Morphological impact of 1-dimensional to 3-dimensional manganese dioxides on catalytic ozone decomposition correlated with crystal facets and lattice oxygen mobilities

Abstract

Ozone is a pollutant that has received widespread attention in recent years, and manganese dioxide (MnO₂) has been widely used for catalytic ozone decomposition. However, few studies have described the structure–activity correlation of different morphological types of MnO₂. In this study, a series of MnO₂ crystals (α-, β-, γ-, δ-, ε- and λ-MnO₂) were synthesized, and their catalytic activities on ozone decomposition (25 °C, dry air) were comparatively studied, which exhibited the order ε-MnO₂ > α-MnO₂ > γ-MnO₂ > β-MnO₂ ≈ δ-MnO₂ > λ-MnO₂. XRD and HRTEM results confirmed their diversities on the exposed crystal planes. It was confirmed that ε-MnO₂ with the (1 0 2) plane had the largest number of oxygen vacancies and the best oxygen mobility. These findings elucidated the favorable performance of ε-MnO₂ in the aforementioned tests. DFT calculations revealed the reaction mechanism, showing that ε-MnO₂ has the lowest energy barrier for the rate-determining O₂²⁻ desorption step (2.04 eV). This work illustrated the crucial role of oxygen vacancies and the mobility of lattice oxygen, shedding light on the strategies for rational design and control synthesis of effective catalysts for ozone elimination.

---

Environmental significance

As the level of indoor pollution continues to increase, ozone existing in the enclosed environment has received considerable attention and its effective control has become an unavoidable objective. Catalytic oxidation at ambient temperature is considered one of the most promising routes for ozone removal because of its high efficiency, easy operation, and environmental friendliness. In this study, different types of MnO₂ were prepared and used for catalytic ozone removal, and the differences among them were investigated with regard to the exposure facet, content of oxygen vacancies, and mobility of oxygen species. The mechanism of catalytic ozone decomposition by MnO₂ was further studied by means of DFT calculations. This work could provide a guidance for the reasonable design of highly efficient catalysts for ozone elimination.

1. Introduction

In recent years, ozone pollution is attracting more and more attention from the government and academic community due to its hazardous characteristics towards the biological environment and human beings. On Oct 1, 2015, the U.S. Environmental Protection Agency (EPA) updated the National Ambient Air Quality Standards (NAAQS) for ground-level ozone from 75 parts per billion (ppb) to 70 ppb with an aim to protect the public health by reducing premature deaths and morbidity of lung diseases, such as asthma or chronic obstructive pulmonary disease.¹ In China, the Ministry of Ecology and Environment of the People's Republic of China also proposed to take actions to build a precise ozone prevention and control system in 2022.

Ozone in the troposphere is formed through the reactions among volatile organic compounds (VOCs), carbon monoxide (CO), methane (CH₄), and nitrogen oxides (NO_x) in the presence of sunlight. In addition, the rate of ozone formation can be significantly accelerated by fine particulate matter with an aerodynamic diameter of 2.5 mm or less (PM_2.5).^2,3 Moreover, outdoor ozone might readily transfer indoors, which can cause direct harm to health. Furthermore, higher-than-normal levels of ozone can be generated in enclosed spaces, such as airplane cabin³ and room where laser printer⁴ and electrostatic precipitator⁵ are used.

For ozone removal, physisorption and catalytic post-treatment are mainly applied.^6–11 Physisorption involves just a physical capture, where ozone is never decomposed. In contrast, direct catalytic decomposition of O₃ into O₂ at room temperature is promising due to its high efficiency, ease of operation, and eco-friendliness. The key to the excellent performance of O₃ depletion at room-temperature is the catalyst design from an atomic or molecular aspect. Among all the latest catalysts reported in the literature, manganese oxides have exhibited diverse morphologies and promising applications for O₃ purification. Tang et al.¹² reported that α-MnO₂ with (0 0 1) facets had outstanding activity for ozone decomposition due to their abundant active sites and down-shifted lowest unoccupied Mn 3d orbitals. Usually, the use of β-MnO₂ alone does not work well in the decomposition of ozone¹³ due to its (1 × 1) tunnels, high Mn average oxidation state (AOS), low specific surface area (S_BET) and few surface adsorbed oxygen species resulting from the oxygen vacancy sites. By doping the transition metal Co on β-MnO₂, Zhang et al.¹⁴ increased its specific surface area, decreased the crystallinity and introduced more oxygen vacancies, obtaining a high-performing catalyst for the room-temperature catalytic decomposition of ozone. Li et al.¹⁵ doped transition metals of Ce and Co into γ-MnO₂ and concluded that the decrease in the content of oxygen vacancies was the main reason for the deactivation. Similarly, Xu et al.¹⁶ confirmed that the oxygen vacancy content of γ-MnO₂ is a determinant of the ozone decomposition rate in their study. The above three types of MnO₂ are reported as the mostly used catalysts in ozone elimination as they all belong to one-dimensional crystal forms. Moreover, the two- and three-dimensional configurations of MnO₂ have been recently studied and used as catalysts for a series of redox processes. Wang et al.¹⁷ obtained a series of different δ-MnO₂ by modulating the ratio of Mn²⁺ to Mn⁷⁺ during the preparation process, showing that δ-MnO₂ with the most oxygen vacancy content and excellent oxygen mobility has superior activity. Gopi et al.¹⁸ prepared H⁺-containing δ-MnO₂ using the proton exchange method, which enhanced the S_BET, reduced AOS and introduced more oxygen vacancies. Hong et al.¹⁹ prepared a series of catalysts using acid treatment to introduce lattice defects, which also demonstrated effective catalytic ozone decomposition. It was reported²⁰ that a kind of λ-MnO₂ was synthesized via a hydrothermal tandem synthesis approach. However, the low specific surface area and oxygen vacancy content of the resulting material displayed a relatively weak catalytic oxidation activity towards toluene. Porous λ-MnO₂ with a spinel structure was successfully synthesized from the ZnMn₂O₄via the acid-etching route by Li et al.,²¹ which was applied for benzene oxidation. The acid-etching method increases the specific surface area of λ-MnO₂, enhances its low-temperature reducibility, and increases the oxygen vacancy content, which enables λ-MnO₂ to perform well in benzene oxidation.

Oxygen vacancies have been recognized as the core sites for the catalytic decomposition of ozone.^15,22–29 However, the oxygen vacancy content is directly affected by the distinct facets exposed by different types of MnO₂ crystals. Yang et al.²⁰ synthesized all the six kinds of MnO₂ crystal structures and investigated their activities in the O₃-assisted catalytic oxidation of toluene at room temperature. They demonstrated that different types of MnO₂ with different contents of oxygen vacancies and mobility of oxygen species play important roles in the expression of activity. Through combined computational and experimental studies, Hayashi et al.³⁰ demonstrated that the catalytic oxidation of organic substances is largely dependent on the crystal structures of each type of MnO₂. Tian et al.³¹ synthesized three MnO₂ polymorphs (α-, β-, and ε-MnO₂) and systematically studied their physicochemical properties for CO oxidation. Nevertheless, the morphological-control synthesis of MnO₂ and the correlated structure–performance relationship for O₃ catalytic decomposition are far from being well established. Compared to other oxides, manganese dioxide is much better than other oxides in use for O₃ removal (Fig. S1†).

In this work, we prepared six kinds of MnO₂ crystals (α-, β-, γ-, δ-, ε-, and λ-MnO₂, Fig. 1) with diverse morphologies and investigated their catalytic activities for ozone decomposition. From a series of characterizations and computational results, an attempt to illustrate the structure–performance correlation for these catalysts with diverse morphologies and crystal facets has been done.

Fig. 1 Diverse crystal structures of MnO₂ samples.

2. Experimental and computational details

2.1 Experimental section

Six kinds of MnO₂ were accordingly prepared. All reagents used in this work were analytically pure and used without any treatment. The details of the synthesis methods are described in the ESI.†

2.2 Catalyst characterization

The crystallographic information was investigated by X-ray diffraction (XRD) using a diffractometer equipped with Cu Kα radiation (Bruker, Germany). The specific surface areas (S_BET) were determined using N₂-sorption isothermal analysis (Quantachrome, USA) based on the Brunauer–Emmett–Teller (BET) theory, and the pore size distributions were calculated based on the Barrett–Joyner–Halenda (BJH) method. The temperature-programmed reduction by H₂ (TPR) and temperature-programmed desorption of O₂ (TPD) experiments were carried out on a device with a thermal conductivity detector (TCD, Huasi, China). Temperature-programmed oxygen isotopic exchange (TPOIE) was used to evaluate oxygen mobility in a wide temperature range, and the diagram of the device is shown in Fig. S2.† Scanning electron microscopy (SEM, Gemini Sigma-500, Zeiss, Germany) and high-resolution transmission electron microscopy (HRTEM, JEM-2100, JEOL, Japan) were employed to observe the morphologies and microstructures of MnO₂ samples. An X-ray photoelectron spectroscopy (XPS) instrument (Thermo ESCALAB 250Xi, USA) equipped with a monochromatic Al Kα X-ray source was used to analyze the surface elemental properties. The details are given in the ESI.†

2.3 Computational details

During the theoretical simulation, the slab models were respectively reconstructed using different scales of the supercell according to their structures, and the slabs were periodically repeated with a vacuum spacing of 15 Å for oxygen vacancy formation. All slab calculations were performed using a Monkhorst–Pack grid (k-mesh = 0.030) with a cutoff energy of 500 eV and the smearing parameter set to 0.02 eV. The computational details are given in the ESI.†

2.4 Catalytic activity evaluation

The catalytic performance of the as-prepared samples for ozone decomposition was evaluated by a fixed-bed continuous flow quartz reactor at 25 °C. Gaseous ozone was generated by an UV ozone generator (SNXIN UV STERILIZER) and checked by an ozone monitor (Model 205 Dual Beam Ozone Monitor). The relative humidity (RH) was obtained by flowing air through a water bath and measured with a hygrometer (Center 313, Humidity Temperature Meter). The detailed reaction conditions are described as follows. First, the sample (100 mg, 40–60 mesh) was loaded into a quartz tube (inner diameter = 6 mm). The feed gases, consisting of 40 ppm of O₃, 50% RH, and air, were subsequently introduced into the reaction system. The total gas flow rate was 500 mL min⁻¹, corresponding to a gas hourly space velocity (GHSV) of 5800 h⁻¹. The reaction temperature was 25 °C. The O₃ conversion (X_O3) can be calculated based on the following equation.

Here, C_in is the ozone concentration at the inlet and C_out is the ozone concentration at the outlet.

3. Results and discussion

3.1 Crystal & pore structure and morphology

The XRD patterns (Fig. 2a) of the MnO₂ samples obtained are respectively in good agreement with those of α-MnO₂ (JCPDS No. 44-0141), β-MnO₂ (JCPDS No. 24-0735), γ-MnO₂ (JCPDS No. 14-0644), δ-MnO₂ (JCPDS No. 80-1098), ε-MnO₂ (JCPDS No. 30-0820), and λ-MnO₂ (JCPDS No. 44-0992), indicating that six kinds of MnO₂ with specific crystalline phases have been successfully obtained. As seen from Fig. 2a, the sharp diffraction peaks of β-MnO₂ (1 1 0) and λ-MnO₂ (1 1 1) were visible, indicating their relatively high crystallinities with respect to the other samples (ε-MnO₂, α-MnO₂, γ-MnO₂, and δ-MnO₂). This difference in the crystallinity mainly comes from their synthetic methods. α-MnO₂ was synthesized by the hydrothermal crystallization at 100 °C for 12 h, and the relatively low temperature and short crystallization time endowed the as-prepared α-MnO₂ with a small particle size. Similarly, the spherical MnCO₃ precursor with a small particle size of ε-MnO₂ was synthesized at 80 °C. During the calcination process, this spherical MnCO₃ structure restricted the growth of the crystal so that ε-MnO₂ was able to maintain its small particle size. In addition, their small domains of primary particles result in relatively high specific surface areas. In addition, the positions of characteristic diffraction peaks and the corresponding crystal planes are distinguishable for different crystalline types of MnO₂, indicating that they have various main exposed crystal planes. The crystal size calculated by the XRD patterns are given in Table 1.

Fig. 2 (a) XRD patterns and (b) Raman profiles of six crystal types of MnO₂ samples.

Table 1 Structural characteristic data for different crystalline types of MnO₂ samples

Samples	S BET (m^2 g^−1)	Pore volume (cm^3 g^−1)	Average pore size^b (nm)	Crystal size^c (Å)
a The SBET is calculated by micropore BET assistant. b The average pore size is calculated through BJH method by the desorption data. c The crystal size is calculated by Scherrer's formula.
α-MnO2	58.41	0.28	17.42	3.10
β-MnO2	19.99	0.08	3.05	5.11
γ-MnO2	63.39	0.37	9.55	2.40
δ-MnO2	76.06	0.30	17.47	7.09
ε-MnO2	199.48	0.39	6.55	2.38
λ-MnO2	25.54	0.09	3.82	4.65

---

Raman profiles (Fig. 2b) show that all the MnO₂-type materials synthesized have two diagnostic Raman scattering (RS) bands at 335 and 636 cm⁻¹ belonging to the symmetry of A_g from the vibrations of the [MnO₆] octahedron within the framework of the tetragonal dolomite-type frameworks. Gao et al.^32–34 reported that the vibrational bands are essentially related to the property of the channel structure. The strongest band at 636 cm⁻¹ is assigned with the (1 × 1) tunnel structure formed by the [MnO₆] octahedron of the tetragonal rutile framework, while the two vibrational bands at 182 and 567 cm⁻¹ are related to the large pore structure (δ-, ε-MnO₂: interlayer pore, α-MnO₂: 2 × 2 tunneling structure) formed by the introduction of K⁺ ions.³¹ In particular, the vibrational band at 182 cm⁻¹ is attributed to the (2 × 2) tunneling structure, which is unique to α-MnO₂. The generation of oxygen vacancies leads to the breakage of Mn–O bonds in the samples, resulting in lower peak intensities in the Raman spectra.³⁵ Therefore, the oxygen vacancies can be identified indirectly by examining the peak areas of the characteristic peaks (636 cm⁻¹) in the Raman spectra. Semiquantitative analysis of the peaks (Table S1†) reveals that the peak area corresponding to the signal intensity generated by the vibration of the Mn–O bond of ε-MnO₂ is significantly smaller than those of the other crystalline forms. This observation leads to the inference that its oxygen vacancy content is the highest among all the crystalline forms.

The type and size of tunnels, BET surface areas (S_BET), pore volumes and pore size distributions of the six catalysts are shown in Table 1 and Fig. S3.† Based on the nitrogen adsorption–desorption isothermal curves (Fig. S3†), the S_BET of the six catalysts decreased in the following order: ε-MnO₂ > δ-MnO₂ > γ-MnO₂ > α-MnO₂ > λ-MnO₂ > β-MnO₂. Among them, the S_BET of ε-MnO₂ reached 199.48 m² g⁻¹, which is the largest of the six crystalline types. All six crystal types of MnO₂ have more pronounced hysteresis loops, and their adsorption isotherms are always type IV isotherms. Among them, the hysteresis loops of α-MnO₂, β-MnO₂, γ-MnO₂, and λ-MnO₂, without any saturated adsorption platform, all belong to the H3-type hysteresis loops, indicating that their pore structures are highly irregular. Besides, the typical H2-type hysteresis loop isotherm of ε-MnO₂ shows an obvious saturated adsorption platform, reflecting its large number of mesoporous structures. For δ-MnO₂, its H3-type hysteresis loop has no platform and no clear mesoporous structure, while its hysteresis loop has a wide relative pressure range, implying that its mesoporous structure is a pseudopore formed by the secondary construction of lamellar particles due to the layer structure.³⁶ From the pore-size-distribution curves, it can be found that the pore-size distributions of α-MnO₂, γ-MnO₂, δ-MnO₂, and ε-MnO₂ are dominated by mesopores, with the average pore sizes ranging from 6.55 to 17.47 nm.

Among the six crystalline types, ε-MnO₂ has the highest pore volume (0.39 cm³ g⁻¹). The specific surface area and average pore sizes of β-MnO₂ and λ-MnO₂ are relatively small coupled with the fact that its pore capacity is also extremely low. We infer that β-MnO₂ and λ-MnO₂ have less cavity in their structures. The kinetic diameter of O₃ molecule is about 2.2 Å, which is much smaller than the average pore size of the as-prepared samples, and thus the pore size effect on molecule diffusion is totally excluded. The large pore volume of ε-MnO₂ associated with higher surface area (199.48 m² g⁻¹) usually provides more active sites, whereas the smaller pore volume of β-MnO₂ and λ-MnO₂ associated with a lower surface area (19.99 m² g⁻¹ and 25.54 m² g⁻¹) makes it more difficult to provide sufficient oxygen vacancies.

3.2 Crystal structure

The SEM and HRTEM images of the samples are shown in Fig. 3. As can be seen from the SEM images (Fig. 3a1–f1), the structures of α-MnO₂ (Fig. S4a†), β-MnO₂, and γ-MnO₂ are essentially similar and all of them are spheres composed of straight rods or needles with diverse lengths, which is because they belong to the same one-dimensional structure. On the other hand, δ-MnO₂ and ε-MnO₂ are spheres composed of flower-like flakes or thicker plates, which is due to the fact that δ-MnO₂ and ε-MnO₂ belong to a two-dimensional lamellar structure. In contrast, λ-MnO₂ is a three-dimensional structure formed by the accumulation of irregular small particle spheres. The difference in the morphologies among α-MnO₂, β-MnO₂, and γ-MnO₂ is that the spheres of β-MnO₂ (10–15 μm) exhibit a relatively large size, whereas those of α-MnO₂ and γ-MnO₂ (4–6 μm) display small ones. Besides, α-MnO₂ and γ-MnO₂ are comprised by nanorods with diameters of 20–80 nm and 10–15 nm (Fig. S4a and c†), while β-MnO₂ is composed of rods with large diameters of 100–150 nm and lengths of 1.0 μm (Fig. S4b†). Generally, among these one-dimensional MnO₂, a higher specific surface area is essentially correlated with a smaller diameter of primary rods and size of secondary constructed spheres. Moreover, the petal-like units comprising δ-MnO₂ have diameters in the range of 150–200 nm and thicknesses in the range of 10 nm (Fig. S4d†). On the other hand, the plates constituting ε-MnO₂ have a thickness of about 60 nm (Fig. S4e†). The secondary structure of λ-MnO₂ has a domain in the range of 150–350 nm and a thickness of approximately 30 nm (Fig. S4f†).

Fig. 3 SEM images of (a1) α-MnO₂, (b1) β-MnO₂, (c1) γ-MnO₂, (d1) δ-MnO₂, (e1) ε-MnO₂ and (f1) λ-MnO₂ as well as HRTEM images of (a2 and a3) α-MnO₂, (b2 and b3) β-MnO₂, (c2 and c3) γ-MnO₂, (d2 and d3) δ-MnO₂, (e2 and e3) ε-MnO₂ and (f2 and f3) λ-MnO₂ samples.

Based on the TEM images (Fig. 3a2–f2), the lattice fringes of various types of MnO₂ crystals can be clearly seen. In the case of α-MnO₂, the main lattice fringes with a width of 6.90 Å corresponding to the (1 1 0) crystal planes are clearly observed. As for β-MnO₂, the main lattice fringes with a width of 3.10 Å corresponding to the (1 1 0) crystal planes are visible. For γ-MnO₂, lattice stripes with a width of 2.42 Å corresponding to the (1 3 1) crystal plane are distinguishable in a large number. In the lamellar δ-MnO₂, the main exposed surface is only the (0 0 1) crystal facet with a lattice stripe of 7.06 Å, which is related to its laminar structure, as depicted by its TEM image. In the case of ε-MnO₂, the HRTEM images reveal a multitude of distinctive lattice fringes. The highest number of (1 0 2) crystalline facets is found to correspond with a width of 0.16 nm, which is in alignment with the findings of earlier research.³⁷ Meanwhile, the (1 0 0) crystalline facets, displaying a width of 0.24 nm, can also be found. The lattice fringes of λ-MnO₂ are dominated by the (1 1 1) facet with a width of 4.61 Å. In addition, significant discontinuities can be observed in the lattice fringes in the HRTEM images (Fig. 3a3–f3). When the concentration of defects is sufficiently high, a defect layer is formed, which leads to lattice fringe blurring, as observed in the HRTEM images (Fig. 3a3–f3). In comparison with the other samples, the lattice fringes of ε-MnO₂ shows sever discontinuities and are more blurred, which indicates a large number of defects existing in ε-MnO₂. Meanwhile, β-MnO₂, δ-MnO₂, and λ-MnO₂ have fewer breaks in the lattice fringes in their HRTEM images, indicating that its defect sites are less abundant.

3.3 Valence analysis

Fig. 4 shows the XPS spectra of six MnO₂ crystals obtained by analyzing the chemical state and surface composition of the as-prepared samples using X-ray photoelectron spectroscopy (XPS). It is worth noting that the samples were calcined before this XPS experiment and the chemisorbed water existing in the original sample was completely eliminated; thus, no obvious peaks of chemisorbed water appeared in the position above 532.4 eV. The peaks of O 1s spectra (Fig. 4a) can be divided into two regions. The binding energies in the range from 529.6 to 530.0 eV are attributed to lattice oxygen (O_latt),^38,39 which constitute the bulk-phase oxygen in the crystalline structure. The binding energies in the range from 530.6 to 531.4 eV are attributed to surface adsorbed oxygens (O_ads), a class of oxygen species that can favor the catalytic reaction, including species such as O₂⁻, O⁻, and O₂²⁻, which naturally form chemical bonds around surface oxygen vacancies.^40,41 The amount of O_ads reflects the concentration of oxygen vacancies in the sample. From the quantitative analysis listed in Table 2, it can be seen that the content of surface-adsorbed oxygen (O_ads) of ε-MnO₂ is the most abundant, accounting for 45% of the total oxygen species, indicating that there are abundant oxygen vacancies on the surface of ε-MnO₂.

Fig. 4 O 1s (a) and Mn 2p_3/2 (b) XPS spectra of MnO₂ samples.

Table 2 Quantitative analysis on XPS data of MnO₂ samples

	Mn^3+ (×10^3)	Mn^4+ (×10^3)	Mn^3+/Mn^4+	Oads (×10^4)	Olatt (×10^4)	Oads/Olatt	ΔES	AOS
α-MnO2	10.30	6.00	1.72	7.29	15.33	0.47	5.21	3.09
β-MnO2	8.12	6.94	1.17	1.25	4.01	0.30	4.65	3.72
γ-MnO2	10.33	7.28	1.42	6.24	15.92	0.39	5.11	3.20
δ-MnO2	8.12	4.85	1.40	5.14	15.47	0.33	4.80	3.55
ε-MnO2	10.85	6.28	1.73	8.16	15.00	0.54	5.42	2.85
λ-MnO2	6.34	5.54	1.26	5.07	17.49	0.29	4.88	3.46

---

The peaks of Mn 2p_3/2 (Fig. 4b) near 640.9 eV and 643.1 eV can be attributed to surface Mn³⁺ and Mn⁴⁺, respectively.^42–44 It is known that when Mn³⁺ is included in the oxide skeleton, ionic vacancies are formed on manganese dioxide in order to satisfy the charge balance.⁴⁵ Therefore, the content of Mn³⁺ is often considered as an indicator of surface oxygen vacancies. Among all crystalline types of MnO₂, α-MnO₂ and ε-MnO₂ have the highest content of Mn³⁺ (63%), which indicates a higher content of oxygen vacancies. In addition, due to the formation of oxygen vacancies, the Mn⁴⁺–O bond is more stable than the Mn³⁺–O bond.⁴⁶ The quantitative results indicate that both α-MnO₂ and ε-MnO₂ have abundant Mn³⁺ ions, which means that they have abundant unstable Mn–O bonds on their surfaces. At the same time, the molecular or atomic oxygen species adsorbed upon these surfaces with more Mn³⁺ are more likely to be released and participate in the catalytic decomposition of ozone. Similarly, the presence of lowly valent manganese on the surface can promote the dissociation and activation of surrounding oxygen species.⁴⁷

The average oxidation state (AOS) of manganese was calculated based on the bimodal split spacing (ΔE_S) of the Mn 3s spectra (Fig. S5†) by the following equation.

AOS = 8.956 − 1.126ΔE_s (2)

The results obtained are shown in Table 3. The AOS sequences of the samples are as follows: ε-MnO₂ (2.85) < α-MnO₂ (3.09) < γ-MnO₂ (3.20) < λ-MnO₂ (3.46) < δ-MnO₂ (3.55) < β-MnO₂ (3.72). This shows that ε-MnO₂ has the lowest AOS, in correspondence with its content of Mn³⁺, indicating that it may have the most abundant oxygen vacancies as well as the best mobility of lattice oxygen. In order to ensure the validity of the structures, iodometric titration was used to check the AOS of the accurate bulk manganese. As listed in Table S3,† the AOS in the bulk measured by iodometric titration is close to that over the surface determined by the Mn 3s spectrum. This implies the good crystallization of these as-prepared samples based on the similar AOS in the bulk and on the surface. Accordingly, the low density of XRD peaks of some samples might be ascribed to the essentially small particle sizes.

Table 3 Quantitative analysis on H₂-TPR profiles of MnO₂ samples

	Temperature/°C			H2 consumption (mmol g^−1)
	Peak α	Peak β	Peak γ	Peak α	Peak β	Peak γ	Total
α-MnO2	298.0	315.2	337.6	2.29	3.91	2.95	9.16
β-MnO2	316.1	333.7	426.6	3.94	3.14	4.33	11.42
γ-MnO2	299.1	336.7	443.3	0.81	4.62	3.85	9.28
δ-MnO2	306.0	—	338.9	3.88	—	7.44	11.32
ε-MnO2	296.2	314.2	407.5	3.46	1.42	3.29	8.18
λ-MnO2	334.4	359.1	462.3	4.54	2.53	4.04	11.10

---

3.4 Temperature-programmed analysis

The H₂-TPR curves of the six MnO₂ samples are shown in Fig. 5a. As the temperature continues to rise, the reduction process of MnO₂ can be divided into three parts. (α): MnO₂ → Mn₂O₃ (Mn⁴⁺ → Mn³⁺), (β): Mn₂O₃ → Mn₃O₄ (Mn³⁺ → Mn³⁺, Mn²⁺), (γ): Mn₃O₄ → MnO (Mn³⁺, Mn²⁺ → Mn²⁺). The reduction peaks are due to the gradual reduction of MnO₂ to Mn₂O₃/Mn₃O₄ and MnO, and the lower temperature of the first reduction peak indicates the better redox properties of the catalyst. The order of the low-temperature reduction peaks for the six types of crystals is ε-MnO₂ (296.2 °C) < α-MnO₂ (298.0 °C) < γ-MnO₂ (299.1 °C) < δ-MnO₂ (306.0 °C) < β-MnO₂ (316.1 °C) < λ-MnO₂ (334.4 °C). The hydrogen consumption of each crystal type can be calculated by quantitative analysis, as shown in Table 3. The theoretical hydrogen consumption of reduction from MnO₂ to MnO is 11.5 mmol g⁻¹. However, the hydrogen consumption of all the as-prepared samples is lower than the theoretical hydrogen consumption of MnO₂ to MnO, indicating the original presence of a certain amount of Mn³⁺ in the samples,^47,48 which is related to the oxygen vacancies. Among them, ε-MnO₂ has the lowest hydrogen consumption, indicating that its Mn³⁺ content is the highest among all the samples, which is in accordance with the findings of previous XPS analysis.

Fig. 5 H₂-TPR (a) and O₂-TPD (b) curves for the MnO₂ samples.

The adsorbed oxygen species over each type of crystal MnO₂ were investigated by O₂-TPD, and the results are shown in Fig. 5b. The desorption peaks below 400 °C are usually considered to be due to the desorption of α-O₂, which is related to surface molecular oxygen species.^49–53 The desorption peaks from 400 to 700 °C are due to the liberation of β-O₂, which is associated with surface lattice oxygen. The desorption peaks above 700 °C are due to the release of lattice oxygen diffused from the bulk. It is generally accepted that active physisorbed oxygen can be removed from the surface at low temperatures.⁵⁴ Herein, α-O₂ desorption with temperature below 400 °C was hardly observed. According to the β-O₂ peaks from 400 °C to 700 °C, the six types of MnO₂ are in the order ε-MnO₂ (501 °C) < γ-MnO₂ (543 °C) < α-MnO₂ (548 °C) < λ-MnO₂ (596 °C) < δ-MnO₂ (598 °C) < β-MnO₂ (679 °C). Earlier peaks indicate it is easier for the Mn–O bond to be broken up, thus leading to the greater mobility of surface lattice oxygen for oxygen vacancy formation. With the lowest β-O₂ peak, ε-MnO₂ has a greater mobility of lattice oxygen, which may lead to the more formation of oxygen vacancies and surface-active species to participate in the reaction.⁴⁹ The quantitative analysis of O₂-TPD is listed in Table S2.† It shows that ε-MnO₂ corresponds to the β-O₂ peak with the largest area, indicating the highest exchange of lattice oxygen. According to a former study,⁵⁵ the transformation of α-, β-, γ-, δ- and λ-MnO₂ to Mn₂O₃ occurs at about 500 °C. However, the initiation of the β-O₂ peak of ε-MnO₂ is at about 350 °C, which supports its high oxygen mobility.

To further investigate the oxygen exchange behaviors, we performed TPOIE experiments. The as-prepared samples followed a heterogeneous exchange mechanism, which means the exchange of one oxygen atom at a time.³⁶ The isotopic exchange curves of each sample is shown in Fig. S6.† The reaction of the TPOIE process is shown in the following formulas.

¹⁸O¹⁸O(g) + Mn–¹⁶O(s) ⇔ ¹⁸O¹⁶O(g) + Mn–¹⁸O(ads) (3)

¹⁸O¹⁶O(g) + Mn–¹⁶O(s) ⇔ ¹⁶O¹⁶O(g) + Mn–¹⁸O(ads) (4)

Here, (s), (g) and (ads) refer to the lattice oxygen in solid phase, gaseous oxygen and the adsorbed molecular oxygen, respectively. The showing curves of ¹⁸O¹⁶O resulted in the exchange between the lattice O of Mn–¹⁶O and ¹⁸O₂. Additionally, the curves of ¹⁶O₂ resulted in the exchange between ¹⁸O¹⁶O and Mn–¹⁶O₂. The ¹⁸O¹⁶O isotopic exchange curves of α-MnO₂ showed that it has a fast exchange rate among all the samples, proving its best oxygen mobility. Meanwhile, β-, γ- and ε-MnO₂ showed the maximum value of ¹⁸O¹⁶O, which meant they can also reach the isotopic exchange balance. It is noteworthy that the lower corresponding temperature of the maximum value of ¹⁸O¹⁶O indicates the better oxygen mobility, suggesting that oxygen mobility at low temperatures follows the order ε-MnO₂ > γ-MnO₂ > β-MnO₂ > λ-MnO₂. In addition, the total ¹⁸O¹⁶O exchange of λ-MnO₂ and δ-MnO₂ is lower than that of ¹⁶O₂, indicating that they have relative weak oxygen mobility and are insufficient to reach the exchange equilibrium of the first step.

The number of oxygen atoms exchanged (Ne) and the rate of oxygen exchange (Re) with increasing temperature are shown in Fig. 6. As shown in Fig. 6a, the Ne is different for various crystal types. Among them, ε-, γ- and α-MnO₂ have the maximum exchange volume, while β-, λ- and δ-MnO₂ have a relatively low exchange ability. Generally, a larger specific surface area allows for the optimal release and utilization of both surface adsorbed oxygens and lattice oxygens, resulting in an enhanced exchange capacity,⁵⁶ which may be one of the reasons for the higher exchange capacity of ε-MnO₂.

Fig. 6 (a) The total number of exchanged ¹⁸O (Ne) curves and (b) the exchange rate (Re) with the temperature of MnO₂ samples.

The disparity in oxygen migration ability of different crystalline catalysts can be further investigated based on the Re curve in Fig. 6b. In the experimental temperature range, the maximum exchange rate Re_max occurs when the exchange of gas and solid phases appears to be in equilibrium. Consequently, the oxygen mobility of the samples can be evaluated based on the temperature corresponding to Re_max. This implies that the lower temperature for Re_max associated with the higher oxygen-mobility rate. Therefore, the oxygen migration rates in these samples are in the following order: α-MnO₂ > ε-MnO₂ > γ-MnO₂ > δ-MnO₂ > λ-MnO₂ > β-MnO₂. As the temperature increases, a second extreme point will exist for some of the crystalline forms, while the corresponding temperature exhibits the mobility of the lattice oxygen, and it can be clearly seen that α-MnO₂ and ε-MnO₂ correspond to better mobility of the lattice oxygens, which is in good agreement with the previous results of O₂-TPD.

3.5 DFT calculation

As mentioned earlier, various MnO₂ correspond to different exposed surfaces. We therefore constructed separate models for the exposed crystal facets of manganese oxide, as shown in Fig. S8.† It is suggested that the formation of oxygen vacancies may be due to the easy loss of lattice oxygen and thereafter the generation of large amounts of surface reactive oxygen species adsorbed upon these ionic vacancies, which facilitates the corresponding redox reactions.⁴⁰ Therefore, the ease of oxygen vacancy formation theoretically supports the presence of large amounts of surface reactive oxygen species. Combined with the previous characterization, the Mn³⁺ content of different crystalline forms can be observed by XPS and H₂-TPR, which indicated the oxygen vacancies content. However, whether the oxygen vacancies are easy to form or not depends essentially on the oxygen formation energy. In order to assess the difficulty of oxygen vacancy formation in each crystal type of MnO₂, we calculated the oxygen-vacancy-formation energy of each crystal type based on density functional theory crystallography up to the first-principle calculations.where E_vf denotes the energy of formation of oxygen vacancies, E_(slab+vo) denotes the energy of the structure containing oxygen vacancies, E_O2 denotes the energy of O₂ molecule, and E_slab denotes the energy of the intact crystal cell. The calculation results of the oxygen vacancy content for each crystal type are shown in Table 4. The oxygen-vacancy-formation energy corresponded well with the O_ads content deduced from the XPS results.

Table 4 Oxygen vacancy formation energies for different crystal types of MnO₂

	Crystal plane	E vf (eV)
α-MnO2	(1 1 0)	0.46
β-MnO2	(1 1 0)	2.12
γ-MnO2	(1 3 1)	0.82
δ-MnO2	(0 0 1)	3.35
ε-MnO2	(1 0 2)	0.20
λ-MnO2	(1 1 1)	0.67

---

The order of oxygen vacancy formation energy is ε-MnO₂ < α-MnO₂ < λ-MnO₂ < γ-MnO₂ < β-MnO₂ < δ-MnO₂. The lower the absolute value of the oxygen vacancy formation energy is, the easier it is to form oxygen vacancies becomes. This is basically consistent with the previous characterization results.⁵⁷ Compared with other samples, the ε-MnO₂ sample is more prone to produce oxygen vacancies. Meanwhile, for δ-MnO₂ and β-MnO₂, the formation energies of oxygen vacancies are very large, which would make it difficult to generate oxygen vacancies in the reaction. It is noteworthy that the primary cell structure of δ-MnO₂ and ε-MnO₂ is from the same precursor,^19,31,58 but the different crystal planes dominantly exposed [δ-MnO₂ (1 1 0), ε-MnO₂ (1 0 2)] (Fig. S9†) lead to a huge difference in the energies for oxygen-vacancy formation. It suggests that the distinct surface facets can have a great influence on the catalytic behaviors in ozone decomposition.

Additionally, as shown in Fig. 7, by comparing the structures before and after the calculation, it can be found that after the oxygen vacancies are created, there is a significant changing of the Mn–O bonds and Mn–Mn bonds around the oxygen vacancies, showing a clear Jahn–Teller effect.⁵⁹ This disorder change of the crystal structure makes it easier for activated lattice oxygen to participate in the reaction.

Fig. 7 Changes in the structural and bond length between two neighboring Mn atoms before and after oxygen vacancy formations: (a1 and a2) α-, (b1 and b2) β-, (c1 and c2) γ-, (d1 and d2) δ-, (e1 and e2) ε- and (f1 and f2) λ-MnO₂.

In order to further investigate the interaction of O₃ with MnO₂, we used the nudged elastic band (NEB) method to carry out a search for the transition state of O₃ from adsorption to bond cleavage. Based on the methodology of NEB transition state search, we performed calculations for the top three samples with the superior activity. The results of the calculation are shown in Fig. 8.

Fig. 8 Free energy during the reaction coordinate and the specific process of the reaction of O₃ with MnO₂.

The reaction paths are essentially the same for α-MnO₂ and ε-MnO₂. However, there is a little difference in the first O₃ adsorption. The reaction of the first O₃ with MnO₂ is illustrated by Fig. 8. First, O₃ adsorbed on the surface shifts to some extent, and after reaching a certain position, one of the O–O bonds elongates and breaks after a certain length, forming O₂· and O·. Subsequently, O· binds to the crystal and O₂· escapes. The reaction of γ-MnO₂ with O₃ is different: first, the O₃ adsorbed on the indicated rotates, the bond angle initially becomes smaller, and O–O bond breakage occurs after the displacement until it is suitable. This phenomenon should be attributed to their difference in pore tunnels. γ-MnO₂ has predominantly (1 × 2) and (1 × 1) tunnels, where the extension of the O₃ bond is limited by the pore structure at this exposed crystal surface; thus, the O–O bond, which should have been elongated and then broken, led to a change in the bond angle. Similarly, Chen et al.⁶⁰ observed that MnO₂ with the (2 × 2) tunnel structure is much more active than MnO₂ with the (1 × 1) structure since the effective diameter of the (2 × 2) tunnel is more suitable. Galliez demonstrated that the hydrogen trapping reaction by γ-MnO₂ is also limited by the tunnels via DFT calculations.⁶¹ This suggests that the appropriate pore aperture of MnO₂ for ensuing O₃ molecules to access the active sites through diffusion is an important factor in promoting the catalytic activity. Structurally, the (1 0 2) phase of ε-MnO₂ exposed showed sufficient reaction sites, while the (1 3 1) phase of γ-MnO₂ brings the limitation of bond-breaking of ozone. During the reaction, it can be seen that different crystal types differ in their reaction energy barriers. The height of the energy barrier (ΔE_a) indicates the magnitude of the energy barrier that needs to be overcome to transform the reactants into products. The higher the energy barrier is, the more difficult the reaction becomes. According to our calculations, ε-MnO₂ has the lowest reaction potential barrier (ΔE_a = 0.95 eV), demonstrating that its reaction with O₃ is rather easier to carry out. The energy barrier of γ-MnO₂ (ΔE_a = 1.55 eV) is significantly higher than those of α-MnO₂ (ΔE_a = 0.98 eV) and ε-MnO₂, indicating that the reaction pathway of γ-MnO₂ needs to overcome more energetic barriers and that the reaction is relatively more difficult to take place. We also performed transition state calculations for the O₂²⁻ desorption process. As shown in Fig. 8, the transition state in the desorption of O₂²⁻ showed higher energy barriers for all the three types of MnO₂, indicating that the desorption of the O₂²⁻ should be the decisive step. The O₂²⁻ desorption energy is in order ε-MnO₂ (2.04 eV) < α-MnO₂ (2.30 eV) < γ-MnO₂ (2.76 eV). It can be seen that for both ozone adsorption and O₂²⁻ desorption, ε-MnO₂ has the minimal energy barrier, suggesting its high activity in catalytic ozone decomposition.

The mechanism of MnO₂ in catalytic ozone decomposition is described as follows. First, O₃ absorbed on the bulk with oxygen vacancies (O_v), filled the oxygen vacancies to form a perfect bulk and O₃ rapidly broke bonds. In order to satisfy the charge balance, electrons from Mn³⁺ are transferred to the O atoms to form O²⁻ (eqn (6)). After the desorption of O₂, another O₃ is absorbed on the perfect bulk. Subsequently, the O–O bond in O₃ breaks to form O₂ and gets out of the bulk, forming a perfect cell. Moreover, O²⁻ in the lattice loses an electron, which transfers to O*, forming O₂²⁻ (eqn (7)). Finally, each O₂²⁻ gives two electrons to the two nearby Mn⁴⁺ into oxygen desorption while restoring the oxygen vacancies (eqn (8)). If the O₂²⁻ formed during the reaction is not released in time to restore the oxygen vacancies, the reaction rate would be slowed down. Actually, the activation energy (ΔE_a) for O₂²⁻ desorption is much higher than that for O₃ dissociation for the same crystalline form (ε-MnO₂: 2.04 eV > 0.95 eV, α-MnO₂: 2.30 eV > 0.98 eV, γ-MnO₂: 2.76 eV > 1.55 eV). Therefore, the rate-determining step is O₂²⁻ desorption to restore the surface oxygen vacancies.^13,62,63 In the catalytic decomposition of ozone, molecular oxygen is primarily adsorbed on the oxygen vacancies on the surface of the samples, where it is successively decomposed to produce atomic oxygen and other surface activated oxygen species, including O₂⁻ and O₂²⁻, which participate in the overall reaction process. From the above mechanism, it is clear that oxygen vacancy is the key to promote the reaction.

[Mn³⁺–O_v–Mn³⁺] + O₃ → [Mn⁴⁺–O²⁻–Mn⁴⁺] + O₂(g) (6)

[Mn⁴⁺–O²⁻–Mn⁴⁺] + O₃ → [Mn⁴⁺–O₂²⁻–Mn⁴⁺] + O₂(g) (7)

[Mn⁴⁺–O₂²⁻–Mn⁴⁺] → [Mn³⁺–O_v–Mn³⁺] + O₂(g) (8)

3.6 Catalytic activity test

The catalytic performance of ozone decomposition over each type of MnO₂ at room temperature was tested using a fixed-bed continuous flow reactor, and the results are shown in Fig. 9. It was seen that at 25 °C and 200 min reaction time, the activities of diverse crystal types for ozone decomposition are significantly different, decreasing in the order ε-MnO₂ > α-MnO₂ > γ-MnO₂ > β-MnO₂ ≈ δ-MnO₂ > λ-MnO₂. As seen from Fig. 9, ε-MnO₂ has the highest activity, reaching 100% ozone depletion and this efficiency is maintained within 200 minutes. The catalytic activity of α-MnO₂ was slightly stronger than that of γ-MnO₂ after 200 minutes of the reaction. The activity for α-MnO₂ and γ-MnO₂ respectively decreased to 80% and 76%. The activities of β-MnO₂ and δ-MnO₂ exhibited diminished performance at a comparable level, with the activity declining to less than 30% after 200 minutes of reaction. For λ-MnO₂, the rate of activity declined sharply at the initial stage of reaction and reached approximately 60% by 30 minutes. Thereafter, the rate of deactivation slowed and its reactivity dropped to 40% after 200 minutes. In order to study the stability of ε-MnO₂, the long-term performance of ε-MnO₂ was investigated in Fig. S7.† This best-performing ε-MnO₂ can maintain complete O₃ removal lasting for above 14 h.

Fig. 9 Reaction curves of catalyzed ozone over time at room temperature for each crystal type of the MnO₂ samples (reaction conditions: initial ozone concentration: 25 °C, 40 ppm, RH = 0, catalyst mass 100 mg, GHSV = 5800 h⁻¹).

3.7 Correlation between structure and performance

As discussed in section 3.6, the catalytic activity test yielded the following order: ε-MnO₂ > α-MnO₂ > γ-MnO₂ > β-MnO₂ ≈ δ-MnO₂ > λ-MnO₂. The activity of MnO₂ in the catalytic decomposition of ozone is the result of a multifactorial determination. Among them, oxygen vacancies are decisive for the fast reaction. Suitable exposed crystalline surfaces provide more oxygen vacancies and facilitate the catalytic reaction. The (1 0 2) facet on ε-MnO₂ provides good exposure orientation, and with this crystal surface as the primary exposed surface, it is better able to participate in the catalytic removal of O₃. In order to avoid the unobjective effect of discussing only by per surface area or per volume on the catalyst activity, we used the normalized reaction rate (R_norm, the calculation details are displayed in the ESI,† Fig. S10) to evaluate the special activity of the catalyst with respect to surface area, and the results are given in Fig. S10.† The order of R_norm for each catalyst is consistent with the order of activity with respect to volume. In addition, by comparing the specific surface areas of β- and δ-MnO₂, we can infer that although the surface area of β-MnO₂ (19.99 m² g⁻¹) is significantly lower than that of δ-MnO₂ (76.06 m² g⁻¹), the activity of these two catalysts is close to each other, which suggests that the specific surface area is not the decisive factor for the catalytic activity. Raman mapping shows that ε-MnO₂ has the most defective sites (Table S1†), indicating that it has the highest content of oxygen vacancies. The XPS results show that ε-MnO₂ is the best both in terms of the content in Mn³⁺ (Mn³⁺/Mn⁴⁺ = 1.73) and O_ads (O_ads/O_latt = 0.54), which corresponds with the Raman results. By calculating the oxygen-vacancy-formation energy, it was found that the oxygen-vacancy-formation energy of ε-MnO₂ (0.20 eV) is significantly lower than those of other crystalline types, indicating that it is very susceptible to the formation of oxygen vacancies.

The decisive step in the catalytic removal of ozone by MnO₂ is the desorption of the peroxygenated species O₂²⁻. If the oxygen vacancies in one reaction cycle are not recovered in time, they will affect O₃ adsorption in the next circular step. DFT calculations show that ε-MnO₂ has the lowest activation energy required for the desorption of O₂²⁻, suggesting that it is more likely to desorb O₂²⁻ and restore oxygen vacancies. The process of O₂²⁻ desorption to form oxygen vacancies is also directly related to oxygen mobility. O₂-TPD and TPOIE showed that ε-MnO₂ exhibits the highest oxygen mobility, enabling the rapid removal of O₂²⁻ to restore the oxygen vacancies.

As the rate-determining step, the desorption energy of O₂²⁻ determines the whole reaction rate. Through the DFT calculation, we found that the best three samples have a significant discrepancy in the O₂²⁻ desorption potentials, which is in order ε-MnO₂ (2.04 eV) < α-MnO₂ (2.30 eV) < γ-MnO₂ (2.76 eV). This can well explain the active performance of ε-MnO₂: due to the lower potential barrier (2.04 eV) in the rate-decisive step, the O₂²⁻ generated during its reaction can be desorbed in time to restore the oxygen vacancies and start the cycle of the next reaction.

In addition to the decisive step of O₂²⁻ desorption, the decomposition of molecular O₃ on MnO₂ is also crucial for the reaction. By calculating the transition states via the NEB method, we found that the decomposition mechanism of O₃ by MnO₂ of different crystalline types is different, which largely depends on its pore structure and exposed surface. When the pores are large enough (ε-MnO₂ and α-MnO₂), one of the O–O bonds in O₃ elongates before breaking the bond. When the pore structure is limited (γ-MnO₂), the decomposition of O₃ leads to a reduction of the bond angle before its breakage. The activation energy of the decomposition process of adsorbed O₃ is in the order ε-MnO₂ < α-MnO₂ < γ-MnO₂. This is also the same order as the activity and the energy barrier of the rate-determining step.

Elsewhere, the large specific surface area of ε-MnO₂ (199.48 m² g⁻¹) leads to a high exposure of the crystal surface, resulting in the formation of more oxygen vacancies. On the other hand, the S_BET of δ-MnO₂ (76.06 m² g⁻¹) is higher than that of β-MnO₂ (19.99 m² g⁻¹), but there is no significant activity difference between them; thus, S_BET is not the decisive factor in ozone decomposition. The large pore volume of ε-MnO₂ (0.39 cm³ g⁻¹) is able to accommodate more reactants and products, and its suitable pore size (6.55 nm) provides a high specific surface area without limiting the diffusion of O₃.

4. Conclusion

In this study, six types of MnO₂ (α-, β-, γ-, δ-, ε- and λ-MnO₂) catalysts were synthesized for ozone removal. Compared to other Mn-based catalysts reported in the literature (Table S4†), the as-prepared ε-MnO₂ in this study performs outstanding behaviors in ozone removal. In summary, the room-temperature decomposition of ozone over MnO₂ is dominated by a combination of multiple factors. The different facets of crystals have various structures, which also determines that the exposed surfaces are essentially distinguishable, which has important implications in catalytic O₃ decomposition. The HRTEM images showed their respective exposed crystal surfaces, which enclose different capacities for ozone adsorption and reaction sites. Among all types of MnO₂, ε-MnO₂ shows the best O₃ decomposition activity under the ambient conditions (25 °C, dry air). Compared with the other samples, it has the largest specific surface area, which provides a large number of adsorption sites. Moreover, its abundant oxygen vacancies play a crucial role in the ozone-decomposition process. The formation of oxygen vacancies associated with (1 0 2) facet leads to the partial distortion of the structure, weakening the Mn–O bonding energy; therefore, the lattice oxygen mobility is substantially increased. Raman analysis showed that a large number of oxygen vacancies existed in ε-MnO₂. Its low average oxidation state (2.85), high Mn³⁺ content (Mn³⁺/Mn⁴⁺ = 1.73) and adsorbed oxygen content (O_ads/O_latt = 0.54) were also verified by the Raman spectroscopy. In addition, good reducibility and strong oxygen mobility are also crucial for its high activity. DFT calculations reveal that it is easy for ε-MnO₂ (E_vf = 0.20 eV) to form oxygen vacancies. The calculations of the reaction pathway also showed that ε-MnO₂ has minimal reaction activation energy whether at the decisive step of O₂²⁻ decomposition (ΔE_a = 2.04 eV) or O₃ adsorption (ΔE_a = 0.95 eV), thus supporting its high activity for ozone decomposition.

This work elucidated the structure–performance relationship of MnO₂ toward the catalytic decomposition of ozone at ordinary conditions and clarified the origin of high catalytic activity, and the reaction mechanism was also revealed by DFT calculations, which provides guidance for future research on MnO₂-based materials toward the ozone purification.

Data availability

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

Author contributions

Haotian Wu: conceptualization, methodology, investigation, validation, computation, writing – original draft. Runduo Zhang: writing – review and editing, supervision, and funding acquisition. Bin Kang: formal analysis. Xiaonan Guo: software. Zhaoying Di: methodology. Kun Wang: data curation. Jingbo Jia: validation. Ying Wei: methodology. Zhou-jun Wang: methodology.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This work was supported by National Key Research and Development Program of China (2023YFC3905400), National Natural Science Foundation of China (No. 22176010), and the Fundamental Research Funds for the Central Universities (JD2405).

References

1. J. Jia, P. Zhang and L. Chen, The effect of morphology of α-MnO₂ on catalytic decomposition of gaseous ozone, Catal. Sci. Technol., 2016, 6, 5841–5847.
2. R. Atkinson, Atmospheric chemistry of VOCs and NO_x, Atmos. Environ., 2000, 34, 2063–2101.
3. M. O. Fadeyi, Ozone in indoor environments: Research progress in the past 15 years, Sustain. Cities Soc., 2015, 18, 78–94.
4. H. Wang, C. He, L. Morawska, P. McGarry and G. Johnson, Ozone-initiated particle formation, particle aging, and precursors in a laser printer, Environ. Sci. Technol., 2012, 46, 704–712.
5. A. Zukeran, Y. Ikeda, F. Mashimo, K. Yasumoto, M. Matsumoto and N. Matsumoto, Effect of collection efficiency and reducing ozone generation in an electrostatic precipitator with knife edge disk type electrode, J. Electrost., 2023, 125, 103842.
6. S. W. Benson and A. E. Axworthy, Jr., Mechanism of the Gas Phase, Thermal decomposition of ozone, J. Chem. Phys., 1957, 26, 1718–1726.
7. B. Dhandapani and S. T. Oyama, Gas phase ozone decomposition catalysts, Appl. Catal., B, 1997, 11, 129–166.
8. J. H. Hibben, I. The effect of radiation on the decomposition of ozone and nitrous oxide, J. Am. Chem. Soc., 1928, 50, 937–940.
9. P. Lee and J. Davidson, Evaluation of activated carbon filters for removal of ozone at the PPB level, Am. Ind. Hyg. Assoc. J., 1999, 60, 589–600.
10. C.-C. Lin, C.-Y. Chao, M.-Y. Liu and Y.-L. Lee, Feasibility of ozone absorption by H₂O₂ solution in rotating packed beds, J. Hazard. Mater., 2009, 167, 1014–1020.
11. Y. S. Mok, D. J. Koh, D. N. Shin and K. T. Kim, Gaseous ozone decomposition using a nonthermal plasma reactor with adsorbent and dielectric pellets, Korean J. Chem. Eng., 2009, 26, 1613–1619.
12. Y. Chen, W. Qu, C. Li, J. Chen, Z. Ma and X. Tang, Ultra-low-temperature ozone abatement on α-MnO₂ (001) facets with down-shifted lowest unoccupied orbitals, Ind. Eng. Chem. Res., 2018, 57, 12590–12594.
13. J. Jia, P. Zhang and L. Chen, Catalytic decomposition of gaseous ozone over manganese dioxides with different crystal structures, Appl. Catal., B, 2016, 189, 210–218.
14. L. Zhang, S. Wang, C. Ni, M. Wang and S. Wang, Ozone elimination over oxygen-deficient MnO_x based catalysts: Effect of different transition metal dopants, Chem. Eng. Sci., 2021, 229, 116011.
15. X. Li, J. Ma, L. Yang, G. He, C. Zhang, R. Zhang and H. He, Oxygen vacancies induced by transition metal doping in γ-MnO₂ for highly efficient ozone decomposition, Environ. Sci. Technol., 2018, 52, 12685–12696.
16. Z. Xu, W. Yang, W. Si, J. Chen, Y. Peng and J. Li, A novel γ-like MnO₂ catalyst for ozone decomposition in high humidity conditions, J. Hazard. Mater., 2021, 420, 126641.
17. Z. Wang, T. Li, S. Zhang, R. Zhang, Y. Zhang and Q. Zhong, One-step synthesis of δ-MnO₂ with rich defects for efficient ozone decomposition under humid conditions, Chem. Eng. J., 2024, 488, 150693.
18. T. Gopi, G. Swetha, S. Chandra Shekar, C. Ramakrishna, B. Saini, R. Krishna and P. V. L. Rao, Catalytic decomposition of ozone on nanostructured potassium and proton containing δ-MnO₂ catalysts, Catal. Commun., 2017, 92, 51–55.
19. W. Hong, M. Shao, T. Zhu, H. Wang, Y. Sun, F. Shen and X. Li, To promote ozone catalytic decomposition by fabricating manganese vacancies in ε-MnO₂ catalyst via selective dissolution of Mn-Li precursors, Appl. Catal., B, 2020, 274, 119088.
20. R. Yang, Z. Guo, L. Cai, R. Zhu, Y. Fan, Y. Zhang, P. Han, W. Zhang, X. Zhu, Q. Zhao, Z. Zhu, C. K. Chan and Z. Zeng, Investigation into the phase–activity relationship of MnO₂ nanomaterials toward ozone-assisted catalytic oxidation of toluene, Small, 2021, 17, 2103052.
21. L. Li, Q. Yang, D. Wang, Y. Peng, J. Yan, J. Li and J. Crittenden, Facile synthesis λ-MnO₂ spinel for highly effective catalytic oxidation of benzene, Chem. Eng. J., 2021, 421, 127828.
22. Y. Liu, P. Zhang, J. Zhan and L. Liu, Heat treatment of MnCO₃: An easy way to obtain efficient and stable MnO₂ for humid O₃ decomposition, Appl. Surf. Sci., 2019, 463, 374–385.
23. Y. Li, J. Liu and D. He, Catalytic synthesis of glycerol carbonate from biomass-based glycerol and dimethyl carbonate over Li-La₂O₃ catalysts, Appl. Catal., A, 2018, 564, 234–242.
24. J. Ma, C. Wang and H. He, Transition metal doped cryptomelane-type manganese oxide catalysts for ozone decomposition, Appl. Catal., B, 2017, 201, 503–510.
25. H. Deng, S. Kang, J. Ma, L. Wang, C. Zhang and H. He, Role of structural defects in MnO_x promoted by Ag doping in the catalytic combustion of volatile organic compounds and ambient decomposition of O₃, Environ. Sci. Technol., 2019, 53, 10871–10879.
26. Y. Liu and P. Zhang, Catalytic decomposition of gaseous ozone over todorokite-type manganese dioxides at room temperature: Effects of cerium modification, Appl. Catal., A, 2017, 530, 102–110.
27. L. Yang, J. Ma, X. Li, C. Zhang and H. He, Enhancing oxygen vacancies of Ce-OMS-2 via optimized hydrothermal conditions to improve catalytic ozone decomposition, Ind. Eng. Chem. Res., 2020, 59, 118–128.
28. H.-L. Yuan, Y.-L. Zhao, X.-J. Qin, Y.-P. Liu, H.-F. Yu, P.-F. Zhu, Q. Jin, X.-W. Yang and X.-D. Luo, Anti-inflammatory and analgesic activities of Neolamarckia cadamba and its bioactive monoterpenoid indole alkaloids, J. Ethnopharmacol., 2020, 260, 113103.
29. W. Hong, T. Zhu, Y. Sun, H. Wang, X. Li and F. Shen, Enhancing oxygen vacancies by introducing Na⁺ into OMS-2 tunnels to promote catalytic ozone decomposition, Environ. Sci. Technol., 2019, 53, 13332–13343.
30. E. Hayashi, Y. Yamaguchi, K. Kamata, N. Tsunoda, Y. Kumagai, F. Oba and M. Hara, Effect of MnO₂ crystal structure on aerobic oxidation of 5-Hydroxymethylfurfural to 2,5-furandicarboxylic acid, J. Am. Chem. Soc., 2019, 141, 890–900.
31. F.-X. Tian, H. Li, M. Zhu, W. Tu, D. Lin and Y.-F. Han, Effect of MnO₂ polymorphs' structure on low-temperature catalytic oxidation: Crystalline controlled oxygen vacancy formation, ACS Appl. Mater. Interfaces, 2022, 14, 18525–18538.
32. T. Gao, H. Fjellvåg and P. Norby, A comparison study on Raman scattering properties of α- and β-MnO₂, Anal. Chim. Acta, 2009, 648, 235–239.
33. C. Lume-Pereira, S. Baral, A. Henglein and E. Janata, Chemistry of colloidal manganese dioxide. 1. Mechanism of reduction by an organic radical (a radiation chemical study), J. Phys. Chem., 1985, 89, 5772–5778.
34. S. Ma, X. Ye, X. Jiang, W. Cen, W. Jiang and H. Wang, First principles calculation of mechanical, dynamical and thermodynamic properties of MnO₂ with four crystal phases, J. Alloys Compd., 2021, 852, 157007.
35. L. Fan, J. Zhao, F. Jing, Z. Ma, Y. Fan, X. Qin and G. Shao, Fabrication of oxygen-vacancy abundant MnO₂ nanowires@NiMn_xO_y-δ nanosheets core-shell heterostructure for capacity supercapacitors, J. Energy Storage, 2022, 52, 104845.
36. B. Shi, Z. Di, X. Guo, Y. Wei, R. Zhang and J. Jia, Facet control of manganese oxides with diverse redox abilities and acidities for catalytically removing hazardous 1,2-dichloroethane, Mater. Adv., 2022, 3, 1101–1114.
37. L. Zhang, S. Yang, W. Fu, Y. Cui, J. Wang, D. Zhao, C. Yang, X. Wang and B. Cao, Plasma-induced ε-MnO₂ based aqueous zinc-ion batteries and their dissolution-deposition mechanism, J. Mater. Sci. Technol., 2022, 127, 206–213.
38. P. Hu, Z. Amghouz, Z. Huang, F. Xu, Y. Chen and X. Tang, Surface-confined atomic silver centers catalyzing formaldehyde oxidation, Environ. Sci. Technol., 2015, 49, 2384–2390.
39. J. Wang, H. Zhao, J. Song, T. Zhu and W. Xu, Structure-activity relationship of manganese oxide catalysts for the catalytic oxidation of (chloro)-VOCs, Catalysts, 2019, 9, 726.
40. M. Li, R. Zhang, H. Wang, H. Chen and Y. Wei, Role of the exposure facets upon diverse morphologies of cobalt spinels on catalytic deN₂O process, Catal. Today, 2021, 376, 177–187.
41. K. Xiang, Z. Xu, T. Qu, Z. Tian, Y. Zhang, Y. Wang, M. Xie, X. Guo, W. Ding and X. Guo, Two dimensional oxygen-vacancy-rich Co₃O₄ nanosheets with excellent supercapacitor performances, Chem. Commun., 2017, 53, 12410–12413.
42. W. Liu, H. Ge, Z. Gu, X. Lu, J. Li and J. Wang, Electrochemical deposition tailors the catalytic performance of MnO₂-based micromotors, Small, 2018, 14, 1802771.
43. Z. Liu, K. Xu, P. She, S. Yin, X. Zhu and H. Sun, Self-assembly of 2D MnO₂ nanosheets into high-purity aerogels with ultralow density, Chem. Sci., 2016, 7, 1926–1932.
44. C. Ma, Y. Wen, C. Rong, N. Zhang, J. Zheng and B. H. Chen, δ-MnO₂ with an ultrahigh Mn⁴⁺ fraction is highly active and stable for catalytic wet air oxidation of phenol under mild conditions, Catal. Sci. Technol., 2017, 7, 3200–3204.
45. B. Peng, W. Bao, L. Wei, R. Zhang, Z. Wang, Z. Wang and Y. Wei, Highly active OMS-2 for catalytic ozone decomposition under humid conditions, Pet. Sci., 2019, 16, 912–919.
46. Y. Zhang, Electronegativities of elements in valence states and their applications. 1. Electronegativities of elements in valence states, Inorg. Chem., 1982, 21, 3886–3889.
47. W. Yang, Z. A. Su, Z. Xu, W. Yang, Y. Peng and J. Li, Comparative study of α-, β-, γ- and δ-MnO₂ on toluene oxidation: Oxygen vacancies and reaction intermediates, Appl. Catal., B, 2020, 260, 118150.
48. F. Shi, F. Wang, H. Dai, J. Dai, J. Deng, Y. Liu, G. Bai, K. Ji and C. T. Au, Rod-, flower-, and dumbbell-like MnO₂: Highly active catalysts for the combustion of toluene, Appl. Catal., A, 2012, 433–434, 206–213.
49. B. Kang, M. Guo, H. Wu, X. Guo, Z. Di, Y. Wei, J. Jia, Z.-j. Wang and R. Zhang, Effect of alkali/alkaline-earth-metal doping on the Co₃O₄ spinel structure and N₂O decomposition, Catal. Sci. Technol., 2024, 14, 2825–2837.
50. T. Bu, J. Li, Q. Lin, D. P. McMeekin, J. Sun, M. Wang, W. Chen, X. Wen, W. Mao, C. R. McNeill, W. Huang, X.-L. Zhang, J. Zhong, Y.-B. Cheng, U. Bach and F. Huang, Structure engineering of hierarchical layered perovskite interface for efficient and stable wide bandgap photovoltaics, Nano Energy, 2020, 75, 104917.
51. H. Li, P. Zhang, J. Jia, X. Wang and S. Rong, A cobalt-vacant Co₃O₄ as a stable catalyst for room-temperature decomposition of ozone in humid air, Appl. Catal., B, 2024, 340, 123222.
52. X. Liang, L. Wang, T. Wen, H. Liu, J. Zhang, Z. Liu, C. Zhu and C. Long, Mesoporous poorly crystalline α-Fe₂O₃ with abundant oxygen vacancies and acid sites for ozone decomposition, Sci. Total Environ., 2022, 804, 150161.
53. T. Liu, Y. Yao, L. Wei, Z. Shi, L. Han, H. Yuan, B. Li, L. Dong, F. Wang and C. Sun, Preparation and evaluation of copper–manganese oxide as a high-efficiency catalyst for CO oxidation and NO reduction by CO, J. Phys. Chem. C, 2017, 121, 12757–12770.
54. W. Bao, H. Chen, H. Wang, R. Zhang, Y. Wei and L. Zheng, Pt nanoparticles supported on N/Ce-doped activated carbon for the catalytic oxidation of formaldehyde at room temperature, ACS Appl. Nano Mater., 2020, 3, 2614–2624.
55. T. Hatakeyama, N. L. Okamoto and T. Ichitsubo, Thermal stability of MnO₂ polymorphs, J. Solid State Chem., 2022, 305, 122683.
56. S. Royer, D. Duprez and S. Kaliaguine, Oxygen mobility in LaCoO₃ perovskites, Catal. Today, 2006, 112, 99–102.
57. B. Baldassarri, J. He, A. Gopakumar, S. Griesemer, A. J. A. Salgado-Casanova, T.-C. Liu, S. B. Torrisi and C. Wolverton, Oxygen vacancy formation energy in metal oxides: High-throughput computational studies and machine-learning predictions, Chem. Mater., 2023, 35, 10619–10634.
58. Y. Yang, Y. Li, Q. Zhang, M. Zeng, S. Wu, L. Lan and X. Zhao, Novel photoactivation and solar-light-driven thermocatalysis on ε-MnO₂ nanosheets lead to highly efficient catalytic abatement of ethyl acetate without acetaldehyde as unfavorable by-product, J. Mater. Chem. A, 2018, 6, 14195–14206.
59. L. Chen, Y. Liu, X. Fang and Y. Cheng, Simple strategy for the construction of oxygen vacancies on α-MnO₂ catalyst to improve toluene catalytic oxidation, J. Hazard. Mater., 2021, 409, 125020.
60. T. Chen, H. Dou, X. Li, X. Tang, J. Li and J. Hao, Tunnel structure effect of manganese oxides in complete oxidation of formaldehyde, Microporous Mesoporous Mater., 2009, 122, 270–274.
61. K. Galliez, P. Deniard, C. Payen, D. Lambertin, F. Bart, H.-J. Koo, M.-H. Whangbo and S. Jobic, Pair distribution function and density functional theory analyses of hydrogen trapping by γ-MnO₂, Inorg. Chem., 2015, 54, 1194–1196.
62. B. Xu, H. Qin, X. Wu and Y. Sun, Ozone decomposition mechanism at different structural oxygen vacancies on manganese dioxide, J. Phys. Chem. C, 2022, 126, 17076–17083.
63. L. Wu, Z. Wang, J. Liu, C. Liu, X. Li, Y. Zhang, W. Wang, J. Ma and Z. Sun, Selective ozone catalyzation modulated by surface and bulk oxygen vacancies over MnO₂ for superior water purification, Appl. Catal., B, 2024, 343, 123526.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4en00857j

---