WorksPhysics and Astronomy
Modelling the photocatalytic oxidation of methane and other air pollutants for applications in ventilation systems
Abstract
Photocatalytic oxidation (PCO) is a promising strategy for indoor air purification and outdoor pollutant abatement, potentially offering treatment for climate- and health-relevant pollutants such as methane (CH), nitrogen oxides (NO) and volatile organic compounds (VOCs). In this work, we present experiments evaluating the PCO of CH (2 to 10 ppm) under varying UV-C light intensities (4 to 59 W/m), using titanium dioxide (TiO) as the photocatalyst. At 2 ppm CH, TiO achieves a maximum conversion efficiency of 24.4% and a maximum apparent quantum yield of % over the tested UV-C light intensities, demonstrating activity at environmentally relevant concentrations. We develop a model to interpret the experimental results and assess the potential of PCO for ventilation applications. The model is validated against our CH data and literature results for formaldehyde (HCHO) and NO. While laboratory-scale configurations achieve high conversions (e.g., 24.4% for CH), ventilation-scale performance is predicted to be limited by thin concentration boundary layers and short residence times, with conversion efficiencies dropping to around %. Finally, we estimate the climate impact of CH removal in terms of COe emission rates, demonstrating that TiO-based PCO in ventilation applications can yield a net climate benefit (i.e., a net-negative COe emissions rate) when the modelled COe removal rate exceeds the emissions from catalyst material production and UV operation, particularly when pre-existing UV-C irradiation is leveraged.
Introduction
Air pollution is among the leading environmental and health risks worldwide, contributing significantly to respiratory disease, cardiovascular morbidity and premature mortality [1, 2, 3]. While outdoor air pollution has historically drawn more attention, due in part to its connection with climate change [4, 5], recent studies highlight that indoor air quality can be equally hazardous, especially as people now spend over 90% of their time indoors, in homes, workplaces, schools and transport systems [6]. This shift places indoor air quality at the forefront of public health and building design. Indoor environments can be polluted with volatile organic compounds (VOCs), nitrogen oxides (NO), ozone (O), methane (CH) and nitrous oxide (NO), originating from building materials, cleaning products, combustion sources and infiltration from outdoor air [1, 2]. Some of these pollutants are harmful at parts-per-billion levels and have been linked to long-term health effects, including asthma, neurotoxicity and cancer. Outdoor air, meanwhile, contains mixtures of anthropogenic and biogenic species. CH and NO are potent greenhouse gases (GHGs) with global warming potentials (GWPs) many times greater than carbon dioxide (CO) [4, 7]. NO and VOCs serve as precursors to secondary pollutants, including tropospheric O and fine particulate matter (PM) [8, 9]. Despite regulatory efforts, progress in reducing emissions of these pollutants from sources (e.g., vehicles, industry, agriculture) has been slow. To address these diffuse emissions, there is interest in integrating removal technologies (e.g., catalysts) into built environments that experience substantial airflow (e.g., building exteriors, ventilation systems), as part of a distributed approach to air-quality management [10, 11, 12].
Photocatalytic oxidation (PCO) has emerged as a potentially low-energy and scalable strategy to degrade air pollutants [13, 14]. In these systems, a photocatalyst, typically titanium dioxide (TiO), is irradiated with UV or visible light, generating reactive oxygen species such as hydroxyl radicals (OH·) and superoxide anions (O) [6]. These reactive oxygen species can oxidise adsorbed pollutants into less harmful or less radiatively active products, without requiring chemical reagents or frequent replacement. PCO technologies have been explored in diverse applications, including building exteriors and indoor air purifiers [12, 15, 16, 17]. Their promise lies in continuous pollutant degradation under natural or artificial illumination. However, widespread adoption remains limited by uncertainties in performance under realistic conditions, where pollutant levels are low, airflows are complex and operational constraints such as humidity and light intensity vary [14, 15]. Beyond catalyst material selection and operating conditions, the overall feasibility of pollutant-removal technologies also depends on fabrication, installation and energy inputs, all of which influence the net climate impact of deployed systems.
The performance of PCO has been shown to be sensitive to operating conditions [6, 13, 14, 15, 17], including air pollutant concentration, light intensity, humidity, catalyst properties and flow rate. Bridging the gap between laboratory-scale performance and effectiveness in application will require models that incorporate both reaction kinetics and mass transport and are validated against multiple experimental measurements for different air pollutants. It is critical to distinguish between reaction-limited and mass-transfer-limited regimes [13, 14], which can exhibit similar behaviour but require different design strategies (e.g., improving catalyst activity in the former, or enhancing flow conditions in the latter). Overly simplified models may obscure these distinctions, whereas over-parameterised models or complicated numerical simulations can lack physical interpretability and may not readily support optimisation.
Recent theoretical and experimental studies have begun to address some of these issues [6, 13, 14, 15, 17]. Zhang et al. [13] developed a model coupling the Navier–Stokes equations, advection–diffusion transport and a Langmuir–Hinshelwood kinetic model, to predict NO removal using UiO-66-NH across varying light intensities, pollutant concentrations and humidity levels. Timmerhuis et al. [14] investigated the role of the Péclet and Damköhler numbers in characterising Bisphenol A removal via TiO, demonstrating that neglecting mass transport can lead to errors in rate constant predictions. Malay eri et al. [6] reviewed reactor configurations and photocatalysts for VOC removal, highlighting the role of catalyst exposure, photon-flux distribution and boundary-layer effects in system performance. Yu et al. [15] investigated a Trombe-wall PCO system for formaldehyde (HCHO) removal using a TiO-coated ceramic and developed a kinetic model accounting for UV intensity, temperature, humidity and concentration. Ballari et al. [17] combined experimental and theoretical studies of removal using -coated concrete pavers and used a Langmuir–Hinshelwood kinetic model to capture the effects of irradiance, humidity and flow conditions. Collectively, these studies highlight the variety of operational conditions and the effectiveness of PCO systems at laboratory scale.
Russell et al. [16] conclude that deployments of coatings in outdoor applications typically yield ambient pollutant concentration reductions of around 2% in the immediate vicinity of treated surfaces. In addition, formation of secondary species has been observed in some studies, such that air-quality impacts depend on surface reactivity and atmospheric chemistry. Cost analyses suggest that using photocatalytic oxidation for atmospheric removal is prohibitively expensive due to the high flow rates required for effective removal [10, 18, 19]. Further analysis shows that targeting higher-concentration sources can yield meaningful climate benefits at source concentrations above 10 ppm, with cost-effective operation using current technologies requiring concentrations above ppm [20]. Chlorine and OH· radicals can oxidise in the gas phase, and laboratory photochemical systems have reported relatively low energy inputs for low-concentration pollutant removal [10, 21]. Photochemical destruction approaches have also been investigated for and fluorinated anaesthetic gases in exhaust streams [22].
Building on these foundations, the present study combines modelling and experiments to assess the performance of PCO systems for ventilation applications. We conduct -oxidation experiments under varying UV-C light intensities at environmentally relevant concentrations for indoor and outdoor air. These measurements inform the development of a model, solved using asymptotic and numerical methods, that couples non-linear Langmuir–Hinshelwood kinetics with advection–diffusion transport under UV-C illumination. We validate the model using our experimental data, as well as literature data for and HCHO oxidation using , over a range of flow rates, concentrations and light intensities. The model allows us to quantify the extent to which laboratory- and application-scale PCO systems are constrained by concentrations, boundary-layer thickness, residence time and geometry. We estimate the potential removal rates and the associated emissions rates for the PCO of , enabling assessment of PCO performance under realistic operating conditions and providing a framework for the design and optimisation of potential air-purification technologies.
The rest of the paper is organised as follows. In § 2, we describe the experimental setup and measurements for . § 3 presents the model, discusses the numerical methods and parameter fitting. In § 4, we present model and experimental results, investigating performance at lab and application scales, and estimating the potential climate benefits of PCO technologies for removal. § 5 summarises the results and applications of this modelling and measurement framework, and discusses potential future extensions.
Experiments
To evaluate the PCO of at environmentally relevant concentrations (typical of indoor and outdoor air), we designed a set of experiments informed by previous studies on catalyst performance [23, 24]. Comparative data for and HCHO are taken from Ballari et al. [17] and Yu et al. [15], respectively, and are discussed further in § 4.1. Experimental conditions, PCO measurements and the flow–reactor system are described in § 2.1. Figure 1(a) shows the experimental setup used to measure conversion efficiency. The kinetic parameters fitted to each dataset are described in § 3 and their values are listed in Table 1.

Figure 1. (a) Schematic of the laboratory setup used for the PCO experiments, showing the gas delivery system, UV-C illumination, reactor, GC, FID and TCD for CH removal over TiO. (b) Two-dimensional reactor model, representing laminar flow with advection, diffusion and reaction at the UV-C illuminated photocatalyst wall. (c) A potential ventilation application of this UV-C pollutant-removal technology and its connection to our laboratory setup.
| Ballari et al. [17] | Yu et al. [15] | Present work | |
| Flow and geometric quantities | |||
| Pollutant | NO | HCHO | CH |
| Concentration (ppm) | 0.11 to 1.09 | 0.17 to 0.9 | 2 to 10 |
| Centreline height (mm) | 1.5 | 5 | 10 |
| Length (mm) | 200 | 240 | 60 |
| Flow rate (L/min) | 3 to 5 | 1 to 5 | 0.1 |
| Temperature (°C) | 20 | 15 to 45 | 20 |
| Humidity (%) | 10 to 80 | 20 to 90 | 0 |
| UV-C (W/m) | 0.3 to 13 | 2.1 to 9 | 4 to 59 |
| Fitted constants | |||
| Light | 0.23 | 0.38 | 0.21 |
| Adsorption (m/mol) | 3.9 × 10 | 1.1 × 10 | 4.6 × 10 |
| Rate (mol·m(a)/(s·W)) | 5.22 | 1.48 | 0.084 |
| Non-dimensional numbers | |||
| Péclet | 2.50 to 4.17 | 2.50 | 0.085 |
| Damköhler | 3.45 | 0.26 to 0.46 | 0.016 to 0.027 |
| Langmuir | 0.54 to 5.71 | 0.26 to 1.36 | 0.17 to 0.84 |
Table 1. Summary of parameters for each of the PCO studies on CH, HCHO and NO removal using TiO. Quantities marked with an asterisk are fitted to the experimental data. Parameters held fixed across all studies are: catalyst (TiO), molecular diffusivity (1.8 × 10 m/s), dynamic viscosity (1.85 × 10 Pa·s) and pressure (1 atm).
Experimental setup and conditions
To evaluate the PCO of CH at low concentrations relevant to air-purification applications, titanium dioxide (TiO, Titanium (IV) oxide, anatase, nanopowder, nm, 99.7%, trace metals basis, Sigma-Aldrich) was selected as a representative photocatalyst. TiO is widely studied for its strong photocatalytic activity under UV-C illumination [6, 23], making it a suitable benchmark for evaluating CH oxidation rates. In addition to PCO performance, the broader climate benefit of PCO systems in potential ventilation applications is quantified by considering the energy consumption of the UV-C illumination and the catalyst’s embodied carbon, as discussed in § 4.6 and detailed in Appendix § 7. Practical deployment of PCO systems would be most effective when high photocatalytic activity is combined with low overall carbon intensity.
The PCO experiments were performed using a reactor system designed to evaluate CH removal at environmentally relevant concentrations under UV-C irradiation (Figure 1a). For each experiment, 2 g of TiO was deposited as a thin layer covering the bottom surface of the cylindrical reactor (38 cm total height, 6 cm diameter) and irradiated by a UV-C lamp delivering light intensities between 4 and 59 on the catalyst surface. The geometry adopted in the model (§ 3) is an idealisation of this configuration, with the centreline height (1 cm) chosen to preserve the residence time of the experimental setup. CH and O served as the active reactants, with the carrier gas. Gas flows were regulated using mass flow controllers (MFCs), with the total volumetric flow rate maintained at 100 mL/min for all experiments. This value falls within the range of flow rates reported for dilute methane abatement systems in recent literature [18]. Input concentrations ranged from 2 to 10 ppm, reflecting environmentally relevant levels for indoor and outdoor air. Experiments were conducted at ambient temperature (20 °C) under dry conditions (no added humidity). UV-C illumination was selected because the narrow spectral output of the lamp simplifies photon-flux calibration and AQY calculations, and because UV-C germicidal lamps represent the most widely deployed UV infrastructure in existing HVAC systems, making the results relevant to the deployment scenario described in § 4.4. is also active under UV-A illumination; the model structure is expected to carry over, but extrapolation to UV-A conditions would require separate calibration of model parameters, since ’s photon absorption efficiency is wavelength-dependent.
Measurement and performance metrics
Gas-phase products were quantified using an Agilent 8890 gas chromatograph (GC) equipped with a flame ionisation detector (FID) and a thermal conductivity detector (TCD). Peak areas for and were extracted from the chromatograms and converted to concentrations in parts per million (ppm). The conversion efficiency was calculated assuming complete oxidation to . This assumption is supported by the absence of additional FID peaks in our chromatograms. The conversion efficiency is given by
where and are the concentrations measured at the outlet before and during UV-C irradiation, respectively. In the absence of UV-C irradiation, no catalytic reaction occurs and the concentration is uniform throughout the reactor at steady state, so serves as an experimental proxy for the inlet concentration , making directly comparable to the model prediction in (8). This stoichiometric approach provides a measure of -degradation efficiency under the tested concentrations and light intensities. The apparent quantum yield (AQY) is defined as the ratio of the reaction rate to the incident UV-C photon flux (both in moles/) and is computed for each combination of concentration and light intensity as
In the above, is determined from the measured conversion efficiency at each concentration and is the incident UV-C photon flux (moles/), where is the incident light intensity (), is the energy of photons (J) and is Avogadro’s constant (mol).
In addition to the surface reaction at the catalyst, UV-C irradiation at the wavelengths emitted by our UV-C lamp can photolyse to generate reactive oxygen species, including OH·, which are capable of oxidising in the bulk [18, 21, 25]. To isolate the catalytic contribution from the bulk reaction and enable comparison with the reactor PCO model (§ 3), a control experiment was performed for each combination of concentration and UV-C light intensity, repeating the measurement with no catalyst present. The conversion efficiency is then defined as
where is the total conversion efficiency measured with catalyst present and is the conversion efficiency measured without catalyst present, both evaluated using (1). We use to characterise catalyst performance and to compute the AQY via (2); these quantities are compared with reactor-model predictions in § 4.1. A detailed quantification of the total conversion efficiency, including the bulk and surface reactions, as well as a description of the bulk-phase reaction mechanism, is left for future work.
PCO mechanism
The PCO mechanism on TiO proceeds through band-gap excitation, generating electron–hole pairs. Photogenerated holes react with surface-adsorbed water and hydroxyl groups to form hydroxyl radicals (OH·), while electrons react with O to form superoxide anions (O) [6]. These reactive oxygen species subsequently attack adsorbed CH molecules. For CH, the initial C–H bond activation by OH· is rate-limiting owing to the high bond dissociation energy; successive oxidation steps produce methanol (CHOH), formaldehyde (HCHO), formic acid (HCOOH) and ultimately CO and HO [26].
Experimental scope and limitations
A schematic of the PCO setup, including the gas delivery system, UV-C lamp alignment, reactor chamber, GC, FID and TCD analysis configuration, is shown in Figure 1(a). Table 1 summarises the parameters for the experiments, as well as for HCHO [15] and [17] PCO systems. All experiments were conducted under controlled laboratory conditions using dry gas feeds, clean reactant streams and an idealised reactor geometry. While these conditions allow for reproducible catalyst performance, they do not fully replicate real-world environments, where humidity, contaminants, fluctuating flow and light-intensity profiles may affect the catalyst oxidation state and long-term stability [27, 28]. Accordingly, the reported performance metrics reflect catalytic behaviour under laboratory conditions rather than application environments.
The dry experimental conditions used in the present study do not capture the humidity dependence reported for photocatalytic oxidation, where competition for adsorption sites and enhanced generation produce a non-monotonic effect on conversion efficiency [29]; by contrast, the and HCHO datasets used for validation here [15, 17] report a monotonic decrease in conversion efficiency with increasing relative humidity. These effects will be incorporated in future studies at representative indoor humidity levels. Catalyst mass and coated length were held fixed across all experiments; the effect of catalyst area is captured through the reactor length and fitted parameters and in (7). Conversion efficiency increases with at fixed channel height and flow rate, and is investigated in the reactor-to-duct scaling in § 4.3–§ 4.4. Residence time effects are captured by the Péclet number and reactor aspect ratio ; the experimentally tested flow rate of 100 mL/min corresponds to a low- regime (Table 1), and the ventilation-scale extrapolation in § 4.4 evaluates performance at high- conditions representative of building airflows.
Model
With the PCO experiments for established in § 2, we develop a modelling framework to interpret these results and selected literature data for HCHO and [15, 17]. The modelled domain is outlined in Figure 1(b). In § 4.4, we scale these results up to ventilation applications to assess potential climate benefits, as illustrated in Figure 1(c). We begin by presenting the governing equations and outlining the model assumptions (§ 3.1–§ 3.2). Then, we describe the solution approach, which combines asymptotic methods (Appendix § 6) and numerical methods (§ 3.3). Finally, we describe the fitting of model parameters to experimental data (§ 3.4).
Governing equations
We model the transport and removal of a gaseous pollutant in a 2D, steady-state, laminar-flow reactor (Figure 1b). The channel has centreline height and length , with the bottom wall () coated with a photocatalyst and uniformly illuminated by UV-C light. We assume that the flow is fully developed in the streamwise direction and incompressible, with a parabolic streamwise velocity profile and negligible wall-normal velocity.
Under these assumptions, the pollutant concentration field, , satisfies the steady advection–diffusion equation:
where is the velocity profile and is the molecular diffusivity. The velocity profile for fully-developed laminar flow between parallel plates is
where is the mean streamwise velocity. The pollutant enters the reactor at a uniform concentration . The top wall () is impermeable, while the bottom wall () undergoes photocatalytic reaction. The corresponding boundary conditions are
where we assume, following Zhang et al. [13], that the surface reaction rate follows a light-modulated Langmuir–Hinshelwood expression:
where is the rate constant (molm/(sW)), is the adsorption equilibrium constant (m/mol), is the UV irradiance (W/m) and is the light-intensity exponent. The conversion efficiency is defined as the reduction in the average pollutant concentration across the channel at the outlet:
The unknown parameters in (7), , and , are fitted to three experimental datasets, as described in § 3.4. The model captures both reaction saturation at high pollutant concentrations and a sub-linear dependence on light intensity, as will be observed in the experiments (§ 2).
Non-dimensionalisation
To compare PCO performance across reactors and against results from prior experiments (§ 2; Ballari et al. [17]; Yu et al. [15]), we introduce the non-dimensional variables
Substituting (9) into the advection–diffusion equation (4) yields the transport equation
where the Péclet number, , quantifies the ratio of advective to diffusive transport. The velocity field (5) becomes
The boundary conditions (6) become
where the Damköhler number, , measures the ratio of surface reaction rate to mass transport, and the Langmuir parameter, , quantifies surface saturation at the inlet concentration. The conversion efficiency (8) is given by
Asymptotic solutions to the governing equations and conversion efficiency (10)–(13) are derived in the limits of strong and weak diffusion (, ), reaction (, ) and adsorption (, ) in Appendix A.
Numerical methods
The non-dimensional system of equations, (10)–(12), is solved numerically using MATLAB’s pdepe [30]. The domain is discretised using finite differences, and the resulting system is integrated in to obtain the concentration field . Mesh independence is verified by grid refinement, and standard grid sizes of are used for all results in Section 4. The numerical solution is benchmarked against the asymptotic results derived in Appendix A and against literature data from Ballari et al. [17] and Yu et al. [15].
Model fitting to experiments
We compare the PCO of three representative indoor and outdoor pollutants: CH, HCHO and NO, across three experimental datasets, each employing TiO as the photocatalyst. These systems differ in geometry, flow rate, pollutant concentration, operating temperature and UV intensity, as summarised in Table 1. The Ballari et al. [17] and Yu et al. [15] datasets are the most appropriate for model validation at the concentration range and geometry considered here. Each PCO system is modelled using the 2D advection–diffusion equation for pollutant concentration (4), coupled to a surface reaction following light-modulated Langmuir–Hinshelwood kinetics (6). Kinetic parameters (, , ) are estimated by minimizing the sum of squared errors (SSE) between experimental and modelled removal efficiencies across experimental realisations:
The PDE system (4)–(6) is solved numerically using MATLAB’s pdepe, and the kinetic parameters are obtained by minimising the SSE using MATLAB’s fmincon [30].
Results
With the CH PCO experiments described in Section 2 and the model outlined in Section 3, we compare model and experimental trends across CH, NO and HCHO (Section 4.1), we compare experimental trends for CH with others found in the literature (Section 4.2), we scale trends from laboratory reactors to ventilation systems (Section 4.3–4.4, fig:1b–c), and we estimate the potential climate benefits of PCO technologies in ventilation systems (Section 4.6). All conversion efficiencies, AQYs, and climate-benefit estimates for CH reported in this section characterise only the catalytic contribution, isolated from the bulk reaction through the control experiments described in § 2; total values, including the bulk reaction, will be characterised in future work. Error bars on experimental data points in Figure 2 indicate the range of conversion efficiencies measured across repeated experiments.

Figure 2. Conversion efficiency (%) as a function of (a, d, g) inlet concentration (ppm) and (b, e, h) UV light intensity (W/m), with the corresponding experimental locations in (, ) space and contours of from 0% (blue) to 100% (red) shown in (c, f, i). Data correspond to (a–c) (§ 2), (d–f) [17], and (g–i) HCHO [15]. Solid lines denote numerical solutions of the advection–diffusion equation with Langmuir–Hinshelwood kinetics (4)–(6), while dashed lines indicate asymptotic solutions in the small- (18) and large- limits (43).
Pollutant removal under variable conditions
Using the PCO model in § 3, we simulate the conversion efficiency of , and HCHO across a range of inlet concentrations (0.11 to 10 ppm) and UV-C light intensities (0.1 to 59 W/m), consistent with the experimental conditions (§ 2, [17], [15]). All experiments employ as the photocatalyst, at standard atmospheric pressure, with fixed physical constants (diffusivity, viscosity). The systems differ in chemical properties, reactor design and fitted kinetic parameters (Table 1). exhibits the highest adsorption constant (), while () and HCHO () have lower values, reflecting weaker adsorption onto the catalyst surface. has the highest rate constant (), followed by HCHO () and (), indicating that reacts most rapidly once adsorbed on the catalyst surface, although a direct comparison is complicated by the units of depending on , which differs across pollutants. Light-intensity exponents vary across experiments (0.23 for , 0.38 for HCHO, 0.21 for ). Values of indicate that the systems operate in a regime where mass-transport limitation contributes to the photon-utilisation efficiency, in addition to electron–hole recombination losses that are typically associated with at moderate-to-high irradiance [6].
For all PCO systems, conversion efficiency decreases non-linearly with concentration due to Langmuir–Hinshelwood kinetics (7), with the steepest decline observed for (Figure 2d). and HCHO show more gradual declines in conversion efficiency due to the combined effects of adsorption and reaction at the catalyst surface (Figure 2a, Figure 2g). The conversion efficiency increases with light intensity and begins to saturate at high irradiance for and (Figure 2b, Figure 2e). In contrast, HCHO operates at lower light intensities and exhibits less saturation within the investigated range (Figure 2h). Agreement between model predictions and experimental data is good across all cases, with root-mean-square errors of 1.02% for , 2.46% for and 0.36% for HCHO. Deviations are attributable to measurement uncertainty, humidity effects not captured in the dry experiments, and simplifications in the assumed reactor geometry. The asymptotic solutions accurately reproduce the numerical results in their respective regions of validity, providing a closed-form alternative to numerical simulation. Deviations from model predictions may be attributable to low-concentration measurement errors, humidity or flow effects, depending on the specific experimental setup. All experimental realisations are illustrated in non-dimensional Da–Pe space (white symbols in Figure 2c, Figure 2f, Figure 2i), where the magnitudes of Da and Pe place the , and HCHO experiments on the boundary between low and high removal. Asymptotic solutions in this region (Appendix § 6.§ 1) predict where accurately (dashed lines in Figure 2c, Figure 2f, Figure 2i), providing guidance for reactor design to achieve high conversion, i.e., Da Pe.
Apparent quantum yield
To assess PCO performance compared to the literature, we compute the AQY (%) of oxidation across the range of concentrations (2 to 10 ppm) and UV-C light intensities (4 to 59 W/m) considered in § 4.1. The AQY represents the ratio of reacted molecules to incident photons and quantifies photon utilisation efficiency independently of the experimental setup, whereas reflects performance for a given experimental setup. For each experiment, AQY was calculated from the measured reaction rate and incident photon flux as described in § 2.1. We evaluate the AQY for (i) the present experiments, (ii) other recent low-concentration photocatalyst experiments [31] and (iii) the higher-concentration dataset compiled in Tsopelakou et al. [23]. The results are shown in Figure 3.

Figure 3. Apparent quantum yield (%) for CH₄ oxidation for varying inlet concentrations (ppm) and light intensities (W/m²). Symbols denote experimental measurements and solid lines show model predictions using the quantities in Table 1. In (a) we show low-concentration data () and in (b) we show high-concentration data () [31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43].
In Figure 3a, the AQY increases with increasing CH₄ concentration ( to for 2 to 10 ppm and 4 W/m²) and decreases with increasing light intensity ( to for 2 ppm and 4 to 59 W/m²). Our results are consistent with Kessler et al. [31], who recently reported AQYs of – for CH₄ at ppm-level concentrations, suggesting that lower AQYs are intrinsic to dilute-CH₄ photocatalysis rather than an artefact of catalyst or reactor choice. In the high-concentration regime (Figure 3b), recent studies [33, 34, 36, 37, 38] achieve higher AQYs than the low-concentration systems. For the present experiments, the increase with concentration is most pronounced at lower concentrations ( ppm), with the AQY approaching a plateau at higher concentrations ( ppm). Increasing the light intensity enhances the conversion efficiency (Figure 2) but reduces the AQY, reflecting a sub-linear dependence on light intensity (, Table 1). Although higher photon flux accelerates the reaction rate, the reaction rate grows more slowly than the photon supply, resulting in a reduced AQY. At low CH₄ concentrations, the surface coverage is small, so adding more CH₄ molecules substantially increases reaction rates, leading to a steep rise in AQY. As the CH₄ concentration increases further, surface sites approach saturation, and the incremental gain in rate per additional CH₄ molecule diminishes, causing AQY to level off. The model reproduces the trends of the present CH₄-experimental data in the left panel of Figure 3.
Concentration field in the reactor
To examine pollutant transport in the different PCO reactors, we compute steady-state 2D concentration fields for CH₄, NOₓ and HCHO. Pollutant transport occurs predominantly via advection in the streamwise direction and diffusion in the wall-normal direction, with photocatalytic reaction at the bottom surface () and a no-flux condition at the top surface (). Figure 4 shows the concentration field for CH₄ (panel a), NOₓ (b) and HCHO (c) under maximum removal conditions, as identified in Figure 2. The streamwise concentration gradients result from the combined effects of surface kinetics, flow rate, light intensity and pollutant adsorption.

Figure 4. Contours of the concentration field for maximum removal in the laboratory-scale system. (a) CH (Da = 0.027, Pe = 0.085 and = 0.17; § 2), (b) NO (Da = 3.45, Pe = 2.50 and = 0.54; [17]) and (c) HCHO (Da = 0.46, Pe = 2.50 and = 0.26; [15]). The inset shows the location in (Da, Pe) space and contours of from 0% (blue) to 100% (red). Solid lines represent the numerical solution of the model (4)–(6), while dashed lines indicate the asymptotic solution in the small-Pe limit (18).
The CH and HCHO systems exhibit approximately linear concentration profiles along (Figure 4a, Figure 4c) with weak axial gradients. This is consistent with Da/Pe , for which the concentration varies slowly along the reactor and the influence of Langmuir–Hinshelwood nonlinearity (through ) is negligible. In contrast, the NO system shows a sharper inlet decrease followed by slight downstream flattening (Figure 4b), consistent with Da/Pe = . In this regime, adsorption effects (through ) begin to influence the kinetics, leading to the observed curvature. In all cases, Pe , and cross-channel concentration gradients remain weak, indicating that transverse diffusion is sufficiently rapid to maintain near-uniformity in and strong coupling between reaction and cross-channel mass transport.
Concentration field in ventilation applications
To evaluate the scalability of PCO for indoor air treatment, we extend the laboratory-scale model from § 3 to a duct geometry representative of ventilation systems. The duct is assumed to have a rectangular cross-section with a height of 0.2 m, a width of 0.2 m and a 1 m long test section, uniformly illuminated with UV-C light. Air flows at a rate of 0.08 m/s, corresponding to an average velocity of approximately 2 m/s. These operating conditions enhance advective transport and reduce residence time relative to the laboratory-scale setup, resulting in elevated Reynolds numbers and the formation of a concentration boundary layer near the catalyst surface. We assume that the bottom wall of the duct is coated with TiO, while other walls are treated as no-flux boundaries.
Figure 5 presents the simulated concentration fields in the duct under the high-removal con ditions identified in Figure 2. In contrast to the laboratory reactor (Figure 4), the duct-scale concentration field exhibits steep wall-normal gradients near the catalyst, while concentrations remain nearly uniform outside the boundary layer, indicating that the system is mass-transfer-limited. Under these conditions, removal efficiencies are low: 0.010% to 0.017% for CH₄, 0.56% for NOₓ and 0.039% to 0.070% for HCHO. Table 2 summarises these results, showing the estimated mass removed for CH₄, NOₓ and HCHO, together with the associated experimental parameters from § 2, Ballari et al. [17] and Yu et al. [15]. The annual removal for the assumed test section is modest, ranging from to tonnes per pollutant per year across CH₄, NOₓ and HCHO, illustrating the large surface area required to achieve large removal rates in ventilation systems [11].

Figure 5. Contours of the concentration field for maximum removal in the duct-scale system. (a) CH₄ (Da = 0.090, Pe = and = 0.17; § 2), (b) NOₓ (Da = 3.46, Pe = and = 0.57; Ballari et al. [17]) and (c) HCHO (Da = 0.39, Pe = and = 0.26; Yu et al. [15]). The inset shows the location in (Da, Pe) space and contours of from 0% (blue) to 100% (red). Solid lines represent the numerical solution of the model (4)–(6), while dashed lines indicate the asymptotic solution in the large-Pe limit (43).
| Parameter | Ballari et al. [17] | Yu et al. [15] | Present |
| Removal quantities | |||
| Concentration (ppm) | 0.11 to 1.09 | 0.17 to 0.9 | 2 to 10 |
| Removal efficiency (%) | 0.5565 | 0.0389 to 0.0702 | 0.0095 to 0.0165 |
| Max. removal rate ( ppm/s) | 0.612 to 6.07 | 0.066 to 0.632 | 0.190 to 8.25 |
| Max. mass removed ( t/y) | 0.290 to 2.88 | 0.021 to 0.196 | 0.031 to 1.36 |
| Climate benefit for UV light | |||
| COe removal rate ( tCOe/y) | – | – | 0.026 to 1.14 |
| Net COe w/ UV ( tCOe/y) | – | – | 4.27 to 4.38 |
| Net COe w/o UV ( tCOe/y) | – | – | −1.13 to −0.014 |
Table 2. Summary of the predicted removal rates and climate benefit metrics for scaling up photocatalytic studies on CH, HCHO and NO removal using TiO to a ventilation system of height 0.2 m, width 0.2 m and 1 m length. The flow rate is 0.08 m/s and the UV-C lamp is 25 W, which provides the range of light intensities in the laboratory experiments.
Design implications
In the laboratory reactor (Da = 0.016–0.027, Pe = 0.085), the system is in the Da and Pe regime, where Appendix § 6.§ 1 shows that . Hence, the conversion efficiency increases with , and and decreases with . At duct scale (Da , Pe ), Appendix § 6.§ 2 gives and we are in the reaction-limited limit (Da ). Hence, in this regime, an increase in light intensity and catalyst activity, or decrease in flow velocity (increasing residence time), yields an improvement in conversion. However, for Da we have a transport-limited ceiling of , which is independent of light intensity and catalyst activity. Reaching this ceiling requires Da , i.e., an increase in , or of approximately two orders of magnitude from the TiO₂ baseline. This analysis shows that catalyst improvement must be pursued up to a point; catalyst gains alone cannot exceed the 2.6% ceiling at the current ventilation geometry and flow rate. Therefore, reducing the flow velocity is the main way to raise this ceiling and achieve considerably higher conversions.
Climate benefit for UV light in ventilation applications
The PCO system described in Section 4.4 relies on UV-C irradiation to activate TiO surfaces and oxidise CH. This approach achieves high CH conversion (Section 4.1); however, it requires electrical energy for UV illumination, resulting in associated COe emissions. In addition, the embodied carbon of the TiO catalyst contributes to the climate impact and must be included in the assessment [44]. The use of metal-free TiO results in lower embodied carbon compared to metal-loaded catalysts [24]. The net climate impact of CH PCO is assessed for a duct-scale system using the methodology detailed in Appendix B. Note that the catalytic performance reported here represents short-term activity under controlled laboratory conditions. The net COe emissions analysis assumes stable catalyst performance and does not account for potential deactivation or replacement requirements. Evaluation of catalyst durability, structural evolution and lifetime performance in realistic environments is therefore an important direction for future work.
Each system is assumed to operate with a 25 W UV lamp, which we assume provides the maximum irradiance over the catalyst surface, running continuously 24 h/day throughout the year (see Table 3). Under these operating conditions, the system consumes around 219 kWh/y of electrical energy, corresponding to 0.044 tCO/y of emissions for a grid electricity intensity of 200 gCO/kWh. The embodied carbon of the catalyst is based on a TiO mass of 3.2 g and an emission factor of 4 gCOe/gTiO [44]. Assuming an annual replacement cycle for the catalyst (an upper-bound assumption pending lifetime studies), this yields an annual contribution of 1.3 tCOe/y. The pollutant removal efficiency from Section 4.4 was applied to the inlet concentration to calculate the pollutant removal per unit volume, which was then converted to a removal rate using the ventilation flow rate, as summarised in Table 2. Under the specified operating conditions, the system is predicted to remove approximately 3.1 to 1.4 tCH/y. Considering the 20-year global warming potential of CH (GWP = 84 [4]), the annual CH removal corresponds to a climate benefit of approximately to tCOe/y. When the COe emissions from UV electricity consumption and catalyst embodied carbon are included, the net climate effect is predicted to be a net positive emission of approximately to tCOe/y, indicating that the energy cost of UV operation outweighs the climate benefit of CH removal in this configuration. However, if the UV energy is assumed to be already available (e.g., in a pre-existing disinfection systems), the net climate effect is predicted to be approximately to tCOe/y, indicating that the system would be effectively removing COe. Overall, this calculation demonstrates that duct-scale TiO-based PCO may provide a measurable net climate benefit when leveraging pre-existing UV irradiation, even when accounting for catalyst embodied carbon.
| Parameter | Value |
| Catalyst material | |
| Catalyst area (ventilation panel) | |
| UV lamp power | |
| Operation time | |
| Electricity intensity | |
| Catalyst lifetime | |
| emission factor | |
| Catalyst mass |
Table 3. Parameters used in the life-cycle assessment of -based PCO of , including catalyst properties, UV energy demand and emissions factors [4, 44, 51, 52, 53, 54].
For a commercial building HVAC system, assuming the entire ductwork (e.g., 100 m of catalyst-coated duct) operates under pre-existing UV-C irradiation, the predicted maximum annual CH removal scales to approximately tCH/y (0.12 tCOe/y). Ventilation and lighting each account for over 14% of commercial building electricity use [45] At an average electricity use of 203 MWh/y per building [45], this gives Scope 2 emissions of 5.7 tCOe/y at 200 gCO/kWh. The maximum PCO removal benefit for 100 m of catalyst-coated duct is therefore 2.1% of this. However, UV-C illuminated ductwork is not yet widespread, though installation of UV-C disinfection systems in commercial HVAC has been growing since 2020 [46]; where currently installed, light intensities are optimised for microbial disinfection rather than photocatalysis, so the climate-benefit estimates above should be interpreted as an upper bound contingent on this infrastructure becoming more common. Because atmospheric CH is highly dilute, this removal is necessarily small relative to total building-scale emissions regardless of catalyst performance; it is also useful to weigh this benefit against the COe cost of operating the UV-C source itself, as shown above, and meaningful climate impact would require deployment far beyond a single building, consistent with the large-area estimates in Tomlinson et al. [11].
Discussion
The modelling framework developed here provides a description of PCO for CH, NO and HCHO. Comparison of the model with our CH experiments and literature data for NO and HCHO (Figure 2) shows that the fitted kinetic parameters, the rate coefficient , adsorption constant and light exponent , reproduce measured removal efficiencies and differentiate pollutants with fast surface kinetics from those limited by weak adsorption. For CH, strong adsorption increases surface coverage and yields high conversion despite slower surface kinetics. In contrast, NO oxidation is governed by faster surface kinetics even though adsorption is weaker. HCHO exhibits the lowest activity because its adsorption constant and rate coefficient are small, resulting in limited surface coverage and slow reaction. These differences explain the pollutant-specific removal trends predicted by the model and show how surface coverage and saturation constrain PCO performance at environmentally relevant concentrations. Extending the lab-scale results [15, 17] to duct geometries quantifies how reaction and transport mechanisms are affected under ventilation flow conditions. Thin diffusive boundary layers near the walls confine reactions to this region, limiting catalyst utilisation and resulting in removal efficiencies for all pollutants, even under high irradiance (Figure 5). For CH, our analysis predicts that duct-scale operation can achieve a net climate benefit (net-negative COe emissions) when UV energy is already available (e.g., in pre-existing disinfection systems), even after accounting for the embodied carbon of the catalyst. This benefit is primarily enabled by the use of metal-free TiO coatings, which minimise material-related emissions compared to more complex catalyst formulations [24].
The present model assumes laminar flow as an approximation that yields an analytical estimate of mass transfer. In practice, ventilation-scale operating conditions may fall in the turbulent regime, which would enhance mass transfer to the catalyst surface relative to the laminar case [47]. Surface roughness, mixing baffles or other modifications could also be used to enhance mass transfer. However, they would increase pressure drop and fan energy consumption, offsetting part of the climate benefit quantified in § 4.6. A promising configuration is photocatalytic coating of HVAC filters [48]: filters are engineered to maximise fluid–surface interaction, and the resulting higher surface area and longer residence time could increase the mass transfer by several orders of magnitude compared to a duct-wall coating. Furthermore, the concentration boundary-layer analysis presented here for ventilation flow has analogues for other applications, such as -coated building facades, roads and rooftops. In the large- limit (Appendix § 6.§ 3), the near-wall boundary-layer solution depends only on the local streamwise distance and , not on the rest of the geometry, so the same analytical estimates should apply to an external flow over a building facade or rooftop. For a characteristic wind speed of 4 m/s and a building length scale of 30 m, this gives ; using the kinetic parameters fitted to Ballari et al. (Table 1) gives , placing the system firmly in the transport-limited regime. The resulting mass-transfer-limited removal efficiency, , is of a similar order of magnitude to the 2% reduction reported by Russell et al. [16] for outdoor coatings. Our model does not, however, include surface roughness or turbulence, and resolving turbulent transport requires numerical simulation beyond the scope of this manuscript. Extending this work to include surface roughness and turbulent flow regimes, and quantifying the associated pressure-drop penalty in ventilation applications, are left for future work.
Several other limitations require further investigation. The reported metrics isolate the catalytic contribution and do not include the bulk oxidation of by reactive oxygen species and generated by the UV-C lamp [18, 21, 25]; total conversion efficiencies and the bulk reaction mechanism will be quantified in future work. The analysis assumes uniform irradiance, neglecting photon scattering and catalyst shadowing, which can create spatial variations in surface reactivity [6]. Catalyst ageing, deactivation and fouling are not included in the model and may reduce long-term performance. -based photocatalysts can deactivate through accumulation of partially oxidised intermediates [26], poisoning by trace contaminants [49], or sintering under prolonged UV exposure [50]. Catalyst regeneration has been reported to restore activity in some cases [27], but long-term durability under realistic ventilation conditions remains an open question. Mitigation strategies for these effects include catalyst immobilisation and upstream filtration. Thus, the reported metrics reflect intrinsic catalytic behaviour under idealised conditions rather than long-term field performance. Avenues for future work include catalyst durability under humid and contaminant-containing feeds that are more representative of environmental methane sources. The potential formation of secondary species and partially oxidised by-products [26], has not been quantified and requires targeted experimental investigation. Addressing these factors is essential for assessing real-world PCO performance. The PCO of is expected to proceed through successive C–H bond activation steps, producing intermediate species including methanol (), formaldehyde (), formic acid () and ultimately and [26]. In the present experiments, the absence of additional FID peaks in the chromatograms supports the assumption of complete oxidation to ; however, trace intermediates below the detection limit cannot be ruled out and their quantification is an important direction for future work.
Despite these limitations, the framework provides a foundation for the optimisation of PCO ventilation technologies. Integrating experimentally validated kinetics with detailed transport modelling allows quantitative evaluation of design modifications, including increased catalyst area, enhanced UV irradiance, or integration with complementary air-cleaning methods. Alternative ventilation system designs could also be explored [48]. For example, coating ventilation filters may enhance mass transfer and pollutant removal efficiency, as pollutants would be transported through the catalytic medium rather than merely over a coated surface. Future work could integrate sensor-driven control strategies to adjust UV irradiation or airflow dynamically based on real-time pollutant concentrations. A comprehensive life-cycle assessment will also be necessary to quantify the long-term environmental and operational sustainability of PCO systems.
In summary, our experimental and modelling results investigate -based PCO as a potential strategy for improving indoor air quality and, in the case of , predict that it could potentially achieve a net climate benefit if integrated into existing UV-illuminated infrastructure. However, the net climate benefit predicted under the assumption of pre-existing UV-C infrastructure is negligible relative to building-scale emissions. Meaningful atmospheric impact would require large-scale deployment across millions of square metres of catalytic surface, as quantified in [11], which faces substantial economic barriers [10, 19]. We further note that PCO could, in principle, target multiple pollutants simultaneously using the same UV-C infrastructure, which would improve both the health and climate case for deployment; quantifying this multi-pollutant co-benefit is left for future work.
Acknowledgments: We acknowledge Grantham Foundation for supporting this research. SDT acknowledges funding from Research England’s E3 fund via the M4Impact program.
Asymptotic analysis
In Section 4, experimental results are compared with numerical and asymptotic solutions to assess model performance. Asymptotic solutions provide physical insight into pollutant removal and enable simple predictions in parameter regimes where thin boundary layers make full numerical solutions computationally expensive. In this appendix, we derive asymptotic approximations for the pollutant concentration and conversion efficiency in three regimes. These regimes are defined by the relative strength of reaction (), diffusion () and adsorption () compared to advection.
Region I and II: small Pe
We begin by integrating the advection–diffusion equation (10) across the channel and applying the wall flux conditions (12), which gives
In the strong-diffusion limit, , cross-channel concentration gradients are expected to be small, so the leading-order concentration is nearly uniform in and depends only on , i.e., . Substituting this approximation into the cross-channel integrated equation (15) yields the leading-order ordinary differential equation (ODE)
For , (16) reduces to a linear ODE, whose solution is
Substituting (17) into (13) gives the leading-order conversion efficiency
For , integrating the ODE (16) with respect to gives the implicit relation
Expanding in (19) and matching terms at yields the first-order correction
Therefore, the concentration field can be approximated as
Substituting the concentration field (21) into (13) yields the leading-order conversion efficiency
Region III: small Da, large Pe
When , advection dominates cross-channel diffusion and a thin concentration boundary layer develops adjacent to the reactive wall at . Retaining the leading-order terms in the advection–diffusion equation (10) in the near-wall region yields
with boundary conditions given by (12). For , the concentration field is expanded as
where satisfies the linear boundary-layer equation (23). For , the wall-flux condition reduces to , and we seek a similarity solution of the form
Substituting the asymptotic expansion (24) and similarity form (25) into the advection–diffusion equation (23), together with the boundary conditions (10)–(12), yields the following boundary-value problem
The solution to the boundary-value problem (26) is
Substituting the solution (27) into the similarity solution (25) gives the first-order concentration field
Using the definition of conversion efficiency (13), the leading-order contribution is
Substituting the similarity solution (28) into (29) and changing variables to the similarity coordinate (with ) gives
Evaluating the integral in (30) gives the leading-order conversion efficiency
Note that , so the leading-order expression is positive, as required for a physical conversion efficiency. For , the expansion (24) remains valid, but the wall-flux condition (12) reduces to
The corresponding similarity solution, satisfying the wall-flux condition (32), is
Substituting the asymptotic expansion (24) and similarity solution (33) into the advection–diffusion equation (23), along with the boundary conditions (12), yields the same ODE and boundary conditions as in (26). The solution of the boundary-value problem is therefore identical to (27), yielding the first-order concentration field
and the leading-order conversion efficiency
Region IV: large Da, large Pe
When , the concentration at the reactive surface is nearly zero, consistent with the wall-flux condition (12). We therefore consider the boundary layer near once more. The concentration field is expanded in inverse powers of as
where the leading-order concentration field satisfies the boundary-layer equation (23) with boundary conditions
To reduce the boundary-layer problem to an ODE, we introduce the similarity variable
Substituting the similarity solution (38) into the boundary-layer equation and the boundary conditions (23)–(37) reduces the problem to the ODE
The solution to the boundary value problem (39) is
Using the asymptotic expansion (36) and the leading-order similarity solution (38), the conversion efficiency (8) at leading order is
Evaluating the integral
where the change of variables has been used to transform the integral to the similarity variable. Substituting the integral result (42) into the conversion efficiency expression (41) yields the leading-order conversion efficiency
Methodology for calculating net carbon dioxide equivalent emissions rates from photocatalytic oxidation
This appendix describes the methodology used to quantify the net climate impact of PCO on UV-illuminated . The net emission rate (g/s) associated with PCO is defined as the sum of four terms: (i) direct produced by the stoichiometric conversion of , (ii) the avoided climate forcing associated with removal, (iii) indirect emissions from UV energy use, and (iv) the embodied carbon of the photocatalyst material. The parameter values used in these calculations are listed in Table 3. The total net balance is given by:
with each term defined in Sections B.1–B.4.
Carbon dioxide production rate
The PCO of follows the reaction , which produces one mole of per mole of converted. The experimentally measured conversion rate (§ 2), reported in molecules per second, is converted to moles per second using Avogadro’s number and multiplied by the molar mass of () to obtain the production rate in .
Carbon dioxide equivalent removal rate
Because has a high GWP, its conversion results in a net reduction in . Using the 20-year GWP value (GWP [4]), the avoided emissions per mole of converted are , where the subtraction of 1 accounts for the single mole of produced during oxidation. Multiplying this factor by the conversion rate yields the removal rate in .
Carbon dioxide UV emission rate
Continuous UV illumination of the photocatalyst requires electrical power. The electrical energy consumption is calculated as , where represents the typical power of UV lamps used in ventilation-integrated PCO systems [51], excluding high-power lamps () not applicable in this context, and . The annual electrical demand is converted to emissions using an electricity emissions factor of [52]. Dividing the annual emissions by the total number of seconds in a year yields the emission rate from UV operation in .
Embodied carbon in photocatalyst materials
The embodied associated with production is accounted for using a material emissions factor of 4 [44]. The total mass is g, corresponding to a coating density of 80 [53] applied over a 0.04 ventilation panel. The total embodied emissions, , are considered as an annual material burden for the system, yielding a material emission rate in g/s.
Calculation of total net carbon dioxide equivalent emissions
The net climate impact of the photocatalytic system is calculated by summing all contributions in (44). This expression accounts for the climate benefit of removal as well as the contributions from UV operation and catalyst material production. Results for the UV intensities, inlet concentrations, and flow rates investigated experimentally (§ 2) are presented in Table 2. A negative net emission rate indicates that the photocatalytic system provides a net climate benefit.
References
- [1]I. Manisalidis et al. “Environmental and health impacts of air pollution: A review”. In: Front. Public Health 8 (2020), p. 14.DOI
- [2]World Health Organization. WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide. World Health Organization, 2021.
- [3]B. S. Sosa et al. “Human health risk due to variations in PM10–PM2.5 and associated PAHs levels”. In: Atmos. Environ. 160 (2017), pp. 27–35.DOI
- [4]P. Kumar. “Climate change and cities: Challenges ahead”. In: Front. Sustain. Cities 3 (2021), p. 645613.DOI
- [5]M. W. Jones et al. “National contributions to climate change due to historical emissions of carbon dioxide, methane, and nitrous oxide since 1850”. In: Sci. Data 10.1 (2023), p. 155.DOI
- [6]M. Malayeri, F. Haghighat, and C.-S. Lee. “Modeling of volatile organic compounds degradation by photocatalytic oxidation reactor in indoor air: A review”. In: Build. Environ. 154 (2019), pp. 309–323.DOI
- [7]I. C. Prentice et al. “The carbon cycle and atmospheric carbon dioxide”. In: Climate Change 2001: The Scientific Basis, Intergovernmental Panel on Climate Change (2001).
- [8]T. Boningari and P. G. Smirniotis. “Impact of nitrogen oxides on the environment and human health: Mn-based materials for the NOₓ abatement”. In: Curr. Opin. Chem. Eng. 13 (2016), pp. 133–141.
- [9]J. Zhang, Y. Wei, and Z. Fang. “Ozone pollution: A major health hazard worldwide”. In: Front. Immunol. 10 (2019), p. 2518.DOI
- [10]R. Randall, R. B. Jackson, and A. Majumdar. “Cost modeling of photocatalytic decomposition of atmospheric methane and nitrous oxide”. In: Environ. Res. Lett. 19.6 (2024), p. 064015.DOI
- [11]S. D. Tomlinson et al. “Harnessing natural and mechanical airflows for surface-based atmospheric pollutant removal”. In: arXiv (2025), arXiv:2503.11803.arxiv.org/abs/2503.11803
- [12]J. Ma et al. “Environmental catalytic city: New engine for air pollution control”. In: J. Environ. Sci. 156 (2025), pp. 576–583.DOI
- [13]G. Zhang et al. “Multivariate modeling of intrinsic kinetics for gas-solid heterogeneous photocatalytic reaction: A general method for different pollutant-photocatalyst systems”. In: Chem. Eng. J. 479 (2024), p. 147651.DOI
- [14]N. A. B. Timmerhuis, J. A. Wood, and R. G. H. Lammertink. “Connecting experimental degradation kinetics to theoretical models for photocatalytic reactors: The influence of mass transport limitations”. In: Chem. Eng. Sci. 245 (2021), p. 116835.DOI
- [15]B. Yu et al. “Experiments and kinetics of solar PCO for indoor air purification in PCO/TW system”. In: Build. Environ. 115 (2017), pp. 130–146.DOI
- [16]H. S. Russell et al. “A review of photocatalytic materials for urban NOx remediation”. In: Catalysts 11.6 (2021), p. 675.DOI
- [17]M. M. Ballari et al. “Modelling and experimental study of the NOx photocatalytic degradation employing concrete pavement with titanium dioxide”. In: Catal. Today 151.1–2 (2010), pp. 71–76.DOI
- [18]L. Pennacchio et al. “Physical and practical constraints on atmospheric methane removal technologies”. In: Environ. Res. Lett. 19.10 (2024), p. 104058.DOI
- [19]C. Hickey and M. Allen. “Economics of Enhanced Methane Oxidation Relative to Carbon Dioxide Removal”. In: Environ. Res. Lett. 19.6 (2024), p. 064043.DOI
- [20]S. Abernethy, M. I. Kessler, and R. B. Jackson. “Assessing the potential benefits of methane oxidation technologies using a concentration-based framework”. In: Environ. Res. Lett. 18.9 (2023), p. 094064.DOI
- [21]M. Krogsbøll et al. “Efficient mitigation of dilute methane, ammonia, and odor in ventilation air from cow and pig barns and a biogas plant: photoreactor field demonstration”. In: ACS ES&T Air 2.8 (2025), pp. 1648–1655.
- [22]V. Rauchenwald et al. “New method of destroying waste anesthetic gases using gas-phase photochemistry”. In: Anesth. Analg. 131.1 (2020), pp. 288–297.DOI
- [23]A. M. Tsopelakou et al. “Exploring the bounds of methane catalysis in the context of atmospheric methane removal”. In: Environ. Res. Lett. 19.5 (2024), p. 054020.DOI
- [24]A. M. Tsopelakou et al. “Evaluating the potential of thermal catalysis for environmental methane mitigation”. In: Sustain. Sci. Technol. 3.2 (2026), p. 024002.DOI
- [25]N. Iversen and P. Roslev. “Mitigation of atmospheric and elevated methane by photochemical oxidation at ambient conditions”. In: Sci. Total Environ. 976 (2025), p. 179338.DOI
- [26]R. Pitchai and K. Klier. “Partial oxidation of methane”. In: Catal. Rev. Sci. Eng. 28.1 (1986), pp. 13–88.
- [27]I. M. S. Anekwe and Y. M. Isa. “Unlocking catalytic longevity: a critical review of catalyst deactivation pathways and regeneration technologies”. In: Energy Adv. 4.9 (2025), pp. 1075–1113.DOI
- [28]D. Dissanayake et al. “Partial oxidation of methane to carbon monoxide and hydrogen over a Ni/Al2O3 catalyst”. In: J. Catal. 132.1 (1991), pp. 117–127.DOI
- [29]J. Ying et al. “Unraveling humidity’s dual effects on trace methane photocatalytic ozonation: Enhanced radical generation versus active site blockade”. In: J. Environ. Chem. Eng. (2025), p. 120330.DOI
- [30]MATLAB. The MathWorks, Inc. Natick, Massachusetts, 2025.
- [31]M. I. Kessler et al. “A Humidity-Tolerant Photocatalyst for Methane Removal”. In: Environ. Sci. Technol. 60.7 (2026), pp. 5442–5452.DOI
- [32]B. Han et al. “A thermo-photo hybrid process for steam reforming of methane: highly efficient visible light photocatalysis”. In: Chem. Commun. 55.54 (2019), pp. 7816–7819.DOI
- [33]L. Luo et al. “Binary Au–Cu reaction sites decorated ZnO for selective methane oxidation to C1 oxygenates with nearly 100% selectivity at room temperature”. In: J. Am. Chem. Soc. 144.2 (2021), pp. 740–750.DOI
- [34]Y. Fan et al. “Selective photocatalytic oxidation of methane by quantum-sized bismuth vanadate”. In: Nat. Sustain. 4.6 (2021), pp. 509–515.
- [35]H. Yoshida et al. “Hydrogen production from methane and water on platinum loaded titanium oxide photocatalysts”. In: J. Phys. Chem. C 112.14 (2008), pp. 5542–5551.DOI
- [36]Y. Jiang et al. “Enabling specific photocatalytic methane oxidation by controlling free radical type”. In: J. Am. Chem. Soc. 145.4 (2023), pp. 2698–2707.DOI
- [37]H. Song et al. “Selective photo-oxidation of methane to methanol with oxygen over dual-cocatalyst-modified titanium dioxide”. In: ACS Catal. 10.23 (2020), pp. 14318–14326.DOI
- [38]D. Hu et al. “Thermocatalysis enables photocatalytic oxidation of methane to formic acid at room temperature beyond the selectivity limits”. In: Cell Rep. Phys. Sci. 4.2 (2023).DOI
- [39]H. Li et al. “Photocatalytic degradation of lignin on synthesized Ag–AgCl/ZnO nanorods under solar light and preliminary trials for methane fermentation”. In: Bioresour. Technol. 175 (2015), pp. 494–501.DOI
- [40]H. Song et al. “Visible-light-mediated methane activation for steam methane reforming under mild conditions: a case study of Rh/TiO2 catalysts”. In: ACS Catal. 8.8 (2018), pp. 7556–7565.DOI
- [41]J. Wei and E. Iglesia. “Reaction pathways and site requirements for the activation and chemical conversion of methane on Ru-based catalysts”. In: J. Phys. Chem. B 108.22 (2004), pp. 7253–7262.DOI
- [42]J. Wei and E. Iglesia. “Mechanism and site requirements for activation and chemical conversion of methane on supported Pt clusters and turnover rate comparisons among noble metals”. In: J. Phys. Chem. B 108.13 (2004), pp. 4094–4103.DOI
- [43]L. Li et al. “Efficient Sunlight-Driven Dehydrogenative Coupling of Methane to Ethane over a Zn+-Modified Zeolite”. In: Angew. Chem. 123.36 (2011).DOI
- [44]Rio Tinto. Our approach to climate change: 2020 Climate Change Report. Accessed: 2025-08-07. 2020. URL: https://www.riotinto.com/-/media/Content/Documents/Invest/Reports/Climate
- [45]U.S. Energy Information Administration. 2018 Commercial Buildings Energy Consumption Survey (CBECS): Consumption and Expenditure Highlights. Tech. rep. U.S. Department of Energy, Office of Energy Statistics, Dec. 2022.
- [46]G. M. Thornton et al. “The impact of heating, ventilation, and air conditioning design features on the transmission of viruses, including the 2019 novel coronavirus: A systematic review of ultraviolet radiation”. In: PLoS One 17.4 (2022), e0266487.DOI
- [47]F. P. Incropera. Incropera’s Principles of Heat and Mass Transfer. 2017.
- [48]S. D. Tomlinson et al. “Modelling laminar flow in V-shaped filters integrated with catalyst technologies for atmospheric pollutant removal”. In: Int. J. Heat Mass Transfer 257 (2026), p. 128206.arxiv.org/abs/2506.00603
- [49]Z. Zhou et al. “Effect of sulfur poisoning during worldwide harmonized light vehicles test cycle on NOₓ reduction performance and active sites of selective catalytic reduction filter”. In: Catalysts 15.7 (2025), p. 682.DOI
- [50]M. D. Argyle and C. H. Bartholomew. “Heterogeneous catalyst deactivation and regeneration: a review”. In: Catalysts 5.1 (2015), pp. 145–269.
- [51]S. Escobedo and H. de Lasa. “Photocatalysis for air treatment processes: Current technologies and future applications for the removal of organic pollutants and viruses”. In: Catalysts 10.9 (2020), p. 966.DOI
- [52]European Environment Agency. Greenhouse Gas Emission Intensity of Electricity Generation (g CO₂e/kWh). European Environment Agency, European Topic Centre on Climate Change Mitigation and Energy. Chart and country-level data, EU average emission intensity. June 2024. URL: https://www.eea.europa.eu/en/analysis/maps-and-charts/co2-emission-inter
- [53]Johnson Matthey Catalysts Ltd. SCR Catalyst. U.S. Patent Application No. US2020/0108376. 2020. URL: https://www.freepatentsonline.com/20200108376.pdf.
- [54]P. S. Hammershøi, A. D. Jensen, and T. V. W. Janssens. “Impact of SO₂-poisoning over the lifetime of a Cu-CHA catalyst for NH₃-SCR”. In: Appl. Catal. B 238 (2018), pp. 104–110.DOI