO₂⁻, see eqn (2) and (3)).

Fig. 2 Typical concentration profile of NO₂⁻ hydrogenation (A), NH₄⁺ selectivity as a function of NO₂⁻ conversion (B) and NH₂OH concentration as a function of time in which arrows indicate full NO₂⁻ conversion (C). Reaction conditions: 300 mL H₂O, 10 mg Pd/Al₂O₃, 0.8 mmol L⁻¹ KNO₂, 10 mL min⁻¹ CO₂, 5–80 mL min⁻¹ H₂ and He to balance to 100 mL min⁻¹ total flow, RT, 600 rpm.

Despite the intermediary character of NH₂OH, monitoring NH₂OH is crucial as its maximum concentration at RT is typically observed around full NO₂⁻ conversion (Fig. 1C), thus representing a point where the reaction could otherwise be mistakenly considered complete. While the NH₂OH selectivity ranges from 3–8% at H₂ partial pressures ranging from 0.2 to 0.8 bar (SI, Fig. 3), the NH₂OH production is minimized when using very low H₂ partial pressure (0.05 bar, Fig. 2C). This drastic change in selectivity, however, comes at the expense of reduced NO₂⁻ hydrogenation. This behavior is consistent with the higher H₂ demand for NH₂OH formation compared to N₂ formation (2 vs. 1.5H₂ per NO₂⁻, see eqn (2) and (4)).

Varying the initial NO₂⁻ concentration had little influence on NH₄⁺ and NH₂OH formation (Fig. 3A and B, respectively) which is consistent with an apparent reaction order of zero in NO₂⁻, implying that the surface coverages are only mildly influenced by variation of the NO₂⁻ concentration. At the highest initial NO₂⁻ concentration, the formation of NH₄⁺ and NH₂OH appears to be slightly reduced. This can be rationalized by increased N-coverage on the catalyst surface resulting in a higher N–N coupling probability which is in line with findings of Xu et al.²⁷

Fig. 3 NH₄⁺ concentration (A) and NH₂OH concentration (B) at RT as a function of time at different initial NO₂⁻ concentrations. Arrows indicate the time of full NO₂⁻ conversion, if reached. Reaction conditions: 300 mL H₂O, 10 mg Pd/Al₂O₃, 80 : 10 : 10 mL min⁻¹ H₂ : CO₂ : He, RT, 600 rpm.

3.2 NO₂⁻ hydrogenation on Pd/Al₂O₃ at 40 °C

The same set of experiments was performed at 40 °C to investigate the effect of reaction temperature. The H₂ reaction order remained ∼0.2 while the NO₂⁻ hydrogenation activity is more sensitive to changes in the initial NO₂⁻ concentration at 40 °C illustrated by a positive reaction order of 0.24 below 0.8 mmol L⁻¹ NO₂⁻ and a negative reaction order of −0.47 above 0.8 mmol L⁻¹ (SI, Fig. 4).

A typical concentration profile at 40 °C is shown in the SI, Fig. 5. At 40 °C, the NH₄⁺ selectivity increases from 3–8% (RT, Fig. 2B) to 10–15% or even to 25% (0.6 bar, Fig. 4A) and the U-shape is less pronounced, or at 0.6 bar even disappeared entirely. In contrast, the maximum NH₂OH concentration at 40 °C decreased by ∼50% compared to RT and the maximum NH₂OH concentration is no longer observed near full NO₂⁻ conversion (Fig. 4B). Instead, the NH₂OH concentrations declined already before completion of NO₂⁻ hydrogenation. This indicates that elevated temperatures favor NH₂OH decomposition, lowering the measurable bulk NH₂OH concentration.

Fig. 4 NH₄⁺ selectivity as a function of NO₂⁻ conversion (A), the NH₂OH concentration profile (B) both at 40 °C and selectivity distribution at ∼50% NO₂⁻ conversion at RT and 40 °C and at different H₂ partial pressures (C). Arrows indicate full NO₂⁻ conversion (B) or the difference in NH₄⁺ selectivity. Reaction conditions: 300 mL H₂O, 10 mg Pd/Al₂O₃, 0.8 mmol L⁻¹ KNO₂, 10 mL min⁻¹ CO₂, 5–80 mL min⁻¹ H₂ and He to balance to 100 mL min⁻¹ total flow, RT or 40 °C, 600 rpm.

As NH₂OH primarily decomposes to NH₄⁺, faster NH₂OH decomposition explains the higher NH₄⁺ selectivity at 40 °C. Notably, the N₂ selectivity, calculated as the remainder of NH₄⁺ and NH₂OH selectivity, is largely unchanged across most H₂ partial pressures at both RT and 40 °C (Fig. 4C). Thus, the additional NH₄⁺ at 40 °C can be attributed to accelerated NH₂OH decomposition. Only at 0.6 bar of hydrogen the nitrogen selectivity is substantially lower while NH₄⁺ selectivity reaches a higher value, suggesting increased NH₄⁺ formation from NO₂⁻.

This kinetic study indicates that the formation of NH₄⁺ can be suppressed under H₂-deficient conditions, albeit at the expense of lower overall NO₂⁻ hydrogenation rates. The selectivity toward NH₂OH depends on both hydrogen partial pressure and temperature. At low H₂ partial pressures, the selectivity to hydroxylamine is low due to hydrogen starvation on the catalyst surface, which limits its formation rate. With increasing temperature, NH₂OH selectivity also declines because of its accelerated decomposition to ammonia. This intricate interplay between surface coverage and activation barriers for hydroxylamine formation and decomposition highlights the inherent difficulty in controlling the selectivity toward hydroxylamine and ammonia in this reaction.

3.3 NO₂⁻ hydrogenation on Pd/Al₂O₃ with NH₂OH co-feeding

To evaluate the effect of NH₂OH on the selectivity and the NO₂⁻ hydrogenation activity, NH₂OH was co-fed in concentrations ranging from 0.1–0.85 mmol L⁻¹ under H₂ deficient (0.05 bar) and H₂ rich (0.8 bar) conditions at both RT and 40 °C. As expected, higher NO₂⁻ hydrogenation rates were observed at elevated temperature and higher H₂ partial pressure. Surprisingly, the presence of co-fed NH₂OH did not affect the NO₂⁻ hydrogenation activity regardless of temperature or H₂ partial pressure as illustrated by apparent reaction orders close to zero (Fig. 5).

Fig. 5 Effect of NH₂OH on the NO₂⁻ hydrogenation rate on Pd/Al₂O₃ at different H₂ partial pressures and different temperatures. Reaction conditions: 0.8 mmol L⁻¹ KNO₂, 300 mL H₂O, 10 mg Pd/Al₂O₃, 10 mL min⁻¹ CO₂, 5 or 80 mL min⁻¹ H₂ and He to balance to 100 mL min⁻¹ total flow, RT or 40 °C, 600 rpm.

Reporting the NH₄⁺ and NH₂OH selectivity in NH₂OH co-feeding experiments is complex as co-fed NH₂OH represents an additional N-source resulting in an ill-defined denominator for the selectivity calculation. For instance, the NH₂OH selectivity could formally reach negative values and might therefore be misleading. To avoid this complexity, we report the NH₄⁺ and NH₂OH concentrations as a function of time. Figures with the same data as a function of NO₂⁻ conversion deliver the same trends and are shown in the SI, Fig. 6.

At RT, the NH₂OH concentration remained essentially constant at its initial co-feed level regardless of the H₂ partial pressure (Fig. 6A for 0.8 bar and SI, Fig. 6 for 0.05 bar). Because NH₂OH formation and decomposition occur simultaneously, only the net concentration change can be tracked. Thus, from the concentration profile alone, it cannot be determined whether NH₂OH formation and decomposition are in balance or whether no NH₂OH is formed in the first place. The NH₄⁺ concentrations, however, increase with increasing NH₂OH co-feed at RT as illustrated by the first two clustered columns in Fig. 6D. This suggests that NH₂OH is not a passive spectator but NH₂OH formation and decomposition are in balance with the NH₂OH decomposition contributing to increased NH₄⁺ formation. Under H₂-deficient conditions (0.05 bar), the NH₄⁺ concentrations were consistently lower, as expected and discussed in Sections 3.1 and 3.2.

Fig. 6 NH₂OH concentration as a function of time at different NH₂OH co-feed concentrations at RT or 40 °C and at 0.8 or 0.05 bar H₂ (as stated in the chart, A–C) and the NH₄⁺ concentration at different rounded NH₂OH co-feed concentrations after 60 min of reaction (D). Reaction conditions: 300 mL H₂O, 0.8 mmol L⁻¹ KNO₂, 10 mg Pd/Al₂O₃, 80 or 5 mL min⁻¹ H₂, 10 mL min⁻¹ CO₂ and He to make up to 100 mL min⁻¹, RT or 40 °C, 600 rpm.

At 40 °C and low H₂ partial pressure the NH₂OH concentrations decrease over time resulting in slightly higher NH₄⁺ concentrations (Fig. 6B and D (third clustered columns), respectively). The decrease in the NH₂OH concentration supports the earlier observations that the NH₂OH decomposition is promoted at elevated temperatures. At 40 °C and high H₂ partial pressure, substantially higher NH₄⁺ concentrations are obtained than under all other conditions which underlines faster NH₂OH decomposition at elevated temperature and the need for sufficient H₂. Interestingly, the NH₄⁺ concentrations show a non-linear behavior with respect to the NH₂OH co-feed levels under these conditions. While all co-feed experiments deliver higher NH₄⁺ concentrations than the baseline experiment without any NH₂OH co-feed, the increase in the NH₄⁺ concentration was more pronounced at lower co-feed levels (Fig. 6D, last clustered columns). A possible explanation could be that at high NH₂OH concentrations the surface is increasingly covered with NH₂OH species. This may result in lower H-coverage at high NH₂OH co-feed than at low NH₂OH co-feed leading to higher N : H ratios and therefore more N–N coupling and less NH₄⁺ formation. More broadly, this underscores the complex interplay of the different surface species and its relevance for the reaction mechanism.

As summarized in Table 2, these findings indicate that the NO₂⁻ hydrogenation activity is unaffected by NH₂OH co-feeding. Instead, the hydroxylamine co-feeding led to higher NH₄⁺ formation rates, indicating that this intermediate regulates the product selectivity. This increased NH₄⁺ formation correlates with faster NH₂OH decomposition, which is more pronounced at elevated temperatures.

Table 2 Overview of the main trends of the NH₂OH co-feeding experiments with respect to NO₂⁻ hydrogenation activity, NH₂OH concentration and NH₄⁺ formation under different reaction conditions

Effect of NH2OH co-feeding	RT		40 °C
	0.05 bar	0.8 bar	0.05 bar	0.8 bar
NO2^− hydrog. activity	Activity unchanged
[NH2OH]	Constant		Decreasing
NH4^+ formation	Increasing with [NH2OH]0		Increased at high [NH2OH]0	Increasing non-linearly

---

3.4 Reviewing the NO₂⁻ hydrogenation reaction mechanism on Pd/Al₂O₃

Only very few studies propose a reaction mechanism considering elementary steps of the surface reaction and, to the best of our knowledge, none of them considers the desorption of NH₂OH.^27,28,33 Thus, earlier published mechanistically relevant findings based on ATR-IR spectroscopy,^29–31 DFT^28,33–35 and reaction kinetics^27,28 are critically discussed and a revised reaction mechanism under consideration of desorbed NH₂OH is provided. Mechanistic studies on NO₂⁻ over metals other than Pd³⁶ or NO₃⁻ hydrogenation^37–41 were not considered as the initial hydrogenation from NO₃⁻ to NO₂⁻ is the RDS of this reaction and the surface coverages in NO₃⁻ and NO₂⁻ are thus not comparable.

3.4.1 Role of NH₂OH. Ebbesen et al. conducted ATR-IR studies on the decomposition of NH₂OH on Pd/Al₂O₃. They showed that in the absence of H₂ and NO₂⁻, NH₂OH decomposes into NH₄⁺ and another species that initially was assumed to be .³¹ In following studies using isotope labelling, Rao et al. showed that this species contains oxygen and was therefore defined as NO_xH_y which represents HNO*, NOH*, NHOH* and/or NH₂OH* (unpublished work, SI, Fig. 7). The three species were indistinguishable from one another. Subsequent titration of the species with H₂ in the ATR-IR cell showed that is converted into NH₄⁺.³¹ Notably, NO* was not observed in these experiments suggesting at least one irreversible reaction step from NO* towards on the Pd catalyst. In our latest work we reported NH₄⁺ selectivity <100% in batch experiments converting NH₂OH. This indicates that N₂ or N₂O can be formed from NH₂OH even in the absence of NO₂⁻ and H₂ via catalytic disproportionation.¹⁷ Generally, these results suggest that there is a direct pathway from NH₂OH towards NH₄⁺, while NH₂OH can also merge into the mechanistic pathway that leads to N₂, but NH₂OH cannot form NO*.

3.4.2 Langmuir–Hinshelwood or Eley–Ridel mechanism for N₂ formation. Xu et al. formulated a NO₂⁻ hydrogenation mechanism based on extensive kinetic studies.²⁷ All possible Langmuir–Hinshelwood (LH) equations were derived and compared with the apparent reaction orders in a broad concentration regime. Four of these LH type reaction mechanisms match the observed reaction orders. Two of these mechanisms proceed via a N* species (see SI, Fig. 8C and D) and were excluded as earlier work of Zhao et al. claimed N* species to be kinetically irrelevant.²⁹ Zhao observed increasing NH₄⁺ concentrations although NO₂⁻ was entirely converted and suggested N* as strongly adsorbed surface species that only very slowly converts into NH₄⁺.²⁹ While this seemed a reasonable explanation as N* species are IR inactive and could therefore not be observed in ATR-IR studies, it cannot be excluded that NH₂OH was formed in these experiments and the increase in the NH₄⁺ concentration arose from NH₂OH decomposition. Nevertheless, DFT calculations showed that the N–O dissociation has a high activation barrier^28,34 and a high surface diffusion barrier of N*.³⁴ This disfavors the N* formation and subsequently reduces the probability of coupling with other surface species. Thus, N₂ formation based on a surface reaction pathway via N* appears unlikely.

Xu's two remaining mechanisms are shown in Fig. 7. The first one (A) was previously excluded based on ATR-IR results of Ebbesen et al.³¹ that claim that NH₂OH cannot form N₂ since no N₂O* as an N₂ precursor was detected when feeding NH₂OH into the ATR-IR cell.³¹ As mentioned, we observed NH₄⁺ selectivity <100% in batch experiments converting NH₂OH indicating that N₂ can be formed from NH₂OH.¹⁷ Thus, both mechanisms are an option based on the current knowledge. Huang et al. reported similar kinetic studies including microkinetic modelling and DFT calculations that consider proton shuttling effects of the solvent to further deepen the understanding of the reaction mechanism.²⁸ Essentially, the suggested reaction mechanism follows the same elementary steps as those reported by Xu et al.²⁷ shown in Fig. 7B.

Fig. 7 Suggested NO₂⁻ hydrogenation mechanisms by Xu et al.²⁷

In contrast to this LH-type mechanism, Lee et al. suggested an Elye Ridel (ER) like mechanism based on DFT in which N* reacts with H* and NO₂⁻ and H⁺ from the liquid phase to form N₂O that is then quickly converted to N₂.³⁴

N* + H* + NO₂⁻ + H⁺ → N₂O* + H₂O + * (8)

This reaction pathway is reported energetically favorable, but a reaction in which four molecules from two different phases meet and react in a concerted manner appears kinetically unlikely. Notably, this study did not consider transition state calculations so the authors themselves encourage careful interpretation of the suggested decomposition pathway.³⁴ While Wong et al.³³ adopt this EL mechanism, this elaboration will focus on the LH-type mechanism.

3.4.3 Revised NO₂⁻ hydrogenation mechanism: the N–N coupling pathway and RDS. Unraveling the exact coupling species is not realistically possible due to the complex interplay of different surface species which are to a great extent not even distinguishable (NO_xH_y). Nevertheless, we attempt to streamline the most probable reaction pathway, RDS and selectivity-determining intermediates based on our experimental observations, ATR-IR findings, DFT calculations and reaction kinetics.

The suggested reaction pathway (Fig. 8) is inspired by the proposed mechanism by Xu et al.²⁷ as the results of the reaction kinetics experiments are similar in the studies by Xu et al.²⁷ and Huang et al.²⁸ and the present work. The revised version includes the desorption of NH₂OH that is evident due to the detection of NH₂OH species regardless of the reaction conditions.¹⁷ Additionally, the hydrogenation equilibrium between NH₂OH* and HNOH*, which allows NH₂OH_(aq) to merge into the N₂ formation pathway, is included. This is supported by NH₄⁺ selectivity <100% in NH₂OH decomposition experiments¹⁷ and represents a crucial difference with respect to the reaction mechanism suggested by Kim³⁴ and adopted by Wong.³³

Fig. 8 Revised NO₂⁻ hydrogenation mechanism under consideration of NH₂OH desorption.

Under all investigated reaction conditions, N₂ was observed as the main product of the NO₂⁻ hydrogenation reaction. While N–N coupling is a prerequisite for N₂ formation, Xu et al.²⁷ demonstrated that this cannot be the rate determining step (RDS) of the reaction as the observed negative reaction orders at high NO₂⁻ concentrations and low H₂ partial pressures Xu observed do not match any rate expression considering N–N coupling. Instead, second order rate dependencies with respect to NO₂⁻ would have to be observed experimentally to support such N–N coupling as the RDS.²⁷ Direct coupling of surface abundant N-species (NO* and ) can be rejected, since significant coupling of surface abundant species is intrinsically contradicting as the coverage would be drastically reduced as a consequence of their coupling. ATR-IR studies revealed the NO* and the ill-defined species are surface abundant.^29–31 Thus, N₂ cannot be formed by NO–NO coupling, which is also supported by DFT calculations that suggest that dipole-repulsions between NO* species would lead to high activation barriers.³⁴ The NH₂OH co-feeding experiments (Section 3.3) delivered zeroth order kinetics in NH₂OH; thus it is likely that the ill-defined surface abundant represents a highly hydrogenated species such as HNOH* or NH₂OH*. Therefore, HNOH* and NH₂OH* can also be excluded for coupling with itself or NO* to produce N₂. Also, participation of N* in the N–N coupling is unlikely due to high surface diffusion barriers for N* as discussed in Section 3.4.2, leaving NO*–NH*, NHO*–NH* or NOH*–NH* as possible pathways for N₂ formation.

Huang et al.²⁸ suggested a co-limitation of the hydrogenation of NO* to HNO* and the consecutive hydrogenation of HNO* to HNOH* as the RDS based on their kinetic model.²⁸ DFT calculations in their work delivered lower activation barriers for HNOH* formation via HNO* than via NOH*. While NOH* might still form and couple with NH* to form N₂, we focus on HNO* for the subsequent pathway to HNOH*. Huang et al.²⁸ suggested that the degree of rate control shifts based on the reaction conditions but is under typical reaction conditions (RT, 0.8 bar H₂ partial pressure) at about 90% for the NO* hydrogenation to HNO*.

While the observed reaction kinetics in this work are similar and therefore confirm the previous model not considering NH₂OH, the model cannot explain the RDS in the presence of NH₂OH sufficiently. If NO* hydrogenation was also the dominant RDS under NH₂OH rich conditions, all following steps were quasi-equilibrated and thus NH₂OH_(aq) was equilibrated with NH* according to the 0^th law of thermodynamics. NH* can be coupled with NO* to form N₂ and NO* is easily formed from NO₂⁻ and surface abundant; thus an acceleration of the NO₂⁻ hydrogenation with increasing NH₂OH co-feed (NH₂OH reaction order > 0) would be expected. In other words, the NO* → NOH* RDS could be bypassed owing to the abundance of NH*. However, the observation is 0^th order in NH₂OH; thus, there must be another RDS under NH₂OH rich conditions.

The H-assisted dissociation of the N–O bond in HNOH* to NH* can be a RDS based on the expected apparent reaction orders derived from the LH equations by Xu et al.²⁷ without considering NH₂OH. A similar derivation under NH₂OH rich conditions was derived (see the SI) and resulted in a reaction order range of [−1, 0] for NO₂⁻ and [−1, 1] for NH₂OH which covers the observed orders. Notably, in ATR-IR titration experiments of the surface abundant NO* and species it was observed that NO* is more quickly converted than ,^29,30 supporting that HNOH* → NH* is a realistic RDS.²⁴ Controversially, if HNOH* → NH* is the RDS under NH₂OH rich conditions, one might expect the reaction to be accelerated with increasing NH₂OH concentration due to increased HNOH* coverage. However, if the surface is saturated with NH₂OH (and its equilibrated species including HNOH*), changes in NH₂OH in the solution would not affect the reaction rate. This surface saturation of NH₂OH species under NH₂OH rich conditions agrees with ATR-IR experiments that showed that NH_xO_y is a surface abundant species when feeding NH₂OH.^30,31

Thus, we suggest that the RDS varies even more on the surface coverages than reported by Huang et al.²⁸ In the absence of NH₂OH the RDS lies dominantly on the NO* → NOH* hydrogenation while it shifts to HNOH* → NH* under NH₂OH abundant conditions. In between these two extreme cases the degree of rate control shifts non-linearly and therefore accounts for intriguing trends such as decreasing NH₄⁺ formation with increasing NH₂OH co-feed (Fig. 6D) or higher NH₄⁺ formation at 0.6 bar H₂ and elevated temperature (Fig. 4C). NH* remains the last common reaction intermediate in the pathways towards N₂ and NH₄⁺. Thus, the surface coverage of H* and various N* species determines the probability whether NH* proceeds via the N₂ or NH₄⁺ pathway. Clearly, higher H-coverage increases the probability for NH₄⁺ formation, which is consistent with increased NH₄⁺ selectivity at the beginning of a reaction and approaching full NO₂⁻ conversion and lower NH₄⁺ formation under H₂ deficient conditions as discussed in Section 3.1.

In conclusion, the recent discovery of NH₂OH as a desorbed reaction intermediate clearly highlights the need for the incorporation of NH₂OH into the NO₂⁻ hydrogenation mechanism. In our revised mechanistic scheme we (1) account for the adsorption equilibrium of NH₂OH, (2) include the hydrogenation equilibrium of HNOH* and NH₂OH* representing the possibility for NH₂OH to merge into the N₂ formation pathway, and (3) suggest a changing degree of rate control depending on the surface coverages with NO hydrogenation as the dominant RDS in the absence of NH₂OH and the H-assisted N–O dissociation of HNOH* to NH* as the RDS under NH₂OH rich conditions. This mechanism emphasizes that the selectivity is determined by NH* conversion, either reacting with H* to form NH₄⁺ or by coupling with an N-species to form N₂, depending on the surface coverage of H and HNO*/NOH* or NO*. Our proposed mechanism provides a coherent picture of the reaction mechanism consistent with the experimental data and previous results obtained from our group. Nonetheless, the dominant surface mechanism will inevitably depend on catalyst characteristics and operating conditions, and other mechanistic interpretations may also be reasonable.

3.5 NO₂⁻ hydrogenation on SnPd/Al₂O₃ – activity, selectivity and NH₂OH co-feeding

The same set of experiments, varying the NO₂⁻ concentration from 0.3–8 mmol L⁻¹ and the H₂ partial pressure from 0.05–0.8 bar, was conducted on a SnPd/Al₂O₃ catalyst. This catalyst was prepared from the previously tested Pd/Al₂O₃ by Controlled Surface Deposition (CSD), introducing Sn corresponding to 0.4 monolayers relative to the accessible Pd surface atoms. Remarkably, the overall NO₂⁻ hydrogenation activity remained largely unaffected by Sn doping. Since the activity was unchanged, we assume, as for Pd/Al₂O₃, that mass transport limitations are absent (SI, Fig. 1). The derived reaction orders are likewise similar to those observed for Pd/Al₂O₃: the order in NO₂⁻ remains close to zero (Fig. 9A), while the order in H₂ increases from 0.2 (Pd/Al₂O₃) to 0.5 (SnPd/Al₂O₃, Fig. 9B), indicating a higher sensitivity to changes in H₂ partial pressure. This can be rationalized by the partial blockage of Pd sites by Sn, which cannot dissociate H₂, thereby reducing the effective H availability. In NH₂OH co-feeding experiments, the NO₂⁻ hydrogenation activity also remained unchanged, resulting in an apparent reaction order in NH₂OH close to zero at both 0.8 and 0.05 bar H₂ (Fig. 9C).

Fig. 9 Effect of NO₂⁻ (A), H₂ (B) and NH₂OH concentration (C) on the NO₂⁻ hydrogenation rate on SnPd/Al₂O₃. Reaction conditions: 300 mL H₂O, 10 mg SnPd/Al₂O₃, 10 mL min⁻¹ CO₂, 5–80 mL min⁻¹ H₂ and He to balance to 100 mL min⁻¹ total flow, RT, 600 rpm.

The selectivity trends on SnPd/Al₂O₃ mirror those on Pd/Al₂O₃: NH₂OH concentrations increase over the course of the reaction, peaking near full NO₂⁻ conversion (Fig. 10B), while NH₄⁺ selectivity exhibits a U-shaped dependence on conversion (Fig. 10A) as discussed in Section 3.1. As for Pd/Al₂O₃, variations in the initial NO₂⁻ concentration have only minor effects on product distribution (SI, Fig. 9), and both NH₄⁺ and NH₂OH concentrations are lower under H₂-deficient conditions (Fig. 10).

Fig. 10 NH₄⁺ (A) and NH₂OH (B) concentrations as a function of NO₂⁻ conversion and time, respectively, and selectivity distribution at ∼50% NO₂⁻ conversion on Pd and SnPd at different H₂ partial pressures (C). Pd data are the same as in Fig. 4 and replotted for intuitive comparison. The arrows in (C) indicate the difference in N₂ selectivity. Reaction conditions: 300 mL H₂O, 10 mg SnPd/Al₂O₃, 10 mL min⁻¹ CO₂, 5–80 mL min⁻¹ H₂ and He to balance to 100 mL min⁻¹ total flow, RT or 40 °C, 600 rpm.

Notably, the NH₂OH and NH₄⁺ selectivity are mildly increased on SnPd/Al₂O₃ for all H₂ pressures ≥0.2 bar resulting in 4–16% lower N₂ selectivity upon Sn doping (Fig. 10C). While this appears contradictory to the notion of reduced H coverage caused by Sn blocking of Pd sites, this suggests that Sn also shifts the balance of surface reactions away from N–N coupling and towards NH₄⁺ formation. Formation of N₂ requires N–N bond formation via coupling between a reduced N adsorbate (e.g., NH*) and a more oxidized N adsorbate (e.g., NO*/NOH*/HNOH*), which therefore requires co-adsorption of these intermediates in close proximity on Pd ensembles. Introducing Sn partially dilutes Pd, reducing the probability for such coupling configurations and thus slightly suppressing the N₂ pathway.⁴² In contrast, sequential hydrogenation steps toward NH₄⁺ rely primarily on H* availability and are less sensitive to the availability of larger Pd ensembles.⁴² This is consistent with the much higher mobility (lower diffusion barrier) of H* on Pd compared with N-species,³⁴ which favors hydrogenation when coupling upon Pd dilution. The same argumentation applies to NH₂OH formation, since it does not require N–N coupling but follows from stepwise hydrogenation. However, we cannot rule out that the presence of Sn modifies the electronic structure of neighboring Pd surface atoms, increasing the rate of hydrogenation to NH₄⁺.

The NH₂OH co-feeding experiments at high H₂ partial pressure (0.8 bar) on SnPd/Al₂O₃ reproduce the trends observed on Pd/Al₂O₃: the NH₂OH concentration remains essentially constant, while more NH₄⁺ is formed when NH₂OH is co-fed (SI, Fig. 10 and 11B). Interestingly, the same non-linearity is observed in decreasing NH₄⁺ formation with increasing NH₂OH co-feed concentration, as on Pd at elevated temperature. Under H₂-deficient conditions, NH₂OH depletion is more pronounced on SnPd than on Pd (Fig. 11A and SI, Fig. 6F) and the NH₄⁺ concentration increases with increasing NH₂OH co-feed concentration (Fig. 11B). The NH₂OH co-feeding results can be rationalized by the ability to disproportionate into NH₄⁺ and N₂ without net consumption of H₂. Under H₂-deficient conditions this disproportionation pathway gains relative importance, because pathways that rely on H₂ uptake are suppressed. On SnPd/Al₂O₃ this is more pronounced; thus the presence of Sn facilitates the NH₂OH decomposition. This is in line with the enhanced N–O bond activation ability of Sn also allowing to break the first N–O bond in NO₃⁻. As a result, NH₂OH is more readily consumed on SnPd/Al₂O₃ than on Pd/Al₂O₃, consistent with the stronger decrease in the NH₂OH concentration observed at low H₂ partial pressure.

Fig. 11 NH₂OH concentration as a function of NO₂⁻ conversion at different NH₂OH co-feed concentrations at 0.05 bar H₂ (A) and the NH₄⁺ concentration at different rounded NH₂OH co-feed concentrations after 60 min of reaction (B). Reaction conditions: 300 mL H₂O, 0.8 mmol L⁻¹ KNO₂, 10 mg SnPd/Al₂O₃, 80 or 5 mL min⁻¹ H₂, 10 mL min⁻¹ CO₂ and He to make up to 100 mL min⁻¹, RT, 600 rpm.

In conclusion, SnPd/Al₂O₃ exhibits NO₂⁻ hydrogenation activity and selectivity trends very similar to Pd/Al₂O₃. The main differences are a higher sensitivity to H₂ in both activity and selectivity, likely due to lower H affinity on SnPd. Also, the NH₂OH and NH₄⁺ selectivities are increased resulting in a lower N₂ selectivity on SnPd.

In the bigger picture, these results show the demanding challenges for NO₃⁻ hydrogenation aiming for high N₂ selectivity. The NO₃⁻ hydrogenation requires a bimetallic catalyst such as SnPd/Al₂O₃ for the initial NO₃⁻ hydrogenation to NO₂⁻ which represents the RDS of this reaction. As a result, NO₂⁻ is commonly detected in traces or not detected in the liquid bulk at all. Thus, the surface coverage of N-species in the NO₃⁻ hydrogenation on the Pd sites is lower and shifts the coverage towards higher H : N ratios favoring NH₄⁺ and NH₂OH formation. This in combination with the higher NH₄⁺ production affinity of SnPd itself highlights the fundamental challenges of selective NO₃⁻ hydrogenation for drinking water purification.

4 Conclusion

In this study, we have demonstrated that NH₂OH is a key intermediate in the catalytic hydrogenation of NO₂⁻. The selectivity towards NH₄⁺ is strongly conversion-dependent, with both NH₂OH and NH₄⁺ formation substantially suppressed only under H₂-deficient conditions, albeit at the expense of lower overall NO₂⁻ hydrogenation activity. Elevated temperature accelerates NH₂OH decomposition, thereby lowering the NH₂OH concentration. However, rather than enhancing N₂ formation, this shift promotes NH₄⁺ selectivity leaving N₂ selectivity largely unaffected. Co-feeding experiments confirmed that the presence of NH₂OH does not influence the intrinsic NO₂⁻ hydrogenation rate and that NH₂OH formation and decomposition are balanced. The latter suggests that a continuous process could in principle be designed in a way that NH₂OH would not excessively accumulate due to its intermediary nature. The activity and selectivity patterns for Pd/Al₂O₃ and SnPd/Al₂O₃ were very similar with higher NH₂OH and NH₄⁺ formation on SnPd/Al₂O₃ being the main differences.

We critically reviewed the NO₂⁻ hydrogenation mechanism and provided a revised LH-based scheme including NH₂OH as desorbed species. The revised mechanism highlights that NH₂OH can merge into the N₂ formation pathway and suggests the RDS to be strongly dependent on the surface coverages.

Author contributions

J. Betting: methodology, investigation, writing – original draft; Y. Preedawichitkun: investigation; T. Sooknoi: funding acquisition, writing – review & editing; J. Faria Albanese: funding acquisition, supervision, writing – review & editing; L. Lefferts: supervision, writing – review & editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6ta00106h.

Acknowledgements

We gratefully acknowledge the financial support of the LNG-Zero project (grant number 20002675). We also thank the Royal Golden Jubilee PhD Program (RGJ-PhD Program Grant Number N41A640277) and the National Research Council of Thailand (NRCT Grant Number N42A680527).

References

1. N. Lehnert, H. T. Dong, J. B. Harland, A. P. Hunt and C. J. White, Nat. Rev. Chem., 2018, 2, 278–289.
2. D. E. Canfield, A. N. Glazer and P. G. Falkowski, Science, 2010, 330, 192–196.
3. X. Xia, S. Zhang, S. Li, L. Zhang, G. Wang, L. Zhang, J. Wang and Z. Li, Environ Sci Process Impacts, 2018, 20, 863–891.
4. X. Zhang, B. B. Ward and D. M. Sigman, Chem. Rev., 2020, 120, 5308–5351.
5. M. Le Moal, C. Gascuel-Odoux, A. Ménesguen, Y. Souchon, C. Étrillard, A. Levain, F. Moatar, A. Pannard, P. Souchu, A. Lefebvre and G. Pinay, Sci. Total Environ., 2019, 651, 1–11.
6. C. S. Bruning-Fann and J. B. Kaneene, Vet Hum Toxicol, 1993, 35, 521–538.
7. WHO, Guidelines for Drinking-Water Quality, 2022.
8. I. Sanchis, E. Diaz, A. H. Pizarro, J. J. Rodriguez and A. F. Mohedano, Sep. Purif. Technol., 2022, 290, 120750.
9. K. G. N. Quiton, M. C. Lu and Y. H. Huang, Chemosphere, 2021, 262, 128371.
10. N. Barrabés and J. Sá, Appl. Catal., B, 2011, 104, 1–5.
11. G. Tokazhanov, E. Ramazanova, S. Hamid, S. Bae and W. Lee, Chem. Eng. J., 2020, 384, 123252.
12. J. Martínez, A. Ortiz and I. Ortiz, Appl. Catal., B, 2017, 207, 42–59.
13. European Union, Directive on the Quality of Water Intended for Human Consumption, 2020.
14. P. Liao, J. Kang, R. Xiang, S. Wang and G. Li, Angew. Chem., Int. Ed., 2024, 63, e202311752.
15. S. R. Udayasurian and T. Li, Nanoscale, 2024, 16, 2805–2819.
16. C. L. Rooney, Q. Sun, B. Shang and H. Wang, J. Am. Chem. Soc., 2025, 147, 9378–9385.
17. J. Betting, L. Lefferts and J. Faria Albanese, Chem. Commun., 2025, 61, 12147–12150.
18. G. Yang, P. Zhou, J. Liang, H. Li and F. Wang, Inorg. Chem. Front., 2023, 10, 4610–4631.
19. G. Centi and S. Perathoner, Appl. Catal., B, 2003, 41, 15–29.
20. V. I. Parvulescu, F. Epron, H. Garcia and P. Granger, Chem. Rev., 2022, 122, 2981–3121.
21. B. P. Chaplin, M. Reinhard, W. F. Schneider, C. Schüth, J. R. Shapley, T. J. Strathmann and C. J. Werth, Environ. Sci. Technol., 2012, 46, 3655–3670.
22. F. Ruiz-Beviá and M. J. Fernández-Torres, J. Clean. Prod., 2019, 217, 398–408.
23. European Commission, Off. J. Eur. Union, 2009, 207.
24. V. D. Moesdijk, TU Eindhoven, 1979.
25. A. Garron, K. Lázár and F. Epron, Appl. Catal., B, 2005, 59, 57–69.
26. P. Munnik, P. E. De Jongh and K. P. De Jong, Chem. Rev., 2015, 115, 6687–6718.
27. P. Xu, S. Agarwal and L. Lefferts, J. Catal., 2020, 383, 124–134.
28. P. Huang, Y. Yan, A. Banerjee, L. Lefferts, B. Wang and J. A. Faria Albanese, J. Catal., 2022, 413, 252–263.
29. Y. Zhao, N. Koteswara Rao and L. Lefferts, J. Catal., 2016, 337, 102–110.
30. S. D. Ebbesen, B. L. Mojet and L. Lefferts, J. Catal., 2008, 256, 15–23.
31. S. D. Ebbesen, B. L. Mojet and L. Lefferts, Langmuir, 2008, 24, 869–879.
32. R. S. Postma, R. Brunet Espinosa and L. Lefferts, ChemCatChem, 2018, 10, 3770–3776.
33. H. Li, S. Guo, K. Shin, M. S. Wong and G. Henkelman, ACS Catal., 2019, 9, 7957–7966.
34. H. Shin, S. Jung, S. Bae, W. Lee and H. Kim, Environ. Sci. Technol., 2014, 48, 12768–12774.
35. L. Y. Huai, C. Z. He, H. Wang, H. Wen, W. C. Yi and J. Y. Liu, J. Catal., 2015, 322, 73–83.
36. C. A. Clark, C. P. Reddy, H. Xu, K. N. Heck, G. Luo, T. P. Senftle and M. S. Wong, ACS Catal., 2020, 10, 494–509.
37. S. Guo, K. Heck, S. Kasiraju, H. Qian, Z. Zhao, L. C. Grabow, J. T. Miller and M. S. Wong, ACS Catal., 2018, 8, 503–515.
38. U. Prüsse, M. Hähnlein, J. Daum and K.-D. Vorlop, Catal. Today, 2000, 55, 79–90.
39. O. M. Ilinitch, L. V Nosova, V. V Gorodetskii, V. P. Ivanov, S. N. Trukhan, E. N. Gribov, S. V Bogdanov and F. P. Cuperus, J. Mol. Catal. A Chem., 2000, 158, 237–249.
40. J. Wärna, I. Turunen, T. Salmit and T. Maunula, Chem. Eng. Sci., 1994, 49, 5763–5773.
41. I. Mikami, Y. Sakamoto, Y. Yoshinaga and T. Okuhara, Appl. Catal., B, 2003, 44, 79–86.
42. H. Berndt, I. Mönnich, B. Lücke and M. Menzel, Tin Promoted Palladium Catalysts for Nitrate Removal from Drinking Water, 2001, vol. 30.

---