Hydroxylamine: an overseen intermediate that brings into question nitrogen selectivity in metal-catalyzed nitrate and nitrite reduction

Abstract

In decades of nitrate and nitrite hydrogenation research, nitrite, ammonia, and nitrogen gas were assumed to be the only relevant products. However, we have discovered hydroxylamine on several metal catalysts under various reaction conditions using a simple derivatization strategy based on the oximation of benzaldehyde with hydroxylamine. This previously overlooked intermediate challenges pervasive assumptions of nitrogen gas selectivity and compels a reexamination of the reaction mechanism. Additionally, the hydroxylamine presence represents a major setback for the application of catalytic nitrate and nitrite reduction in drinking water purification.

---

Nitrogen oxyanions are ubiquitous intermediates in the natural nitrogen cycle, playing a crucial role in sustaining life on Earth. The development of synthetic nitrate fertilizers has been instrumental in supporting global population growth by enabling ammonia synthesis from dinitrogen and hydrogen, which is later oxidized into nitro oxyanions. Unfortunately, the unintended leakage and accumulation of these compounds in the environment pose serious threats to public health due to its toxicity and ecological damage via eutrophication of water bodies.^1–7

In nature, various bacteria—such as those within the Proteobacteria phylum—facilitate denitrification in oxygen-deprived environments leveraging nitrate (NO₃⁻) and nitrite (NO₂⁻) reductase enzymes, provided that sufficient carbon sources are available to drive the metabolic machinery.^1,3 While natural bacteria can contribute to reducing NO₃⁻ and NO₂⁻, the rates achieved through biological denitrification are insufficient to counteract the continuous release of nitrogen oxyanions into the environment.^8,9

Inspired by natural reductase enzymes, the catalytic reduction of NO₃⁻ and NO₂⁻ to N₂ using hydrogen on metals has been proposed as an alternative to bio-based processes. Its simplicity and higher reaction rates have driven extensive research for the past thirty years. This higher activity, however, often comes at the expense of ammonia formation.^10–12 This is highly undesirable due to stringent concentration limits of ammonia (0.5 mg L⁻¹) compared to nitrate (50 mg L⁻¹) in drinking water.¹³ Therefore, substantial research has been devoted to developing catalysts with high selectivity to unleash its practical use.

Two distinct catalysts are required for NO₃⁻ and NO₂⁻ reduction (see eqn (1)–(3)). While nanoparticles of palladium (Pd) suffice for NO₂⁻ reduction, bimetallic Pd–Cu, Pd–Sn or Pd–In catalysts arose as best alternatives for NO₃⁻ reduction.^10,11,14,15

NO₃⁻ + H₂ → NO₂⁻ + H₂O (1)

2NO₂⁻ + 3H₂ + 2H⁺ → N₂ + 4H₂O (2)

NO₂⁻ + 3H₂ + 2H⁺ → NH₄⁺ + 2H₂O (3)

The standard procedure to record the reaction rate and product distribution is ion chromatography (IC) or high-performance liquid chromatography (HPLC) to determine NH₄⁺, NO₃⁻ and NO₂⁻ concentrations. A fundamental assumption in the literature is that no other products are formed and, thus, the remainder in the products is dinitrogen (N₂). Thus, commonly a mass balance closure has not been proven. This method is broadly accepted as closing the mass balance is a major challenge due to (1) N₂ contamination from the surrounding atmosphere is difficult to suppress, (2) the low concentration of gaseous products (e.g. NO and N₂O) under typical reaction conditions, and (3) complexity of sampling and quantification of the gas- and liquid phase streams during reaction at the low concentrations of NO₃⁻ and NO₂⁻ reactants typically employed (50 ppm). In samples from the stationary liquid phase, the product amounts can be directly calculated, while the gaseous products are dependent on liquid-gas mass transfer, headspace of the reactor and the gas flow rate that continuously flushes out the gaseous products.

Careful inspection of the literature revealed that only very few studies have presented a closed mass balance. Werth et al. achieved a closed mass balance over the full reaction time using isotope labelled N species.¹⁶ Vorlop et al. analyzed both liquid and gas phases in a batch reaction but missed up to ∼30% in the mass balance during the reaction. A closed mass balance could only be achieved by elongating the reaction time by a factor of ∼1.5 beyond full NO₃⁻ and NO₂⁻ conversion.¹⁷ They suggested that strong adsorption of intermediates on the catalyst and dissolved nitrous oxide acted as reservoirs during the reaction, leading to the incomplete mass balance closure at low conversions. In earlier works, our group also suggested adsorbed intermediates as a reason for an increase of the NH₄⁺ concentration after full NO₂⁻ conversion.¹⁸ Pintar et al. mentioned that no NH₂OH was found in a few of their studies^19–21 without presenting any proof to support this claim. Wong et al. mentioned NH₂OH as an adsorbed species in their reaction mechanism but assumed that its desorption was not favorable. Thus, NH₂OH as a dissolved intermediate was not reported.²² The same group detected hydrazine (N₂H₂) over both Pd/Al₂O₃ and Rh/Al₂O₃ catalysts.²³ However, this was observed only at unusually high pH values (>7) where low catalytic activities are recorded, and the maximum yield of hydrazine remained limited to ∼1.5% (0.5 ppm in solution).²³ Here, one would wonder if this simplification of the nitrogen mass balance would be of any importance from the scientific and application perspective. In the present contribution, we challenge this postulate and explore if other relevant species beyond NO₂⁻, NH₄⁺ and N₂ are formed during the reaction.

To address this question, we conducted a series of experiments using Pd-based catalysts for the reduction of NO₃⁻ and NO₂⁻ in aqueous environments and quantified NH₂OH. Measuring NH₂OH, however, is not trivial. This species can undergo degradation at high pH.^24,25 Fortunately, hydroxylamine reacts quantitatively with aldehydes, ketones, and acids. These reactions are fast, chemoselective, and thermodynamically favorable at room temperature, making them an attractive proxy for the NH₂OH formation during NO₃⁻ and NO₂⁻ reduction. In fact, Lee et al. leveraged this chemistry to produce benzaldehyde oxime from benzaldehyde using NO₃⁻ and NO₂⁻ as a N-source over nanoscale zero-valent iron, which is only possible if NH₂OH is formed during the reaction.²⁶ While in situ C–N bond formation from NO₃⁻ and NO₂⁻ was not followed in thermo-catalysis, it is an emerging research field in electro-catalysis.^27–29

To detect and quantify NH₂OH, we added 1 μL benzaldehyde to the liquid aliquot right after sampling the reaction mixture, suppressing decomposition of hydroxylamine to other products than benzaldehyde-oxime (Fig. 1). By using a 5-fold excess of benzaldehyde with respect to the maximal possible NH₂OH concentration, full oximation was realized without significantly changing the sample volume (1.5 mL). As the catalyst is separated from the liquid aliquot to terminate the reaction in the sample, before the benzaldehyde addition, the native product distribution remains unchanged by this strategy. The resulting benzaldehyde oxime can be quantified by liquid chromatography (HPLC) due to its strong UV-vis absorbance at 248 nm (see details in the ESI†).

Fig. 1 Aldehyde oximation strategy for detection of NH₂OH intermediates using benzaldehyde in the aqueous phase.

To establish the experimental space in which NH₂OH is formed as a reaction intermediate, we varied the reaction conditions and the catalyst composition in the NO₃⁻ and NO₂⁻ hydrogenation. In a typical experiment, carbon dioxide (CO₂) was flushed through the reactor to buffer the media (pH ∼ 6), while the temperature was set at room conditions (22 °C), and the nitrate concentration was ∼0.8 mM (∼50 mg L⁻¹ NO₃⁻), representing well the reaction conditions widely used in the literature.^10,11,14,15Fig. 2 illustrates that the NH₄⁺ concentration increases with increasing NO₃⁻ conversion while NO₂⁻ occurs in trace amounts throughout the entire experiment. Surprisingly, the protocol herein proposed revealed the formation of NH₂OH during the reaction. Notably, the rate of NH₂OH decomposition is slower than that of NO₃⁻ reduction, leading to substantial NH₂OH accumulation in the system.

Fig. 2 Typical concentration profile of NO₃⁻ reduction (entry 6 in Table 1). Reaction conditions: 50 mg SnPd/Al₂O₃, 300 mL, 80 : 10 : 10 mL min⁻¹ H₂ : CO₂ : He, 600 rpm, and RT.

The point of full NO₃⁻ and NO₂⁻ conversion would typically be considered as the end of the reaction and, therefore, be the reference for determination of NH₄⁺ and N₂ selectivity. Since this calculation would disregard NH₂OH formation, the resulting mass balance and the N₂ selectivity would be erroneous. The NH₂OH yield at full NO₃⁻ and NO₂⁻ conversion, which we report as Y_comp quantifies this error. Meanwhile, the maximal yield of NH₂OH (Y_max) reflects the interplay between the rates of NO₃⁻ reduction to NO₂⁻, N₂, NH₄⁺, and NH₂OH. In this system, a high Y_max would suggest that this intermediate is quickly formed during the reaction and subsequently desorbs from the catalyst surface. This fundamental step in the reaction mechanism is neglected in the literature, emphasizing the importance of transient quantification of NH₂OH during the reaction to unravel the fundamental mechanism of NO₃⁻ and NO₂⁻ reduction.

Table 1 shows the Y_max and Y_comp of different catalysts, for NO₃⁻ as well as NO₂⁻ reduction, at 40 °C and, to some extent, at varied N : H ratios. While it is true that the fundamental details are not fully understood yet, we here report a ubiquitous phenomenon since NH₂OH was detected in all experiments. The NO₂⁻ reduction reaction was studied at a typical H₂ partial pressure (0.8 bar, entry 1), as well as at a lower H₂ partial pressure (0.2 bar, entry 2), a lower initial concentration (0.2 mM, entry 3) and an elevated temperature (40 °C, entry 4, Fig. S8, ESI†) resulting in maximal NH₂OH yields that varied between 9 and 56%. Two commercial Pd/Al₂O₃ catalysts were used to confirm that NH₂OH formation is a widely occurring phenomenon and does not arise from custom-prepared catalysts (entries 7 and 8). Doping the Pd/Al₂O₃ catalysts with Sn, In and Cu showed substantial concentration of NH₂OH in the reaction mixtures, reaching maximum yields that varied from 4 to 23% (entries 6, 9–11). More importantly, only in a few experiments the NH₂OH amounts after full NO₃⁻ and NO₂⁻ conversion were ≤1% (entries 4 and 11). For all other experiments, however, NH₂OH ranged from 4–56%, highlighting the importance of reporting the concentration of this species when reporting catalyst performance. This is especially relevant when applying this for drinking water purification as hydroxylamine is even more toxic than the NO₃⁻, NO₂⁻, and NH₄⁺ counterparts.

Table 1 Overview of nitrate and nitrate reduction experiments with several catalysts and conditions with maximum NH₂OH yield (Y_max) and NH₂OH yield at full NO₃⁻/NO₂⁻ conversion (Y_comp). Bold entries denote the variation with respect to the standard conditions or the previous entry. Catalysts denoted with * and ** are different commercial Pd/A₂O₃ catalysts or based on them

No.	Substrate	Catalyst	C 0/mmol L^−1	H2 : CO2 : He/mL min^−1	T/°C	m cat/mg	Y max	Y comp
1	NO2^−	Pd/Al2O3	0.8	80 : 10 : 10	22	50	34	32
2	NO2^−	Pd/Al2O3	0.8	20 : 10 : 70	22	20	17	14
3	NO2^−	Pd/Al2O3	0.2	80 : 10 : 10	22	20	56	56
4	NO2^−	Pd/Al2O3	0.8	80 : 10 : 10	40	20	9	0
5	NO2^−	SnPd/Al 2 O 3	0.8	80 : 10 : 10	22	50	16	16
6	NO 3 ^−	SnPd/Al2O3	0.8	80 : 10 : 10	22	50	23	12
7	NO 2 ^−	Pd/Al 2 O 3 *	0.8	80 : 10 : 10	22	20	16	10
8	NO2^−	Pd/Al 2 O 3 **	0.8	80 : 10 : 10	22	20	4	4
9	NO3^−	SnPd/Al2O3**	0.8	80 : 10 : 10	22	100	15	14
10	NO3^−	InPd/Al2O3**	0.8	80 : 10 : 10	22	100	23	14
11	NO3^−	CuPd/Al2O3**	0.8	80 : 10 : 10	22	100	9	1

---

The selectivity of the NH₂OH decomposition after full NO₃⁻ and NO₂⁻ conversion in the presence of H₂ is important as it can lead to higher NH₄⁺ concentrations, which is highly undesirable in drinking water. The NH₂OH decomposition can either proceed via catalytic hydrogenation resulting in 100% NH₄⁺ or via catalytic disproportionation resulting in the 1 : 1 formation of N₂ and NH₄⁺.³⁰

NH₃OH⁺ + H₂ → NH₄⁺ + H₂O (4)

4NH₂OH + 2H⁺ → 2NH₄⁺ + N₂O + 3H₂O (5)

The formed N₂O under reaction conditions is quickly converted to N₂.¹⁶ The same cannot be said about NH₂OH. When decomposition experiments of hydroxylamine were conducted in the absence of a catalyst and any NO₃⁻ or NO₂⁻, the observed reaction rates were negligible (Fig. S7, ESI†). In stark contrast, in the presence of a catalyst substantial hydroxylamine conversion was observed (Fig. S9 and S10, ESI†). To visualize this one could compare the concentration increase of NH₄⁺ due to the conversion of hydroxylamine, after completion of the NO₃⁻ and NO₂⁻ conversion. As shown in Fig. 3, two lines can be plotted to indicate the NH₄⁺ concentrations that would be obtained if the conversion takes place via either hydrogenation (100% selectivity NH₄⁺) or disproportionation (50% selectivity NH₄⁺). The data points spread in-between both boundaries indicating that both hydrogenation and disproportionation contribute (see the source data Table S2, ESI†). While it is true that in these experiments the extent of NH₂OH conversion was different for each data point, the results suggest a complex interplay between the NH₄⁺ and N₂ formation. Here, one may hypothesize that a major part of the NH₄⁺ formed during NO₃⁻ and NO₂⁻ reduction is generated via a NH₂OH intermediate owing to the high NH₄⁺ selectivity of the NH₂OH decomposition. Considering that NH₄⁺ formation is the major roadblock for widespread implementation of this technology in drinking water purification, it is crucial to consider NH₂OH in the catalyst and process design as well as in mechanistic studies.

Fig. 3 Amount of NH₄⁺ formed in relation to the amount of NH₂OH converted upon full conversion of NO₃⁻ and NO₂⁻ for experiments 1 to 11.

In this study, we have shown unequivocal evidence that hydroxylamine is an essential intermediate product in the reduction of NO₃⁻ and NO₂⁻ in aqueous environments. These results have profound implications on the calculation of selectivity to N₂ based on the concentrations of NO₃⁻, NO₂⁻ and NH₄⁺ exclusively. More importantly, this discovery reshapes our understanding of this critical reaction and provides a simple and accurate strategy for hydroxylamine quantification via oximation that is relevant for thermo-catalytic and potentially enzymatic processes^31,32 for nitrogen oxyanions reduction in aqueous environments.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the ESI.†

Notes and references

1. M. Hu, Y. Liu, Z. Yao, L. Ma and X. Wang, Front. Environ. Sci. Eng., 2018, 12, 1–18.
2. WHO, Guidelines for drinking-water quality, 2022.
3. A. J. Lecloux, Catal. Today, 1999, 53, 23–34.
4. D. E. Canfield, A. N. Glazer and P. G. Falkowski, Science, 1979, 2010(330), 192–196.
5. U. Prüsse, M. Hähnlein, J. Daum and K.-D. Vorlop, Catal. Today, 2000, 55, 79–90.
6. K. N. Heck, S. Garcia-Segura, P. Westerhoff and M. S. Wong, Acc. Chem. Res., 2019, 52, 906–915.
7. C. S. Bruning-Fann and J. B. Kaneene, Vet. Hum. Toxicol., 1993, 35, 521–538.
8. J. S. Stamler, D. J. Singel and J. Loscalzo, Science, 1979, 1992(258), 1898–1902.
9. J. N. Galloway, A. R. Townsend, J. W. Erisman, M. Bekunda, Z. Cai, J. R. Freney, L. A. Martinelli, S. P. Seitzinger and M. A. Sutton, Science, 1979, 2008(320), 889–892.
10. G. Tokazhanov, E. Ramazanova, S. Hamid, S. Bae and W. Lee, Chem. Eng. J., 2020, 384, 123252.
11. I. Sanchis, E. Diaz, A. H. Pizarro, J. J. Rodriguez and A. F. Mohedano, Sep. Purif. Technol., 2022, 290, 120750.
12. F. Ruiz-Beviá and M. J. Fernández-Torres, J. Clean Prod., 2019, 217, 398–408.
13. European Union, Directive on the quality of water intended for human consumption, 2020.
14. K. G. N. Quiton, M. C. Lu and Y. H. Huang, Chemosphere, 2021, 262, 128371.
15. J. Martínez, A. Ortiz and I. Ortiz, Appl. Catal., B, 2017, 207, 42–59.
16. R. Zhang, D. Shuai, K. A. Guy, J. R. Shapley, T. J. Strathmann and C. J. Werth, ChemCatChem, 2013, 5, 313–321.
17. U. Prüsse, J. Daum, C. Bock and K.-D. Vorlop, Stud. Surf. Sci. Catal., 2000, 130, 2237–2242.
18. Y. Zhao, N. Koteswara Rao and L. Lefferts, J. Catal., 2016, 337, 102–110.
19. A. Pintar, J. Batista and I. Muševič, Appl. Catal., B, 2004, 52, 49–60.
20. A. Pintar and J. Batista, Appl. Catal., B, 2006, 63, 150–159.
21. A. Pintar and J. Batista, J. Hazard. Mater., 2007, 149, 387–398.
22. H. Li, S. Guo, K. Shin, M. S. Wong and G. Henkelman, ACS Catal., 2019, 9, 7957–7966.
23. 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.
24. Q. Wang, C. Wei, L. M. Pérez, W. J. Rogers, M. B. Hall and M. S. Mannan, J. Phys. Chem. A, 2010, 114, 9262–9269.
25. C. Wei, S. R. Saraf, W. J. Rogers and M. Sam Mannan, Thermochim. Acta, 2004, 421, 1–9.
26. M. A. Kumar, J. K. Choe, W. Lee and S. Yoon, Environ. Nanotechnol. Monit. Manag., 2017, 8, 97–102.
27. P. Liao, J. Kang, R. Xiang, S. Wang and G. Li, Angew. Chem., Int. Ed., 2024, 63, 1–12.
28. S. R. Udayasurian and T. Li, Nanoscale, 2024, 16, 2805–2819.
29. C. L. Rooney, Q. Sun, B. Shang and H. Wang, J. Am. Chem. Soc., 2025, 147, 9378–9385.
30. V. D. Moesdijk, TU Eindhoven, 1979.
31. N. Lehnert, H. T. Dong, J. B. Harland, A. P. Hunt and C. J. White, Nat. Rev. Chem., 2018, 2, 278–289.
32. N. Jawali, J. K. Saws and P. V. Sane, Phytochemistry, 1978, 17, 1527–1530.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5cc02803e

---