Kinetics of carbon monoxide oxidation with Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, Co) catalysts

Abstract

Base metal substituted Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, Co) catalysts were synthesized by the solution combustion method and characterized by XRD, XPS, TEM and BET surface area analysis. The catalytic activities of these materials were investigated by performing CO oxidation. The rates and the apparent activation energies of the reaction for CO oxidation were determined for each catalyst. All the substituted catalysts showed high rates and lower activation energies for the oxidation of CO as compared to unsubstituted SnO₂. The rate was found to be much higher over copper substituted SnO₂ as compared to other studied catalysts. 100% CO conversion was obtained below 225 °C over this catalyst. A bifunctional reaction mechanism was developed that accounts for CO adsorption on base metal and support ions and O₂ dissociation on the oxide ion vacancy. The kinetic parameters were determined by fitting the model to the experimental data. The high rates of the CO oxidation reactions at low temperatures were rationalized by the high dissociative chemisorption of adsorbed O₂ over these catalysts.

---

1. Introduction

Carbon monoxide, emitted from many industrial processes and transportation, is considered as an important air pollutant. Catalytic oxidation is an efficient way to convert CO to CO₂ and has been studied using various simple and mixed oxide catalysts.^1–6 Precious metals such as Pt, Pd, Au supported on various metal oxides or combination of metal oxides are well known oxidation catalysts.^7–9 The supports used are mainly Al₂O₃, CeO₂, ZrO₂, TiO₂ and mixed oxides with high dispersion of Pt, Pd and Au. However, the high cost of precious metals and their sensitivity to sulfur poisoning have motivated the research into base metal substituted catalysts. It has been well documented that high activity and resistance to poisoning can be obtained by doping Cu, Co, Ni oxides.^10,11 The reaction rate and mechanism are highly influenced by the identity of supports.^12,13 For instance, non-reducible Al₂O₃ shows lower activity than reducible oxides such as CeO₂. Reducible supports such as CeO₂, TiO₂, Fe₂O₃ are often used for CO oxidation studies and have shown higher activity at low temperatures as compared to their non-reducible oxide counterparts.¹⁴ The difference in the reducibility of supports was found to be responsible for the stabilization of various intermediate species and different reaction pathways.^15,16

SnO₂ has been extensively used as an oxidation catalyst as it can reversibly undergo the Sn⁴⁺ ↔ Sn²⁺ reaction at relatively lower temperatures (∼200 °C). It is one of the most widely used semiconductor oxides for the manufacturing of sensors.¹⁷ A significant lowering in the activation energy for CO oxidation reaction over SnO₂ as compared to CeO₂ is indicative of the ease of availability of lattice oxygen in SnO₂. It has been reported¹⁸ that the apparent activation energy for tin oxide and cerium oxide is 4.7 kcal mol⁻¹ and 13.2 kcal mol⁻¹, respectively. This shows that tin oxide is an interesting material having higher activity as compared to ceria and can catalyze the oxidation reaction at relatively low temperatures. Hence, it is of interest to study the catalytic activity of SnO₂ towards CO oxidation. Attempts have been made to improve the low temperature reducibility and enhance the oxygen storage capacity of SnO₂ by substituting a part of Sn⁴⁺ by other suitable cations. Pt, Pd, Au impregnated SnO₂ were reported as active catalysts for CO or CH₄ oxidation.^19,20 Grass and Lintz¹⁵ suggested a probable pathway of CO oxidation over the Pt/SnO₂ catalyst, which assumed adsorption of oxygen on Pt followed by its migration to the reaction sites situated at the periphery between oxide and Pt metal. Based on the comparative evaluation of the catalytic activity of Fe₂O₃, Mn₂O₃ and SnO₂ and their mixed oxides, Kulshreshtha et al. showed that mixed oxides (Fe₂O₃ + SnO₂ and Mn₂O₃ + SnO₂) show strong synergistic effects for CO oxidation.²¹ Depending on the support, different reaction mechanisms have been proposed for Pt, Rh and Pd.^12,22 The reaction mechanism over a noble metal is well understood and the oxidation of CO over Pt is assumed to proceed via a reaction between adsorbed oxygen and adsorbed CO species in a Langmuir–Hinshelwood type of model. However, it is still unclear if CO reacts in a similar way over base metal catalysts in the adsorbed state or in the gas phase in an Eley–Rideal type of model.

Although catalysts for the oxidation of carbon monoxide have been widely investigated over supported base metals, there are very limited data on the kinetics of catalyticCO oxidation over these catalysts. The mechanistic understanding of correlation between catalyst properties and metal–support interaction is still missing. The oxidation of carbon monoxide over SnO₂ catalysts was postulated to follow the mechanism that involves the adsorption of carbon monoxide, desorption of carbon dioxide and oxygen regeneration of the catalyst. The kinetics was represented by the Langmuir type rate equation. Sedmak et al. studied the kinetics of CO oxidation in excess hydrogen over a nanostructured Cu_0.1Ce_0.9O_2−δcatalyst under preferential oxidation conditions and showed that the kinetics of the reaction was found to follow the redox mechanism represented by the Mars and van Krevelen type of rate equation.²³ Dekker et al. proposed a model that involves the oxidation of reduced sites by O₂ and/or N₂O, followed by a reaction with CO, yielding a surface intermediate that releases CO.²⁴ In a recent study, Kocemba and Rynkowski²⁵ explained how oxygen adsorption affects the catalytic activity of tin dioxide towards CO oxidation.

In order to lower the cost and improve activity of catalysts, novel non-stoichiometric Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, Co) catalysts were synthesized by the solution combustion method using a tin oxalate precursor. The activities of these catalysts at various temperatures were determined and compared. A detailed kinetic model was proposed for describing the kinetics of the CO oxidation. Spectroscopic insights were used in determining the possible surface processes and rate mechanisms were proposed based on elementary pathways. The kinetic parameters were obtained from fitting the experimental data.

2. Experimental section

2.1. Synthesis and characterization

Transition metal ions (M = Cu, Fe, Mn, and Co) substituted SnO₂ catalysts were prepared by a novel, single step solution combustion method using a tin oxalate precursor. The combustion mixture for the preparation of 5 atom% Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, and Co) catalysts contained stoichiometric amounts of tin oxalate (SnC₂O₄) and respective metal nitrates like Mn(NO₃)₂·4H₂O, Fe(NO₃)₃·9H₂O, Co(NO₃)₂·6H₂O and Cu(NO₃)₂·3H₂O and glycine (CH₂NH₂COOH) as fuel (all from S.D. Fine Chem, India). SnC₂O₄ was prepared from SnCl₂ and oxalic acid. In a typical procedure, 50 ml of 0.5 M solution of SnCl₂·2H₂O was slowly added to 50 ml of 0.5 M solution of oxalic acid at 70 °C under stirring, giving SnC₂O₄ crystals. The precipitate was filtered off, washed with distilled water several times until chloride ions were completely removed, and finally dried at 75 °C for 15 h. The formation of pure crystalline SnC₂O₄ was confirmed by X-ray diffraction (XRD). SnC₂O₄ was dissolved in nitric acid and Sn²⁺ was oxidized to Sn⁴⁺ (as confirmed by the HgCl₂ test).²⁶

In the solution combustion method, SnC₂O₄, Cu(NO₃)₂·3H₂O and CH₂NH₂COOH were taken in the following molar ratio 0.95 : 0.05 : 2.166 to make 5 atom% Sn_0.95Cu_0.05O_2−δ. The solution was introduced into a muffle furnace maintained at 400 °C. The solution boiled with frothing and foaming with concomitant dehydration. At the point of its complete dehydration, the redox mixture ignited yielding a voluminous finely dispersed solid product.

All the compounds were characterized by X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), transmission electron microscopy (TEM) and BET surface analysis. The XRD patterns of the compounds were obtained from a Philips X'pert diffractometer with Cu-Kα radiation in a 6 h range of 10–80 with an interval of 0.02. TEM analysis was performed on a JEOL JEM-200CX, operating at 200 kV. The samples were dispersed in ethanol and treated with ultrasound for 5 min, and then deposited on a copper grid coated with a preformed holey carbon film. BET surface area was determined by a nitrogen adsorption–desorption method at liquid nitrogen temperature using a Quantachrome NOVA 1000 surface area analyzer. X-Ray photoelectron spectra of transition metal substituted oxides were recorded on a Thermo Fisher Scientific Multilab 2000 (England) instrument with Al K radiation (1486.6 eV). The binding energies reported are with reference to graphite at 284.5 eV. Oxide samples were ground with 30 wt% graphite powder, made into thin pellets at room temperature, and XPS spectra were recorded. There was no charging of oxide samples. The composition of the copper substituted catalyst was confirmed by inductively coupled plasma-optical emission spectroscopy (ICP-OES, Thermo-iCAP 6000 series). The compounds were dissolved in acidic solution and heated in a water bath at 80 °C for 24 h. The clear solutions were then diluted with deionized water and used for the analysis.

2.2. CO oxidation experiments

CO oxidation experiments were carried out in an 8.5 mm ID and 35 cm long quartz tube reactor. The powdered catalyst of different weights was diluted with glass beads (180 μm) to make a bed of length 1.1 cm. The catalysts were packed between two glass wool plugs in the center of the reactor. The temperature of the catalytic section was maintained using a PID controller with the thermocouple placed at the center of the catalyst bed. Before the experiments, the as-prepared catalyst was heated in O₂ flow at 300 °C for 1 h followed by degassing in N₂ flow to the experimental temperature to remove the residual moisture. A mixture of CO, O₂, and N₂ (all from Chemix Speciality Gases, Bangalore, India) was sent through flow controllers to the reactor. Reactions were carried out with a feed mixture consisting of 2.4 vol% CO, 2.4 vol% O₂ and balance N₂. All of the reactions were carried out under isothermal conditions with a total flow of 80 cm³ min⁻¹ (GHSV of 28 210 h⁻¹ based on the catalyst bed volume of 0.17 cm³ for 200 mg of the catalyst). The spent catalyst was characterized after the reaction using XPS. The products of the reaction were analyzed using an online gas chromatograph (Mayura Analytical, India). CO and CO₂ were separated using a molecular sieve column and were detected using a flame ionization detector. Because the signal of CO₂ in FID is poor, CO₂ is converted to CH₄ using a methanizer in the presence of Rh catalyst at 350 °C. This peak is measured and calibrated with standard gases. The nonlinear regression technique was used to determine the optimized kinetic parameters.

3. Results

3.1. Structural studies

X-Ray diffraction spectra of all base metal substituted catalysts along with unsubstituted SnO₂ are recorded at room temperature and shown in Fig. 1a. SnO₂ based catalysts were found to crystallize in TiO₂ like rutile structure. No additional phases such as the SnO₂ orthorhombic phase, corresponding transition metal oxide or other SnO based phases were observed. This showed the formation of single phase solid solutions that can be represented by the formula Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, Co) where M represents the transition metal. The peaks corresponding to metals were absent showing substitution of metals in the lattice. The patterns show broad X-ray line width. The substitution of metals in the lattice results in a change in lattice parameters due to the difference in ionic radii of Sn⁴⁺ and M²⁺/M³⁺. Therefore, profile refinement analysis of the spectra was carried out to determine the structural parameters.

Fig. 1 (a) XRD of SnO₂ and Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, Co). (b) Profile refinement of Sn_0.95Cu_0.05O_2−δ.

The XRD data were refined using the JANA 2000 suite program. The profile refinement of Sn_0.95Cu_0.05O_2−δ is shown in Fig. 1b. The symbols in the Fig. 1b show the experimental XRD data and the solid lines show the predicted XRD pattern. The difference between the actual and theoretical patterns is shown by the light line at the bottom. Fitting of the data to the tetragonal structure gave satisfactory values of the reliability data. The rest of profile refinements of the catalyst are given in Fig. S1a–c (see ESI†). The presence of wide peaks is an indication of nanometre sized crystallites. The crystallite sizes of the compounds were calculated using the Scherrer formula from the most intense peak. The crystallites were found to be in the range of 5 to 10 nm. The crystallite size of Sn_0.95Cu_0.05O_2−δ was the smallest while the crystallite size of Sn_0.95Co_0.05O_2−δ was the largest among studied compounds. Table 1 shows the cell parameters, crystallite sizes and the refinement reliability data for all the compounds. It was found that a decrease in lattice parameters occurred on substitution for all compounds.

Table 1 Profile refined parameters of Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, Co)

Catalysts	a/b (Å)	c (Å)	Cell volume (Å^3)	wRp	R p	χ ^2	Crystallite size (nm)	Surface area (m^2 g^−1)
Sn0.95Cu0.05O2−δ	4.7356	3.1854	71.43	7.2	5.7	1.02	5.5	33
Sn0.95Fe0.05O2−δ	4.7187	3.1735	70.66	9.1	6.8	1.05	8.6	31
Sn0.95Mn0.05O2−δ	4.7330	3.1862	70.94	10.4	7.9	0.84	7.8	32
Sn0.95Co0.05O2−δ	4.7263	3.1776	70.98	7.5	6.8	0.95	9.1	31

---

The TEM images and the ring type diffraction patterns of Sn_0.95Cu_0.05O_2−δ and unsubstituted SnO₂ are shown in Fig. 2 and Fig. S2 (see ESI†), respectively. In the TEM image of Sn_0.95Cu_0.05O_2−δ, the absence of Cu metal/oxide particle fringes confirmed the substitution of Cu ions in the SnO₂ matrix. Pure SnO₂ synthesized by the solution combustion method crystallizes in the rutile phase with particle sizes of 13 to 27 nm. On substitution of copper into the SnO₂ matrix, the particle size decreased and the particle sizes for Sn_0.95Cu_0.05O_2−δ measured from the image are in the range of 9 to 18 nm. The BET surface areas for all substituted SnO₂ catalysts are in the range of 31 to 33 m² g⁻¹, as shown in Table 1. ICP measurements gave the composition of Sn_0.95Cu_0.05O_2−δ and the amount of Cu in the catalyst was found to be 4.8 at%.

Fig. 2 (a) TEM image and (b) the electron diffraction pattern of Sn_0.95Cu_0.05O_2−δ.

The oxidation states of the base metals in the compounds were determined using XPS. The spectra were recorded both before and after the reaction. After the experiment, the flow of the reactant gases was stopped and the catalysts were allowed to stabilize at room temperature in a stream of N₂. The catalyst was quickly transferred to the vacuum chamber of the spectrometer to avoid the formation of possible surface oxide species. Different spent samples of the same catalyst obtained after different reaction runs were analyzed in the same manner and reproducible results were obtained.

Fig. 3 shows the Cu 2p spectra and Sn 3d spectra in Sn_0.95Cu_0.05O_2−δ before and after the reaction. The binding energies were calibrated with respect to the binding energy of graphite observed at 284.5 eV for all catalysts. Core level Cu (3d_{3/2, 1/2}) spectra of the as-prepared catalyst were resolved into sets of spin orbit doublets. Accordingly, the Cu (2p_{3/2, 1/2}) peaks at 934.0 and 953.8 eV with satellites at 9 eV below the main peak in Sn_0.95Cu_0.05O_2−δ can only be attributed to Cu in the 2+ oxidation state. Core level Sn (3d) spectra of the as-prepared catalyst have binding energy peaks at 486.3 eV, which correspond to the Sn⁴⁺ state and a small percentage of Sn is in the 2+ oxidation state. Sn (3d) peaks at lower binding energy of 485.3 eV are due to the 2+ oxidation state of Sn.²⁷ Taking into consideration the full width at half maxima, it was found that Sn is present partially in the 2+ oxidation state in the as-prepared sample. This is possible because SnO₂ can easily be reduced to SnO.²²

Fig. 3 Core level XPS of (a) Cu 2p and (b) Sn 3d in Sn_0.95Cu_0.05O_2−δ.

The presence of Sn²⁺ can be explained by the substitution of Sn⁴⁺ ions by base metal ions whose oxidation state is different from the parent metal oxide. During this process, oxygen vacancies may arise to maintain the charge neutrality when some Sn⁴⁺ reduces to Sn²⁺. A small amount of cerium was observed in the 3+ state when 2% atom Pt was substituted in the CeO₂ matrix.²⁸ The XPS spectra of the used sample after reaction were very wide. The presence of wide peaks spanning binding energy more than 10 eV clearly showed the presence of multiple oxidation states. The spectra having non-zero intensities from 930 to 940 eV showed the presence of Cu²⁺ and Cu¹⁺ states. For each oxidation state, two peaks corresponding to Cu 2p_{3/2, 1/2} having an intensity ratio of 3 : 2 should be present. These peaks were approximated as the standard Gaussian peaks. The peak positions corresponding to the different oxidation states of Cu were found from the literature.²⁹ An increase in the Sn²⁺ oxidation state relative to the as-prepared catalyst was observed in all compounds. Further reduction to the metallic Sn⁰ state was not observed under the mentioned reaction conditions.

Fig. 4 shows the XPS spectra of the Sn 3d and Mn 2p electrons for the Sn_0.95Mn_0.05O_2−δ after and before reaction. The Mn 2p core level spectra, as seen in Fig. 4a, are split into 2p_3/2 and 2p_1/2 components due to the spin orbit coupling. From the smoothed spectra, the core level binding energies of 2p_3/2 and 2p_1/2 for Sn_0.95Mn_0.05O_2−δ were estimated to be about 641.7 and 653.4 eV, respectively, and the difference in binding energies between Mn 2p_3/2 and Mn 2p_1/2 is 11.7 eV.³⁰ Based on above analysis, we confirmed that Mn ions have a chemical valence of 3+ in as-synthesized samples. The binding energy values of Sn 3d show that all Sn is in the 4+ state. During the oxidation reaction, several oxidation states can be formed and coexist or progressively change one into the other. Mn 2p_3/2spectra are quite broad and the binding energy shifts are not large enough to clearly distinguish between the different oxides, mainly when two or more species are simultaneously present. From the analysis of the broad Mn 2p main peaks, it is not possible to have detailed information on the intermediate oxidation steps. The corresponding Mn 2p_3/2 satellites cannot be used, since for Mn₂O₃, they obviously overlap with the Mn 2p_1/2 peak resulting in a high intensity of the Mn 2p_1/2 peak. However, only the binding energy shifts are sufficient to identify chemical states. The peak position corresponding to binding energies of Mn 2p_3/2 was assigned to 3+ and 2+ oxidation states.³⁰

Fig. 4 Core level XPS of (a) Mn 2p and (b) Sn 3d in Sn_0.95Mn_0.05O_2−δ.

The Co 2p spectrum, displayed in Fig. S3 (see ESI†), shows a main binding energy peak of Co 2p_3/2 corresponding to 780.5 eV and a satellite peak at 786.4 eV in the as-synthesized catalyst.³¹ The binding energy corresponding to the 778.10 eV peak represents metallic Co. Co²⁺ and Co³⁺ have very similar Co 2p binding energy, which makes identification of the oxidation state on account of the binding energy alone unreliable. There are, however, clear features in the 2p core level photoemission spectra that allow for an unambiguous differentiation between Co²⁺ and Co³⁺. Firstly, the presence of a strong shake-up satellite peak of Co²⁺-2p towards higher binding energy in contrast to Co³⁺-2p that exhibits only a weak satellite eliminates possibility of cobalt being in the 2+ oxidation state. The XPS spectrum of Co 2p region suggests that Co is in the 2+ oxidation state.³¹ The spectra of Sn manifest clearly that most of Sn is in the 4+ state along with a small amount of the 2+ state. After reaction, the presence of the broad peak indicates multiple oxidation states.

The Fe 3p_1/2XPS spectral region of as-synthesized Sn_0.95Fe_0.05O_2−δ is shown in Fig. S4 (see ESI†). The binding energy corresponding to the different oxidation states of Fe was found from the literature.³² The XPS peak of hematite phases occurs at 710.8 eV; however, the core level peak of Fe in Sn_0.95Fe_0.05O_2−δ showed a slight shift to higher binding energies compared to the Fe oxides, indicating the difference in the atomic environment surrounding the incorporated Fe ions. A detailed analysis of the spectrum demonstrated that the peak is broadened, which can be attributed to the presence of multiple oxidation states of Fe. The binding energies of a peak corresponding to Fe 2p_3/2 are located at 711.30 eV and 710.30 eV, respectively; suggesting that Fe exists mainly in the 3+ state along with a small amount of the 2+ state. The positions of the Sn 3d_5/2 and Sn 3d_3/2 peak maxima correspond to SnO₂. The presence of a small amount of Sn²⁺ was also observed. After reaction, the binding energies of 710.9 eV for Fe 2p_3/2 can be attributed to iron being in the tervalent states (Fe³⁺, Fe²⁺ and Fe⁰).

3.2. Catalytic reactions

CO oxidation by O₂ over base metal substituted SnO₂ was carried out with the inlet gas flow consisting of 2.4% vol CO, 2.4% vol O₂ with total gas flow of 80 cm³ min⁻¹ made up with N₂ over 200 mg catalysts. Fig. 5 shows the comparison of the catalysts for CO conversion. 100% CO conversion was obtained at below 225 °C over the Sn_0.95Cu_0.05O_2−δ and below 350 °C over Sn_0.95Co_0.05O_2−δ. The T₅₀ (corresponding to 50% CO conversion temperature) over catalysts Sn_0.95Cu_0.05O_2−δ, Sn_0.95Fe_0.05O_2−δ, Sn_0.95Mn_0.05O_2−δ and Sn_0.95Co_0.05O_2−δ are 177, 279, 282 and 296 °C, respectively. Experiments were also carried out with a mixture of gases and different amounts of catalysts loading. The reactant mixture had a gas composition of 2.4% CO, 2.4% O₂ and balance N₂ with the total gas flow of 80 cm³ min⁻¹. Fig. 6 shows the variation of W/F_CO (W is the weight of the catalyst and F_CO is the flow rate of CO) with the fractional conversion of CO for all catalysts. In the cases of Sn_0.95Cu_0.05O_2−δ and Sn_0.95Mn_0.05O_2−δ, it can be observed that at higher temperatures, the conversions did not vary linearly with W/F_CO. This shows that the reaction is mass transfer limited at high W/F_CO and shows two different reaction regimes for these catalysts.³³ In all cases, the rates of CO oxidation were calculated at various temperatures from the slope of the linear portion. Thus the slope was calculated from the plot (Fig. 6) of conversion with W/F_CO, where the conversions were lesser than 15%. The rate of CO oxidation reaction over the copper substituted catalyst is found to be very high at low temperatures among the reported catalysts^34–39 (Table 2). The apparent activation energies of CO oxidation over different catalysts were calculated by the Arrhenius plot (Fig. 7) for all substituted catalysts. The activation energies for Cu, Fe, Mn, Co doped catalysts are 50, 53, 59 and 69 kJ mol⁻¹, respectively.

Fig. 5 Effect of temperature on CO conversion over Sn_0.95M_0.05O_2−δ.

Fig. 6 Variation of fractional conversion of CO with the weight of catalyst for (a) Sn_0.95Cu_0.05O_2−δ, (b) Sn_0.95Fe_0.05O_2−δ, (c) Sn_0.95Mn_0.05O_2−δ and (d) Sn_0.95Co_0.05O_2−δ.

Table 2 Comparison of rates and activation energies for the CO oxidation reaction over various catalysts

Catalysts	Rate (μmol g^−1 s^−1) (°C)	E a (kJ mol^−1)	Ref.
5 wt% Ru/SiO2	1.00 (110)	94	34
5 wt% Rh/SiO2	0.0251 (110)	103	34
5 wt% Pd/SiO2	0.316 (143)	103	34
5 wt% Pt/SiO2	0.32 (115)	56	34
25Au75Pd/SiO2	0.128 (140)	—	35
0.014 wt% Rh/Al2O3	0.6 (196)	115.5	35
Pd/CeO2/Al2O3	38.0 (250)	84	36
Ce0.98Pd0.02O1.98	3.9 (120)	121	37
12 wt% Cu/δ-Al2O3	0.09 (150)	90	38
10% wt CuO/SiO2	0.04 (250)	59	39
Sn0.95Cu0.05O2−δ	15 (225)	50	Present study

---

Fig. 7 Arrhenius plot of CO oxidation to determine the activation energy.

3.3. Kinetic model

Ionic substitution into a reducible support enhances the oxygen and hydrogen storage capacities⁴⁰ due to the formation of solid defects. This is given by the formula Sn_1−X⁴⁺M²⁺_XO²⁻_2−δ□_δ where δ < x at room temperature and □ is an oxygen vacancy. These metals (M) have variable and lower oxidation states than that of Sn in SnO₂. These metals induce redox couples in the substituted base metals as well as in the SnO₂ matrix and render strong metal–support interactions. An ionic substitution in the form of M²⁺/M³⁺ ions for the Sn⁴⁺ site in the SnO₂ lattice also creates an oxide ion vacancy. These oxide vacancies in SnO₂ are crucial for oxygen adsorption and dissociation.⁴¹ The oxide vacancies are formed around the M²⁺/M⁺³ ions offering sites for O₂ adsorption.^42,43 However, there is experimental evidence that oxygen adsorbs in a molecular form like the superoxide O²⁻ species, preferentially on oxygen vacancies.⁴⁹ These sites are abundant in the proximity of the M²⁺/M³⁺ particles due to the Schottky junction between the M²⁺/M³⁺ and the n-type semiconducting oxides.⁴⁴ Therefore, the sites along the perimeter of metal ions–support interface would be the best sites for the reaction between CO adsorbed on oxide and the oxygen adsorbed on the vacancies. During the reduction of this solid solution, the reduced ionic substituted metal cations (M) are reoxidized by reduction of Sn⁴⁺ in the vicinity of Sn²⁺. The redox equilibrium can be represented as Sn⁴⁺ + M¹⁺ ⇔ Sn²⁺ + M²⁺. The reducibility of Sn⁴⁺ to Sn²⁺ enhances the flexibility of metal ions to adapt to a different oxidation state by maintaining the electronic neutrality of lattice.

Several explanations of observed synergism between supported noble metals and support are given in the literature. Bifunctional mechanisms with or without spillover of adsorbed CO and O₂ from noble metals to tin oxide have been proposed.^45,46 The (i) role of metal promoter and support, (ii) active site for adsorption of CO and O₂, and (iii) interaction of support and promoter (active phase) are not clearly understood. In the case of base metal supported catalysts, it has been observed that the rate of CO oxidation has weak dependence on the partial pressure of oxygen (P_O2) and has positive order dependence from 0 to 1 in the partial pressure of CO (P_CO).^24,47–49 Both Eley–Rideal and Langmuir–Hinshelwood models have been proposed for CO oxidation over Cu based catalysts.

Fig. 8 shows the variation in the rates of CO oxidation over Sn_0.95Cu_0.05O_2−δ under CO and O₂ partial pressures at different temperatures. The reaction rates increase with CO and O₂ partial pressures, but saturate at high partial pressures. Fig. 8 shows that the reaction rate rapidly increases with P_O2 than with P_CO at low partial pressures. The reaction order of P_CO seems to decrease from one to zero as P_CO increases. Since there are no maxima in the reaction rate, when either CO or O₂ partial pressures were varied, this indicates that these two reactants do not adsorb competitively on the surface of catalyst. The reaction rate has the shape of a Langmuir isotherm when CO and O₂ partial pressures are varied. CO oxidation mainly occurs through the lattice oxygen incorporation and the O₂ depleted oxide surface get rejuvenated due to the presence of O₂ in the reactant mixture. These results are consistent with our XPS results in which substituted base metals and Sn⁴⁺ ions undergo reduction, releasing lattice oxygen to maintain electrical neutrality. Based upon the above arguments, we proposed several plausible mechanism steps for CO reaction over base metals substituted SnO₂. Similar kinds of bifunctional mechanisms with or without spillover have been proposed for supported noble metals over reducible supports.^{12,22,48,51,52} Various possible steps involved in the mechanism are

CO + S ⇔ CO_S (1)

O₂ + 2“V” → 2“O” (2)

CO_S + “O” → CO₂ + S + “V” (3)

S, “V”, “O” refer to the vacant sites on support, oxide ion vacancy on support and oxygen species on the support, respectively. In this mechanism, it is assumed that CO is adsorbed on metal ions (Cu, Fe, Mn, Co) and Sn ions and oxygen is dissociatively chemisorbed on the oxide vacancies. The chemical nature of adsorbed CO species on metal ions and Sn ions has been treated to be the same. Fig. 8 suggests that CO adsorption can be described by the Langmuir isotherm. In the case of supported tin oxide on alumina, the best correlation could be obtained⁵⁰ using the Langmuir type rate equation that involved irreversible CO oxidation and CO₂ desorption steps and fast oxygen regeneration of the catalyst. Kulshreshtha et al. showed better mobility of oxygen in the case of mixed oxide (Mn₂O₃ + SnO₂) as compared to SnO₂.²¹ It was shown that the rate of CO oxidation on the SnO₂ catalyst under steady state conditions is proportional to the fraction of (OSnOCO) sites.³

θ_CO = K₁C_COθ_V1 (4)

where , θ_CO, θ_V1 represent the adsorption equilibrium for CO on the catalyst, the fraction of sites occupied by CO molecules on the metal surface, fraction unoccupied by CO molecules on the metal surface and SnO₂, respectively. The site balance on the metal support is given by

θ_V1 + θ_CO = 1 (5)

Note that oxygen adsorption is not included in this balance because CO and O₂ do not adsorb competitively. The CO coverage isThe rate of adsorption of O₂ on the support can be written asθ_V2 is the fraction of unoccupied oxygen vacancies. The steady state balance over θ_O species is,Oxygen atom and oxide ion vacancy balance on the support can be written as

θ_V2 + θ_O = 1 (9)

where θ_O is a fraction of O₂ on oxide vacancy. The rate of CO₂ formation is

r_CO2 = K₃θ_COθ_O (10)

After substitution of θ_CO and θ_O, the final form of rate equation is given byThis mechanism is a bifunctional mechanism in which the adsorbed CO on metal sites and oxygen adsorbed on lattice vacancies react to form CO₂. No step was assumed to be the rate determining step while deriving this expression. Note that for this mechanism, it is implicitly assumed that the reaction takes place in adjacent metal sites and tin active sites (interfacial sites). This model is tested for all catalysts and the values of rate coefficient were determined by the nonlinear regression. The optimized rate parameters are given in Table 3. Fig. 9a shows the variation of the rate with temperature and this satisfactorily correlates the experimental data.

Fig. 8 Variation of the rate of CO₂ formation with the (a) partial pressure of CO at constant P_O2 and (b) partial pressure of O₂ at constant P_CO.

Table 3 Rate parameters used in the model for the CO oxidation reaction

Parameter	Sn0.95Cu0.05O2−δ	Sn0.95Fe0.05O2−δ	Sn0.95Mn0.05O2−δ	Sn0.95Co0.05O2−δ
K 1	96√Texp(4400/T)	86√Texp(4300/T)	83√Texp(4330/T)	99√Texp(4400/T)
K 2	3.6 × 10^6√T	2 × 10^6√T	2 × 10^6√T	3.5 × 10^4√T
K 3	7.3exp(−3730/T)	9.3exp(−4410/T)	8.9exp(−5000/T)	11.7exp(−5620/T)
K 4	2.9 × 10^5exp(−4989/T)	1 × 10^5exp(−5387/T)	2.54 × 10^5exp(−6100/T)	4.85 × 10^5exp(−6724/T)
K 5	1.0 × 10^4exp(6196/T)	1.12 × 10^4exp(5997/T)	2.9 × 10^4exp(6203/T)	2.9 × 10^4exp(6724/T)
n	0.18	0.19	0.21	0.23
K 6	1022exp(−5869/T)	1114exp(−5962/T)	3754exp(−7459/T)	4631exp(−7795/T)
K 7	5011exp(1003/T)	2160exp(741/T)	5969exp(729/T)	6451exp(819/T)
m	0.21	0.24	0.22	0.25

---

Fig. 9 Variation of the rate of CO oxidation with temperature. The lines represent the fit based on the (a) proposed model, (b) MVK model and (c) LFS model.

Though the proposed model satisfactorily fits the experimental data, it was compared to existing models. Among the current models available in the literature, the Mars–van Krevelen model (MVK) and Langmuir–Hinshelwood type (LFS) model proposed by Liu and Flytzani-Stephanopoulos are often used to describe the kinetics of CO oxidation over Cu/CeO₂ surfaces.^23,55

The MVK model consists of two steps namely the reduction of oxidized catalyst surface along with formation of CO₂ and reoxidation of the catalyst surface with O₂. Thus, in this case, the catalyst undergoes reduction while the reductant is oxidized. The kinetic equation derived on the basis of above mentioned redox steps is²³

The parameters K₄ and K₅ in the above equation are the reaction rate constants for the reduction of catalyst by CO and reoxidation of the catalyst by O₂, respectively. The parameters K₄, K₅ and n obtained by fitting the reaction rate with eqn (12) are given in Table 3. The model fit and the experimental data are shown in Fig. 9b.

The LFS model assumes that there is no involvement of lattice oxygen in the reaction and atomic O₂ is provided by the amount adsorbed onto the catalyst surface. The equation derived based on this model⁵⁵ is

The parameters K₆ and K₇ represent the surface reaction rate constant and CO adsorption equilibrium constant, respectively. The parameters obtained by fitting eqn (13) are given in Table 3 and the experimental data are compared with the model fit in Fig. 9c.

4. Discussion

The value of the parameter, K₁, over Sn_0.95Cu_0.05O_2−δ is comparable to that of CO adsorption on metallic copper.^53,54 The smaller value of K₁ over Sn_0.95Fe_0.05O_2−δ and Sn_0.95Mn_0.05O_2−δ is manifestation of lower CO adsorption on these catalysts. The Cu⁺ species for the Sn_0.95Cu_0.05O_2−δcatalyst after reaction were observed by XPS. This indicates that CO adsorption sites are highly dispersed on the catalyst surface and the Cu⁺ surface provides strong CO adsorption sites.^23,58

Several spectroscopic observations were used to characterize various oxygen species chemisorbed on SnO₂.²⁵ In the case of transition metal oxides supported on the reducible metal oxide, surface lattice oxygen is involved in the oxidation reaction.⁵⁵ Our XPS studies also show involvement of lattice oxygen in the reaction. If CO reacts with adsorbed oxygen rather than oxygen, the formation of oxygen vacancies by CO will not be the rate determining step.⁵⁶ This is indeed observed in our proposed mechanism. If all steps in the mechanism are compared, O₂ dissociation on oxide vacancy is the slowest step and confirms the involvement of lattice oxygen during reaction. The oxygen dissociation over Sn_0.95Cu_0.05O_2−δ is much faster than any other studied catalyst. This shows that reducibility of Sn⁴⁺ to Sn²⁺ enhances the flexibility of copper ions to adapt to different oxidation states by maintaining the neutrality of lattice much more easily than any other catalyst. Though the adsorption of CO on Sn_0.95Co_0.05O_2−δ is comparable to Sn_0.95Cu_0.05O_2−δ, the reducibility of Sn_0.95Co_0.05O_2−δcatalyst is poor (observed from the value of K₂) resulting in the lower catalytic activity of Sn_0.95Co_0.05O_2−δ.

Fig. 9a–c represent the model fit for the proposed model, MVK model and LFS model for CO oxidation over Sn_0.95M_0.05O_2−δ (M = Cu, Fe, Mn, Co). It can be observed that all of the models satisfactorily fit the experimental data, with the regression correlation coefficient of above 0.98. Therefore, it is not possible to discriminate among the three models on the basis of the correlation with experimental data. The derivation of the MVK rate equation for redox reactions experiences a number of inconsistencies and this rate equation must be viewed only as a mathematical data-fitting function.⁵⁷ Further, this equation is applicable only for a reaction involving molecular oxygen adsorbed on a single site and not for lattice O ions/atoms that react with the reductant. The main assumption of the LFS model is that no lattice oxygen is involved in the CO oxidation reaction over the Cu/CeO₂ catalyst. This is in contrast with the findings of the spectroscopic studies,^58,59 where involvement of lattice oxygen was observed. Our XPS results also support the utilization of lattice oxygen during reaction. Based on the above discussion, though all the three models correlate the experimental data reasonably well, the proposed model gives insights into the rates of the actual processes that are likely to occur during the reaction.

5. Conclusions

The ionically substituted base metals (Cu, Fe Mn, CO) in the SnO₂ matrix were synthesized by a single step solution combustion method and these materials showed high catalytic activity compared to unsubstituted SnO₂. The order of activity for CO oxidation followed the order: Sn_0.95Cu_0.05O_2−δ > Sn_0.95Fe_0.05O_2−δ > Sn_0.95Mn_0.05O_2−δ > Sn_0.95Co_0.05O_2−δ. Oxide ion vacancy was found to be the main factor that governs the high rate of CO oxidation over Sn_0.95Cu_0.05O_2−δ. The significant enhancement of catalytic activity in the case of Sn_0.95Cu_0.05O_2−δ is mainly because of the efficiency of the Cu²⁺/Cu¹⁺ couple in this process. A reaction model based on a bifunctional mechanism was proposed and used to correlate experimental data. The reaction rate parameters were determined from this model.

Acknowledgements

The authors thank the Department of Science and Technology, India, for financial support.

References

1. B. Mguig, M. Calatayud and C. Minot, J. Mol. Struct. (THEOCHEM), 2004, 709, 73–78.
2. J. L. Ayastuy, M. P. González-Marcos, J. R. González-Velasco and M. A. Gutiérrez-Ortiz, Appl. Catal., B, 2007, 70, 532–541.
3. M. J. Fuller and M. E. Warwick, J. Catal., 1973, 29, 441–450.
4. H. Zhu, Z. Qin, W. Shan, W. Shen and J. Wang, Catal. Today, 2007, 126, 382–386.
5. P. M. Sreekanth and P. G. Smirniotis, Catal. Lett., 2008, 122, 37–42.
6. P. Ciambelli, S. Cimino, G. Lasorella, L. Lisi, S. De Rossi, M. Faticanti, G. Minelli and P. Porta, Appl. Catal., B, 2000, 37, 231–241.
7. X. S. Huang, H. Sun, L. C. Wang, Y. M. Liu, K. N. Fan and Y. Cao, Appl. Catal., B, 2009, 90, 224–232.
8. M. Luo, Z. Y. Hou, X. X. Yuan and X. M. Zheng, Catal. Lett., 1998, 50, 205–209.
9. J. L. Cao, Q. F. Deng and Z. Y. Yuan, J. Mater. Sci., 2009, 44, 6663–6669.
10. T. J. Huang, T. C. Yu and S. H. Chang, Appl. Catal., 1989, 52, 157–163.
11. W. Liu, A. F. Sarofim and M. F. Stephanopoulos, Appl. Catal., B, 1994, 4, 167–186.
12. S. H. Oh, G. B. Fisher, J. E. Carpenter and D. W. Goodman, J. Catal., 1986, 100, 360–376.
13. M. M. V. M. Souza, N. F. P. Ribeiro and M. Schmal, Int. J. Hydrogen Energy, 2007, 32, 425–429.
14. P. Bera, K. C. Patil, V. Jayaram, G. N. Subbanna and M. S. Hegde, J. Catal., 2000, 196, 293–301.
15. K. Grass and H. G. Lintz, J. Catal., 1997, 172, 446–452.
16. U. Oran and D. Uner, Appl. Catal., B, 2004, 54, 183–191.
17. W. Fliegel, G. Behr, J. Werner and G. Krabbes, Sens. Actuators, B, 1994, 18–19, 474.
18. R. Sasikala, N. M. Gupta and S. K. Kulshreshtha, Catal. Lett., 2001, 71, 69–73.
19. R. Schryer, B. T. Upchurch, B. D. Sidney, K. G. Brown, G. B. Hoflund and R. K. Herz, J. Catal., 1991, 130, 314–317.
20. T. Matsui, T. Okanishi, K. Fujiwara, K. Tsutsui, R. Kikuchi, T. Takeguchi and K. Eguchi, Sci. Technol. Adv. Mater, 2006, 7, 524–530.
21. S. K. Kulshreshtha, M. M. Gadgil and R. Sasikala, Catal. Lett., 1996, 37, 181–185.
22. S. Roy, A. Marimuthu, M. S. Hegde and G. Madras, Appl. Catal., B, 2007, 73, 300–310.
23. G. Sedmak, S. Hočevar and J. Levec, J. Catal., 2003, 213, 135–150.
24. N. J. J. Dekker, J. A. A. Hoorn, S. Stegenga, F. Kapteijn and J. A. Moulijn, AIChE J., 1992, 38, 385–396.
25. I. Kocemba and J. M. Rynkowski, Catal. Today, 2011, 169, 192–199.
26. T. Baidya, A. Gupta, P. A. Deshpande, G. Madras and M. S. Hegde, J. Phys. Chem. C, 2009, 113, 4059–4068.
27. R. O. Ansell, T. Dickinson, A. F. Povey and P. M. A. Sherwood, J. Electron Spectrosc. Relat. Phenom., 1977, 11, 301–313.
28. A. Gupta and M. S. Hegde, Appl. Catal., B, 2010, 99, 279–288.
29. J. G. Jolley, G. G. Geesey, M. R. Hankins, R. B. Wright and P. L. Wichlacz, Appl. Surf. Sci., 1989, 37, 469–480.
30. V. D. Castro and G. Polzonetti, J. Electron Spectrosc. Relat. Phenom., 1989, 48, 117–123.
31. T. J. Chuang, C. R. Brundle and D. W. Rice, Surf. Sci., 1976, 59, 413–429.
32. N. S. McIntyre and D. G. Zetaruk, Anal. Chem., 1977, 49, 1521–1529.
33. L. A. Petrov, Y. Alhamed, A. Al-Zahrani and M. Daous, J. Catal., 2011, 32, 1085–1112.
34. N. W. Cant, P. C. Hicks and B. S. Lennon, J. Catal., 1978, 54, 372–383.
35. A. M. Venezia, L. F. Liotta, G. Pantaleo, V. L. Parola, G. Deganello, A. Beck, Z. Koppány, K. Frey, D. Horváth and L. Guczi, Appl. Catal., A, 2003, 251, 359–368.
36. Y. Y. Yung-Fang, J. Catal., 1984, 87, 152–162.
37. S. Roy, A. Marimuthu, M. S. Hegde and G. Madras, Appl. Catal., B, 2007, 71, 23–31.
38. K. I. Choi and M. A. Vannice, J. Catal., 1991, 127, 465–488.
39. R. A. Prokopowicz, P. L. Silveston, R. R. Hudgins and D. E. Irish, React. Kinet. Catal. Lett., 1988, 37, 63–70.
40. G. Wrobel, C. Lamonier, A. Bennani, A. D'Huysser and A. Aboukaïs, J. Chem. Soc., Faraday Trans., 1996, 92(11), 2001–136.
41. C. T. Campbell, Science, 2003, 299, 657.
42. P. Bera, K. R. Priolkar, P. R. Sarode, M. S. Hegde, S. Emura, R. Kumashiro and N. P. Lalla, Chem. Mater., 2002, 14, 3591–3601.
43. L. Guczi, D. Horváth, Z. Pászti, L. Tóth, Z. E. Horváth, A. Karacs and G. Petõ, J. Phys. Chem. B, 2000, 104, 3183–3193.
44. H. Liu, A. I. Kozlov, A. P. Kozlova, T. Shido, K. Asakura and Y. Iwasawa, J. Catal., 1999, 185, 252–264.
45. G. C. Bond, L. R. Molloy and M. J. Fuller, J. Chem. Soc., Chem. Commun., 1975, 796–797.
46. G. S. Zafiris and R. J. Gorte, J. Catal., 1993, 143, 86–91.
47. A. Bielanski and J. Haber, Oxygen in catalysis, Dekker, New York, 1991.
48. G. I. Golodets, Heterogeneous catalytic reactions involving molecular oxygen, Studies in Surface Science and Catalysis, vol. 15, Elsevier, Amsterdam, 1983.
49. K. I. Choi and M. A. Vannice, J. Catal., 1991, 131, 22–35.
50. P. T. Wierzchowski and L. W. Zagorsk, Appl. Catal., B, 2003, 44, 53–65.
51. P. Mannila, T. Salmi, H. Haario, M. Luoma, M. Härkönen and J. Sohlo, J. Catal., 1986, 100, 360–376.
52. B. K. Cho, J. Catal., 1992, 138, 255–266.
53. D. B. Clarke, I. Suzuki and A. T. Bell, J. Catal., 1993, 142, 27–36.
54. M. J. Sandoval and A. T. Bell, J. Catal., 1993, 144, 227–237.
55. W. Liu and M. F. Stephanopoulos, J. Catal., 1995, 153, 317–332.
56. A. Tschope, W. Liu, M. F. Stephanopoulos and J. Y. Ying, J. Catal., 1995, 157, 42–50.
57. M. A. Vannice, Catal. Today, 2007, 123, 18–22.
58. P. G. Harrison, I. K. Ball, W. Azelee, W. Daniell and D. Goldfarb, Chem. Mater., 2000, 12, 3715.
59. A. Martinez-Arias, M. Fernandez-Garcia, O. Galvez, J. M. Coronado, J. A. Anderson, J. C. Conesa, J. Soria and G. Munuera, J. Catal., 2000, 195, 207.

---

Footnote

† Electronic supplementary information (ESI) available: Fig. S1 shows the profile refinement of (a) Sn_0.95Fe_0.05O_2−δ, (b) Sn_0.95Mn_0.05O_2−δ and (c) Sn_0.95Co_0.05O_2−δ. Fig. S2 shows the TEM image and the electron diffraction pattern of SnO₂. Fig. S3 shows the core level XPS of Co 2p and Sn 3d in Sn_0.95Co_0.05O_2−δ. Fig. S4 shows the core level XPS of Fe 2p and Sn 3d in Sn_0.95Fe_0.05O_2−δ. See DOI: 10.1039/c1cy00421b

---