Ni-substituted LaMnO₃ perovskites for ethanol oxidation

Abstract

The B-site substitution of LaMnO₃ perovskites by Ni was investigated under the oxidation of ethanol. Proper characterization techniques, including BET surface area measurement, XRD, FTIR, XPS, TPR, O₂-TPD, and FE-SEM experiments were performed to survey the physicochemical properties of perovskites. The results reveal that up to 25% of Mn can be replaced by Ni; beyond this limit, segregated NiO_x can be synthesized. Inserting Ni into the solid solution of perovskite yields unique bridging lattice oxygen sites (Ni–O–Mn) in Ni-doped LaMnO₃. Based on catalytic performance in ethanol oxidation, the Ni–O–Mn sites are likely to promote ethanol conversion and the oxidation of acetaldehyde to CO₂ at low reaction temperatures. The abatement of intermediates over Ni–O–Mn sites is hypothesized and a plausible reaction pathway is proposed. Moreover, the time on-stream testing revealed that the interaction between Ni and Mn is likely to enhance perovskite's thermal stability in ethanol oxidation.

---

Introduction

Ethanol can be derived from renewable energy crops and lignocellulosic biomass. Using ethanol as a fuel additive is a potential way to alleviate our fossil-fuel addiction. For example, 10% and 85% ethanol–gasoline blended fuels, i.e., E10 and E85, are commercially available for drivers in the US. As demand for energy continues to increase, the market growth for ethanol fuel is inevitable.¹ However, burning ethanol-containing fuels yields different vehicle exhaust compounds from those produced by burning fossil fuels.² More volatile organic compounds (VOCs) are generated using 10% and 15% ethanol–diesel blends in existing engines when compared to certified diesel fuel.³ Unburned ethanol and its derivatives, such as acetaldehyde, a carcinogen in humans,⁴ are harmful pollutants. Hence, the development of suitable catalysts for the abatement of ethanol and ethanol derivatives is urgently required.

Perovskite (ABO₃) is an effective catalyst for the removal of CO, NO_x, VOCs, and hydrocarbons.^5,6 In some cases, perovskites perform comparably to, or even better than, conventional (precious metal-based) three-way catalysts, such as strontium-containing perovskites in NO_x abatement.⁷ Daihatsu Motors has successfully implemented trace platinum group metal (pgm)-doped perovskites in their exhaust converter systems.^8–11 The self-regeneration behavior of pgm-promoted perovskites under oxidative and reductive conditions makes them more active and durable than platinum-based catalysts. Our group has recently reported on the effectiveness of similar perovskites in methanol partial oxidation^12,13 and combustion.¹⁴ The unique lattice structure and oxygen nonstoichiometry of perovskite are believed to be important in oxidation reactions.⁵

Surprisingly, only a few studies of ethanol combustion using perovskites are available. The pioneering study by the Hitachi research group¹⁵ tested LnBO₃ (Ln = La, Pr, Sm, and Gd; B = Fe, Co and Ni) as ethanol sensors in ethanol oxidation. The short response time at low ethanol concentration makes perovskite a promising candidate in ethanol conversion. This inspires follow-up studies to utilize various compositions and elements at the A- or B-site position.^16–22 Wang et al.¹⁷ and Najjar and Batis²¹ reported relatively high reactivities (low T₅₀ and T₉₅, where the former denotes the temperature for 50% ethanol conversion, and the latter is the temperature for 95% ethanol conversion) of their catalysts in ethanol oxidation. Wang et al.¹⁷ used Ag-supported La_0.6Sr_0.4MnO₃ in both methanol and ethanol oxidations, and reported that anchored Ag⁺ on the surface of perovskite not only acts as a catalytic promoter, but also increases the surface adsorbed oxygen-to-lattice oxygen (O₂⁻²/O⁻) ratio; both factors are essential in the conversion of alcohols. Najjar and Batis²¹ investigated gelling agents (glycine and citric acid) for LaMnO₃ synthesis using a combustion method, and discovered that LaMnO₃ made by using glycine as the gelling agent had a high thermal stability and activity in ethanol combustion. This is attributed to the decrease of surface La/Mn ratio and the increase of the adsorbed oxygen species on the perovskite's surface.

This study investigates the use of Ni-incorporated LaMnO₃ in ethanol oxidation. LaMnO₃ is known to have a wide range of oxidative (excess oxygen) and reductive (oxygen vacancy) non-stoichiometries.²³ The non-stoichiometric oxygen content markedly affects various oxidation reactions, such as methane²⁴ and propane²³ oxidations. This can be ascribed to the oxygen adsorption–activation step on catalyst surface and the enhancement of oxygen mobility.¹⁷ Partial substitution of Mn with cations with a valence less than three (such as Cu) is another way to improve the oxidation chemistry either in the bulk or on the surface of LaMnO₃-based perovskites.²³ In this work, Ni is chosen as the B-site dopant because of its bivalence and reactivity in oxidative environments.^25–27 Incorporating Ni in LaMnO₃ may have a synergistic effect on the conversion of ethanol and its derivatives in oxidative environments. This study attempts to explore the effect of Ni-substituted LaMnO₃ perovskite in ethanol combustion. Physicochemical properties of catalysts were examined, and the relationship between the Ni content and the catalytic behaviors of LaMnO₃-based perovskites was elucidated.

Experimental

Catalyst synthesis

Substituted LaMn_1−yNi_yO₃ (y = 0.1, 0.25, and 0.4) and plain LaMnO₃ and LaNiO₃ were synthesized by the Pechini method.^28,29 Metal nitrates precursors (La(NO₃)₃·6H₂O (Merck, 99%), Mn(NO₃)₂·4H₂O (Merck, 99%), and Ni(NO₃)₂·6H₂O (Merck, 99%)) were used during the synthesis of the respective perovskites. These precursors were dissolved in water in the desired stoichiometric ratios. The obtained solution was then added dropwise to a 1 M citric acid solution, which contained approximately four times as many citric acid ions in relation to the metal ions of the precursors. The mixture was stirred and heated to 80 °C in a water bath. After stirring for an hour, an equivalent molar ratio of ethylene glycol (Alfa Aesar, 99%) using citric acid as the reference was poured into the solution. The resulting solution was then stirred at 90 °C for more than four hours until a gel was formed. The gel was transferred to a furnace, and dried at 80 °C for 24 h. The remaining paste was then calcined in air from 80 to 400 °C at a heating rate of 1 °C min⁻¹, and kept isotherm for two hours. Finally, the precursors were calcined from 400 to 900 °C (3 °C min⁻¹) for four hours.

Characterization

Powder X-ray diffraction patterns were obtained using a Shimadzu Labx XRD-6000 with Cu Kα radiation (0.15418 nm). The scan rate was set to 4° min⁻¹ and the 2θ range was 20–80°. The Scherrer relation was employed to estimate the crystallite size of each sample, based on the strongest peak from the (1 1 0) plane of perovskite.

Fourier transform infrared spectroscopy (FTIR) spectra were recorded in the 400–4000 cm⁻¹ region with a Perkin-Elmer Spectrum 100 Optica using the KBr pellet method. The catalyst powder was diluted to about 1 wt% within KBr. The mixture was then pulverized and pelletized. Prior to the test, the pellet was dehydrated at 130 °C for 30 min in a N₂ stream (25 mL min⁻¹). All the spectra were obtained with an average of 16 scans and 4 cm⁻¹ resolution at room temperature.

X-ray photoelectron spectroscopy (XPS) was performed using a Thermo Scientific K-Alpha system equipped with a 180° hemispherical sector analyzer and an Al monochromator (Al Kα X-ray sources, 1486.6 eV). The diameter of the X-ray spot was 400 μm. The C 1s signal (284.5 eV) of adventitious carbon was used to correct the energy shift.

Single-point BET surface area measurement, temperature-programmed reduction (TPR), and O₂-temperature-programmed desorption (O₂-TPD) were carried out using a Micromeritics apparatus model Autochem II 2920 with a thermal conductivity detector (TCD). Approximately 100 mg of sample was consumed per trial. Prior to each test, the sample was dehydrated at 150 °C under an N₂ stream (30 mL min⁻¹) for 30 min. In the BET test, a 30% N₂/He stream was used for N₂ physisorption at −196 °C. The specific surface area of each sample was estimated from the desorbed N₂ area using the single-point BET equation. In TPR analysis, a 10% H₂/Ar (30 mL min⁻¹) stream was passed through the sample and the temperature was raised from 50 to 900 °C with a ramp of 5 °C min⁻¹. For O₂-TPD, the sample was treated in a He stream (30 mL min⁻¹) from 50 to 1000 °C at 5 °C min⁻¹ followed by an isotherm at 1000 °C for half an hour.

The surface morphologies of as-synthesized and used samples were determined using a field-emission scanning electron microscope (FE-SEM, JEOL JSM-6701F). The acquisition time per location was 10 min. The SEM accelerating voltage was 10 kV. The scanned area was ∼4 μm × 3 μm. Both as-synthesized and used catalysts were examined.

Activity evaluation

Catalytic performances were examined in a continuous fixed-bed system.³⁰ Ethanol was injected into the system by a HPLC pump (Jasco PU-2080) at a feeding rate of 3 × 10⁻³ mL min⁻¹. Oxygen and nitrogen were regulated using mass flow controllers (Brooks 5850E). The molar composition of the feed was C₂H₅OH/O₂/N₂ = 1.7/18.1/80.2 and the gas hourly space velocity (GHSV) was 10 446 h⁻¹. About 10 mg of catalyst with a 40–80 mesh size was used in each trial. To maintain an isothermal condition, the catalyst was diluted with 150 mg SiO₂ (40–80 mesh). SiO₂ was inert in each trial. Internal and external mass transfer limitations were examined prior to the tests. The former was probed by measuring the catalytic performances using two different catalyst particle sizes (20–40 mesh and 40–80 mesh); the latter was tested by comparing two catalyst bed loadings, 10 mg catalyst mixed with 150 mg SiO₂ and 15 mg catalyst mixed with 220 mg SiO₂, with different flow rates but the same contact time (GHSV = 10 446 h⁻¹). Both internal and external mass transfer resistances could be ignored because the differences of ethanol conversions in these trials were trivial (less than 3%).

The system outlet was connected to a gas chromatograph (GC, SRI 8610) equipped with a TCD, a methanizer, and a flame ionization detector (FID). Two packed columns, 5 Å molecular sieve and Porapak Q, were used to separate reactants and products. The detected compounds were ethanol, acetaldehyde, CO₂, CO, CH₄, O₂, and N₂.

Nitrogen also served as the internal standard. All products were measured relative to GC calibration standards. Catalytic results were reported with carbon atom mass balances less than 10% errors. Three to four runs were made for each data point, providing 95% confidence intervals (error bars) for conversion and selectivity. Ethanol conversion was calculated as moles of ethanol reacted divided by moles of ethanol injected. Selectivity of carbon-containing product was defined as 100 × (moles of ethanol converted to product Xi)/(moles of ethanol converted). The data of conversion or selectivity was interpolated with a B-spline function to allow a better comparison.

Results and discussion

Table 1 lists the surface areas of the LaMn_1−yNi_yO₃ catalysts. All catalysts displayed similar surface areas, ranging from 7.6 to 10.5 m² g⁻¹. Fig. 1 shows the XRD patterns of LaMn_1−yNi_yO₃. Those of LaMnO₃ ([JCPDS: 35-1353]) and LaNiO₃ ([JCPDS: 33-0711]) were included for reference. All samples yielded the characteristic peaks of ABO₃-type perovskite phase. The main diffraction peak of NiO ([JCPDS: 89-5881]) at 2θ = 43.3° was observed on LaNiO₃. It reveals the segregation of NiO particles from perovskite solid solution. However, NiO was not identified on Ni-substituted catalysts, possibly because no segregated NiO was present or the NiO clusters were too small to be detected. The crystallite size decreased from 15.1 to 13.0 nm with increasing y values, suggesting the formation of disorder or deformation in LaMnO₃ perovskite by adding Ni.

Table 1 Surface area, mean crystallite size, and XPS-derived surface composition of LaMn_1−yNi_yO₃

LaMn1−yNiyO3	SAA (m^2 g^−1)	dXRD (nm)	Mn/La	Ni/La	Ni/Mn	La/(Mn + Ni)	Ni/(Mn + Ni)	(Mn^4+/Mn^3+)	Ox/Olatt
y = 0	8.7	15.1	0.83	—	0	1.21	0	1.03	0.63
y = 0.1	9.3	14.8	1.21	0.16	0.13	0.73	0.12	1.10	0.66
y = 0.25	10.5	14.5	1.08	0.28	0.26	0.74	0.21	1.21	0.70
y = 0.4	8.2	13.2	0.79	0.37	0.47	0.86	0.32	1.06	0.64
y = 1	7.6	13.0	—	0.67	—	1.49	1	—	2.00

---

Fig. 1 XRD patterns of LaMn_1−yNi_yO₃ perovskites. The diffraction of (1 1 0) plane is zoomed in to show the 2θ angle shift with increasing Ni loading.

A close inspection of the XRD patterns showed that the 2θ angle of each peak increased slightly with increasing Ni amount. This correlation is attributed to the decrease of interplanar distance by Ni substitution: the more Ni incorporated, the more Mn³⁺ (low-spin 0.58 Å, high-spin 0.65 Å) could be replaced by Ni³⁺ (low-spin 0.56 Å, high-spin 0.60 Å) ions, resulting in a decrease of the unit cell volume.³¹ Blasco et al.³² have also observed a similar tendency regarding unit cell volume with increasing Ni content in LaMn_1−yNi_yO₃ perovskites.

Fig. 2 shows (1 1 0), (1 1 1), (2 0 0), and (2 1 1) diffraction shifts as functions of Ni content. The shift gradually increased with Ni content from y = 0 to y = 0.25 then leveled off. Provendier et al.²⁶ and Pecchi et al.³¹ have observed the same trend of XRD peak shift in the LaFe_1−yNi_yO₃ system. The shift has been ascribed to the degree of Ni substitution, and a certain proportion of segregation has been noted in the high substitution region with the maximum shift value. LaMnO₃ has an orthorhombic unit cell. Substituting Mn with smaller Ni ions destabilizes the MnO₆ octahedra and enhances the rhombohedral distortion (due to the Jahn–Teller effect).³³ This structural phase transition of LaMn_1−yNi_yO₃ may lead to the coexistence of multiple phases (e.g., La₂NiMnO₆ and LaNiO₃) at a high Ni-loading.^32,33 Therefore, the complete insertion of Ni in perovskite structure should occur at y < 0.25; NiO_x phases should segregate at y = 0.25 and above.³¹

Fig. 2 High angle shifts of the diffraction peaks as functions of Ni substitution content.

Fig. 3 shows the FTIR spectra of the LaMn_1−yNi_yO₃ catalysts. Besides LaNiO₃, all perovskites displayed an adsorption band at approximately 600 cm⁻¹. This band is related to Mn–O stretching in octahedral site.³⁴ The band of Mn–O stretching was shifted from 610 to 601 cm⁻¹ (red shift) with an increase of Ni loading. The red shift may be explained by the photon response theory³⁵ since Ni is heavier than Mn in the solid solution. Therefore, the downward-shifted wavenumber suggested the presence of Ni–O–Mn bonding in Ni-containing LaMnO₃. Moreover, at y = 0.25 and 0.4, a band at 411 cm⁻¹ and a shoulder at 638 cm⁻¹ were observed. The former is the stretching vibration of Ni–O bond;^36,37 the latter, Ni–OH bending.³⁸ This is consistent with the XRD diffraction shifts: NiO_x could be formed on the surface at y = 0.25 and above. The IR spectrum of LaNiO₃ was not well resolved; a close inspection within 400–600 cm⁻¹ was inserted in Fig. 3. The unclear vibrational modes of LaNiO₃ may be related to its low electrical resistivity (10⁻¹ Ω cm).³⁹ At such a low resistivity, the residence time (10⁻¹² s) of charge carriers at a site can be comparable to the lattice vibrational period (10⁻¹³ s), at which a localized vibrational mode may not be specified.⁴⁰ Two tiny bands at 428 cm⁻¹ and 502 cm⁻¹ were identified, in accord with an earlier work.⁴¹ However, the assignment of these two responses is unclear.

Fig. 3 FTIR spectra of LaMn_1−yNi_yO₃ perovskites.

Fig. 4 displays the XPS spectra of LaMn_1−yNi_yO₃ perovskites. The La 3d spectra included two doublet peaks at 830 eV to 840 eV and 850 eV to 860 eV, corresponding to La 3d_5/2 and La 3d_3/2 signals, respectively. Notably, the responses of La 3d_3/2 and Ni 2p_3/2 overlap each other, and are difficult to be differentiated. A small hump at approximately 870 eV, which is associated with Ni 2p_1/2, was identified in the spectrum of LaNiO₃ (y = 1). Multiple states of Ni (Ni³⁺/Ni²⁺) have been noted in LaNiO₃-based perovskites.⁴² The spin-orbit splitting of the La 3d level for each sample was close to 16.8 eV;⁴² moreover, the La 3d_5/2 and La 3d_3/2 responses were close to the index peaks of pure La₂O₃,⁴³ indicating that La ions are in a trivalent state. Mn 2p_3/2 and Mn 2p_1/2 signals were located at approximately 641.7 eV and 653.3 eV, respectively. Mn²⁺, Mn³⁺, and Mn⁴⁺ could coexist in the Mn 2p_3/2 region;⁴⁴ however, the lack of a satellite peak of Mn²⁺ (648.8 eV)⁴⁵ eliminates the possibility of the presence of Mn²⁺ ions. The responses of Mn 2p_3/2 and Mn 2p_1/2 can be deconvoluted into four peaks. The responses at 641.5 eV and 653.0 eV were assigned to Mn³⁺ cations, while the peaks at 642.9 eV and 654.5 eV were attributed to Mn⁴⁺ ions.⁴² To provide quantitative analysis, the surface atomic ratios (see Table 1) were estimated from the intensities of La 3d_5/2, Ni 2p_1/2, and Mn 2p_3/2 peaks using the atomic sensitivity factors provided by Wagner et al.⁴⁶ The Mn⁴⁺/Mn³⁺ molar ratio was reported in Table 1 as well.

Fig. 4 La 3d, Ni 2p, Mn 2p, and O 1s core level spectra of LaMn_1−yNi_yO₃ perovskites.

The surface enrichment of La can be speculated for plain LaMnO₃ and LaNiO₃ because the Mn/La and Ni/La ratios at y = 0 and 1 were all less than unity. However, this is not the case for the y = 0.1 to 0.4 samples: the La/(Mn + Ni) ratios were all less than unity. This suggests the surface of Ni-containing LaMnO₃ was enriched by Ni and Mn cations. To provide a clear view of the distribution of La, Mn, and Ni on the surface of perovskite, Fig. 5 displays measured surface atomic ratios as functions of nominal surface atomic ratios of Mn/La, Ni/La, Ni/Mn, and Ni/(Mn + Ni). Apparently, the surface became Mn-enriched for Ni-containing LaMnO₃ because their experimental Mn/La values were all greater than nominal ones. The measured Ni/La, Ni/Mn, and Ni/(Mn + Ni) all displayed a descending trend with growing y value. Segregation of NiO_x on perovskite's surface at y = 0.25 and above is a possible explanation. This is because segregated NiO_x particles contained Ni cations in the bulk that cannot be detected by XPS, thereby lowering the surface atomic ratio for Ni with respect to La or Mn. The y = 0.1 sample had all its experimental Ni/La, Ni/Mn, and Ni/(Mn + Ni) ratios greater than nominal ones, indicating the presence of highly dispersed Ni on its surface. The Mn⁴⁺/Mn³⁺ molar ratios were close to 1, decreased followed the order as: y = 0.25 (1.21) > y = 0.1 (1.10) > y = 0.4 (1.06) > y = 0 (1.03). This is in accordance with Zhang et al.,⁴² who have reported similar values of LaMnO₃ at 0.975 and LaMn_0.8Ni_0.2O₃ at 1.216. Apparently, Ni substitution in the B-site position can increase the composition of Mn⁴⁺ in LaMnO₃ perovskites.

Fig. 5 Experimental Mn/La, Ni/La, Ni/Mn, and Ni/(Mn + Ni) atomic ratios versus nominal ones.

Combined signals were obtained from the O 1s state, suggesting the presence of different oxygen species on the surface layer. Lattice oxygen (e.g., La–O–Mn at 529.1 eV),¹³ (O₂)⁻ species or low coordinated oxygen ions located at special site or domains of the surface (531.0 eV),⁴⁷ and hydroxyl/carbonate groups (532.3 eV)³¹ are believed to coexist on the surface of perovskite. Therefore, the O 1s spectra were deconvoluted based on these three responses. As the Ni content increased, the BE of lattice oxygen shifted from 529.2 eV to 528.1 eV. Tabata et al.⁴⁵ observed a similar O 1s chemical shift in LaMn_1−yCu_yO₃ (y = 0, 0.1, 0.2, 0.3, and 0.4). They proposed that more defects and oxygen vacancies could be generated as the substitution of Cu ions at the B-site increased. This effect alters the bond strength of the surface lattice oxygen. Similarly, replacing Ni in LaMnO₃ generates defects and/or segregated NiO particles, which weakening the lattice oxygen bonds in perovskite. Therefore, the BEs of lattice oxygen declined as the Ni content increased. The molar ratio of (O₂)⁻ species or low coordinated oxygen ions to lattice oxygen (O_x/O_latt) was presented in Table 1. The O_x/O_latt ratio showed a same trend as that of (Mn⁴⁺/Mn³⁺), declined as: y = 0.25 (0.70) > y = 0.1 (0.66) > y = 0.4 (0.64) > y = 0 (0.63). Accordingly, the more Mn⁴⁺ on the surface, the more (O₂)⁻ species or low coordinated oxygen ions could exist.⁴² Plain LaNiO₃ displayed higher O_x/O_latt value (2.00) than the other samples because La–Ni oxides are prone to form oxygen-deficient perovskites, which can be described as a superstructure with a formula La_nNi_nO_3n−1.⁴⁸ Hence, less lattice oxygens were available on LaNiO₃ than other perovskites.

The C 1s spectra were presented in Fig. S1 (ESI†). The C 1s signals of all perovskites showed responses of adventitious carbon (284.5 eV),⁴⁹ lanthanum carbonate (∼286.2 eV),⁵⁰ and carbonyl group (∼288.6 eV).^49,51 This is in agreement with the literature that lanthanum-based perovskites are often contaminated by carbonate species.

Fig. 6 presents the TPR profiles of tested perovskites and Table S1 (ESI†) lists the amounts of H₂ consumed by them. The TPR patterns of pure LaMnO₃ and LaNiO₃ included multiple peaks. The reduction of LaMnO₃ is complicated: Mn⁴⁺ and Mn³⁺ can coexist at the outset of TPR, and a full reduction of Mn ions to metallic Mn is not possible.^52,53 Therefore, the low-temperature responses (<400 °C) are attributed to the reduction of Mn⁴⁺ to Mn³⁺; the high-temperature signals (>400 °C) are referred to the reduction of Mn³⁺ to Mn²⁺. Some reduction of Mn³⁺ to Mn²⁺ at low temperatures cannot be ruled out.⁵⁴ The reduction of LaNiO₃ was similar to that in an earlier study:⁵⁵ the reduction of Ni³⁺ to Ni²⁺ (i.e., La₂Ni₂O₅) at 347 °C was followed by that of Ni²⁺ to metallic Ni at 469 °C. The XRD patterns of post-TPR LaMnO₃ and LaNiO₃ (see Fig. S2 of ESI†) further support the aforementioned assertions: the index peaks of La₂O₃ and MnO were with the reduction of LaMnO₃; those of La₂O₃ and Ni were associated with the reduction of LaNiO₃.

Fig. 6 TPR profiles of LaMn_1−yNi_yO₃ perovskites.

The first reduction of LaMn_1−yNi_yO₃ occurred in the range of 250 °C to 500 °C, in which the LaMnO₃ and LaNiO₃ phases were reduced. Hence, Mn and Ni ions were reduced concurrently in this range of temperatures. The second reduction of LaMn_1−yNi_yO₃ may be related to Mn and Ni ions since the amount of H₂ consumed in the high temperature range does not increase in proportional to the extent of Ni replacement (see Table S1†).

To provide a comparison, Fig. S3 (ESI†) presents the TPR results for physical mixtures of LaNiO₃ and LaMnO₃ with LaNiO₃-to-LaMnO₃ ratios of 0.1, 0.25, and 0.4. In Fig. 6, Ni-incorporated LaMnO₃ exhibited two TPR responses at 200 °C to 500 °C and 650 °C to 850 °C. The high-temperature responses of LaMn_1−yNi_yO₃ shifted to higher temperatures than those of plain LaMnO₃ and LaNiO₃, especially for the y = 0.1 sample. Tailing of reduction response at 800 °C and above could be observed for the y = 0.1 and 0.25 samples. This suggests an existence of Ni–O–Mn interaction, which retards the reduction process. The TPR profiles of physically mixed LaNiO₃ and LaMnO₃ further support the above-mentioned hypothesis: three distinct peaks at 331, 466, and 731 °C appear to be the merged responses of LaNiO₃ and LaMnO₃. Notably, LaMn_0.9N_0.1O₃ exhibited the highest reduction signal at 783 °C, indicating that the interaction within the solid solution of LaMn_0.9N_0.1O₃ is the strongest among all samples. Conversely, the weakest interaction between Ni and Mn was found in LaMn_0.6N_0.4O₃ since it had the lowest high-temperature reduction response at 736 °C.

Fig. 7 shows the O₂-TPD profiles. For clarity, the signal of LaNiO₃ was multiplied by a factor of 0.1. O₂-TPD can be used to estimate the lability of oxygen in tested perovskites. The outset of desorption signal increased in the order of y = 0 (82 °C) < y = 0.1 (114 °C) < y = 0.25 (132 °C) < y = 0.4 (211 °C) < y = 1 (677 °C), suggesting that more substituted Ni at B-site resulted in lower oxygen lability in LaMn_1−yNi_yO₃. Desorbed oxygen is frequently categorized as α- and β-oxygen, using 500 °C as the demarcation. α-oxygen is adsorbed and near-surface oxygen, while β-oxygen is lattice oxygen in the perovskite framework.⁵⁶ Negligible α-oxygen desorption was detected in LaNiO₃. Table 2 summarizes the amounts of desorbed oxygen and the temperatures of peak desorption of tested perovskites. The desorption temperatures of α- and β-oxygen increased with Ni content for LaMn_1−yNi_yO₃. Plain LaNiO₃ exhibited a distinct peak at 948 °C. Ni-containing LaMnO₃ perovskites had higher α-oxygen-to-β-oxygen ratios than pure LaMnO₃ and LaNiO₃. The α-oxygen-to-β-oxygen ratio followed the sequence of y = 0.25 (0.47) > y = 0.1 (0.39) > y = 0.4 (0.32) > y = 0 (0.30), in consistent with the trend of the O_x/O_latt ratio observed by XPS. Among all catalysts, LaMn_0.9Ni_0.1O₃ released the most α-oxygen (72.2 μmol/g cat.); LaNiO₃ contained the highest β-oxygen (695.9 μmol/g cat.).

Fig. 7 O₂-TPD profiles of LaMn_1−yNi_yO₃ perovskites.

Table 2 Oxygen desorption characterization of LaMn_1−yNi_yO₃ catalysts

Catalyst		O2-TPD
		α-O2^a	β-O2^b	α-O2/β-O2
a Maximum temperature of oxygen desorption peak below 500 °C.b Maximum temperature of oxygen desorption peak above 500 °C.
y = 0	T (°C) μmol/g cat.	131	686
		69.8 (23.0%)	233.2 (77.0%)	0.30
y = 0.1	T (°C) μmol/g cat.	136	697
		72.2 (28.1%)	184.5 (71.9%)	0.39
y = 0.25	T (°C) μmol/g cat.	205	928
		68.1 (32.1%)	144.0 (67.9%)	0.47
y = 0.4	T (°C) μmol/g cat.	255	1000
		24.8 (24.1%)	78.0 (75.9%)	0.32
y = 1	T (°C) μmol/g cat.	—	948
		—	695.9 (100%)	0

---

Ni substitution negatively affects oxygen lability/oxygen availability in LaMnO₃-based perovskites. This can be explained by estimating the oxygen content from the charge balance. According to the XPS results, trivalent and tetravalent Mn ions could coexist in LaMnO₃; therefore, its chemical formula is LaMn³⁺_aMn⁴⁺_1−aO_3+δ (δ = (1 − a)/2). Divalent and trivalent Ni ions were present in LaNiO₃, yielding the formula LaNi²⁺_bNi³⁺_1−bO_3+λ (λ = −(1 − b)/2). Since δ is positive and λ is negative, LaMnO₃ should host excess oxygen while oxygen vacancies are formed in LaNiO₃. Substituting Ni in LaMnO₃ should satisfy the charge balance condition, yielding an oxygen content between δ and λ. The only exception is LaNiO₃, which released the most oxygen in O₂-TPD. This finding is consistent with the XPS results concerning the inhomogeneity of the surface and unstable Ni³⁺ in perovskite, which yields weak lattice oxygen bonding in LaNiO₃. Still, doping Ni has a positive effect on α-oxygen content, which is optimal at y = 0.1.

Fig. 8 plots ethanol conversions of LaMn_1−yNi_yO₃ as functions of temperature. Each catalyst yields an S-shaped curve besides LaNiO₃, which displays an abrupt increase of ethanol conversion at 250 °C. At temperatures below 200 °C, all catalysts converted less than 20% ethanol, except for LaMn_0.9Ni_0.1O₃, which achieved 21% conversion at 200 °C. The T₅₀ increased in the order of LaMn_0.9Ni_0.1O₃ (226 °C) < LaMn_0.75Ni_0.25O₃ (243 °C) < LaMnO₃ (249 °C) < LaMn_0.6Ni_0.4O₃ (255 °C) < LaNiO₃ (266 °C); the T₉₅, LaMn_0.9Ni_0.1O₃ (252 °C) < LaMn_0.75Ni_0.25O₃ (269 °C) < LaMnO₃ (288 °C) < LaMn_0.6Ni_0.4O₃ (309 °C) < LaNiO₃ (314 °C). This indicates that LaMn_0.9Ni_0.1O₃ is the most active catalyst in ethanol combustion while LaNiO₃ is the least active. Ling et al.⁵⁷ have observed a sudden increase of conversion at 250 °C for LaNiO₃ in ethanol oxidation. By a conductivity test, surface reaction of chemisorbed ethanol and oxygen species was discovered to be promoted at 250 °C. This leads to an increase of LaNiO₃ conductivity due to oxygen consumption, and ethanol conversion can be greatly improved on partially reduced surface. They also pointed out that the maximum conductivity change of LaNiO₃ occurs in the range of 250 to 350 °C, in accordance with our conversion trend.

Fig. 8 Conversion of ethanol as a function of temperature over LaMn_1−yNi_yO₃ perovskites.

Fig. 9 shows acetaldehyde and carbon dioxide selectivities as functions of temperature. Negligible amounts of CO and CH₄, both less than 1%, were detected in each catalyst, and are ignored for clarity. At temperatures below 225 °C, the selectivity of acetaldehyde exceeded 80% while that of CO₂ was less than 20%. With increasing temperature, the selectivity of acetaldehyde declined and that of CO₂ increased. At 235 °C, LaMn_0.9Ni_0.1O₃ generated about 38% acetaldehyde and 62% CO₂. The other catalysts produced more than 80% acetaldehyde and less than 20% CO₂ at this temperature. At 250 °C, LaMn_0.75Ni_0.25O₃ yielded equal amounts of acetaldehyde (30%) and CO₂ (70%) to those of LaMn_0.9Ni_0.1O₃. The remaining catalysts at 260 °C and above yielded less than 40% acetaldehyde and over 60% CO₂. LaNiO₃ seems to be the least effective in acetaldehyde abatement because more than 40% acetaldehyde selectivity was detected at 275 °C.

Fig. 9 Acetaldehyde and carbon dioxide selectivities as functions of temperature over LaMn_1−yNi_yO₃ perovskites.

The oxidation activity and chemistry of alkane and alcohol are known to be strongly associated with the transport and storage of oxygen.^23,58 The above catalytic outcomes indicate that adsorbed and near-surface oxygen (α-oxygen) are more effective than lattice oxygen (β-oxygen) in ethanol oxidation. Hence, the α-oxygen-lean LaNiO₃ was the least active catalyst whereas the activities of α-oxygen-rich LaMn_0.9Ni_0.1O₃ and LaMn_0.75Ni_0.25O₃ were enhanced. The high onset temperature of O₂-TPD and low concentration of α-oxygen of LaMn_0.6Ni_0.4O₃ reflected low reactivity, implying the critical role of α-oxygen in ethanol oxidation. α-oxygen is known to act as an activated oxygen species, which initiates the oxidation of chemisorbed intermediates into carbon oxides and water.⁵⁹ The role of α-oxygen in ethanol oxidation is as proposed elsewhere.^17,21 To provide a better comparison, Fig. S4 (ESI†) displays the variations of XPS-derived Ni loading, amounts of α- and β-oxygen, and T₅₀ and T₉₅ of tested LaMn_1−yNi_yO₃.

LaMn_0.9Ni_0.1O₃ possesses the highest activity among all catalysts, even greater than that of LaMn_0.75Ni_0.25O₃, which has the highest O_x/O_latt ratio. This implies that besides α-oxygen effect, other factors should be considered for catalyst's efficacy in ethanol oxidation. Presumably, at a lower Ni loading, Ni can be distributed more evenly in perovskite, generating more Ni–O–Mn linkages on the surface. Increasing the Ni loading in LaMnO₃ produces segregated NiO_x clusters on the surface of perovskite, reducing the amounts of Ni–O–Mn bonds. The bridging oxygen of Ni–O–Mn should possess an unique interaction, which is effective in promoting ethanol oxidation and intermediate conversion to carbon dioxide. Similar bridging sites have been noted on different perovskites (e.g., Ni²⁺–O–Ni³⁺ species of La_2−xSr_xNiO₄,⁶⁰ Mn³⁺–O–Mn⁴⁺ bonds of LaMn_1−yCu_yO₃,⁵² and Fe³⁺–O–Fe⁵⁺ pairs of SrFeO₃⁶¹). These sites have been proposed to be effective in promoting charge transfer and oxygen mobility of the catalysts in N₂O decomposition⁶⁰ and CH₄ combustion.^52,61

Fig. 10 displays the proposed reaction pathway of ethanol conversion over LaMn_1−yNi_yO₃. The strongly basic O–La–O sites are known to be active in ethanol activation;²¹ however, the pathway for the possibility of ethanol chemisorption on La is excluded when considering only the effect of B-site substitution on LaMnO₃. At the outset of the reaction, ethanol chemisorbs on the surface, and forms an ethoxide at the metal site (Mn or Ni) and a hydroxyl at the neighboring lattice oxygen site. Subsequent dehydration yields an acetaldehyde-derived intermediate and an oxygen vacancy. At low temperatures, the reduced surface is re-oxidized, and acetaldehyde is desorbed. At high temperatures, the acetaldehyde intermediate may associate with the surface bridging oxygen (Mn–O–Mn, Ni–O–Mn, or Ni–O–Ni) to form an acetyl, which is then converted to carbon oxides, water, and methane.⁶² Ethanol dehydrogenation to acetaldehyde was excluded in Fig. 10 because negligible conversion (less than 5%) of ethanol was identified without feeding O₂ in our system.

Fig. 10 Proposed reaction pathway of ethanol oxidation.

The catalytic results show that acetaldehyde selectivities are nearly identical (above 80%) at temperatures below 225 °C. Acetaldehyde is known to be derived through a redox cycle in ethanol oxidation.^21,63 Therefore, the redox properties for LaMnO₃, LaNiO₃, and LaMn_1−yNi_yO₃ perovskites should be similar. Carbon dioxide selectivities of these catalysts increase differently with increasing temperature, implying changing oxidation power in converting the chemisorbed intermediate (herein acetyl species). In the proposed pathway, acetyl oxidation proceeds on the bridging oxygen in Mn–O–Mn, Ni–O–Mn, or Ni–O–Ni. The amount of CO₂ increased abruptly over LaMn_0.9Ni_0.1O₃ at the lowest temperature (235 °C) of any of the tested perovskites, suggesting that its bridging oxygen was the most strongly oxidizing. This is attributed to the unique Ni–O–Mn interaction of LaMn_0.9Ni_0.1O₃, as revealed by the TPR test. To ascertain whether the Ni–O–Mn interaction in perovskite is uniquely responsible for its reactivity and acetaldehyde oxidation, catalytic tests of physically mixed LaNiO₃ and LaMnO₃ in LaNiO₃-to-LaMnO₃ ratios of 0.1, 0.25, and 0.4 were conducted (see Fig. S5 and S6 of ESI†). The T₅₀ values of the physically mixed samples were in the range of 245 °C to 254 °C – about 20 °C higher than that of LaMn_0.9Ni_0.1O₃ but close to those of LaMn_0.75Ni_0.25O₃ and LaMn_0.6Ni_0.4O₃. However, the T₉₅ values of the physically mixed samples exceeded their respective counterparts, ranging from 300 to 321 °C. For all samples, below 250 °C, acetaldehyde selectivity exceeded 80% and CO₂ selectivity was less than 20%. These results demonstrate that the Ni–O–Mn interaction in Ni-doped LaMnO₃ promotes the conversion of ethanol and the oxidation of acetaldehyde to CO₂.

Fig. 11 illustrates the evolution of time on-stream ethanol conversion over a 36 h period at 300 °C. Initially, ethanol was fully converted over LaMnO₃, LaMn_0.9Ni_0.1O₃, and LaMn_0.75Ni_0.25O₃, while 93.7% and 91.9% conversions were achieved by LaMn_0.6Ni_0.4O₃ and LaNiO₃, respectively. All catalysts suffered of ageing but at different extents. Ethanol conversion decreased from 100% to 96.2% for LaMnO₃ (a decrease of ∼3.8%), 100% to 98.0% for LaMn_0.9Ni_0.1O₃ (a decrease of ∼2.0%), 100% to 97.5% for LaMn_0.75Ni_0.25O₃ (a decrease of ∼2.5%), 93.7% to 89.7% for LaMn_0.6Ni_0.4O₃ (a decrease of ∼4.4%), and 91.9% to 78.6% for LaNiO₃ (a decrease of ∼14.5%). Moreover, LaMn_0.9Ni_0.1O₃ remained its activity for more than 25 h, whereas deactivation could be observed within 20 h for other catalysts. Apparently, LaMn_0.9Ni_0.1O₃ has the highest thermal stability while LaNiO₃ has the least among tested catalysts. Since coking is less likely to occur under oxygen-rich environments, agglomeration of catalyst particle and/or deconstruction of perovskite structure should be responsible for catalyst deactivation.⁶⁴ The SEM micrographs of freshly prepared and used LaMn_1−yNi_yO₃ after on-stream testing (see Fig. S7 of ESI†) further supported this claim: some welding and particle agglomeration could be identified on used samples; however, no carbon deposition (e.g., carbon nanotubes) was specified. The high thermal stability of LaMn_0.9Ni_0.1O₃ may, again, ascribe to the unique interaction between Ni and Mn, thereby suppressing the destruction of its perovskite framework.

Fig. 11 Conversion of ethanol as a function of time on-stream over LaMn_1−yNi_yO₃ perovskites at 300 °C.

Conclusions

Doping Ni into LaMnO₃ perovskite can substantially alter its physicochemical properties, and thereby changing its catalytic performances. A limited amount of Ni (less than 25%) can be dispersed in LaMnO₃ framework; adding more Ni forms segregated NiO_x clusters. Incorporating Ni generates Ni–O–Mn sites, which are associated with a unique interaction, is assumed to play a key role in enhancing catalytic activity, acetaldehyde abatement, and stability in ethanol oxidation. A plausible mechanism based on the surface bridging oxygen site was thereby proposed. Future study should seek to clarify the nature of the bridging oxygen site, possibly through computational chemistry such as density functional theory,⁶⁵ to provide a molecular-level understanding of perovskite catalyst design.

Acknowledgements

The authors acknowledge the National Science Council (Taiwan) for the financial support under Contract Number 102-2221-E-155-060-MY2. The authors also thank Oscar Kerkenaar for the English language editing.

Notes and references

1. A. Hira, Energy Policy, 2011, 39, 6925–6935.
2. A. C. Hansen, Q. Zhang and P. W. L. Lyne, Bioresour. Technol., 2005, 96, 277–285.
3. M. D. Kass, J. F. Thomas, J. M. Storey, N. Domingo, J. Wade and G. Kenreck, SAE Technical Paper 2001-01-2018 (SP-1632), 2001.
4. US EPA, http://www.epa.gov/chemfact/s_acetal.txt.
5. M. A. Peña and J. L. G. Fierro, Chem. Rev., 2001, 101, 1981–2018.
6. J. Zhu and A. Thomas, Appl. Catal., B, 2009, 92, 225–233.
7. C. H. Kim, G. Qi, K. Dahlberg and W. Li, Science, 2010, 327, 1624–1627.
8. Y. Nishihata, J. Mizuki, T. Akao, H. Tanaka, M. Uenishi, M. Kimura, T. Okamoto and N. Hamada, Nature, 2002, 418, 164–167.
9. H. Tanaka, Catal. Surv. Asia, 2005, 9, 63–74.
10. H. Tanaka, M. Uenishi, M. Taniguchi, I. Tan, K. Narita, M. Kimura, K. Kaneko, Y. Nishihata and J. Mizuki, Catal. Today, 2006, 117, 321–328.
11. H. Tanaka, M. Taniguchi, M. Uenishi, N. Kajita, I. Tan, Y. Nishihata, J. Mizuki, K. Narita, M. Kimura and K. Kaneko, Angew. Chem., Int. Ed., 2006, 45, 5998–6002.
12. C.-L. Li, C.-L. Wang and Y.-C. Lin, Catal. Today, 2011, 174, 135–140.
13. C.-L. Li and Y.-C. Lin, Appl. Catal., B, 2011, 107, 284–293.
14. C.-L. Li, B.-S. Jiang, W.-L. Fanchiang and Y.-C. Lin, Catal. Commun., 2011, 16, 165–169.
15. H. Obayashi, Y. Sakurai and T. Gejo, J. Solid State Chem., 1976, 17, 299–303.
16. T. Shimizu, Appl. Catal., 1986, 28, 81–88.
17. W. Wang, H. Zhang, G. Lin and Z. Xiong, Appl. Catal., B, 2000, 24, 219–232.
18. N. A. Merino, B. P. Barbero, P. Ruiz and L. E. Cadus, J. Catal., 2006, 240, 245–257.
19. B. P. Barbero, J. A. Gamboa and L. E. Cadus, Appl. Catal., B, 2006, 65, 21–30.
20. B. Bialobok, J. Trawczynski, W. Mista and M. Zawadzki, Appl. Catal., B, 2007, 72, 395–403.
21. H. Najjar and H. Batis, Appl. Catal., A, 2010, 383, 192–201.
22. M. Zawadzki and J. Trawczynski, Catal. Today, 2011, 176, 449–452.
23. T. Nitadori, S. Kurihara and M. Misono, J. Catal., 1986, 98, 221–228.
24. I. Rossetti, C. Biffi and L. Forni, Chem. Eng. J., 2010, 162, 768–775.
25. T. Hayakawa, A. Andersen, M. Shimizu, K. Suzuki and K. Takehira, Catal. Lett., 1993, 22, 307–317.
26. H. Provendier, C. Petit, C. Estournes, S. Libs and A. Kiennemann, Appl. Catal., A, 1999, 180, 163–173.
27. F. B. Noronha, M. C. Durao, M. S. Batista and L. G. Appel, Catal. Today, 2003, 85, 13–21.
28. M. P. Pechini, US Pat., 3330697, 1967.
29. P. Courty, H. Ajot, C. Marcilly and B. Delmon, Powder Technol., 1973, 7, 21–38.
30. C.-L. Li and Y.-C. Lin, Catal. Lett., 2010, 140, 69–76.
31. G. Pecchi, P. Reyes, R. Zamora, L. E. Cadus and J. L. G. Fierro, J. Solid State Chem., 2008, 181, 905–912.
32. J. Blasco, M. C. Sanchez, J. Perez-Cacho, J. Garcia, G. Subias and J. Campo, J. Phys. Chem. Solids, 2002, 63, 781–792.
33. N. Van Minh, S.-J. Kim and I.-S. Yang, Phys. B, 2003, 327, 208–210.
34. F. Gao, R. A. Lewis, X. L. Wang and S. X. Dou, J. Alloys Compd., 2002, 347, 314–318.
35. D. W. Taylor, Photon response theory and the infrared and Raman experiments, Elsvier Science Publishers, Amsterdam, 1988.
36. G. Zhou, D.-W. Wang, L.-C. Yin, N. Li, F. Li and H.-M. Cheng, ACS Nano, 2012, 6, 3214–3223.
37. S. Mustansar Abbas, S. Tajammul Hussain, S. Ali, N. Ahmad, N. Ali, S. Abbas and Z. Ali, J. Solid State Chem., 2013, 202, 43–50.
38. N. Bayal and P. Jeevanandam, J. Alloys Compd., 2012, 537, 232–241.
39. T. Vaz and A. V. Salker, Mater. Sci. Eng., B, 2007, 143, 81–84.
40. P. Ganguly and N. Y. Vasanthacharya, J. Solid State Chem., 1986, 61, 164–170.
41. M. Panneerselvam and K. J. Rao, J. Mater. Chem., 2003, 13, 596–601.
42. C. Zhang, C. Wang, W. Zhan, Y. Guo, Y. Guo, G. Lu, A. Baylet and A. Giroir-Fendler, Appl. Catal., B, 2013, 129, 509–516.
43. M. O'Connell, A. K. Norman, C. F. Huttermann and M. A. Morris, Catal. Today, 1999, 47, 123–132.
44. M. Kantcheva, M. U. Kucukkal and S. Suzer, J. Mol. Struct., 1999, 482–483, 19–22.
45. K. Tabata, Y. Hirano and E. Suzuki, Appl. Catal., A, 1998, 170, 245–254.
46. C. D. Wagner, L. E. Davis, M. V. Zeller, J. A. Taylor, R. H. Raymond and L. H. Gale, Surf. Interface Anal., 1981, 3, 211–225.
47. S. Rousseau, S. Loridant, P. Delichere, A. Boreave, J. P. Deloume and P. Vernoux, Appl. Catal., B, 2009, 88, 438–447.
48. M. T. Anderson, J. T. Vaughey and K. R. Poeppelmeier, Chem. Mater., 1993, 5, 151–165.
49. M. Cinar, L. M. Castanier and A. R. Kovscek, Energy Fuels, 2011, 25, 4438–4451.
50. C. K. Jorgensen and H. Berthon, Chem. Phys. Lett., 1972, 13, 186–189.
51. H. Taguchi, A. Sugita, M. Nagao and K. Tabata, J. Solid State Chem., 1995, 119, 164–168.
52. L. Lisi, G. Bagnasco, P. Ciambelli, S. De Rossi, P. Porta, G. Russo and M. Turco, J. Solid State Chem., 1999, 146, 176–183.
53. S. Royer, H. Alamdari, D. Duprez and S. Kaliaguine, Appl. Catal., B, 2005, 58, 273–288.
54. H. Ziaei-Azad, A. Khodadadi, P. Esmaeilnejad-Ahranjani and Y. Mortazavi, Appl. Catal., B, 2011, 102, 62–70.
55. H. Chen, H. Yu, F. Peng, G. Yang, H. Wang, J. Yang and Y. Tang, Chem. Eng. J., 2010, 160, 333–339.
56. J. N. Kuhn and U. S. Ozkan, J. Catal., 2008, 253, 200–211.
57. T.-R. Ling, Z.-B. Chen and M.-D. Lee, Appl. Catal., A, 1996, 136, 191.
58. J. Choisnet, N. Abadzhieva, P. Stefanov, D. Klissurski, J. M. Bassat, V. Rives and L. Minchev, J. Chem. Soc., Faraday Trans., 1994, 90, 1987–1991.
59. B. Levasseur and S. Kaliaguine, Appl. Catal., A, 2008, 343, 29–38.
60. A. K. Ladavos and P. J. Pomonis, J. Chem. Soc., Faraday Trans., 1991, 87, 3291–3297.
61. V. C. Belessi, A. K. Ladavos and P. J. Pomonis, Appl. Catal., B, 2001, 31, 183–194.
62. K. Rintramee, K. Fottinger, G. Rupprechter and J. Wittayakun, Appl. Catal., B, 2012, 115–116, 225–235.
63. B.-S. Jiang, R. Chang and Y.-C. Lin, Ind. Eng. Chem. Res., 2013, 52, 37–42.
64. H. Najjar, J.-F. Lamonier, O. Mentre, J.-M. Giraudon and H. Batis, Catal. Sci. Technol., 2013, 3, 1002–1016.
65. M. Pishahang, C. E. Mohn, S. Stolen and E. Bakken, RSC Adv., 2012, 2, 10667–10672.

---

Footnote

† Electronic supplementary information (ESI) available: The reduction properties, C 1s core level spectra, and post-TPR XRD patterns of LaMn_1−yNi_yO₃, TPR profiles of physically mixed LaMnO₃ and LaNiO₃ particles, ethanol conversion and product selectivity as functions of temperature of physically mixed LaMnO₃ and LaNiO₃ particles, the summary of XPS-derived Ni loading, amounts of α- and β-oxygen, and T₅₀ and T₉₅ of tested LaMn_1−yNi_yO₃, the SEM images of as-synthesized and used samples. See DOI: 10.1039/c3ra46323k

---