Activation of peroxymonosulfate by metal (Fe, Mn, Cu and Ni) doping ordered mesoporous Co3O4 for the degradation of enrofloxacin

Various transition metals (Fe, Mn, Cu and Ni) were doped into ordered mesoporous Co3O4 to synthesize Co3O4-composite spinels. Their formation was evidenced by transmission electronic microscopy (TEM), X-ray diffraction (XRD) and Brunauer–Emmett–Teller (BET) analysis. It was found that Co3O4-composite spinels could efficiently activate peroxymonosulfate (PMS) to remove enrofloxacin (ENR) and the catalytic activity followed the order Co3O4–CuCo2O4 > Co3O4–CoMn2O4 > Co3O4–CoFe2O4 > Co3O4–NiCo2O4. Moreover, through the calculation of the specific apparent rate constant (ksapp), it can be proved that the Co and Cu ions had the best synergistic effect for PMS activation. The Co3O4-composite spinels presented a wide pH range for the activation of PMS, but strong acidic and alkaline conditions were detrimental to ENR removal. Higher reaction temperature could promote the PMS activation process. Sulfate radical was identified as the dominating reactive species in Co3O4-composite spinel/PMS systems through radical quenching experiments. Meanwhile, the probable mechanisms concerning Co3O4-composite spinel activated PMS were proposed.


Introduction
During the past years, sulfate radical (SO 4 À c) based advanced oxidation processes (SR-AOPs) have attracted an increasing interest among researchers owing to their great potential in degradation or even mineralization of recalcitrant organic pollutants. 1 Compared with the hydroxyl radical (cOH), SO 4 À c possesses a longer lifespan, higher independence of pH and higher selectivity of oxidation. 2,3 Peroxymonosulfate (PMS), a precursor of SO 4 À c, is deemed as a cost-effective and environmental-friendly oxidant. 4 PMS remains stable in aqueous solution and barely decomposes into SO 4 À c by itself, but it can be activated to produce SO 4 À c by the use of UV, transition metals, and some nonmetal catalysts. [5][6][7] Among different activation technologies, transition metals have attracted much attention due to their lower energy consumption and higher activation efficiency. Actually, many transition metal ions such as Co 2+ , Mn 2+ , Ni 2+ , Fe 2+ , Ru 3+ , Ce 3+ and so forth, have been proved as qualied catalysts for PMS activation. 2 Of note, Co 2+ has been found to possess the highest reactivity. 8 Unfortunately, the Co 2+ /PMS process is unfavorable in practical application because of the toxicity of Co 2+ . In order to relieve the secondary pollution, heterogeneous cobalt-based catalysts have become a research hotspot.
Anipsitakis et al. rstly employed Co 3 O 4 to activate PMS and found Co 3 O 4 presented an excellent catalytic behavior in the activation of PMS. 9 Chen et al. successfully prepared nanoscale Co 3 O 4 and tested its catalytic performance in PMS solution, results showed that 0.2 mM acid orange 7 (AO7) can be completely degraded within 30 min by 2 mM PMS in the presence of 0.5 g L À1 Co 3 O 4 . 10 Pu et al. fabricated three types of Co 3 O 4 using different metal organic frameworks, and found that all the Co 3 O 4 exhibited outstanding catalytic activity and the difference in catalytic ability can be attributed to the difference in specic surface area. 11 Consequently, Co 3 O 4 /PMS system is quite acceptable from the view of application due to the high activation efficiency and limitation of cobalt leaching. However, on the basis of the underlying threat of cobalt ions, it is essential to take measures to further limit the cobalt leakage during PMS activation.
It is reported that bimetallic oxides may be desirable catalysts to ease the conict between catalytic performance and metal ions leaching, because intimate interactions between two metals can effectively suppress the leakage of metal ions, such as Fe-Co interactions in CoFe 2 O 4 . 2 Moreover, bimetallic oxides are also prominent PMS activators. Su et al. synthesized a series of Co x Fe 3Àx O 4 nanoparticles and found that the higher cobalt content in Co x Fe 3Àx O 4 showed the higher catalytic activity towards PMS. 12 The high catalytic behavior of CoFe 2 O 4 was also illustrated in our previous study. 13 Yao 14 Similarly, CuCo 2 O 4 also exhibited high catalytic performance and low metal leachability in PMS solution. 15 In our previous research, order mesoporous Co 3 O 4 (OM-Co 3 O 4 ) was fabricated and showed superior catalytic ability toward PMS than its spinel counterpart, but the leakage of cobalt was up to 77.74 mg L À1 which was higher than conventional Co 3 O 4 nanoparticles. 16 Therefore, it can be reasonably speculated that cobalt leaching will reduce if some transition metals are doped to OM-Co 3 O 4 to form mixed spinels with cobalt.
Herein, diverse transition metals (i.e., Fe, Mn, Cu and Ni) was introduced into OM-Co 3 O 4 to synthesized a series of Co 3 O 4composite spinels which were characterized by transmission electronic microscope (TEM), X-ray diffraction (XRD), Brunauer-Emmett-Teller (BET) and Zeta potential analysis. Due to the ubiquitous detection in aquatic environment, 17 enrooxacin (ENR) was selected as target pollutant in this study. The catalytic activities of as-prepared Co 3 O 4 -composite spinels were systematically compared through apparent rate constant, PMS consumption, intensity of electron paramagnetic resonance (EPR) signal and specic apparent rate constant. Moreover, the effects of initial pH and reaction temperature during PMS activation were also investigated. Finally, a possible mechanism of PMS activation was proposed through quenching tests. To be best of our knowledge, it is the rst time to apply order mesoporous Co 3 O 4 -composite spinels as effective PMS activators for the control of organic pollutants.

Catalysts preparation and characterization
OM-Co 3 O 4 was synthesized using nanocasting route with KIT-6 as hard template, and the procedure was conducted as described before. 16  The crystal structures of catalysts were characterized by X'Pert PRO diffractometer (PANalytical, Holland) with Cu Ka radiation. The morphologies and structures of Co 3 O 4 -composite spinels were observed using transmission electron microscopy (TEM) (Philips, Holland). N 2 adsorption and desorption isotherms were measured using ASAP 2010 analyzer (Micromeritics, USA) at liquid nitrogen temperature (À196 C). The pH at point of zero charge (pH pzc ) was determined by Zetasizer Nano analyzer (Malvern, UK).

Catalytic experimental procedure
The catalytic degradation experiments were performed with a 100 mL ENR solution at 10 mg L À1 in 250 mL brown glass bottles, which were installed in a controlled temperature water bath stirring apparatus. In a typical run, specic amount of catalysts was added to ENR solution to receive adsorptiondesorption equilibrium, followed by pH adjustment with H 2 SO 4 and NaOH solution (100 mM) to ensure a desirable pH value aer PMS addition. Subsequently, an appropriate amount of PMS was charged into the reaction solution to initiate experiment. At dened time intervals, 1 mL samples were collected and quenched by 0.1 mL Na 2 SO 3 (100 mM). The resulting mixtures were immediately ltered by a 0.22 mm syringe lter for further analysis. All experiments were carried out in duplicates and the mean values were reported (with error bar).

Analytical methods
ENR concentrations were measured through a highperformance liquid chromatograph (HPLC, Agilent 1200, USA) with an Eclipse XDB-C18 column (5 mm particle, 150 Â 4.5 mm), the concentrations were measured at l ¼ 278 nm using a mobile phase consisting of a mixture of acetonitrile and phosphoric acid (pH ¼ 2.5) (v/v ¼ 20 : 80) at a ow rate of 1.0 mL min À1 . The PMS concentrations were measured by the method of Waclawek et al. 18 EPR analysis were performed on a Bruker A300 spectrometer (Germany) with DMPO as a spin-trapping agent. The parameters of EPR spectrometer were center eld was 3360.67 G, sweep width was 100 G, static eld was 3310.66 G, microwave frequency was 9.42 GHz, microwave power was 2.03 mW, modulation amplitude was 1.0 G and sweep time was 30.72 s.

Characterization
The crystalline phases of OM-Co 3 O 4 and Co 3 O 4 -composite spinels were displayed in Fig. 1 The TEM and HR-TEM images of OM-Co 3 O 4 and Co 3 O 4composite spinels were showed in Fig. 2. It can be clearly seen that OM-Co 3 O 4 showed a highly ordered mesoporous structure, and the spacing distances between two fringes are 0.285 and 0.467 nm, which were in conformity with (220) and (111) planes, respectively. Obviously, aer the introduction of metal dopants, ordered mesoporous structure was partly or completely destroyed, which may be attributed to the formation of Co 3 O 4composite spinels. The lattice fringes can be clearly observed in HR-TEM images, indicating that highly crystalline nature of Co 3 O 4 -composite spinels, which was corresponded to the strong and sharp diffraction peaks in XRD analysis. Similarly with OM-Co 3 O 4 , the spacing distances between two fringes in Co 3 O 4 -CoFe 2 O 4 were 0.281 and 0.471 nm, corresponding to (220) and (111) planes, respectively. And that in Co 3 O 4 -CoMn 2 O 4 were 0.279 and 0.469 nm, which were also assigned to (220) and (111) planes, respectively. However, as for Co 3 O 4 -CuCo 2 O 4 and Co 3 O 4 -NiCo 2 O 4 , the (111) plane was not observed, and the lattice spacing of 0.275 and 0.286 nm was corresponded to (220) plane.
The surface areas and pore size distributions of OM-Co 3 O 4 and composite materials were investigated by N 2 adsorptiondesorption isotherms. As shown in Fig. 3 CoM 2 O 4 , which can also be observed from TEM images. The changes of pore volume and pore diameter also demonstrated the conclusion, as seen in the Fig. 3(b), the pore volume and pore diameter all followed the order of OM- The textural parameters of OM-Co 3 O 4 and Co 3 O 4 -composite spinels were summarized in Table 1.
XPS analysis can be used to determine the surface composition and chemical oxidation states of OM-Co 3 O 4 and Co 3 O 4composite spinels. In XPS spectra of OM-Co 3 O 4 ( Fig. 4(a)), the sharp peak emerged at 779.6 eV was assignable to Co 2p 3/2 , which could be deconvoluted into octahedral Co 3+ at 779.4 eV and tetrahedral Co 2+ at 780.7 eV. 20 The proportions of Co 2+ and Co 3+ were determined to be 63.01% and 36.99%, respectively. The O 1s envelope ( Fig. 4(b)) could be deconvoluted into two parts, namely the lattice oxygen (O latt ) at 529.2 eV and surface adsorbed oxygen (O ads ) at 530.8 eV. 16 Based on this deconvolution, the proportions of O latt and O ads were found to be 50.86% and 49.14%, respectively. Aer the doping of iron into OM-Co 3 O 4 , the content of Co 2+ increased from 63.01% to 65.34%, which could be ascribed to the substitution of Co 3+ with Fe 3+ in OM-Co 3 O 4 . It was worth noting that the content of O ads increased from 50.86% to 51.74%, which was conductive to the PMS activation. 4 As seen in Fig. 4(c), the doped iron existed in the form of positive trivalent.

Catalytic activity of Co 3 O 4 -composite spinels
ENR removal in different systems was presented in Fig. 5 where [ENR] 0 is the initial ENR concentration, [ENR] is the concentration of ENR at time t, and k app is the apparent rate constant. As seen in Fig. 4 (Fig. 5(b)), further authenticating the order of catalytic activity of Co 3 O 4 -composite spinels. It was suggested that the catalytic performance of Co 3 O 4composite spinels not only depended on the specic surface area, but relied on dopant itself. In order to eliminate the difference in the specic surface area, the specic apparent rate constant k sapp which dened as the ratio of k app to the BET surface area was introduced: where k sapp is the specic apparent rate constant, and S BET is the specic surface area of composite spinels. As shown in Fig. 5(

Effect of initial pH
The inuence of initial pH ranging from 3 to 11 on ENR degradation was investigated in the PMS activation process, and the results were displayed in Fig. 6. From Fig. 6(a)-(e), it can be seen that four Co 3 O 4 -composite spinels all showed a wide pH range for PMS oxidation and higher ENR removals were obtained in pH range of 5 to 9 while lower removals occurred at strong acidic and alkaline conditions. Similar results were also reported by the previous investigations, such as degradation of orange II in MnFe 2 O 4 /PMS process and removal of acetaminophen in Fe 3 O 4 /PMS system. 25,26 The ENR degradation was signicantly inhibited at strong acidic condition might be originated from the attachment of H + to the peroxide bond (O-O) of PMS (eqn (3)) and the change of catalyst surface charge (eqn (4)), so that the interfacial repulsion would result in a weaker catalytic performance. 27 The retardation of ENR removal at strong alkaline condition can be ascribed to the following reasons: ( ) ¼ 1.22 V) was less oxidative and more difficult to react. 28 Additionally, SO 5 2À could also lead to a stronger electrostatic repulsion between catalyst surface and PMS anions.
The values of k sapp were also calculated and the results were presented in Fig. 6(f). It could be more intuitionistic to compare the catalytic performances of four Co 3 O 4 -composite spinels in different conditions due to the elimination of difference in specic surface area. As shown in Fig. 5(f)

Effect of temperature
The effect of reaction temperature (25,35,45 and 55 C) on ENR removal in the process of PMS activation was studied. As displayed in Fig. 7(a-d) 29,30 In addition, higher reaction temperature was benecial for reactant molecules to overcome activation energy barrier. 17 The activation energy (E a ) could be determined by plotting ln k app against 1/T based on Arrhenius equation (Fig. 7(e) activated PMS systems, respectively. The lower E a value signied the higher catalytic reactivity, and the order of E a was well corresponded to the sequence of catalytic activity. Moreover, all the E a values were much higher than that of the diffusion-controlled reactions, which usually ranged from 10 to 13 kJ mol À1 . 31 This implied that the apparent reaction rate for ENR removal during Co 3 O 4 -composite spinels activated PMS processes was dominated by the rate of intrinsic chemical reactions on the catalyst surface. It was reported that out-sphere interactions were usually diffusion-controlled reactions, thus PMS activation by Co 3 O 4 -composite spinels was most likely an inner-sphere electron-transfer process. 32 The consumption of PMS during ENR oxidation processes was monitored (Fig. 7(f)). Similar with the trend of k app , PMS consumption also increased as the reaction temperature increased. with the reaction temperature increased from 25 to 55 C, suggesting that higher reaction temperature was conducive to PMS activation and ENR degradation, which was well correspond to the conclusions by the observations of k app values. In addition, the higher PMS consumption also reected higher catalytic reactivity. From Fig. 7(f)

Radical identication and catalytic mechanism
Three different scavengers, tert-butyl alcohol (TBA), ethanol (EtOH) and phenol were employed to identify the dominant radical species in Co 3 O 4 -composite spinels/PMS systems. TBA can rapidly react with cOH (k cOH ¼ 3.8-7.6 Â 10 8 M À1 s À1 ) but has a much lower reactivity with SO 4 À c (k SO 4 À c ¼ 4-9.1 Â 10 5 M À1 s À1 ), 33 and EtOH is a well scavenger for cOH and SO 4 À c (k cOH ¼ 1.2-2.8 Â 10 9 M À1 s À1 , k SO 4 À c ¼ 1.6-7.7 Â 10 7 M À1 s À1 ). 34 Phenol can also react with cOH and SO 4 À c at a high rate (k cOH ¼ 6.6 Â 10 9 M À1 s À1 , k SO 4 À c ¼ 8.8 Â 10 9 M À1 s À1 ). 35 In view of the difference in reaction rates, using TBA, EtOH and phenol as scavengers was a feasible program for the identication of primary active species. As presented in Fig. 8(a-d), only a slight reduction of ENR removal could be obtained in the presence of 10 or 100 mM TBA, implying that cOH was involved in Co 3 O 4 -composite spinels activated PMS processes. With the addition of 10 mM EtOH, the ENR degradation was signicantly inhibited and the removal efficiencies in Co 3  composite spinels and cOH was also involved in these processes. XPS analysis of Co 3 O 4 -CoFe 2 O 4 before and aer catalytic oxidation was also performed to illustrate the heterogeneous catalytic mechanism (Fig. 4). As shown in Fig. 4(a), before catalytic oxidation, the contents of Co 2+ and Co 3+ was determined to be 65.34% and 34.66%. Aer catalytic oxidation, the proportions of Co 2+ and Co 3+ were changed to 61.93% and 38.07%, respectively. The partial increase of Co 3+ was ascribed to the electrons donating of Co 2+ during the oxidation process. In the case of O 1s spectra ( Fig. 4(b)), the content of O latt decreased from 48.26% to 46.35%, and the proportion of O ads increased from 51.74% to 53.65%. The increment of O ads can be attributed to the generation of Co-OH or O 2 adsorbed on the surface of Co 3 O 4 -CoFe 2 O 4 . It has been reported that^Co 2+ À À OH was the critical species for the generation of radicals during the process of PMS activation. 36 Of note, the Fe 2p 3/2 envelope could be deconvoluted into Fe 2+ at 709.5 eV, which indicated that the redox reactions between Co and Fe were involved in the PMS activation.
Based on the results of quenching experiments and XPS analysis, the plausible mechanisms of Co 3 O 4 -composite spinels activated PMS were put forward. Taking Co 3 O 4 -CoFe 2 O 4 /PMS system as example, H 2 O molecules were rstly physically absorbed on the part of^Co 2+ sites to generate^Co 2+ -À OH. Then,^Co 2+ À À OH would react with HSO 5 À to form SO 4 À c aer introduction of PMS (eqn (5)), and could regenerate through the reaction between formed^Co 3+ À À OH species and HSO 5 À (eqn (6)). Similarly,^Fe 3+ could also combine with dissociative adsorption of H 2 O molecules to form^Fe 3+ -À OH, which would transform to^Fe 2+ À À OH (eqn (7)) and generate SO 4 À c by reacting with HSO 5 À (eqn (8)). In addition, due to the standard redox potential of^Co 3+ /^Co 2+ was 1.92 V, 8 reduction of^Co 3+ by^Fe 2+ was thermodynamically feasible (eqn (9)). The efficient regeneration of surface^Co 2+ by this process may be able to remain the stability and high efficiency of Co 3 O 4 -CoFe 2 O 4 . Besides,^Co 2+ or^Fe 2+ on catalyst surface could also react with PMS to produce cOH (eqn (10) and (11)), which could also be generate by the transformation of SO 4 À c (eqn (12) and (13) Table 2.

Conflicts of interest
No conict of interest exists in the submission of this manuscript and manuscript is approved by all authors for publication.