DEMS-monitoring liquid | gas interfacial ammonia oxidation at carbon nanofibre membranes

Abstract

Electrochemical ammonia oxidation is of interest in waste treatment as well as in electrochemical sensing applications and demonstrated here at a carbon nanofibre (“bucky-paper”) electrode. The electrode is placed at the aqueous electrolyte | gas interface, and current (cyclic voltammetry) as well as ambient differential electrochemical mass spectrometry (DEMS, cyclic voltbarometry) data are recorded as a function of solution composition and pH. The oxidation of oxalate to CO₂ is employed as a test and calibration system. Anodic polarization of the carbon nanofibre membrane in inert aqueous electrolyte is shown to result in direct sustained anodic CO₂ evolution. In alkaline aqueous media (starting at pH 9) significant levels of nitrogen from ammonia are produced in competition to CO₂ formation from carbon nanofibres without the need for additional catalysts. However, for applications with low level ammonia, catalysts will be required to minimize current losses, carbon nanofibre corrosion, and side product formation.

---

1. Introduction

Carbon nanofibre electrodes have found wide applications in electrochemical energy storage,^1,2 desalination,³ sensing devices,⁴ and in the form of composites⁵ or nano-carpets⁶ for electroanalysis. Beneficial effects derive from (i) high surface area, (ii) good electronic conductivity, and (iii) a diameter only slightly bigger than the interfacial double layer. Carbon nanofibre materials are produced in bulk, but can be oxidized in mineral acids and compacted via filtration to give stable “bucky paper” membranes for application in electrochemistry.⁷ Here, a 50 μm thick carbon nanofibre membrane is employed as an electrode at the interface between gas and aqueous electrolyte to allow online monitoring of gas evolution processes under ambient pressure conditions. The oxidation of ammonia to nitrogen is observed at the carbon nanofibre membrane electrode.

Koper and coworkers have reviewed the electrochemistry of nitrogen species⁸ and in particular the oxidation of ammonia in alkaline aqueous media which is of relevance in waste water treatment, fuel cells, and sensing. The electro-oxidative removal of ammonia (in terms of total ammonia nitrogen TAN) from reverse osmosis concentrates⁹ has been proposed. Relatively little is known about the ammonia oxidation on carbon electrodes, although carbon often features as a catalyst support for metal nano-catalysts.¹⁰ Several reports have appeared on the direct oxidation of ammonia at boron-doped diamond electrodes¹¹ where the presence of NH₂˙ radical intermediates may lead to hydrazine intermediates and other electrolysis products.

Differential electrochemical mass spectrometry (or DEMS) methods have been pioneered by Brukens¹² and Baltruschat¹³ and are now widely applied in electro-synthesis,¹⁴ mechanistic studies in waste water treatment,¹⁵ and fuel cell process analysis.^16,17 In DEMS studies, porous electrodes based on thin Teflon® or other such membrane materials are usually employed to separate the low pressure MS inlet nozzle from the aqueous solution phase. Here, a carbon nanofibre membrane is employed directly without applied pressure gradient. An ambient pressure purge gas, usually argon, is flowed over the electrode to pick up product gases for MS analysis.

In this study, a highly porous carbon nanofibre membrane (“bucky paper”) is employed to investigate the oxidation of ammonia in aqueous electrolyte media. In exploratory experiments, both current responses (cyclic voltammograms) and product partial pressure responses (cyclic voltbarograms) are recorded simultaneously to provide mechanistic insight and to allow process monitoring. Potential-dependent competition or N₂ evolution from ammonia and CO₂ evolution from carbon nanofibres is observed and discussed in terms of water treatment processes.

2. Experimental

2.1. Reagents

Chemical reagents such as Na₂SO₄, (NH₄)₂SO₄, Na₂(C₂O₄), and NaOH were obtained from Sigma-Aldrich in analytical reagent grade and used without further purification. Demineralised and filtered water was taken from a Millipore water purification system with no less than 18 MΩ cm resistivity.

2.2. Instrumentation

Voltammetric experiments were performed with an Autolab PGSTAT12 system (Ecochemie, Netherlands) in staircase voltammetry or galvanostatic mode. The counter and reference electrode were platinum gauze and KCl-saturated calomel (SCE, Radiometer), respectively and immersed into the aqueous electrolyte (see Fig. 1). The working electrode was a 4 mm diameter carbon nanofibre membrane disc (ca. 50 μm thick “Bucky paper”, Nanolabs US, with low resistivity (∼0.1 Ohm cm) and relatively low impurity levels (Fe 0.36, Si 0.31, Al 0.23, Na 0.32, S 0.23 at.%) mounted with Ambersil silicone (Silicoset 151) on a glass capillary of 3.5 mm inner diameter and 5 mm outer diameter (Fig. 1). The electrical contact was made with a 1 mm stripe of pyrolytic graphite film (Goodfellow, UK) inside of the glass capillary. Solutions were de-aerated with argon (Pureshield, BOC). The pH was measured with a glass electrode (3505 pH meter, Jenway). All experiments were conducted at a temperature of 22 ± 2 °C.

Fig. 1 (A) Photograph of the micro-reactor cell with carbon nanofibre membrane in a glass capillary with 3.5 and 5.0 mm inner and outer diameter, respectively. (B) Schematic showing the carbon nanofibre membrane working electrode (W), counter (C) and reference (R) electrodes, and gas inlet and outlet.

2.3. Carbon nanofibre membrane electrochemistry with in situ mass spectrometry

The electrochemical cell for carbon nanofibre membrane electrochemistry coupled to ambient differential electrochemical mass spectrometry (DEMS) is shown in Fig. 1. A glass capillary cell with the ca. 50 μm thick carbon nanofibre membrane attached at the end serves as a small volume ambient sampling space with a continuous argon flow of ca. 100 mL min⁻¹. The mass spectrometer is a Universal gas Analyzer UGA 200 (Quadrupole) from Stanford Research Systems. Any gasses produced at the electrode surface can escape into the gas flow with coupled MS detection. When exposed to ambient air the mass readings are for N₂ 5.8 μTorr (mass 28, ambient partial pressure 592 Torr, corresponding to a sensitivity factor 1.0 × 10⁻⁸), O₂ 1.1 μTorr (mass 32, ambient partial pressure 159 Torr, corresponding to a sensitivity factor 0.7 × 10⁻⁸), and CO₂ 87 nTorr (mass 44, ambient partial pressure 30 mTorr, corresponding to a sensitivity factor 3.3 × 10⁻⁶).

The working electrodes were fabricated from a disc of carbon nanofibre that was adhered to the bottom of a glass tube in contact with a graphite electrical connection.¹⁸ A gas inlet and outlet were sealed into the top of the tube. A carrier gas, argon, was flowed through the glass capillary (ca. 100 mL min⁻¹) and the mass spectrometer inlet continuously sampled the gas space (see Fig. 1B) under ambient conditions.

3. Results and discussion

3.1. Voltammograms and voltbarograms at carbon nanofibre membrane electrodes (i): oxalate oxidation and background processes

Initial experiments are carried out to characterize the electrochemical cell with carbon nanofibre membrane and the mass spectroscopy-based partial pressure monitoring system. An aqueous test solution is employed containing 0.1 M sodium oxalate. Fig. 2A shows a typical cyclic voltammogram for the oxidation of oxalate (process P1, see eqn (1)) at the carbon nanofibre membrane.

C₂O₄²⁻_(aq) ⇌ 2CO_2(gas) + 2e⁻ (1)

Fig. 2 (A) Cyclic voltammogram and (B) cyclic CO₂ voltbarogram (scan rate 0.005 V s⁻¹) for the oxidation of aqueous 0.1 M sodium oxalate at a carbon nanofibre membrane electrode. (C) As above, but cyclic voltammograms and (D) cyclic CO₂ voltbarograms (scan rate (i) 0.005, (ii) 0.01, and (iii) 0.05 V s⁻¹).

The onset of the oxalate oxidation is observed at 0.8 V vs. SCE with a voltammetric peak response at ca. 1.3 V vs. SCE (see Fig. 2A). Simultaneously, the differential mass spectrometry response for mass 44 corresponding to CO₂ is monitored and the voltbarometric peak response (Fig. 2B) is seen to be shifted positive by approximately 75 mV or 15 s (scan rate 5 mV s⁻¹). This 15 s delay is caused predominantly by the transport of the sample into the MS detection chamber.

The effect of the potential scan rate is shown in Fig. 2B. An increase in the scan rate causes a corresponding increase in the apparent delay time for the voltbarometric response. However, the voltammetric current signal also shows considerable deterioration at 50 mV s⁻¹ due to resistance effects (ca. 500 Ohm based on the rising slope of the voltammetric peak response; mainly due to the graphite–carbon nanofibre contact). When investigated in more dilute oxalate solution, the partial pressure response for the carbon dioxide product is considerably decreased.

Fig. 3A and 3B show voltammetric and voltbarometric data for the oxidation of 10 mM sodium oxalate in 90 mM sodium sulfate. Both the current and partial pressure responses for process P1 are clearly resolved, but also a second more positive process P2 can be seen with an onset potential of ca. 1.2 V vs. SCE. This process is identified here as the background oxidation of carbon nanofibre material to CO₂ (P2, eqn (2)).

C_{(carbon nanofibre)} + 2H₂O_(aq) ⇌ CO_2(gas) + 4e⁻ + 4H⁺_(aq) (2)

Fig. 3 (A) Cyclic voltammogram and (B) cyclic CO₂ voltbarogram (scan rate 0.005 V s⁻¹) for the oxidation of aqueous 0.01 M sodium oxalate in 0.09 M sodium sulfate at a carbon nanofibre membrane electrode. (C) Cyclic voltammogram and (D) cyclic CO₂ voltbarogram for the oxidation of aqueous 0.001 M sodium oxalate in 0.099 M sodium sulfate. (E) Cyclic voltammogram and (F) cyclic CO₂ voltbarogram for the oxidation of aqueous 0.1 M sodium sulfate.

This oxidation process is sustained and not a simple surface oxidation. Formation of CO is not detected. The process can be understood in terms of the formation of edge carboxylate functionalities which undergo oxidative Kolbe-decarboxylation followed by further formation of carboxylate groups. Based on literature reports,¹⁹ the formation of CO₂, CO (at elevated temperatures), and other often non-volatile products such as humics and colloidal carbon are anticipated for acetylene blacks. However, the chemical nature of carbon nanofibre materials is different and possibly beneficial at least in the low current–low corrosion regime employed here. With 1 mM oxalate (Fig. 3C and 3D) and without oxalate (Fig. 3E and 3F) only the background process remains significant.

Next, the linearity of partial pressure response with respect to current response is investigated. Fig. 4A shows data for a galvanostatically controlled chronobarometry experiment for the oxidation of 1 mM oxalate in 99 mM sodium sulfate. The current is changed stepwise from 0 mA to 3.2 mA and back. The chronobarogram for CO₂ (mass 44, see Fig. 4Ai) shows an initial step, but then only poor correlation to the applied current. The chronobarogram for O₂ (mass 32, see Fig. 4Aii) was recorded simultaneously. When taken together and assuming a 1-electron oxidation for CO₂ and a 4-electron oxidation for O₂, a chronobarogram closely matching the applied current trace (see Fig. 4Aiii) is obtained. Therefore both product gases are released into the gas space (see schematic diagram in Fig. 4Aiv) and the mass spectrometry data are linearly correlated to current flow. However, this is not always necessarily the case, as discussed below.

Fig. 4 (A) Chronobarometry data for (i) CO₂ (mass 44) and for (ii) O₂ (mass 32) recorded for the oxidation of 0.001 M sodium oxalate in 0.099 M sodium sulfate at a carbon nanofibre membrane electrode during a sequence of applied current steps. Also (iii) the sum pressure for pCO₂ + 4 pO₂ (assuming a 1-electron and a 4-electron process) is plotted over the current trace to demonstrate good correlation of current and product pressure. The schematic drawing (iv) illustrates a situation where all gaseous products escape into the gas flow to give linear correlation of sum pressure and current. (B) Chronobarometry data for (i) CO₂ (mass 44) recorded for the oxidation of 0.1 M sodium oxalate. The schematic drawing (ii) illustrates CO₂ gas formation into the aqueous phase causing non-linear product pressure responses.

The same experiment repeated in aqueous 0.1 M sodium oxalate solution shows a very different behaviour (see Fig. 4B). There is no significant O₂ evolution (not shown) and the CO₂ evolution chronobarogram is suppressed in the high current region. The schematic drawing in Fig. 4Bii suggests a process in which gas is produced predominantly at the aqueous solution interface of the carbon nanofibre membrane and CO₂ is lost into the solution phase (bubbles can be seen at the working electrode). Therefore, although quantitative correlation of mass and current data is possible in some cases, possible losses of product into the solution phase have to be considered and the mass response may not always correlate to the current response.

3.2. Voltammograms and voltmassograms at carbon nanofibre membrane electrodes (ii): competition of CO₂ and N₂ formation during ammonia oxidation

The anodic conversion of aqueous ammonia to nitrogen is a complex multi-step process (see eqn (3), P3) which has recently been observed on graphitic materials. Carbon as an electrode material often offers a low cost and less polluting alternative compared to transition metal or DSA electrode systems.²⁰ Carbon is usually considered non-catalytic and relatively inert towards complex multi-electron processes such as oxygen, CO₂, or nitrogen evolution. Here it is shown that ammonia oxidation on carbon nanofibre membrane electrodes can compete with carbon dioxide formation, at least for sufficiently high concentrations of ammonia.

2NH_3(aq) + ⇌ N_2(gas) + 6e⁻ + 6H⁺_(aq) (3)

Fig. 5A shows cyclic voltbarograms for N₂ and CO₂ recorded during an applied potential cycle going from 0.0 to 1.5 V vs. SCE. For a solution containing 0.1 M ammonium sulfate adjusted with NaOH to pH 10 and at positive anodic potentials, product peaks for both N₂ and CO₂ are seen. A similar experiment conducted at pH 9 did not show any significant evidence for N₂ (not shown). This onset of ammonia oxidation to N₂ is consistent with literature reports which indicate that an alkaline solution with a pH of 9 or more is required.²¹

Fig. 5 (A) Cyclic voltbarograms (scan rate 0.005 V s⁻¹, time trace) for N₂ (mass 34) and CO₂ (mass 44) for the oxidation of 0.1 M ammonium sulfate at pH 10 (adjusted with NaOH) at a carbon nanofibre membrane. (B) As before at pH 11. (C) As before at pH 12. (D) As before at pH 13. (E) Cyclic voltbarograms (scan rate 0.005 V s⁻¹, time trace) for N₂ (mass 34) for the oxidation of 0.4 M, 0.7 M, 1.3 M, and 2.5 M ammonium sulfate at pH 13 (adjusted with NaOH) at a carbon nanofibre membrane (note relative pressure scale for individual plots, offset applied to enhance visibility).

In a progression of experiments for the oxidation of 0.1 M ammonium sulfate adjusted to pH 11, 12, and 13, the N₂ product response can be seen to systematically increase (see Fig. 5B–D). Interestingly, there is always a shift indicating an earlier onset for N₂ evolution compared to CO₂ evolution. An additional experiment was carried out for the oxidation of ammonia at pH 13 as a function of concentration (see Fig. 5E). Again a systematic increase in the voltbarometric response for N₂ evolution is observed in this case with higher ammonia concentration. It can be concluded that anodic conversion of ammonia to N₂ at carbon nanofibre membrane electrodes is possible (i) at a potential of ca. 1.5 V vs. SCE and (ii) at a solution pH > 9. For a low concentration of ammonia, the competition to carbon dioxide evolution may become unfavorable and sustained use of the carbon nanofibre electrode will result in corrosion damage. In order to avoid or to suppress the concomitant carbon dioxide evolution doped diamond materials²² could be employed, new electro-spun nano-fibre materials employed,²³ catalysts introduced,²⁴ or the effect of the solution composition on the anodic CO₂ evolution process at nano-carbon materials could be investigated in more detail.

4. Summary & conclusion

In exploratory experiments, oxidation of ammonia to nitrogen is observed in situ for the direct electrolysis at carbon nanofibre membrane electrodes. A DEMS system for online mass spectrometry monitoring is used to compare rates for N₂, CO₂, and O₂ evolution as a function of potential and for sufficiently high ammonia levels effective nitrogen formation occurs.

The role of O₂ in the ammonia oxidation process (leading to formation of NO_x) at non-adsorbing electrode surfaces such as boron-doped diamond has been highlighted by Michels et al.²² and effects from these processes are avoided here due to the use of an argon atmosphere and high ammonia concentration. Problems associated with oxygen co-evolution are likely at lower ammonia levels and require more study.

Carbon nanofibre membrane oxidation of ammonia can be viable only if (i) corrosive CO₂ evolution can be minimized, (ii) co-evolution of O₂ is suppressed, and (iii) sufficient alkalinity is induced locally to increase the concentration of free NH₃. Data obtained as a function of ammonia concentration suggests that catalysts will be required for the conversion of low level ammonia to nitrogen without formation of NO_x.

Acknowledgements

R.A.W. thanks Johnson Matthey, GWR, and EPSRC for funding (GWR-DTC studentship for R.A.W.-EP/G03768X/1)

References

1. C. Fongy, S. Jouanneau, D. Guyomard and B. Lestriez, J. Power Sources, 2011, 196, 8494.
2. L. W. Ji, Z. Lin, M. Alcoutlabi and X. W. Zhang, Energy Environ. Sci., 2011, 4, 2682.
3. L. K. Pan, X. Z. Wang, Y. Gao, Y. P. Zhang, Y. W. Chen and Z. Sun, Desalination, 2009, 244, 139.
4. E. Lahiff, C. Lynam, N. Gilmartin, R. O'Kennedy and D. Diamond, Anal. Bioanal. Chem., 2010, 398, 1575.
5. F. Marken, M. L. Gerrard, I. M. Mellor, R. J. Mortimer, C. E. Madden, S. Fletcher, K. Holt, J. S. Foord, R. H. Dahm and F. Page, Electrochem. Commun., 2001, 3, 177.
6. A. G. Guell, K. E. Meadows, P. R. Unwin and J. V. Macpherson, Phys. Chem. Chem. Phys., 2010, 12, 10108.
7. R. A. Webster, F. J. Xia, M. Pan, S. C. Mu, S. E. C. Dale, S. C. Tsang, F. W. Hammett, C. R. Bowen and F. Marken, Electrochim. Acta, 2012, 62, 97.
8. V. Rosca, M. Duca, M. T. de Groot and M. T. M. Koper, Chem. Rev., 2009, 109, 2209.
9. K. Van Hege, M. Verhaege and W. Verstraete, Water Res., 2004, 38, 1550.
10. T. L. Lomocso and E. A. Baranova, Electrochim. Acta, 2011, 56, 8551.
11. N. J. Bunce and D. Bejan, Electrochim. Acta, 2011, 56, 8085.
12. S. Bruckens and R. Raogadde, J. Am. Chem. Soc., 1971, 93, 793.
13. H. Baltruschat, J. Am. Soc. Mass Spectrom., 2004, 15, 1693.
14. E. M. Belgsir, E. Bouhier, H. E. Yei, K. B. Kokoh, B. Beden, H. Huser, J. M. Leger and C. Lamy, Electrochim. Acta, 1991, 36, 1157.
15. A. Kapalka, B. Lanova, H. Baltruschat, G. Foti and C. Comninellis, Electrochem. Commun., 2008, 10, 1215.
16. H. S. Wang and H. D. Abrunña, in Fuel Cells and Hydrogen Storage, Structure and Bonding, A. Bocarsly, D.M.P. Mingos, (ed.) , 2011, 141, 33.
17. H. S. Wang, E. Rus and H. D. Abrunña, Anal. Chem., 2010, 82, 4319.
18. J. D. Watkins, S. D. Ahn, J. E. Taylor, S. D. Bull, P. C. Bulman-Page and F. Marken, Electrochim. Acta, 2011, 56, 6764.
19. P. N. Ross and H. Sokol, J. Electrochem. Soc., 1984, 131, 1742.
20. L. Candido and J. A. C. P. Gomes, Mater. Chem. Phys., 2011, 129, 1146.
21. A. Kapalka, S. Fierro, Z. Frontistis, A. Katsaounis, S. Neodo, O. Frey, N. de Rooij, K. M. Udert and C. Comninellis, Electrochim. Acta, 2011, 56, 1361.
22. N. L. Michels, A. Kapalka, A. A. Abd-El-Latif, H. Baltruschat and C. Comninellis, Electrochem. Commun., 2010, 12, 1199.
23. L. W. Ji, Z. Lin, M. Alcoutlabi, O. Toprakci, Y. F. Yao, G. J. Xu, S. L. Li and X. W. Zhang, RSC Adv., 2012, 2, 192.
24. M. T. M. Koper, Nanoscale, 2011, 3, 2054.

---