Abstract

Low temperature active and stable mesoporous Au (0.1, 0.2, 0.5, and 1.0 wt.%) supported α-Fe2O3 catalysts were prepared via deposition-precipitation method. The H2-pretreated catalyst with 0.5 wt.% Au loading offered CO conversion of 100% at 323 K and showed continual activity for at least 120 h. X-ray diffraction and transmission electron microscopy analysis indicate that Au species were highly dispersed as nanoparticles (20–40 nm) on the surface of α-Fe2O3 support even after thermal treatment at 773 K. The N2-physisorption measurements show that the synthesized α-Fe2O3 support and Au-Fe2O3 nanocomposites possessed mesopores with high specific surface area of about 158 m2 g−1. X-ray photoelectron spectroscopy and H2-TPR results reveal that the Au species exist in metallic and partially oxidized state due to strong interaction with the support. Effective Au-Fe2O3 interaction resulted in a high activity for Au nanoparticles, locally generated by the thermal treatment at 773 K in air.

1. Introduction

The activity of humans in the world led to massive production of carbon monoxide (CO), starting from wood burning for warming houses and utilization of fossil fuels for electricity production, automobile engines, and industrial chemical combustion, which involves incomplete oxidation of carbon compounds [1]. In environmental aspects, CO is a highly toxic gas; high concentration of CO in the atmosphere is hazardous for lives, and it causes many illnesses, suffocation, and sudden death [2]. CO is converted to CO2 which is a less toxic and natural gas that exists in the air; CO2 is also a feedstock for methanol production, which in turn is used to produce many chemical compounds that are valuable for human life, such as plastics [3]. In general, the catalyst efficiency in CO oxidation is determined by its ability to oxidize CO to CO2 fully at a low reaction temperature. For instance, the catalytic converter used in automobiles exhaust is proposed to oxidize CO at a temperature lower than 373 K, as it mainly contains traces of precious metals, for instance, palladium, platinum, and rhodium [4]. Au supported catalyst was demonstrated as a highly active catalyst [5]; precious metals are expensive; therefore, enhancement of the support properties plays a key role in catalytic activity, where a small amount of precious metal is doped on a large amount of support. It was reported that transition-metal oxides such as Fe2O3, TiO2, Co3O4, and NiO are suitable supports [6]. The bulk α-Fe2O3 is a reducible support to provide a large amount of active sites; several researchers used α-Fe2O3 as a support for CO oxidation. We recently synthesized Ag-Fe2O3 nanocomposites and utilized them as catalysts for CO oxidation [7]. It was reported that α-Fe2O3 possessed an advantage in that it has a long lifetime because it could resist poisoning from CO2 [8]. Many research groups [913] utilized Au supported iron oxide materials as a catalyst for CO oxidation. However, the controlled synthesis of Au metal oxide nanocomposites still faces tremendous challenges [14].

The preparation method of α-Fe2O3 support shows a significant change in support properties such as particle size, morphology, and surface area and their interaction with active metals. Synthesized α-Fe2O3 nanoparticles showed a much higher activity than microcrystalline α-Fe2O3 for CO oxidation; in other words, α-Fe2O3 could be fabricated in a desired morphology to obtain an efficient catalyst [15]. Mesoporous support (in which the pore size ranges between 2 and 50 nm) has an advantage to obtain good dispersion of active metal nanoparticles; the mesopores are responsible for capillary condensation and allow active metal nanoparticles to disperse more into the wall of pores [16]. Mesoporous Fe2O3 was prepared by different methods using templates, such as cellulose nanocrystals template [17] and SBA-15 [18]. Park and Lee [12] observed that oxidized gold species are important for CO oxidation and the authors concluded that the oxidized gold species were more active than metallic gold. Further, different opinions were reported with respect to the active catalyst that contained both Au atoms and ions [19, 20]. However, strong evidence of the oxidation states of Au for achieving high activity is one of the major issues that has not yet been resolved.

In this work, we aim to synthesize nanosized mesoporous Au-Fe2O3 composites and use them as catalysts for CO oxidation at low reaction temperatures. We also made an effort to study the physicochemical properties of the catalysts and correlate the properties with catalytic CO oxidation activity.

2. Experimental

2.1. Preparation of Mesoporous α-Fe2O3 Nanomaterials

Single-crystalline mesoporous α-Fe2O3 nanomaterials were obtained by reacting Fe(NO3)3·9H2O with NH4H2PO4 under hydrothermal conditions [21]. Typically, 1.0 g of Fe(NO3)3·9H2O was added to 10 mL of 0.1 N aqueous NH4H2PO4 solution. The added Fe(NO3)3·9H2O was hydrolyzed within a few minutes, and a brown colored precipitate was formed. The obtained precipitate was stirred for 1 h and then transferred into an autoclave lined with a Teflon vessel. Then, the autoclave was heated in an electric oven at a temperature of 353 K for 12 h. After the hydrothermal treatment, the precipitate was washed with distilled water four times, dried in a vacuum oven at 373 K for 12 h, and calcined at 773 K for 2 h.

2.2. Preparation of Au-Fe2O3 Nanocomposites

Gold was loaded on synthesized Fe2O3 support via the deposition-precipitation method. In this method, the aqueous suspension containing 1.0 g of the support material was heated in a water bath at 373 K and the pH value of the solution was adjusted to 9.0 by adding 1 N NH4OH solution. A HAuCl4 solution corresponding to Au loading (0.1, 0.2, 0.5, and 1.0 wt.%) was poured into the suspension. Additional NH4OH solution was added dropwise to the suspension to keep the pH value stable. After stirring for 30 min and aging for 2 h at room temperature, the precipitate was filtrated and washed several times with deionized water until no Cl was detected by a 0.25 mol L−1 AgNO3 solution. The resultant precipitate was dried at 393 K for 3 h and then calcined at 773 K for 4 h. The catalysts were denoted as Au-Fe, where represents the Au loading.

2.3. Characterization

The elemental composition of the catalysts was performed using ICP-AES, Optima 7300DV (PerkinElmer Corporation, USA) instrument. The sample preparation procedure for elemental analysis and a detailed description of the experimental procedure were described in our previous publication [22].

The powder X-ray diffraction measurements for the samples were performed using Bruker D8 Advance target diffractometer. The XRD patterns were obtained using Cu Kα radiation ( = 1.5405 Å) at 40 kV and 40 mA. The identification of phases presented in the samples and crystallite sizes of iron oxide and gold metal were calculated using Scherrer equation as described in our previous publication [22].

Transmission electron microscopy (TEM) and high-resolution TEM images of the samples were obtained using a Philips CM200FEG microscope equipped with a field emission gun operated at 200 kV. Diffuse reflectance UV-Vis spectra for all the samples were collected using a Thermo Scientific evolution 600 UV-visible spectrophotometer conjugated with an integrating sphere in the wavelength range of 200–800 nm.

The textural properties of the synthesized samples were determined from N2 adsorption-desorption measurements at 77 K carried out using Quantachrome NOVA 3200e automated gas adsorption system. A detailed procedure was described in our previous publication [22] to determine the specific surface area, pore volume, and average pore radius of the samples. The X-ray photoelectron spectroscopy measurements for all the samples were carried out using a SPECS GmbH XPS instrument. A standard dual anode excitation source with Mg Kα (1253.6 eV) radiation was used at 13 kV and 100 W. H2-temperature programmed reduction profiles for all the catalyst samples were collected using Quantachrome CHEMBET-3000 instrument. More details of the experimental procedures used in this work can be found in our previous publications [7, 22].

2.4. Catalytic Oxidation of Carbon Monoxide

Catalytic activity measurements for CO oxidation were carried out using a fixed-bed reactor at atmospheric pressure. Quartz spheres (0.5 g) were used to dilute the Au-Fe2O3 catalyst (0.5 g) in a quartz reactor. Prior to the activity tests, the catalysts were treated with 10% H2/N2 (100 mL min−1) at 353 K for 1 h. After the catalysts were cooled down to 273 K under N2 (100 mL min−1), the feed gas containing 100 ppm CO balanced with compressed air was passed through the reactor at 1000 mL min−1. The reaction temperature, measured with a thermocouple, was increased from 273 K to 333 K. On-line gas chromatograph (Agilent) with a TCD detector was employed to measure the reactor inlet and outlet gas streams. TDX-01 column (2 m × 4 mm) was used to separate O2, CO, and CO2.

The conversion of CO was by change in CO concentration:

3. Results and Discussion

The XRD patterns for pure iron oxide and Au-Fe nanocomposite samples are shown in Figure 1. The bulk Fe2O3 sample calcined at 773 K showed diffraction peaks at = 24.4°, 33.5°, 36.0°, 41.1°, 49.8°, 54.7°, 57.3°, 62.7°, and 64.4° corresponding to (012), (104), (110), (113), (024), (116), (122), (214), and (300) planes, respectively. These peaks are clearly indexed to α-Fe2O3 phase [JCPDS number ]. XRD peaks for the α-Fe2O3 sample are significantly broader than those for the micrometer sized sample, indicating that the sample possessed a very small particle size. It is also noteworthy that the synthesized sample does not show any diffraction peaks corresponding to any other crystalline phase, indicating the purity of the sample. XRD patterns of the Au-Fe nanocomposites with 0.1, 0.2, and 0.5 wt.% Au loadings showed only characteristic diffraction peaks of α-Fe2O3. Neither Au oxide nor metallic Au was detected at lower Au loading (<1.0 wt.%). This is possibly due to high dispersion of Au particles on Fe2O3 support in lower loadings. The decrease of intensity of diffraction peaks due to Fe2O3 clearly indicates the decrease of crystallinity of the sample with the increase of Au loading. Additional diffraction peaks can be seen for the 1.0Au-Fe sample: diffraction peaks at = 38.2° and = 44.5° corresponding to (111) and (200) planes [23] for metallic Au crystallites for the 1.0Au-Fe sample.

The TEM analysis was used to determine the morphology and size of the Au and Fe2O3 phases presented in the samples. The TEM images of the Au-Fe nanocomposite samples are presented in Figure 2. It was observed that the synthesized α-Fe2O3 sample possessed some small cylindrically shaped nanoparticles of about 15–25 nm diameter and 150–200 nm length and also many large particles which are composed of several small cylindrically shaped nanoparticles.

This is possibly due to calcination of the samples at a high reaction temperature leading to several particles bundling together via hydrophilic interactions. Some aggregated spherically shaped Fe2O3 nanoparticles of size around 50 nm are also observed along with the cylindrically shaped nanoparticles. The presence of large crystalline particles indicates the formation of polycrystallites, due to the high temperature calcination (773 K). No clear differences in shape, size, and dispersion of Au are observed in the 0.1Au-Fe and 0.2Au-Fe catalysts. However, we must take into account the fact that the loading of Au in these samples is very low and it makes the detection of Au particles difficult due to the low mass and diffraction contrast. TEM images of 0.5Au-Fe and 1.0Au-Fe samples clearly show thick black spherical grain of Au metal on the α-Fe2O3 surface with a size of about 45 nm. Large pores are observed in between the α-Fe2O3 particles, which is further confirmed by N2 adsorption analysis and discussed in a later part of the Results and Discussion.

The particle size for all the samples was calculated using Scherrer’s equation and XRD peak broadening, as shown in Table 1. The half widths of (104) peak of α-Fe2O3 at = 33.5° and of (111) peak of Au at = 38.2° (both Cu Kα) were used. Scherrer’s equation gives the average size in the order of 45 nm for the Fe2O3 and 40 nm for the Au in case of 1.0Au-Fe sample. The XRD determined crystallite sizes of α-Fe2O3 and Au were higher than the crystallite size measured by TEM analysis. Nonuniform particle size distribution could be a reason for such behavior. The average crystal size of α-Fe2O3 was calculated to be 55 nm. The responsible species for the catalytic properties of Au supported catalysts is not clear yet although it was suggested that the interface structures between the Au nanoparticle and metal oxide support act as active sites [24]. In order to understand the interface structure, we carried out high-resolution transmission electron microscopy (HRTEM) analysis for 0.5Au-Fe and 1.0Au-Fe samples. The HRTEM images of both of these samples are shown in Figure 3. It can be seen that there are two lattice fringes with lattice spacings of 0.235 nm and 0.270 nm corresponding to the Au (111) and Fe2O3 (104) planes from different grains, respectively [25], which further confirm the synthesis of Au-Fe2O3 nanocomposite.

In case of 1.0Au-Fe sample, a large size Au particle was observed, which might be formed from several smaller particles, and it was deposited on the plain of Fe2O3 support. The crystal lattices of Au (111) and Fe2O3 (104) are distinct and without strong interaction with each other. On the other hand, in the 0.5Au-Fe sample, it can be seen that the Au crystal lattices were overlapped with each other and the Au particles were surrounded with crystalline Fe2O3 and a distinct interface between Au particles and Fe2O3 formed.

The UV-Vis diffuse reflectance spectra of calcined Fe2O3 and Au-Fe nanocomposites are shown in Figure 4. It was reported that iron oxides generally show three kinds of optical transitions. The peaks between 200 and 400 nm mainly result from the ligand to metal charge transfer transitions and partly from the contribution of the Fe3+ ligand-field transitions. The region between 400 and 600 nm is considered to be a result of pair excitation processes, possibly overlapping the contributions of ligand-field transitions [7]. He et al. [26] reported that the peaks in the 600–750 nm and 750–900 nm regions could be due to the transitions. It is widely known that deposition of the Au on any support leads to a broad absorption band in the range of 500–570 nm and this contribution can be assigned to the well-known plasmon resonance of Au nanoparticles [27].

In case of Au-Fe2O3 nanocomposite samples, the contribution from Au nanoparticles was masked by the extended absorption of α-Fe2O3 in the visible region [28]. However, an increase in the broadness of the Au plasmon absorption band was observed with the increase of Au loading from 0.1 wt.% to 1.0 wt.%, although Au-Fe nanocomposite samples display weak plasmon resonance. Furthermore, a red shift was observed as Au nanoparticles agglomerate on the Fe2O3 surface in case of 1.0Au-Fe sample owing to the relatively big particle size and interparticle interactions. It is possible that the adsorption of oxygen on Au surfaces leads to the formation of surface oxide and forms Fe-Au interactive species in the nanocomposite samples.

The textural properties of calcined α-Fe2O3 and Au-Fe nanocomposites were studied by the N2-physisorption measurements. The N2 adsorption-desorption isotherms of all the samples are shown in Figure 5. For all the samples, the isotherms are identified as type III according to the IUPAC classification, which is a typical characteristic of macroporous materials [29], and a clear H3-type hysteresis loop is observed, thus indicating the presence of plate-like particles giving rise to slit-shaped pores [30]. The hysteresis loop for the 0.1Au-Fe, 0.2Au-Fe, and 0.5Au-Fe samples is similar to the Fe2O3 support, thus indicating that the deposition of Au in low loadings did not modify the macroporous nature of the support. In case of 1.0Au-Fe sample, a similar hysteresis shape was observed; however, the relative pressure where the condensation step occurs shifted to higher values.

The BET surface area, average pore diameter, and pore volume of all the samples estimated from their respective adsorption-desorption isotherms are given in Table 2. The BET surface area of the calcined macroporous α-Fe2O3 is 72 m2 g−1, which is more than 12 times higher than the commercial α-Fe2O3 (6.0 m2 g−1). The pore volume of the α-Fe2O3 was found to be 0.316 cm3 g−1. A decrease of surface area was observed with the increase of Au loading; this is probably due to the fact that Au deposition onto α-Fe2O3 makes the pore walls lower than pure Fe2O3, which is reflected in their surface areas and also confirmed by their respective pore volume and pore radius data.

The pore size distribution (PSD) was obtained by Barrett-Joyner-Halenda (BJH) method as shown in Figure 5(b). The porous Fe2O3 shows unimodal pore size distribution with predominant pore radius of around 87.7 Å. It is likely that some Au nanoparticles crystallized inside the pores of Fe2O3 support during the thermal treatment, thus partially blocking them from the outside, and as a consequence, the mean pore size was shifted to higher values.

X-ray photoelectron spectroscopy (XPS) analysis was utilized to determine the electronic states of elements and composition of the Au-Fe nanocomposites. The XPS spectra of Fe 2p, Au , and O of Au-Fe nanocomposite samples are shown in Figure 6. From the figure, it is clear that the Au peak is weak and Au is not clearly observed for the 0.1Au-Fe sample due to its low Au loading. However, both Au and Au peaks were clearly observed after the increase of Au loading (beyond 0.2 wt.%). It was reported that the and binding energy values for metallic Au0 are 84.1 eV and 87.7 eV. The Au binding energies for the oxidized Au+ and Au3+ ions are 85.6 eV and 86.5 eV, respectively [31]. After peak deconvolution of the Au energy regions, single Au and Au were observed at 83.4 eV and 87.2 eV, respectively, for 0.1Au-Fe, 0.2Au-Fe, and 0.5Au-Fe samples. The binding energy value of Au observed for these samples is lower than the bulk gold (84.1 eV) indicating the formation of Au-Fe2O3 interactive species [11]. A similar observation was reported in the literature [32, 33]. Further, the binding energy of Au peaks in the 0.2Au-Fe sample is slightly lower than in the 0.1Au-Fe and 0.5Au-Fe samples. There is probably a stronger interaction between the Au particles and the Fe2O3 support in 0.2Au-Fe than the other two samples.

It is interesting to note that, after deconvolution of the Au energy regions for the 1.0Au-Fe sample, two components at 83.0 eV (I) and 84.3 eV (II) due to metallic Au and oxidized surface Au species were observed [34]. Lin and Chen [35] also observed two distinct peaks of Au supported iron oxide catalysts. The binding energy value of Au peak (II) is higher than that of Au0 but lower than that of Au+ and Au3+. This observation suggests that the oxidized Au species are in the form of Auδ+, which is a transition state between Au0 and Au+ due to the interaction between nanosized Au particles and the Fe2O3 support [36]. The XRD patterns of 0.1Au-Fe, 0.2Au-Fe, and 0.5Au-Fe samples did not show any diffraction peaks corresponding to metallic Au, but the XPS analysis of these samples shows the presence of metallic Au on the support surface. One possible explanation is that the metallic gold formed on the support surface remains largely amorphous in these low loading samples, as amino or nitroso complexes of gold generally get precipitated during the precipitation-deposition method under basic conditions. Upon heating at 393 K, amino or nitroso complexes of gold start to decompose to metallic gold and gold oxides when we calcine the material at 773 K [11].

It was observed that Fe2O3 sample showed Fe peak maxima at 710.6 eV (not shown in Figure 6); some differences were observed in the Fe binding energy values for Au-Fe nanocomposite samples. The peak was shifted to higher binding energies (711.6 eV for 0.1, 0.2, and 0.5 wt.% Au and 711.9 eV for 1.0 wt.% Au), indicating that the Fe oxidation state of surface iron species is sensitive to the Au composition. Brundle [37] reported that, in bulk iron oxides, Fe (III) has a binding energy of 711 eV, and the binding energy of Fe(II) is centered at 709.7 eV. The Fe 2p deconvoluted spectra of all the Au-Fe samples also showed a minor peak located in between 709.2 eV and 708.7 eV and this peak can be attributed to Fe species with lower oxidation state. The appearance of this peak is due to the formation of Au-Fe interactive species on the support surface. The contribution of this peak to the 0.2Au-Fe sample was 2.2% and increased to 4.3% after the loading was increased to 0.5 wt.%. A further increase of Au loading to 1.0 wt.% caused a decrease in the concentration of this peak to 3.6%. These observations led to the conclusion that the Au nanoparticles are interacting with Fe2O3 due to increased dispersion of the Au metal nanoparticles on the surface of Fe2O3 support. This statement is supported by the increased Au/Fe atomic ratio derived from XPS measurements (Table 2). It is also interesting to note that bulk Au/Fe ratio determined from ICP-AES analysis is higher than Au/Fe ratio derived from XPS analysis results. This observation indicates that the part of the deposited Au is transported into the spaces existing in between the Fe2O3 nanoparticles.

The deconvoluted O 1s spectra for all the samples are shown in Figure 6. Three different peaks can be clearly seen in the spectra, which correspond to three different oxygen species. The 0.1Au-Fe sample showed peaks at 529.3 eV, 531.0 eV, and 533.1 eV. The peaks at 531.0 eV and 529.3 eV can be attributed to oxygen in Fe2O3 and Au-Fe2O3 interactive species, respectively, while the other peak at 533.1 eV can be related to oxygen in the electrophilic state adsorbed on Au metal [38]. The relative proportion of the area under the XPS peaks was changed with Au loading (Table S1). The contribution due to the XPS peak at 533.1 eV was increased with the increase of Au loading from 0.1 to 1.0 wt.%; this observation indicates the increase of Au metal concentration on the Fe2O3 surface. On the other hand, the contribution due to the O 1s peak at 529.3 eV was increased up to 0.5 wt.% but decreased upon further increase to 1.0 wt.%. The XPS results are in accordance with the observations from the TEM analysis, and Au-Fe nanocomposites possess two types of (Au metal and Au-Fe interactive) species and the extent of formation of these species is dependent on the Au loading over the Fe2O3 support.

H2-TPR analysis for all the samples was performed to study the reduction behavior and nature of Au-Fe interaction in the Au-Fe nanocomposites. Figure 7 shows the H2-TPR spectra of α-Fe2O3 and Au-Fe nanocomposite samples calcined at 773 K. The H2-TPR spectrum of α-Fe2O3 showed a broad peak that started at 623 K attributed to the reduction of Fe2O3 to Fe3O4 ( = 723 K) and the other extensive peaks at temperatures beyond 773 K. It was reported that, in the literature [39], the H2-TPR peak that appeared at 573 K can be attributed to the reduction of the hydroxylated Fe2O3 species and the peak at ≈673 K can be attributed to the reduction of Fe2O3 to Fe3O4. The reduction of Fe3O4 to FeO occurs at ≈873 K and finally the reduction to Fe0 happens above 1073 K. We can also observe that the high temperature peak does not reach the baseline in the H2-TPR profile of bulk Fe2O3 sample even at 1073 K, which means the reduction was continued above this temperature.

The H2-TPR patterns for the 0.1Au-Fe and 0.2Au-Fe nanocomposite samples are very similar to the bulk Fe2O3 sample; however, deposition of a small amount of Au resulted in a shift in of the peak due to reduction of FeO to Fe and also the high temperature peak started reaching the baseline in case of 0.2Au-Fe sample. This observation reveals that deposition of small amounts of Au could enhance the reduction of Fe2O3. On the other hand, the H2-TPR spectra of 0.5Au-Fe and 1.0Au-Fe samples substantially differ from the other samples. Increasing of Au loading to 0.5 and 1.0 wt.% resulted in the disappearance of the reduction peak at = 723 K (reduction of Fe2O3 to Fe3O4). Carabineiro et al. [40] reported similar results that Au supported Fe2O3 catalysts showed two reduction peaks above 800°C, which could be due to the reduction of Fe3O4 to FeO (or of FeO to Fe). The deposition of a larger amount of Au leads to the formation of agglomerated Au nanoparticles, which interact with Fe2O3 leading to a shift in reduction temperature. Venugopal and Skurrell [41] also observed a shift in reduction temperature due to the interaction between the Au species and iron oxide support.

The H2 uptake values during the different reduction stages are calculated from integrating the area of the reduction peaks, which are tabulated in Table S2. The overall H2 consumption of the reduction peaks can be assigned to the reduction of α-Fe2O3 to Fe3O4 and also the reduction of Fe3O4 to Fe. In the two stages of reduction, Fe2O3 reacted with hydrogen based on the following reactions.  ; mol H2/mol Fe = 0.166 for complete reduction  ; mol H2/mol Fe = 1.33 for complete reduction

The H2 consumption value observed for bulk Fe2O3 is consistent with the theoretical value. However, the overall H2 consumption observed for Au-Fe nanocomposite samples is higher than the stoichiometric amount required to reduce all the Fe presented in the catalyst, indicating that some degree of surface reduction of catalyst takes place at high temperatures as a result of Au-Fe interaction in these catalysts.

A fixed-bed catalytic reactor was used to evaluate the catalytic activity of synthesized bulk Fe2O3 and Au-Fe nanocomposite samples. The reactions were conducted at the temperature range of 273–333 K isothermally. Figures 8(a) and 8(b) show the catalytic activity in CO oxidation of calcined and H2-pretreated catalysts, respectively. In case of calcined catalysts (Figure 8(a)), 100% CO conversion was observed for 0.5Au-Fe catalyst at 313 K, whereas CO conversion levels corresponded to 97%, 92%, and 41% for 1.0Au-Fe, 0.2Au-Fe, and 0.1Au-Fe catalysts, respectively. The bulk Fe2O3 support did not show any CO conversion at the same temperature.

The CO oxidation catalytic activity results obtained for the H2-pretreated catalysts are shown in Figure 8(b). At lower reaction temperatures, for instance, 283 K, the catalysts showed a very little difference in CO conversion (78%-0.1Au-Fe, 80%-0.2Au-Fe, 83%-0.5Au-Fe, and 78%-1.0Au-Fe). The CO conversion increased to 100% at 323 K for all Au-Fe nanocomposite samples. The much higher performances of the H2-pretreated Au-Fe nanocomposite samples could be due to the well dispersed Au nanoparticles and Au-Fe2O3 interactive species [42]. It is known that the well dispersed Au nanoparticles could weaken C-O and O-O bonds after CO and O2 adsorption, influencing the interaction between the surface and the adsorbents and favoring CO activation [43].

The influence of Au loading on CO conversion for the calcined and H2-pretreated Au-Fe nanocomposite catalysts is shown in Figure 8. It is clear that the CO conversion was enhanced when the Au loading was increased from 0.1 to 0.5 wt.% and then decreased upon further increase of Au loading to 1.0 wt.%. In case of calcined Au-Fe nanocomposite samples, the CO conversion was increased from 30% to 90% when Au loading was increased from 0.1 wt.% to 0.5 wt.% but decreased to 75% upon further increase to 1.0 wt.%. A similar behavior was observed for H2-pretreated catalysts. These results could be explained on the basis of low Au particle size, high dispersion of Au metal active sites for CO adsorption, and activation in case of 0.5Au-Fe catalyst, but agglomeration of Au nanoparticles was observed in case of 1.0Au-Fe catalyst.

Figure 9 represents the influence of GHSV on CO conversion over the calcined and H2-pretreated 0.5Au-Fe catalyst; higher space velocity resulted in lower CO conversion and a similar behavior was observed in all the catalysts. This behavior is expected due to decreased reactants residence time on the catalyst surface with increased GHSV. In case of calcined 0.5Au-Fe catalyst, the CO conversion was decreased from 90% to 74% after the GHSV was increased from 60,000 mL g−1 h−1 to 120,000 mL g−1 h−1; however, the extent of decrease is low in case of H2-pretreated catalyst; the CO conversion was decreased from 100% to 93%. The CO oxidation activity results suggested that H2 pretreatment greatly enhances the reaction rate.

A specific rate for CO oxidation for calcined 0.5Au-Fe catalyst was calculated and compared with reaction rates of different Au supported catalysts reported in the literature under similar reaction conditions (Table 3). The calcined 0.5Au-Fe catalyst showed a relatively high specific rate of 5.3 × 10−2, which was more active than Au-La2O3 catalyst, which possessed a specific rate of 4.0 × 10−2. The H2 pretreatment enhanced the activity further, and the specific rate over the H2-pretreated 0.5Au-Fe catalyst was increased to 12.4 × 10−2, which was higher than that of the Au-Fe2O3 catalyst (9.1 × 10−2) prepared by the deposition-precipitation method [21, 4447]. These results clearly indicate the superior catalytic activity of Au-Fe nanocomposite catalysts for CO oxidation.

The turnover frequencies (TOFs) were also calculated in order to gain an insight into the intrinsic activities of Au-Fe nanocomposite catalysts. The TOF of H2-pretreated 0.5Au-Fe catalyst was 0.038 s−1, exhibiting a significantly higher activity than that of calcined 0.5Au-Fe catalyst (TOF = 0.012 s−1). The TOF of the H2-pretreated 0.5Au-Fe catalyst was comparable to that of the Au/CeO2 catalyst (TOF = 0.047 s−1) [45] and the commercial Au/Fe2O3 catalyst (TOF = 0.04 s−1) [21]. The TOF values of H2-pretreated Au-Fe nanocomposite catalysts are higher than of the nanosized Au-Fe2O3 catalysts reported in the literature [21]. The enhanced activity of the catalyst could be attributed to the small size of the synthesized Au nanoparticles [48] and also the synthesized nanosized mesoporous Fe2O3 support possessed a high surface area. As a result, there is an increase in the amount of active sites for the CO2 to adsorb. The oxygen adsorbs to the support and dissociates [49]. The stability of most active catalysts among the investigated samples was tested at 293 K and the time on stream analysis of the 0.5Au-Fe sample is shown in Figure 10. The CO conversion was readily decreased from 60% to 53% for 0.5Au-Fe calcined catalyst after 3 h and it was further decreased to 50% after 10 h. The initial CO conversion of H2-pretreated 0.5Au-Fe catalyst is 100%, and the CO conversion was decreased with the time on stream. After 10 h, CO conversion was about 94%.

The stability test further evidenced the highest performance of H2-pretreated 0.5Au-Fe catalyst (Figure 11). The XRD analysis of spent catalyst indicated that sintering of Au or Fe2O3 did not occur during the time on stream analysis because the reaction temperature and reduction temperature were lower than the calcination temperature. The slight deactivation observed in the Au-Fe nanocomposite samples is probably due to deposition of carbonate and carbonyl species on the surface of the catalyst.

It is clear that Au-Fe composite samples possessed catalytically active species and they offer a high activity from the beginning and are stable during catalytic tests. The nature of Au and Fe2O3 species responsible for the observed activity could be explained on the basis of characterization results. It was widely reported that properties of catalysts, such as particle size, active metal dispersion, surface area, and strength of interaction, have a great influence on the redox properties and reactivity [50]. The behavior of Au supported catalysts was mainly attributed to the electronic interaction between the Au and support [51] as well as the contribution from H2 pretreatment prior to the reaction.

It is clear that a dispute exists in the literature concerning the role of Fe2O3 support and active Au species in Au/Fe2O3 catalysts for CO oxidation. Liu et al. [52] reported that oxygen atoms of the Fe2O3 lattice do not participate in the reaction at room temperature. The catalytic oxidation takes place via the interaction of CO adsorbed on the surface of the Au particles and O2 adsorbed on the adjacent oxygen vacancies on the surface of the support. However, Tripathy et al. [53] studied the CO oxidation using the microcalorimetry technique and concluded that the lattice oxygen of Fe2O3 does play an important role in the reaction. Later, Guczi et al. [54] proposed that the high activity of Au-Fe2O3 samples for CO oxidation was due to the high O2 adsorption over the Fe2O3, which acted as the oxygen reservoir during the reaction. Okumura and Haruta [9] observed that Au metallic species in Au/α-Fe2O3 catalysts are very active. However, Hutchings et al. [10] and Khoudiakov et al. [11] reported that Au cation species in the Au/Fe2O3 catalysts are responsible for activity in CO oxidation. Park and Lee [12] also reported that oxidized gold species are more active than metallic Au. These authors indicated that reduction of Au oxide to metallic Au could be one of the reasons for the deactivation of Au/Fe2O3 catalysts in CO oxidation. In addition, they observed that Au oxide and iron oxide reduction peaks shifted to a higher temperature in the H2-TPR patterns, indicating that the strong interaction between Au species and Fe2O3 support. Very recently, Luengnaruemitchai et al. [55] reported a decrease of CO conversion from 95% to 44% at 333 K after oxygen pretreatment of the catalyst. It was suggested that the reason for the decrease of activity is the formation of surface oxygen species on the Au surface after the catalyst was treated with oxygen at low temperatures [56]. As we know, the reaction of CO with O2 requires both CO and O2 to be adsorbed on the catalyst surface. However, the reaction temperature for CO oxidation is low (273 K to 333 K), so the surface oxygen species will not be desorbed at these temperatures, and the oxygen species at the surface were occupying reactive sites. Thus, the strongly adsorbed surface oxygen species blocks the adsorption and diffusion of weakly adsorbed surface oxygen resulting in a decrease of CO oxidation activity. For this reason, we pretreated Au-Fe catalysts with H2 at 353 K for 1 h to remove surface oxygen species.

Okumura and Haruta [9] also indicated that spherical gold particles, which are strongly attached to the support by their flat planes, could be one of the reasons for their high activity in catalytic oxidation. Grunwaldt et al. [57] also made a very similar suggestion that the interaction of gold particles with the Fe2O3 support is physically and electronically strong because the shape of gold particles is spherical and flattened on one side. On the basis of catalyst preparation conditions adopted in our study and characterization results, we can suggest that Au3+ ions in the gold precursor deposited on the surface of the support and became Auδ+ due to the interaction between them. In case of catalysts which have lower Au loading (0.1 wt.% to 0.5 wt.%), the extent of interaction is high, which led to the formation of highly dispersive spherical Au particles (XRD, TEM, and XPS results). Further increase of Au loading to 1.0 wt.% resulted in the formation of agglomerated Au particles (XRD results). It was suggested that the agglomeration of Au particles could happen due to atom/clusters diffusion along the support and Ostwald ripening, where large particles grow at the expense of smaller particles due to the atomization and diffusion between particles [58].

The calcined Au-Fe nanocomposite samples contained both metallic Au and oxidic Au-Fe interactive species. It was already proven that both and Au0 species are active for CO oxidation [59]. After the catalyst was exposed to CO/O2 mixture, the Auδ+/Au0 ratio was decreased and after sufficiently long exposure only metallic Au will remain. It appears that the decrease in the activity that we and other researchers [11] observed with just calcined catalysts is due to the to Au0 transformation.

The superior catalytic activity of the 0.5Au-Fe catalyst could be explained as due to the formation of small Au-Fe particles dispersed on the support structure and also possessing of large surface area (Table 1). It was previously reported that the CO oxidation reaction normally takes place at the interface between gold cationic species and the iron sites [60]. The size of the interface between the Au particles and Fe2O3 is an important factor influencing the catalytic response of supported Au catalysts in reaction of CO oxidation [61]. It is generally accepted that the oxidation of CO over transition-metal oxides such as Fe2O3, MnO2, and CuO follows a Mars-van Krevelen mechanism, in which lattice oxygen is involved in CO oxidation and the reduced surface of the catalyst is reoxidized by CO/O2 feed mixture [62]. A high activity for CO oxidation can be achieved due to quick redox cycle through the spillover of reactive oxygen from the support to Au [63]. The XRD pattern of the spent 0.5Au-Fe catalyst (not shown) showed diffraction peaks due to hematite phase as a fresh sample, and no new diffractions peak due to Au oxides appeared. This observation clearly indicated that the Au-Fe catalysts have not undergone any structural changes under adapted reaction conditions.

4. Conclusions

In summary, we have successfully synthesized a mesoporous α-Fe2O3 material by simple hydrothermal method and the obtained α-Fe2O3 was used to prepare Au-Fe nanocomposites with different Au loadings (0.1 to 1.0 wt.%). The calcined and H2-pretreated Au-Fe nanocomposite samples showed a high catalytic activity for CO oxidation at low temperatures. CO was fully converted at reaction temperature as low as 313 K over the H2-pretreated catalysts. Catalytic performance of Au-Fe nanocomposites depends strongly on Au loading, among which 0.5 wt.% Au displayed the best performance. Further increase of the Au loading resulted in a decrease of CO oxidation activity. The characterization results suggest that both metallic Au and partially oxidized Au species are responsible for the catalytic oxidation of CO in addition to the Au particle size. The mesoporous nature with high specific surface area of the Au-Fe nanocomposites is also a contributing factor for the observed high catalytic activity. The Au-Fe nanocomposites also showed good reproducibility and stability in CO oxidation. The Au-Fe nanocomposites showed higher CO oxidation activity than other Au supported catalysts reported in the literature mainly due to enhanced interaction between Au nanoparticles and the mesoporous Fe2O3 support.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

Acknowledgments

This project was funded by the Deanship of Scientific Research (DSR) at King Abdulaziz University, Jeddah, under Grant no. 17/130/1435. The authors therefore acknowledge with thanks the DSR for the technical and financial support.

Supplementary Materials

Table S1: Quantification of elemental species using XPS analysis.

Table S2: Quantitative results of H2 consumption for catalysts in H2-TPR analysis.

  1. Supplementary Material