Potassium permanganate (KMnO₄, ≥99 %), Manganese sulphate (MnSO₄·H₂O, ≥98 %), Cobalt Sulfate (CoSO₄·7H₂O, ≥98 %) were purchased from Sinopharm Chemical Reagent Co., Ltd. Deionized water was obtained from a Millipore Q water purification system. All reagents and chemicals were of analytical grade.

Co-doped α-MnO₂ were synthesized using a modified hydrothermal method, which was showed in Fig. S1. Co-doped α-MnO₂ were synthesized via a one-step hydrothermal method according to a previous report [26]. 36 mmol of MnSO₄·H₂O and certain molar of CoSO₄·7H₂O were dissolved in 50 mL of deionized water under stirring, and then the above mixed solution was added into 50 mL Potassium homologate solution (KMnO₄ was 20 mmol) dropwise, followed by stirring magnetically for about 30 min until the solution became homogeneous. After that, it was transferred into a 200 mL Teflon-lined stainless-steel autoclave. The autoclave was kept at 160 °C for 16 h in an oven and then cooled to room temperature, and then, the product was collected by filtration and fully rinsed several times with deionized water to remove K⁺, followed by drying at 105 °C for 8 h. In order to determine the optimal doping ratio, we prepared catalysts with different initial molar ratio of Co and Mn (0.1, 0.2, 0.3, 0.4 and 0.5 respectively). The sample with Co/Mn = 0.2 had the highest catalytic ozonation of phenol (Fig. S2). Therefore, catalysts with Co/Mn = 0.2 named Co-α-MnO₂ was the materials synthesized. The synthetic procedures for α-MnO₂ was similar to that for Co-doped α-MnO₂ with the exception of adding CoSO₄·7H₂O to the initial solution.

The catalyst was purged 5 h at 120 °C under nitrogen atmosphere protection and then determined the BET surface area and pore-size distribution on a Micromeritics ASAP2020 analyser (Micromeritics TriStar 3000, U.S.) when the sample was cooled. Powder X-ray diffraction (XRD) analysis was carried out on a Bruker D8 ADVANCE Phaser (Rigaku D/MAX-2200PC, Japan) using Cu Ka radiation (k = 0.15418 nm) with a LYNXEYE detector at 40 kV and 30 mA. X-ray photoelectron spectroscopy (XPS) was performed with an ESCALAB250 system (ESCALAB 250, U.S.) equipped with an Al Kα excitation source and operated at 15 kW and 1486.6 eV. Fourier transformed infrared (FTIR) analyses (Nicolet iS10, U.S.) were carried out using a NicoletiS10 FT-IR plus spectrophotometer in a wavelength range of 4000–400 cm⁻¹.

In this study, phenol was selected as the aim pollutant because it is one of the most common industrial wastewater contaminates and would cause serious ecological and environmental problems [27]. The phenol degradated by catalytic ozonation was performed in a self-regulating quartz reaction with 1 L aqueous solution showed in Fig. S3. The concentration of ozone was adjusted via controlling the current and flow rate of the ozone generator and measured by ozone detector. When the ozone concentration is stable and then bubbled into the integrated adsorption-catalytic ozonation reactor to start the reaction. The pneumatic panel was adopted to ensure that gaseous ozone was completely mixed with aqueous solution. The pH_zpc of the prepared catalysts (α-MnO₂, Co-α-MnO₂) in this study was 8.03 and 9.87, respectively. The pH value of the solution was set at 10.00 to achieve the zero potential of the catalyst surface, which enhanced the catalytic ability [28, 29]. During the reaction, ozone concentration and flow rate were set at 3.0 mg/L and 3.0 L/min, respectively. The tail gas was absorbed by potassium iodide solution. 2 g catalyst was mixed with 1 L phenol wastewater (initial concentration was 400 mg/L; pH = 10; 293 ± 1 K) in the reaction system. The residual phenol concentration of water samples were measured by ultraviolet spectrophotometry to evaluate removal performance. Phenol removals were also described with pseudo zero-order, first order and second order kinetics, and the kinetic models were expressed as Eq. (1), Eq. (2) and Eq. (3).

Where K₁, K₂, K₃ are the pseudo zero-order, first-order and second-order rate constant, respectively (min-1); t represents the reaction time; C₀ and C_t stand for the phenol concentration at 0 min and t min, respectively.

Effect of adsorption, ozonation and catalytic ozonation on phenol removal are investigated and compared in this part, which are shown in Fig. 1. It is clear to see that the phenol removal efficiencies are all less than 5% by adsorption (showed in red and black line in Fig. 3). So, the contribution of catalyst adsorption on phenol removal efficiency could be neglectable. The phenol removal efficiencies during 40 min are in order of Co-α-MnO₂ (97.47%) > α-MnO₂ (81.01%) > single ozonation (58.55%).

The parameters of kinetic fitting for the removal phenol are shown in Table S1. The removal efficiency of catalytic ozonation on phenol well accords with the first-order reaction kinetics, while that of single ozonation accords with the zero-order reaction kinetics. The first order reaction kinetics shows that the removal efficiency of phenol by catalytic ozone is proportional to the first order of phenol concentration. The zero-order reaction kinetics shows that the removal efficiency of phenol by single ozone is propotional to the zero power of the phenol concentration. Based on first-order kinetic fitting, Co-α-MnO₂ achieves the highest reaction rate constant for phenol removal (0.092 min⁻¹), which is 2.09 and 4 times higher than that in α-MnO₂ (0.044 min⁻¹) and O₃ (0.023 min⁻¹), respectively. It can be further verified that the doping of Co on α-MnO₂ efficiently improved the catalytic activities. The results might be attributed to the changes of catalyst construction, elementary composition and surface properties, et al., which will be detailly discussed in the following section.

The calculated results of specific surface area and pore volume of α-MnO₂ and Co-α-MnO₂ are showed in Table 1. It can be seen that Co-α-MnO₂ catalyst has a bigger of specific surface area (79.496 m²/g) and pore volume (0.0396 cm³/g) than that of α-MnO₂. The larger the specific surface area and pore volume, the better catalytic activity can be obtained. The reason is that the large specific surface area and pore volume can promote the mass transfer rate between solid-liquid-gas [30]. So, catalytic activity of Co-α-MnO₂ is greater than that of α-MnO_2. N₂ adsorption-desorption isotherm showed that two kinds of catalysts belong to typically III type (as showed in Fig. S4) [31], indicating that the interaction between catalysts and adsorbate is weak. The surface of the catalysts generally adsorbs small molecules such as H₂O molecules, which decomposed to generate hydroxyl radicals to promote the catalytic ozonation [32]. The absorbed H₂O always dissociate into OH⁻ and H⁺ forming the surface hydroxyl groups with surface cations and oxygen anions. In the system, surface hydroxyl groups can decompose ozone through a series of reactions and produce active free radicals to promote catalytic ozonation.

The XRD results of α-MnO₂ and Co-α-MnO₂ catalysts are shown in Fig. 2. All patterns of the prepared α-MnO₂ and Co-α-MnO₂ catalyst samples can be indexed to body-centered tetragonal, which are same as that of the standard α-MnO₂ (JCPDS NO.44-0141). An additional peak at 2θ of 65.8 in the Co-doped catalyst is presumed to be the formation of a similar CoMn₂O₄ structure [33]. In addition, there are no other additional characteristic peaks, indicating that Co mixed in the α-MnO₂ mainly replaced Mn and entered the MnO₂ skeleton structure or embedded in the pore structure of the catalyst tunnel, instead of existing on the surface of the catalyst in the form of cobalt oxide, and the replaced Mn may enter the hydrothermal synthesis solution in the process of reaction. The reason for the decrease of the grain size of Co-α-MnO₂ from 320 Å to 287 Å may be that the doped metal ions enter the original phase, which limits the diffusion rate of Mn²⁺ and Mn7⁺ in the precursor prepared by the catalyst, and finally leads to the change of the lattice parameters and grain size of the catalyst. Also, the lattice parameters a, b, and c of Co-α-MnO₂ become greater compared with that of α-MnO₂ (as shown in Table 1), which show a distortion in the lattice cell. The increased relative intensity of α-MnO₂ diffraction peak after Co doping also indicates the existence of lattice defects, which might promote the migration of adsorbed oxygen to decompose and transform organic matter and finally affected the catalytic performance of the catalyst [34, 35]. Therefore, Co-α-MnO₂ has better catalytic ozonation of phenol than α-MnO₂.

In order to further investigate the elemental valence on the catalyst surface, Mn 2P, O1s and Co2p of both α-MnO₂ and Co-α-MnO₂ characterized by XPS were presented in Fig. 3(b), Fig. 3(c) and Fig. 3(d), respectively. Fig. 3(b) shows that the XPS spectrum of Mn 2p contains two main peaks at binding energies of 657.5 ± 7.5 and 642.5 ± 2.5eV, which could be attributed to Mn2P_1/2 and Mn2p_3/2. The binder energy of the peak at 641.6 eV, 642.5 eV and 643.4 eV are found in Mn2p_3/2, which can be attributed to Mn³⁺ and Mn⁴⁺ [36, 37], respectively. As showed in Table S2, the relative content of Mn³⁺ of Co-α-MnO₂ increases from 8.42% to 41.16% and the ratio of Mn³⁺/Mn⁴⁺ increases from 0.092 to 0.70 compared with α-MnO₂. According to the results of the Table 2, the main valence state of Co in the catalyst is Co³⁺, which indicates that the doped Co²⁺ has a redox reaction in the catalyst preparation process and promotes the increase of the ratio of Mn³⁺/Mn⁴⁺ in Co-α-MnO₂. The higher ratio of Mn³⁺/Mn⁴⁺ in Co-α-MnO₂ is an important reason for the occurrence of more oxygen vacancy, which would promote the generation of reactive oxygen species and facilitate the migration and transformation of oxygen species [38, 39]. In addition, Mn³⁺ and other unsaturated metal ions behave as Lewis acid in the aqueous phase, which is easy to coordinate with water molecules and undergo chemical adsorption to form the surface hydroxyl group. Furthermore, the XPS spectrums of O1s for α-MnO₂ and Co-α-MnO₂ are showed in Fig. 3(c). The binding energy energies at the peak position 530.1 eV and 531.5 eV in the O1s spectrum, corresponding to the lattice oxygen and the adsorbed oxygen respectively [40]. Oxygen presenting form and relative content of oxygen calculated are showed in Table 2. Compared with α-MnO₂, the ratio of adsorbed oxygen (O_ads)/lattice oxygen (O_latt) of Co-α-MnO₂ increases from 0.15 to 0.23, indicating that more O_latt would combine with H₂O to produce surface hydroxyl, which is the active site for the catalyst to adsorb and decompose ozone to produce reactive oxygen, and then improve the efficiency of catalytic ozone oxidation. Surface hydroxyl group is well confirmed by the results of the following FT-IR analysis.

In order to expose the influence of doping Co on surface functional groups of α-MnO₂, the catalysts were characterized with FT-IR, and the results are present in Fig. 4. Although the position is slightly shifted, the intensities of stretching vibration of Mn-O and Mn-O-Mn bonds (467cm⁻¹, 524 cm⁻¹ and 719 cm⁻¹) [41,42] are greatly improved, which is caused by the microstructure regulation due to the doping of Co on α-MnO₂ and has been verified by the results of XRD. The characteristic peak of Mn-OH presents at the wavelength of 1042 cm⁻¹, which is constructed by adsorbing water molecules on the catalyst surface and shedding hydrogen ions [43]. Compared with the α-MnO₂, the peak intensity of Mn-OH in Co-α-MnO₂ increases obviously, indicating that the adsorption capacity of α-MnO₂ is improved by Co doping. The new peaks appeared at 879cm⁻¹, 1387cm⁻¹ corresponding to the stretching vibration of Co-O [44], further confirm the XRD results of Co replacing Mn into the structure of α-MnO₂. The other three new peaks appeared at 1611cm⁻¹, 2969cm⁻¹and 3320cm⁻¹ corresponded to the stretching vibration of the surface hydroxyl (O-H) [45], which are produced by flexural and telescopic vibration of adsorbed water on the catalyst surface. Therefore, in the Co-α-MnO₂ catalytic ozonation system, more surface hydroxyl was generated, and they were showed as acidic or alkaline groups through proton exchange with aqueous solution, which promotes electron transfer and produces reactive oxygen species contributing to higher catalytic ozonation efficiency [46, 47].

Under the same experimental conditions, five repeated experiments of catalytic ozone oxidation of phenol were carried out with the prepared Co-α-MnO₂. The results of reusability of catalyst were shown in Fig. S5. With the increasing of reuse times loading from one to five, the phenol removal results after 40 min were 97.47 %, 93.66%, 91.83%, 86.72%, 83.80%, respectively. It can be found that the removal efficiency of phenol led to a gradually decrease with the increase of reuse times. The collision and friction between catalysts and the deposition of other substances on the surface or inside the pore of the catalyst lead to the disappearance of some surface-active sites, resulting in the loss of catalytic activity of some active sites of Co-α-MnO₂. However, the catalytic activity of Co-α-MnO₂ for ozonation of phenol after repeated use is still higher than that of α-MnO₂. These results suggested that Co-α-MnO₂ has excellent stability and reusability.