Application of Ozone for Toluene Abatement from Gas Streams in a Sustainable, Low-Temperature Catalytic Oxidation Process

Abstract

Gas-phase oxidation of volatile organic compounds (VOCs) with the aid of ozone can be an attractive, energy-efficient way of treating exhaust gas streams in a low-temperature process, enabling the sustainable operation of industrial installations in a natural environment. This work is focused on the efficiency and kinetics of toluene oxidation with ozone towards CO₂ and H₂O in the presence of a SiO₂-supported cobalt catalyst. A kinetic model is proposed based on a simplified reaction mechanism, with the parameters determined from measurements carried out in a fixed-bed reactor at 40–65 °C under conditions ensuring negligible mass transfer resistance. The proposed model provided satisfactory agreement between the predicted and measured toluene and ozone conversion rates and the formation rate of CO₂, as well as in conditions when mass transfer resistance due to internal diffusion in the catalyst pellet was necessary to consider. The discussed results provide an assessment of the space velocity and ozone usage necessary to achieve a given degree of toluene conversion and mineralization to CO₂. The proposed model can be used for the design of a sustainable, low-temperature ozone-assisted catalytic process of VOC abatement.

1. Introduction

Volatile organic compounds (VOCs), indoors or when released into the atmosphere (e.g., with ventilation exhaust gas streams), are hazardous to human health and often harmful to the environment. Some VOCs can also impair quality of life due to their nuisance odor, even in trace concentrations. The methods of VOC removal include physical separation (recovery) from gas streams by means of condensation, adsorption, or absorption, which often allows for their reusage or valorization. If the recovery is not feasible (e.g., due to low VOC concentrations or chemical instability after adsorption or condensation), the VOCs are usually destroyed in a thermal oxidation process such as co-incineration with fuel gas, which is necessary to achieve a high enough temperature of combustion. However, this method raises environmental concerns due to the increased CO₂ emissions from burnt fuel and can be very costly if the volumes of contaminated exhaust gas streams to be treated are high. The application of catalysts allows for the oxidation processes to be carried out at a lower temperature, thus minimizing the total auxiliary fuel consumption and CO₂ production and reducing the operational cost of thermal VOC oxidation. Some review papers have presented the state of the art in low-temperature catalytic removal of VOCs via oxidation [1,2]. Traditionally, supported noble metals (Pt, Pd) were used as catalysts for this purpose, but recently, supported transition metal oxides have been proposed as a less costly alternative. Developments of such catalytic materials, which are active at low temperatures, are the focus of current research. Supported copper oxide seems to be a good candidate for such a purpose and shows high potential in lowering the temperature required for the catalytic oxidation of toluene. The application of silica nanotubes as a novel support for Cu-based catalysts allowed for the complete toluene oxidation to CO₂ with air at a lower temperature (330 °C) than traditional silica-based support materials (450–500 °C) [3]. Copper-impregnated TiO₂ allowed for 90% conversion of toluene in dry air at 225 °C, as reported in [4]. It was also possible to achieve high toluene conversion (90%) in the presence of CO (1:1) using Cu/Al/Ce mixed oxides derived from layered double hydroxides at 275 °C, as reported in [5]. The costs can be reduced further by applying ozone as the primary oxidant in combination with catalysts that are capable of treating diluted gas streams, thus achieving VOC oxidation and ozone decomposition, even at room temperature. Although on-site ozone generation involves additional costs, utilizing electricity from renewable sources may reduce ozone costs and the total carbon footprint connected with the VOC abatement method.

Ozone can oxidize VOCs directly or indirectly via the active oxygen species resulting from ozone decomposition at the catalyst active sites [6,7,8,9]. The catalysts for this process are usually based on metal (Mn, Co, Fe, Cu, Ag, Ni) oxides supported by γ-Al₂O₃, SiO₂ or TiO₂ [10]. An investigation into the catalytic ozonation of benzene vapor over ZSM-5-supported metal oxides revealed that benzene removal efficiency was the highest for MnO₂ and followed the order: MnO₂ > CoO > CuO > NiO > ZnO > CeO₂ [11]. Indeed, many studies are focused on manganese oxides due to their high activity in ozone decomposition and the oxidation of organic compounds [12,13,14,15,16]. Cobalt-based catalysts, although less investigated for this purpose until recently, also seem to be an attractive option for ozone decomposition and VOC removal [17,18].

The catalytic oxidation of VOCs with ozone can be efficiently carried out even at room temperature; however, the accumulation of oxidation by-products of low volatility and accumulation of water, especially if the treated gases are humid, leads to a deterioration in catalytic activity with time. Huang et al. [19] reported catalyst deactivation due to competitive water adsorption during benzene oxidation with ozone over MnO₂/ZSM-5 at ambient temperature and 50% relative humidity. The catalyst activity recovered after lowering the humidity to 5%. When the reactor temperature was increased to 50 °C, catalyst “poisoning” was not observed. Similar behavior was observed with the Co(II)/SiO₂ catalyst, which showed a 90% drop in water sorption capacity at 40 °C compared to that at 20 °C (cf. [18]).

As VOC oxidation is accompanied by ozone decomposition, optimizing ozone use and effectively utilizing the active oxygen species generated during its decomposition are crucial for achieving high degrees of VOC mineralization at low cost. On the other hand, the unreacted ozone might itself pose as a dangerous pollutant, so the catalyst should effectively decompose all residual ozone before releasing the treated gases to the atmosphere. This issue is critical if catalytic ozonation is to be applied indoors for air treatment and purification [20].

Cabrol et al. [21] investigated toluene degradation with ozone on open glass foam impregnated with ruthenium nanoparticles. They simulated the influence of operational parameters on toluene and ozone removal efficiency using the developed model of reaction kinetics. The authors concluded that it was possible to achieve 88% toluene removal and 97% ozone degradation at a moderate reaction length of 0.11 m, at 90 °C, with a high ozone excess (nO3/nT = 46). However, no information was given on the degree of organic carbon mineralization achieved nor on CO₂ selectivity. Nevertheless, in terms of investment and operational costs, it was concluded that catalytic toluene abatement with ozone is an efficient alternative to catalytic thermal oxidation in air. Even higher toluene conversion (95%) with 90% CO₂ selectivity can be achieved at a lower ozone/toluene ratio (nO3/nT = 20 ÷ 30) using a cobalt-impregnated SiO₂ catalyst at the same temperature (90 °C), as reported in [18]. High activity of supported MnOx on γ-alumina or MCM-41 for ozone-assisted toluene conversion (89–97% at 80 °C) was also reported in [8] at an even lower ozone/toluene ratio (nO3/nT = 9), although lower CO₂ selectivity (53–65%) was obtained in this case.

Photocatalytic oxidation (e.g., on TiO₂) is also considered as an energy-efficient and cost-effective alternative for VOC abatement in air, especially if solar radiation can be utilized for this purpose, although the formation of incomplete oxidation by-products, which may be more toxic than the original volatile contaminant and impair catalyst activity, poses a problem for the broader application of this technology [22]

Biofiltration is also regarded as an energy-efficient and sustainable method of VOC abatement, which is economically attractive compared to catalytic (thermal) combustion methods [23]. Considering the requirements for a longer residence time and the investment cost of the larger biofilter, it usually does not require additional energy to heat the treated air. However, a biofilter bed must be periodically refreshed to remove blockages caused by biomass growth. Problems may also arise if the composition of contaminants in the treated air varies over time, especially if toxic components appear, which are harmful to microorganisms inhabiting the biofilter. The remedy for this may be pretreatment of the biofiltered gases with ozone (e.g., in an additional catalytic bed), which may improve the biodegradability of toxic contaminants and help control excessive biomass growth when ozone is injected directly into the biofilter or to remove odorous VOCs downstream of the biofilter. Synergies emerging from the combination of Advanced Oxidation Processes with VOC biofiltration are discussed in [23,24]

The scope of this work included investigating the activity of the Co(II)/SiO₂ catalyst for gas-phase ozone and toluene decomposition and developing a kinetic model of this process for reactor design purposes. Toluene was used as an aromatic test compound (volatile contaminant) because of its widespread use as a reagent/solvent. It is often detected as an unwanted by-product in industrial exhaust gas streams or automobile flue gases. Aside from its direct effect on living organisms, it is regarded as a potential precursor constituent of photochemical smog (c.f. [25]). A comprehensive overview of ozone applications for low-temperature catalytic abatement of various types of VOCs (aromatic and aliphatic hydrocarbons, oxygenated, chlorinated, or sulphur-containing VOCs) is covered in a review paper [26].

2. Materials and Methods

The catalysts were prepared by impregnating silica gel particles in an aqueous solution of CoCl₂ (0.060 mole/dm³), rinsing with distilled water, and subsequently drying at 120 °C for 12 h. The specific surface area of the catalyst and the total volume of pores were determined from the nitrogen adsorption isotherm at 77.4 K and analyzed using a Micromeritics (3Flex) surface analyzer (Micromeritics, Norcross, GA, USA). The estimated total pore volume of the catalyst support was 0.68 cm³/g, with an average pore size of 7.5 nm (BJH method) and a specific surface area of 300 m²/g, determined by the Brunauer–Emmett–Teller (BET) method. The parameters were not affected by the impregnation of silica gel with a solution containing cobalt ions. Cobalt content in the impregnated silica gel outer surface layer and over the cross-section of a catalyst pellet was relatively uniform and fell within a range of 0.7–0.8% mass, as determined with a microanalytic Brucker Energy-Dispersive Spectroscopy (EDS) system (127 eV energetic resolution) linked with Zeiss Ultra Plus scanning electron microscope (Zeiss Group, Jena, Germany). The cobalt concentration in the ground silica gel was confirmed spectrophotometrically after dissolving the cobalt from the impregnated catalyst particles with concentrated hydrochloric acid. The ground catalyst grains fraction of average size, approx. 0.2 mm (mesh 60–80), was used for ozone decomposition and toluene oxidation kinetics measurements to minimize the influence of internal diffusion resistance. The larger catalyst pellets (3 mm in diameter) were used for comparison in several experimental runs to estimate the influence of internal diffusion resistance, which is often significant in industrial conditions.

The experimental setup is shown schematically in Figure 1. Toluene decomposition was carried out in a tubular glass flow reactor (12 mm ID) containing 0.5-1 g of the catalyst, inserted between glass wool plugs placed over a layer of glass beads. A heating coil (H) placed over the inlet section was used to control the temperature, as measured by two thermocouples: one in the upper part of the catalyst bed and the other near the reactor outlet. The temperature was set between 40 and 65 °C during the measurements. Ozone was produced by silent discharge in dehumidified air (RH approx. 10–15%) supplied to the ozone generator (Oz). Before entering the reactor, the ozone–air gas mixture was mixed with the air stream saturated with toluene vapor in a thermostatic bubbler flask containing liquid toluene at 0 °C. Ozone concentration was measured at the reactor inlet (M1) and outlet (M2) with a spectrophotometric meter based on UV absorption at 254 nm (BMT Messtechnik, Stahnsdorf, Germany). Toluene concentration was measured using an Agilent 6890 gas chromatograph equipped with a gas sampling valve and a flame ionization detector. Gas samples were collected via sampling ports (SPs) located at the reactor inlet and outlet. The concentration of evolved CO₂ was measured with an NDIR gas sensor. The gas flow rate through the reactor was set between 40 and 70 dm³/h to maintain toluene or ozone conversion between 10% and 90% in most experimental runs. Toluene concentration at the inlet was set in a range between 3.7 × 10⁻³ mole/m³ and 0.027 mole/m³ by varying the flow rate through the toluene vapor saturator (bubbler flask).

Figure 1. Scheme of experimental setup. F1, F2—flow meters/controllers, Oz—ozone generator, M1, M2—ozone meters, S—desiccant, SP—sampling port, H—heating jacket, T—thermocouple temperature meter.

Ozone concentration at the inlet was set between 5 and 20 g/m³ (0.104–0.416 mole/m³). Ozone and toluene concentrations at the outlet were measured until a steady state was achieved, and the steady-state values observed for 0.5–1 h were used in the subsequent analysis. Ozone decomposition rate (in the absence of toluene) was first investigated in a series of measurements, after initial oxidation of the adsorbed cobalt ions, accompanied by increased ozone consumption and a change in the fresh catalyst color from blue to dark (black) brown. Ozone decomposition measurements were followed by measurements of catalyst activity towards toluene oxidation with ozone. The FTIR spectrum of the catalyst was examined before and after toluene ozonation.

3. Results and Discussion

The results of the measurements were interpreted using the developed kinetic model of ozone and toluene decomposition, presented and discussed in the following sections.

To estimate the rate of gas-phase toluene oxidation with ozone, a series of measurements was carried out in the reactor without the catalyst bed and with a bare SiO₂ bed. The conversion rate of ozone or toluene in the empty reactor was undetectable within the analytical method’s accuracy. Ozone decomposition on bare SiO₂ was found to be approximately 10–100-times slower than on the Co/SiO₂ catalyst over the investigated temperature range (see Table 1) and was, therefore, neglected as a separate reaction path in the model presented below.

Table 1. Comparison of ozone decomposition rates on bare and Co(II)-impregnated silica gel. (inlet c_O3 = 0.2 ÷ 0.21 mol/m³).

Bed Type:	SiO2	Co(II)/SiO2
T [°C]	rO3 [mol/g/h]	rO3 [mol/g/h]
40	3.0 × 10−5	3.62 × 10−4
60	8.3 × 10−5	8.67 × 10−3
72.5	-	1.73 × 10−2
80	6.4 × 10−4	-

3.1. Catalytic Ozone Decomposition in Absence of Toluene

The influence of temperature on the rate of catalytic ozone decomposition was examined in the first part of the measurements, conducted without toluene supply. The observed dependence of the outlet ozone concentration on its inlet value was linear for all the tested samples and reactor temperatures, as shown in Figure 2.

Figure 2. Effect of temperature on outlet ozone concentration as a function of its concentration at the inlet.

Diverse results are reported in the literature regarding the order of the ozone decomposition reaction. Although Dhanpani and Oyama [12] observed that the ozone decomposition rate for a MnO₂/γ-Al₂O₃/cordierite catalyst was proportional to ozone partial pressure to the power of 0.68, measurements carried out by other researchers indicate first-order kinetics regarding ozone decomposition on CuO/MnO₂ [27] or various alumina-supported catalysts containing Pt or Ag, Cu, Mn oxides (cf. [28,29]). These observations are consistent with our observations, i.e., the linearity of the outlet vs. inlet ozone concentration plot in Figure 2, which indicates first-order kinetics of ozone decomposition on the Co(II)/SiO₂ catalyst at all temperatures tested.

Rakitska et al. [30] investigated the catalytic activity of Co(II) ions adsorbed on silica gel for ozone decomposition. By observing the pH drop during SiO₂ impregnation, it was concluded that the Co(II) cations are bound to the silica gel surface via ion exchange as follows:

$$\equiv \mathrm{S}\mathrm{i}\mathrm{O}\mathrm{H}+{\mathrm{C}\mathrm{o}}_{\left(aq\right)}^{2+}\leftrightarrow {\left[\equiv \mathrm{S}\mathrm{i}\mathrm{O}-{\mathrm{C}\mathrm{o}}_{\left(aq\right)}^{2+}\right]}^{+}+{\mathrm{H}}^{+}$$

The mechanism of ozone decomposition on Co(II)-loaded SiO₂, suggested by the authors [30], involves the formation of OH radicals and reversible oxidation of Co(II) ions to Co(III) in the form of silica-bound cobalt hydroxide:

$${\left[\equiv \mathrm{S}\mathrm{i}\mathrm{O}-{\mathrm{C}\mathrm{o}}_{\left(aq\right)}^{2+}\right]}^{+}+{\mathrm{O}}_3\to {\left[\equiv \mathrm{S}\mathrm{i}\mathrm{O}-{\mathrm{C}\mathrm{o}}^{3+}\left(\mathrm{O}\mathrm{H}\right)\right]}^{+}+{\mathrm{O}}_2+{\mathrm{H}\mathrm{O}}^{•}$$

Investigations of ozone decomposition on transition-metal-incorporated zeolites [31], cobalt oxides and Co-based spinels [32] also revealed the formation of other active oxygen and peroxide species (O⁻, O²⁻, O₂²⁻). Based on the above findings, a simplified mechanism of ozone decomposition, assuming the formation of the oxidized metal ion sites and various reactive oxygen species (designated as O*), which may react further on with ozone, can be represented by the following reactions (R1–R2).

$${\mathrm{O}}_3+\mathrm{M}\mathrm{\ast }\to {\mathrm{M}\mathrm{O}}^{\mathrm{*}}+{\mathrm{O}}_2$$

(R1)

$${\mathrm{O}}_3+\mathrm{M}{\mathrm{O}}^{\mathrm{*}}\to \mathrm{M}\mathrm{\ast }+2{\mathrm{O}}_2$$

(R2)

Assuming the Eley–Rideal mechanism for the reaction of ozone with surface-bound oxygen species (reaction R2), the kinetics of ozone decomposition in the absence of VOC based on this reaction scheme is governed by the sum of the corresponding reaction rates:

$$R_1=k_1{c}_{O3}(1-{\mathsf{\Theta }}_{\mathrm{O}\mathrm{\ast }})$$

(1)

$$R_2=k_2{c}_{O3}{\mathsf{\Theta }}_{\mathrm{O}\mathrm{\ast }}$$

(2)

Applying pseudo steady-state approximation for surface-bound reactive oxygen species (O*), the rate of ozone decomposition (in the absence of toluene) resulting from the above scheme agrees with the first-order reaction rate observed in the experiments performed without toluene:

$$R_{O3}=R_1+R_2={\displaystyle \frac{2k_1{k}_2{c}_{O3}}{k_1{+k}_2}}=k_{O3}{c}_{O3}$$

(3)

The first-order rate constant for ozone decomposition was estimated by fitting the predicted outlet ozone concentration to the measured value, assuming plug flow in the catalyst bed and neglecting mass transfer resistances. The calculated values of the rate constant are shown in the Arrhenius plot (see Figure 3), which allowed for estimation of the apparent activation energy and pre-exponential factor for ozone decomposition reaction by linear regression:

$$k_{O3}=52.746\times {10}^6{e}^{-{\displaystyle \frac{48842}{R_{g}T}}}$$

(4)

Figure 3. Effect of temperature on first-order ozone decomposition reaction rate constant.

The resulting value of apparent activation energy E₁ = 48.842 kJ/mol for ozone decomposition in this work is very close to the value for ozone decomposition on Co₃O₄ (48.99 kJ/mol) observed by Zhang et al. [32]. A similar value, 47 kJ/mol, was estimated by Naydenov et al. [33] for ozone decomposition on CeO₂, whereas a higher value of 52.4 kJ/mol was reported for ozone decomposition on Fe₂O₃ (cf. [28]) and 52 kJ/mol on Ag/γ-Al₂O₃ [29].

3.2. Toluene Decomposition with Ozone

In the case of toluene decomposition with ozone, it is assumed that the adsorbed toluene (T_S) reacts with the active oxygen species formed during ozone decomposition on the catalyst surface and then with ozone in a sequence of reactions that ultimately leads to total oxidation to CO₂ and H₂O.

The assumed simplified mechanism for catalytic toluene decomposition with ozone can be represented by Equations (R3)–(R6).

$$\mathrm{T}+\mathrm{s}\leftrightarrow {\mathrm{T}}_{\mathrm{S}}$$

(R3)

$${\mathrm{T}}_{\mathrm{S}}+{\mathrm{O}}^{\mathrm{*}}\to {\mathrm{T}}_{\mathrm{S}}^{*} \stackrel{+O_3}{\to } {\mathrm{T}\mathrm{o}\mathrm{x}}_{\mathrm{S}}$$

(R4)

$${\mathrm{T}\mathrm{o}\mathrm{x}}_{\mathrm{S}}+{\mathrm{O}}^{\mathrm{*}}\to {\mathrm{T}\mathrm{o}\mathrm{x}}_{\mathrm{S}}^{*} \stackrel{{+\mathrm{O}}^{\mathrm{*}}}{\to }\dots \stackrel{+{\mathrm{O}}^{\mathrm{*}}}{\to } {\mathrm{C}\mathrm{O}}_2+{\mathrm{H}}_2\mathrm{O}$$

(R5)

$$\mathrm{T}\mathrm{o}\mathrm{x}+\mathrm{s}\leftrightarrow {\mathrm{T}\mathrm{o}\mathrm{x}}_{\mathrm{S}}$$

(R6)

The decomposition process begins with activation of the adsorbed toluene (Ts + O* → Ts*) via hydrogen abstraction from the methyl group by the active oxygen species (e.g., hydroxyl radical), leading to benzyl radicals and further on to carbonyl and carboxyl intermediates (benzaldehyde, benzoic acid) [34,35,36]. The other possible route of toluene activation is via aromatic ring hydroxylation, resulting from reactions involving active oxygen species [37], which also facilitates the electrophilic reaction with ozone and aromatic ring cleavage. Its further oxidation in reaction with oxygen species and ozone leads to the formation of lower-carbon-number by-products (Tox_S) (reaction R4). These toluene oxidation by-products are supposedly C₂-and C₃-carbonyl and/or carboxylic compounds (e.g., produced by ozonolysis of double bonds in compounds, which may have formed after aromatic ring cleavage, according to the Criegee mechanism). Acetaldehyde, formic acid, oxalic acid, acetic acid, maleic anhydride, glyoxal, and methylglyoxal were detected as typical by-products of toluene decomposition with ozone [5,38].

They are responsible for a clear peak, which is characteristic of aldehyde carbonyl groups (at 1730 cm⁻¹) and evident in the FTIR spectrum of the spent catalyst used in this work for toluene oxidation with ozone (Figure 4). The by-products are eventually oxidized to CO₂ if the ozone/VOC contact time in the catalytic bed is long enough (reaction R5).

Figure 4. FTIR spectra of the fresh catalyst and the used one, taken from the reactor after toluene ozonation reaction.

As with toluene itself, the adsorbed decomposition by-products are also volatile, so the adsorption/desorption equilibrium (R6) was also considered in the kinetic model developed in this work.

Since the reactions of adsorbed toluene or its by-products with reactive oxygen species are very complex and usually relatively fast, it is assumed that the kinetics of toluene and by-product oxidation is determined by the rate of reactive oxygen species supply initiated by ozone decomposition. For simplicity, the rate is represented in the form of equations (Equations (5) and (6)), proportional to ozone concentration and fraction of available active sites necessary for the initiation of ozone decomposition:

$$R_4=k_{VOC}{c}_{O3}{\mathsf{\Theta }}_{Ts}\left(1-{\mathsf{\Theta }}_{Ts}-{\mathsf{\Theta }}_{Toxs}\right)$$

(5)

$$R_5=k_{VOC}{c}_{O3}{\mathsf{\Theta }}_{Toxs}\left(1-{\mathsf{\Theta }}_{Ts}-{\mathsf{\Theta }}_{Toxs}\right)$$

(6)

Assuming that the adsorption and desorption of toluene and its decomposition by-products are much faster than the oxidation reactions, pseudo-steady-state approximation for the adsorbed species allows us to express their fractional coverages as a function of their gas-phase concentrations in the following form:

$${\mathsf{\Theta }}_{Ts}={\displaystyle \frac{c_T{K}_T}{1+c_T{K}_T+c_{Tox}{K}_{Tox}}}$$

(7)

$${\mathsf{\Theta }}_{Toxs}={\displaystyle \frac{c_{Tox}{K}_{Tox}}{1+c_T{K}_T+c_{Tox}{K}_{Tox}}}$$

(8)

Accepting that z₄ moles of ozone are used per mole of activated toluene (Ts*) to decompose its aromatic ring, the total ozone consumption rate can, thus, be approximated as:

$$R_{O3}=z_4{R}_4+k_{O3} c_{O3}\left(1-{\mathsf{\Theta }}_{Ts}-{\mathsf{\Theta }}_{Toxs}\right)$$

(9)

The last term in the above expression is the rate of ozone self-decomposition on active sites unoccupied by the adsorbed toluene and its decomposition by-products. The rate of ozone decomposition given by Equation (9) is equivalent to Equation (3) if toluene is absent, so that the fractional converges (Θ_Ts, Θ_Toxs) are zero.

Since, stoichiometrically, 3.5 moles of the C₂-by-products should be formed per 1 mole of decomposed toluene, the net formation rates of the by-products and carbon dioxide can be expressed as (Equations (10) and (11)):

$$R_{Tox}={3.5R}_4-R_5$$

(10)

$$R_{C{O}_2}={2R}_5$$

(11)

The toluene consumption rate in the above equations is designated as R₄.

Theoretically, only 3 or 4 moles of ozone are necessary to activate 1 mole of toluene molecules and react with each of the double bonds after aromatic ring decomposition, producing C₂- or C₃-carbonyl products; however, the value of z₄ = 5 assumed in this work allowed for a better fit to experimental data.

The rate constant k_VOC and adsorption equilibrium constants K_T and K_Tox were estimated by fitting the predicted ozone, toluene, and carbon dioxide concentrations at the reactor outlet to the measured values, assuming plug flow in the catalyst bed and neglecting mass transfer resistance outside the catalyst particles. Justification for these assumptions is presented in Appendix A. Possible effects of internal diffusion of the reactants and products were considered by numerically integrating species mass balance inside a spherical catalyst pellet, as described further.

The estimated values of the reaction rate and adsorption equilibrium constants as a function of temperature are shown in the Arrhenius plots (see Figure 5 and Figure 6).

Figure 5. Effect of temperature on k_VOC reaction rate constant.

Figure 6. Effect of temperature on toluene (K_T) and its decomposition by-products’ (K_Tox) adsorption equilibrium constants.

The functional dependences of these constants on temperature were identified via linear regression in the following form:

$$\ln \left(k_{VOC}\right)=-{\displaystyle \frac{74256.5}{R_g T}}+ 29.88$$

(12)

$$\ln \left(K_T\right)=-{\displaystyle \frac{-41534.25}{R_g T}}-11912$$

(13)

$$\ln \left(K_{Tox}\right)=-{\displaystyle \frac{-82638.67}{R_g T}}-27.72$$

(14)

Negative values of apparent enthalpy and entropy of adsorption for toluene and the oxidation by-products, evident from the parameters in Equations (13) and (14), indicate that the obtained relationships are also qualitatively meaningful physically. The apparent activation energy for toluene oxidation and its by-products (derived from Equation (12)) corresponds to ~74 kJ/mol, and the adsorption enthalpies for toluene and its by-products are on the order of −41 kJ/mol and −82 kJ/mol, respectively (negative, indicating exothermic adsorption). The activation energy for VOC oxidation is somewhat higher than the value (61 kJ/mol) estimated by Cabrol et al. [21] for toluene oxidation with ozone on open glass foam impregnated with ruthenium nanoparticles or (43.9 kJ/mol) obtained for toluene oxidation with active oxygen species on MnO₂/graphene [14] or (48 kJ/mol) on MnO₂/γ-alumina [16]. However, it must be kept in mind that the value of apparent activation energy obtained in this work pertains not only to the oxidation of adsorbed toluene but also to the subsequent oxidation of less reactive by-products towards CO₂ (Equations (4) and (5)).

The values of ozone, toluene, and carbon dioxide concentrations at the reactor outlet, calculated using Equations (4) and (12)–(14), compare favorably with the measured values in the parity plots (see Figure 7). It can be observed that most points deviate from the perfect-fit line by less than 25%.

Figure 7. Comparison of the measured and calculated outlet concentrations (parity plots).

The concentrations of toluene, ozone and CO₂ measured at 40 °C at the outlet (points) can also be compared in Figure 8 with the values calculated from the model (lines) as a function of inlet ozone and toluene concentrations fed to the catalyst bed containing fine (0.2 mm) particles at gas hourly space velocity (GHSV) 61,000 h⁻¹. The model fits the measurements quite well and predicts the non-monotonic (maximum) dependence of CO₂ formation rate on toluene load in the gas fed to the reactor. The effect of temperature and inlet toluene concentration on the average rate of its decomposition and the rate of CO₂ formation is shown in Figure 9. It can be observed that the toluene decomposition rate tends to constant values or even decreases at inlet toluene concentrations above 0.01 mol/m³ at lower temperatures (i.e., 40–50 °C). The non-monotonic dependence of the CO₂ formation rate on the inlet toluene concentration is even more pronounced and visible at higher temperatures. This effect can be attributed to the blockage of active sites by the adsorbed toluene and its decomposition by-products.

Figure 8. Measured and calculated outlet concentrations at 40 °C as a function of toluene and ozone concentrations at the inlet.

Figure 9. Comparison of the modelled and measured overall rates of toluene decomposition (left) and CO₂ formation (right). Effect of inlet toluene concentration (c_Tin) and temperature at the inlet ozone concentration c_O3in = 0.2–0.25 mol/m³ (dp = 0.2 mm).

The model of reaction kinetics discussed above was also used to predict the rate of the process for larger catalyst pellets, where internal diffusion resistance is more significant. The overall reaction rates calculated for the bed (2 cm long) formed with 3 mm catalyst pellets are compared with the measured values in Figure 10. The calculations assumed plug flow with negligible axial dispersion and negligible mass transfer resistance between the gas phase bulk and the catalyst pellet surface. In order to account for diffusional resistance in the catalyst pellet, species mass balance (diffusion equation) with reaction source/sink terms according to Equations (4)–(10) was solved numerically for ozone, toluene, oxidation by-products and CO₂, assuming the spherical geometry of the pellet. The reacting species mixture was very diluted, and the effective diffusivity of each of the species (D_e,i) was estimated according to Equation (14), where the Knudsen diffusion coefficient (D_K,i) was evaluated based on the values of silica gel porosity ε_S = 0.55 and its tortuosity factor of τ = 1.5 according to Equation (15), where 2ε_S/(A_S ρ_S) represents an estimate of mean pore radius.

$${\displaystyle \frac{1}{D_{e,i}}}={\displaystyle \frac{\tau }{{ϵ}_S}}\left({\displaystyle \frac{1}{D_{K,i}}}+{\displaystyle \frac{1}{D_{i,air}}}\right)$$

(15)

$$D_{K,i}={\displaystyle \frac{8{ϵ}_S}{{3 A}_S {\rho }_S}}\sqrt{{\displaystyle \frac{2 R_{g}T}{\pi M_i}}}$$

(16)

Figure 10. The overall reaction rates of toluene decomposition (left) and CO₂ formation (right) at 50 °C, 65 °C and 85 °C (dp = 3 mm).

Molecular (bulk) diffusivity (D_i,air), although less important here, was estimated according to the Fuller method [39].

The calculated overall rates of toluene decomposition and CO₂ formation in the bed formed with larger (dp = 3 mm) catalyst pellets are shown in Figure 10. The results calculated based on the model (lines) are compared with the measurements (points) taken over the temperature range 50–85 °C in this figure. The overall reaction rate in these conditions is predicted with an average relative error of 25% compared to experimental data. The agreement is quite good despite extrapolating the model predictions to 85 °C, i.e., outside the range (40–65 °C) used to estimate the model parameters.

Comparing the results presented in Figure 9 and Figure 10, it can be observed that the rates of toluene removal and CO₂ formation are significantly lower for 3 mm catalyst pellets than for 0.2 mm catalyst particles, which illustrates the influence of mass transfer resistance due to internal diffusion.

However, larger pellets result in a lower pressure drop across the catalyst bed, which is important in practical applications. Also, the degree of VOC mineralization (CO₂ selectivity) may not be much worse or can be even better than that obtained with very fine (0.2 mm) particles.

The degree of toluene mineralization, i.e., oxidation towards CO₂ and H₂O, can be evaluated based on the CO₂ selectivity, defined as:

$$E_{CO2}={\displaystyle \frac{c_{CO2,out}-c_{CO2,in}}{{7(c}_{T,in}-c_{T,out})}}$$

(17)

The CO₂ selectivity, calculated from Equation (16) based on the measured and simulated changes in CO₂ and toluene concentrations in the reactor, is shown in Figure 11 as a function of the inlet O₃/toluene molar ratio and GHSV value, for beds formed with 0.2 mm and 3 mm catalyst pellets at 65 °C. It can be observed that CO₂ selectivity reaches approximately the same level regardless of catalyst pellet size and GHSV for n_O3/n_Tol ≈ 10. CO₂ selectivity for 3 mm pellets might even exceed that of 0.2 mm catalyst grains, e.g., at the lower O₃/toluene ratios (n_O3/n_Tol < 8) or at higher GHSV values. This can be attributed to the blockage of the more accessible active sites in the fine catalyst particles by unreacted toluene and its oxidation by-products, compared with larger (3 mm) pellets. In the latter case, internal diffusion restricts the inflow of toluene and the outflow of the by-products from the pellet, thereby allowing deeper oxidation of the by-products to CO₂ by ozone (which diffuses faster than toluene).

Figure 11. Carbon dioxide selectivity at T = 65 °C calculated (lines) and measured (points).

The simulated profiles of toluene, ozone and CO₂ concentrations in spherical pellets (dp = 0.2 mm and dp = 3 mm) located close to the inlet at 65 °C are shown in Figure 12. The flat profiles of ozone and toluene concentration in the 0.2 mm particles support the assumption of a kinetic regime in the treatment of the experimental data.

Figure 12. Calculated profiles of ozone, toluene and CO₂ concentrations in the catalyst pellet (T = 65 °C, (n_O3/n_T)_in = 20).

The profiles shown in Figure 12 correspond to a location in the bed very close to the inlet (bed volume → 0 or space velocity GHSV → ∞) so that concentrations at the pellet surface (r/R = 1) are approximately the same for both diameters. The results indicate that toluene is almost completely converted within the 3 mm pellets (i.e., for r/R < 0.7) under these conditions and that the CO₂ concentration is much higher than that in 0.2 mm particles. This may explain why CO₂ selectivity observed for 3 mm pellets may be higher than for 0.2 mm pellets at GHSV > 20,000 h⁻¹ (nO3/nTol = 20) (cf. Figure 13).

Figure 13. Effect of GHSV on carbon dioxide selectivity at T = 55–65 °C, calculated (lines) and measured (points).

However, CO₂ selectivity will exceed that of 3 mm pellets if the bed is long enough or the space velocity (GHSV) is small. Indeed, it can be observed in Figure 13 that, at GHSV < 15,000 h⁻¹, mineralization of toluene (CO₂ selectivity) in a bed composed of smaller (0.2 mm) catalyst particles should be higher.

As catalyst beds for industrial applications should preferably ensure a low-pressure drop, properly designed and optimized catalytically active porous structures composed of smaller Co/SiO₂ catalyst particles should allow for high conversion of both toluene and ozone, with high CO₂ selectivity at moderately low temperatures.

Although the measurements carried out in this work were aimed at determining reaction kinetics, avoiding complete conversion of the reactants (the maximum CO₂ selectivity observed in this work reached 0.7), measurements of the outlet carbon monoxide concentration were attempted in some experimental runs using an electrochemical sensor to evaluate carbon balance.

The CO/CO₂ ratio produced seemed insensitive to catalyst pellet size and decreased with increasing temperature from approximately 0.17 to 0.1 (see Figure 14).

Figure 14. Effect of temperature on produced CO/CO₂ ratio in the reactor for inlet n_O3/n_T = 18 ÷ 25.

Since small amounts of volatile by-products were observed or detected in the gas chromatograms of outlet gas samples (their peaks accounted for less than 20% of the total peak area—see Table 2), this indicates that approximately 10–20% of the organic carbon missing in the carbon balance might have been undetected or remained in the catalyst bed, where it could accumulate over extended operation. However, no significant decrease in the toluene decomposition rate was observed over several hours of measurement. Stabilization of outlet concentrations occurred within approximately 1–2 h after setting inlet concentrations. It was possible to regenerate the catalyst after extended measurement series by burning out the accumulated carbonyl by-products at 120 °C, maintaining the inlet ozone concentration at 20 g/m³ until the outlet CO₂ concentration dropped to the inlet value (typically after 20–30 min).

Table 2. GC-FID peaks report of outlet gas sample (n_O3/n_T = 9, T = 62 °C).

	Name	RT	Area	Amount
		(min)	(pA s)	(%)
1		0.615	67,722	0.194
2		0.917	128,730	0.369
3		1.111	284,468	0.815
4		1.485	194,899	0.558
5	Toluene	2.373	29,232,558	83.75
6		5.328	1,111,781	3.185
7		5.681	689,147	1.974
8		6.602	340,113	0.974
9		7.751	183,230	0.525
10		7.856	271,039	0.777
11		8.041	127,462	0.365
12		8.147	1,381,682	3.958
13		8.183	369,646	1.059
14		8.212	312,315	0.895
15		8.275	210,589	0.603

The transient behavior of the regenerated catalyst is shown in Figure 15, where solid points denote toluene conversion in the ozone–air mixture of lower humidity (RH = 10%), and open circles denote higher-humidity conditions (RH = 30%). It can be observed that increased humidity at 260 min basically did not change the conversion level, which agrees with the findings reported in [18,19] that water “poisoning” can be minimized if the catalyst temperature is higher than 35–40 °C

Figure 15. Approaching steady-state toluene conversion after catalyst regeneration at T = 62 °C (cTin = 0.0243 mol/m³, n_O3/n_T = 9). Horizontal line: previous steady-state level (cTin = 0.0254 mol/m³, n_O3/n_T = 8.5). Vertical (dashed) line designates change in relative humidity (from 10% to 30%).

4. Conclusions

The kinetics of gas-phase ozone and toluene decomposition on a Co/SiO₂ catalyst in a low-temperature process (40–65 °C) were investigated in this work. Ozone decomposition in the absence of toluene follows first-order reaction kinetics.

The model of toluene abatement with ozone based on a simplified reaction mechanism proposed in this work allows for reasonable predictions of the process rate and CO₂ selectivity, even under conditions of considerable mass transfer resistance due to internal diffusion in the catalyst pellet. Although the overall oxidation rate is lower in these conditions, the degree of VOC mineralization (CO₂ selectivity) may be higher, especially at higher GHSV values. The catalyst can be regenerated after prolonged usage by burning out the accumulated incomplete oxidation by-products in an air–ozone mixture at 120 °C