Revisiting Atmospheric Oxidation Kinetics of Nitrogen Oxides: The Use of Low-Cost Electrochemical Sensors to Measure Reaction Kinetics

Abstract

The high cost of equipment is a significant entry barrier to research for smaller organisations in developing solutions to air pollution problems. Low-cost electrochemical sensors have shown sensitivity at parts-per-billion by volume (ppbV) mixing ratios but are subject to variations due to changing environmental conditions, particularly temperature. We have previously demonstrated that under isothermal/isohume conditions such as those found in kinetic studies, very stable electrochemical responses occur. In this paper, we demonstrate the utility of a low-cost IoT-based sensor system that employs four-electrode electrochemical sensors under isothermal/isohume conditions for studying the kinetics of the atmospheric oxidation of nitrogen oxides. The results suggest that reproducible results for NO and NO₂ kinetics can be achieved. The method produced oxidation rates of 7.95 × 10³ L² mol⁻² s⁻¹ (±1.3%), for NO and 7.99 × 10⁻⁴ s⁻¹ (±2.1%) for NO₂. This study suggests that the oxidation kinetics of nitrogen oxides can be assessed with low-cost sensors, which can support a wide range of industrial applications, such as designing biocatalytic coatings for air pollution remediation.

1. Introduction

The thermal oxidation of nitric oxide was first described in 1918 by Bodenstein and Wackenheim [1]. Rate constants for this reaction measured by some 22 authors under “static and nominally water-free systems” have previously been summarised by Tsukahara et al. [2], but these systems are not representative of conditions found in an ambient atmospheric environment where, amongst other species, water vapour is present. The Tsukahara summary reported a reaction rate constant in the range of 1.6 × 10³–7.2 × 10³ L² mol⁻² s⁻¹. Nitric oxide concentration varied in those studies from 1.0 × 10⁻² mol L⁻¹ (343,000 ppmV) down to 8.2 × 10⁻⁸ mol L⁻¹ (2 ppmV). Several of the authors mentioned by Tsukahara also examined the influence of humidity on the reaction rate, with most authors finding no influence on the reaction rate under conditions of up to 90% relative humidity [1,3,4,5]. Treacy and Daniels [6], however, reported an increase in the reaction rate in the presence of water vapour.

England [7] examined the kinetics of the nitric oxide oxidation reaction at atmospheric pressure in the presence of water vapour, with the initial water vapour concentration being varied from 1.24 × 10⁻⁴ to 4.96 × 10⁻⁴ mol L⁻¹ at 25 °C, equating to 9.7% to 38.8% relative humidity respectively. A rate constant of 7.3 × 10³ L² mol⁻² s⁻¹ was determined under dry conditions, and the presence of water vapour had no specific catalytic effect in the gas phase. The NO concentrations examined by England were in the range 2.217 × 10⁻⁶ to 8.508 × 10⁻⁶ mol L⁻¹ (50.3–150.9 ppmV). Under these conditions the concentration of water vapour is some two orders of magnitude greater than that of nitric oxide. This disparity in reactant concentrations is more significant when compared to ambient atmospheric concentrations of nitric oxide generated through both natural and anthropogenic processes. In all of these studies, the concentration of nitrogen oxides and/or water vapour does not represent ambient conditions.

Peak daily concentrations of NO and NO₂ from an industrial city over a four-year period (Wollongong: 34.4248° S, 150.8931° E) [8]. Figure 1 is in the vicinity of 225 ppbV (9.2 × 10⁻⁹ mol L⁻¹) for NO, whilst those of NO₂ are typically lower in the vicinity of 40 ppbV (1.64 × 10⁻⁹ mol L⁻¹). Atmospheric relative humidity varies due to climatic conditions, but at a temperature of 25 °C and relative humidity of 50%, such as that specified for reference conditions in a paint/coatings test laboratory [9], water vapour concentration equates to 6.4 × 10⁻⁴ mol L⁻¹. Under these specific conditions, water vapour concentration is now four orders of magnitude greater than atmospheric NO or NO₂ concentrations.

Since the initial investigations by Bodenstein and Wackenheim [1], various analytical methods have been employed with present-day air quality monitoring stations employing chemiluminescence techniques, which are considered the standard for the measurement of nitrogen oxides for air quality monitoring [10,11,12,13]. Chemiluminescence detectors represent a significant capital investment not only in the actual equipment but also in the supply of ozone for oxidation, maintenance, and suitably trained personnel.

In recent years, the sensitivity of low-cost electrochemical sensors (LCESs) has improved to a point where accuracy at parts per billion by volume (ppbV) levels in the presence of other components of polluted atmospheric air have been achieved [14,15,16], making them suitable for air quality monitoring applications. For example, Meyerhoff et al. [17] found good agreement between a commercially available LCESs and a chemiluminescence detector.

It is well known that electrode responses, particularly those of nitric oxide LCESs, are susceptible to transients due to variations in ambient environmental conditions, in particular, temperature and relative humidity, with significant effort being applied to develop methods to overcome these deficiencies that range from simple post-data corrections to machine learning algorithms [14,15,16,18,19,20,21]. Studies where LCESs have been exposed to highly controlled environmental conditions, i.e., isothermal/isohume conditions within a sealed environmental (reaction) chamber, such as those found in kinetic studies, are beginning to appear in the literature [22,23,24]. We have previously demonstrated that under such isothermal/isohume conditions, very stable electrochemical responses occur. In this study, we assess the use of LCESs under isothermal/isohume conditions to measure the kinetics of the atmospheric oxidation for each nitrogen oxide individually and in isolation to support our work in designing biocatalytic coatings for air pollution remediation related to nitrogen oxides similar to the biocatalytic coatings reported by Estrada et al. for biodegradation of atmospheric VOC’s.

2. Materials and Methods

Instrument-grade air from BOC Gases (Minto, New South Wales, Australia) was used as a source of zero air for purging of the reaction chamber between experiments (

Table 1. Composition of ‘zero air’ used for purging the environmental chamber.

Component	Composition (mol/mole)
Nitrogen	78.1%
Oxygen	20.9%
Argon	0.9%
Water	<25 ppm

).

NO at a mixing ratio of 60 ppm ± 2% in nitrogen and NO₂ at a mixing ratio of 15 ppm ± 2% in nitrogen was obtained from CAC Gas and Instrumentation (Arndell Park, New South Wales, Australia).

The experimental setup using zero air, and NO gas was the same as previously described [24] with the following modification. Four-electrode electrochemical sensors operate in a continuous analysis mode, i.e., they constantly oxidise or reduce the target gas depending on the sensor’s electrochemistry. Were the sensor system to be placed inside the reaction chamber, the continued oxidation or reduction of the target gas by the sensors would deplete the gas concentration at a significantly faster rate than would occur under ambient conditions, thus making the setup unsuitable for kinetic type experiments. To overcome the issue described above, a Teflon hood typically used for sensor calibration is attached to the electrochemical sensors to isolate the LCESs from the chamber environment (Figure 2). The volume of the calibration hood was determined to be 0.011 L, representing 0.14% of chamber volume. At predetermined times, chamber air is pumped through the Teflon hood using a micropump. The outlet of the Teflon hood contains a 2-Position 3-Way Electric Control Solenoid Valve that, when closed, isolates the electrochemical sensors from the bulk chamber environment.

Using this modified system, we first conducted a series of experiments to determine appropriate measurement cycle durations to ensure the gas sample being analysed was representative of the chamber air. Following this, we were then able to measure the reaction kinetics of both NO and NO₂ under varying atmospheric conditions, comparing our results to those previously published.

3. Results and Discussion

3.1. Determination of Appropriate Measurement Cycle Durations

Chamber gas can be drawn through the sensor measurement cell at predetermined times for analysis without affecting the bulk chamber composition. Once gas ceases to flow through the measurement cell and across the sensors, oxidation or reduction of the sample within the measurement cell continues until the target gas is depleted.

Figure 3 displays a situation where varying starting concentrations of NO gas were added to the reaction chamber, and then the chamber gas was continually pumped through the measurement cell across the LCES and analysed, demonstrating the ongoing depletion. It was typically found that a minimum of 3 min purging of the measurement cell with chamber gas using the above method was required to ensure that the sample being analysed was representative of the chamber composition. Following purging, we also determined that averaging a 60 s measurement time with a 1 s frequency provided reproducible results. Once gas flow ceased, we found that NO gas was depleted in approximately 10 min, irrespective of the initial concentration.

3.2. Measurement of Reaction Kinetics Using LCESs

Most recently, Skalska et al. [25] conducted kinetic measurements under anhydrous conditions (Figure 4a), which demonstrated that NO₂ concentration increases in tandem with NO consumption. In their study, the concentration of NO was in the range 6.611 × 10⁻⁶ to 2.988 × 10⁻⁵ mol L⁻¹ (150–530 ppmV), once again significantly higher than ambient. During our preliminary experiments in a humid environment where NO concentration approximated normal ambient levels, we were unable to detect any appreciable NO₂ under the measurement conditions employed (Figure 4b).

Subsequent experiments using the same purge time and measurement intervals as these preliminary experiments, where a stepwise addition of target gas was introduced to the chamber, demonstrated a faster decline in NO₂ concentration compared to that of NO under the same conditions (Figure 5). These stepwise addition experiments confirmed that the NO₂ sensor was able to detect its target gas at the concentrations and chamber conditions employed (Figure 4) and suggests that any NO₂ formed via NO oxidation in our duplication of the Skalska experiment under humid conditions was being consumed by a subsequent reaction at a rate faster than the detection ability of the NO₂ sensor under the measurement conditions.

By following the depletion of NO₂ at the conclusion of the stepwise experiment displayed in Figure 5, we were able to confirm that the NO₂ depletion follows first-order kinetics, confirming previous results published by England et al. [26] (Figure 6a,b). Interpolation of these data allowed a rate constant of 7.99 × 10⁻⁴ s⁻¹ and a half-life of 866 s (14.4 min) to be determined. The half-life of this reaction is significantly faster than that of NO oxidation, explaining the lack of detection of NO₂ in the NO oxidation experiments. England proposed that under the conditions studied, three principal reactions were believed to occur, with Equation (3) occurring at low rates. A rate constant of 1.5 × 10⁵ L² mol⁻² s⁻¹ was determined by England for Equation (2), representing a 20-fold increase compared to Equation (1).

$$2\mathrm{N}{\mathrm{O}}_{\left(\mathrm{g}\right)}+{\mathrm{O}}_{2\left(\mathrm{g}\right)}\to 2\mathrm{N}{\mathrm{O}}_{2\left(\mathrm{g}\right)}$$

(1)

$${\mathrm{N}\mathrm{O}}_{\left(\mathrm{g}\right)}+{\mathrm{N}\mathrm{O}}_{2\left(\mathrm{g}\right)}+{\mathrm{H}}_2{\mathrm{O}}_{\left(\mathrm{g}\right)}\to {2\mathrm{H}\mathrm{N}\mathrm{O}}_{2\left(\mathrm{g}\right)}$$

(2)

$${3\mathrm{N}\mathrm{O}}_{2\left(\mathrm{g}\right)}+{\mathrm{H}}_2\mathrm{O}\left(\mathrm{g}\right)\to {\mathrm{N}\mathrm{O}}_{\left(\mathrm{g}\right)}+{2\mathrm{H}\mathrm{N}\mathrm{O}}_{3\left(\mathrm{g}\right)}$$

(3)

The short half-life of the reaction of the NO₂ when pulse sampled through the measurement cell could potentially lead to inaccurate results due to the combination of “natural” reactions and sensor-induced depletion. To overcome these potential inaccuracies, a series of experiments where the NO₂ LCES was directly exposed to the chamber environment was undertaken to evaluate the efficacy of the NO₂ sensor. In these depletion experiments, NO₂ concentrations in the range 50–250 ppbV (3.13 × 10⁻⁹ to 1.57 × 10⁻⁸ mol L⁻¹) were employed, at a fixed temperature and relative humidity of 25 °C (298 K) and 50% (6.4 × 10⁻⁴ mol L⁻¹ water vapour) respectively (Figure 7). These values more closely approximate ambient conditions.

In addition, three experiments depicted in Figure 8 were conducted at 25 °C (298 K) and a fixed NO₂ concentration of 200 ppbV (1.25 × 10⁻⁸ mol L⁻¹) with relative humidity varying from 30% to 70% (3.84 × 10⁻⁴ to 8.96 × 10⁻⁴ mol L⁻¹ water vapour). The results of these two sets of experiments indicate that firstly, the rate constant decreases with increasing starting NO₂ concentration (Figure 7c) and increases slightly as relative humidity increases given constant starting NO₂ concentrations (Figure 8c).

3.3. Nitric Oxide Kinetics

Nitric oxide oxidation rates are notably slower than those of NO₂, and therefore, pulsed sampling techniques are required. Figure 9a details the change in the measured rate constant with varying sampling intervals and concludes that a sampling interval between 8–12 h is required before stabilisation of the rate constant is achieved (Figure 9b). Having determined the appropriate sampling time for pulsed experiments, we next conducted kinetic experiments for NO depletion at a range of concentrations between 100–2000 ppbV (4.09 × 10⁻⁹ to 8.18 × 10⁻⁸ mol L⁻¹), at fixed temperature and relative humidities of 25 °C (298 K) and 50% (6.4 × 10⁻⁴ mol L⁻¹ water vapour), respectively (Figure 10,

Table 2. Rate constants for NO oxidation in a humid environment (data from Figure 10c).

[NO]/ppbV	[NO]/(mol L−1)	Rate Constant/(L2 mol−2 s−1)
100 500 1000 1500 2000	4.09 × 10−9 2.04 × 10−8 4.09 × 10−8 6.13 × 10−8 8.18 × 10−8	1.37 × 104 1.72 × 104 1.52 × 104 1.22 × 104 7.95 × 103

).

Our measured rate constant at 2000 ppbV (8.18 × 10⁻⁸ mol L⁻¹) using a 12-h sampling time of 7.95 × 10³ L² mol⁻² s⁻¹ is slightly outside the range reported by Tsukahara (7.7 × 10³ L² mol⁻² s⁻¹) but is explainable in terms of either the presence of high concentrations of water vapour in our experiments or more likely, surface area to volume ratios (S/V) of the reaction chamber. This is supported by England [26], who reported a 21% increase in the formation of nitrogen dioxide during humid nitric oxide oxidation when the S/V ratio of their reactor was changed from 0.335 to 1.2 cm⁻¹ following the addition of glass beads.

The rate constant increased with decreasing initial NO concentration, reaching a peak at approximately 500 ppbV before again decreasing. The reason for this decrease at concentrations below 500 ppbV was not apparent and requires further investigation (Figure 10c). Unlike with the NO₂, we detected negligible change in rate constant when relative humidity was varied from 30% to 70% (3.84 × 10⁻⁴ to 8.96 × 10⁻⁴ mol L⁻¹ water vapour).

4. Conclusions

Previous investigations into the oxidation of nitric oxide have been undertaken under conditions that do not represent ambient atmospheric conditions. Using the LCES approach described herein, good agreement with published rate constants was found for NO oxidation rates when concentrations approaching 2 ppmV, as used in those previous studies, were employed. When lower NO and higher water vapour concentrations, such as those found under ambient atmospheric conditions, are considered in experimental conditions, reaction rates that differ from established values were found. The presence of high concentrations of water vapour contributed to side reactions that removed NO₂ formed from the oxidation of NO from the system; however, the LCES device was still able to generate reproducible results given the correct sampling method.

From a cost perspective, the use of LCESs provides a significant advantage over chemiluminescence detectors and their associated costs, allowing smaller research facilities, such as small to medium enterprises in the private sector, to participate in research designed to mitigate atmospheric nitrogen oxide pollutants.

In this study, we have demonstrated that LCESs, when used under isothermal/isohume conditions, are a viable alternative to chemiluminescence detectors to measure the kinetics of atmospheric NO and NO₂ oxidation. Pulse sampling is required for NO gas as spurious results would be obtained as the natural rate of atmospheric oxidation is significantly longer than oxidation resulting from the LCES. In contrast, rapid, non-pulsed sampling is required for NO₂ as the natural reaction rate under humid conditions exceeds that of the gas consumption by the LCES.

We believe that this is the first time that such a system has been used to study the oxidation kinetics of nitrogen oxides and was specifically developed to support our work in designing biocatalytic coatings for nitrogen oxides related air pollution remediation.