α-Fe2O3

Jul 8, 2008 - ... Fe0 nuclei could decompose CO molecules and release O species. ... For a more comprehensive list of citations to this article, users...
0 downloads 0 Views 402KB Size
8576

Langmuir 2008, 24, 8576-8582

Insights into the Oxidation and Decomposition of CO on Au/r-Fe2O3 and on r-Fe2O3 by Coupled TG-FTIR Ziyi Zhong,*,† James Highfield,† Ming Lin,‡ Jaclyn Teo,† and Yi-fan Han† Institute of Chemical Engineering and Sciences, 1 Pesek Road, Jurong Island, Singapore 627833, and Institute of Materials Research and Engineering, 3 Research Link, Singapore 117602 ReceiVed February 4, 2008. ReVised Manuscript ReceiVed May 5, 2008 CO oxidation and decomposition behaviors over nanosized 3% Au/R-Fe2O3 catalyst and over the R-Fe2O3 support were studied in situ via thermogravimetry coupled to on-line FTIR spectroscopy (TG-FTIR), which was used to obtain temperature-programmed reduction (TPR) curves and evolved gas analysis. The catalyst was prepared by a sonicationassisted Au colloid based method and had a Au particle size in the range of 2-5 nm. Carburization studies of H2-prereduced samples were also made in CO gas. According to gravimetry, for the 3% Au/R-Fe2O3 catalyst, there were three distinct stages of CO interaction with the Au catalyst but only two stages for the catalyst support. At low temperatures (e100 °C), only the Au catalyst had a rapid weight loss, which confirmed that CO reacted with highly active absorbed oxygen species and/or OH species which were associated with and promoted by the Au nanoparticles. Around 300 °C, both the catalyst and support samples experienced the reduction of Fe2O3 to Fe3O4, while above 400°C further reduction to FeO and Fe metal took place. Au played no part in the kinetics of Fe3O4 formation because lattice O mobility was rate-limiting. At higher temperature where Fe3O4 was further reduced to FeO and Fe0, the initially formed metallic Fe0 nuclei could decompose CO molecules and release O species. Both this coproduced O species and the lattice oxygen could react with CO molecules. Thus, the CO oxidation was not limited by the mobility of lattice oxygen, and the catalytic function of Au was revealed again. Carburization of metallic Fe, created by prereduction in H2, revealed a distinct weight gain at 350 °C corresponding to Fe3C formation, as subsequently confirmed by X-ray diffraction (XRD). Sustained carbon deposition ensued at 450 °C. In the cases of the 3% Au/γ-Al2O3 and Au/ZrO2 catalysts prepared by the same method, however, after exposure to CO in the same temperature range, no carbon deposit was observed, indicating that although Au nanoparticles could activate the absorbed oxygen molecules at low temperatures, they were not able to activate the lattice oxygen in the three catalyst supports or to dissociate the CO molecules directly.

Introduction In the last two decades, interest has grown dramatically in the application of gold (Au) in catalysis.1–5 This is due mainly to the notable discovery by Haruta et al.2 that nanodispersed Au is highly active in CO oxidation. Follow-up investigations have shown that Au is also useful in a number of catalytically important reactions under mild conditions.6–10 Its performance is not always easy to establish, being dependent on many factors, for example, the size and morphology of the Au particles, the existence of defects, the degree of coordinative unsaturation, oxidation state, type of catalyst support, the population density of surface OH groups and oxygen species, and the concentration of Cl- ion * To whom correspondence should be addressed. Email: Zhong_ziyi@ ices.a-star.edu.sg. † Institute of Chemical Engineering and Sciences. ‡ Institute of Materials Research and Engineering.

(1) Haruta, M.; Kobayashi, T.; Sano, H.; Yamada, N. Chem. Lett. 1987, 405. (2) Haruta, M.; Yamada, N.; Kobayashi, T.; Iijima, S. J. Catal. 1989, 115, 301. (3) Haruta, M. Catal. Today 1997, 36, 153. (4) Hutchings, G. J. Gold Bull. 1996, 29, 123. (5) Hutchings, G. J. Catal. Today 2002, 72, 11. (6) Fu, Q.; Saltsburg, H.; Flytzani-Stephanopoulos, M. Science 2003, 301, 935. (7) Kim, W. B.; Voitl, T.; Rodrigue-Rivera, G. J.; Dumesic, J. A. Science 2004, 305, 1280. (8) Enache, D. I.; Edwards, J. K.; Landon, P.; Solsona-Espiru, B.; Carley, A. F.; Herzing, A. A.; Watanabe, M.; Kiely, C. J.; Knight, D. W.; Hutchings, G. J. Science 2006, 311–362. (9) Chowdhury, B.; Bravo-Sua´rez, J. J.; Date´, M.; Tsubota, S.; Haruta, M. Angew. Chem., Int. Ed. 2006, 45, 412. (10) (a) Hughes, M. D.; Xu, Y. J.; Jenkins, P.; McMorn, P.; Landon, P.; Enache, D. I.; Carley, A. F.; Attard, G. A.; Hutchings, G. J.; King, F.; Stitt, E. H.; Johnston, P.; Griffin, K.; Kiely, C. J. Nature 2005, 437–1132. (b) Date´, M.; Okumura, M.; Tsubota, S.; Haruta, M. Angew. Chem., Int. Ed. 2004, 43, 2129. (c) Burch, R. Phys. Chem. Chem. Phys. 2006, 8, 5483.

contaminant when using HAuCl4 as precursor, and so forth.10,11 Although the number of publications in this area has been remarkable and important progress has been achieved in the last several years, there are still some unresolved issues with respect to the structure of active sites and the catalytic mechanism in Au catalysis. For example, in CO oxidation, while Goodman et al. proposed that partial nonmetallic character in nanosized Au particles may be responsible for the high activity,11 Gates et al. believed that ionized (cationic) Au is the active species.12 In contrast, Kung et al. reported that metallic Au is essential, alone or in combination with oxidized Au, to confer a good catalytic performance.13 The difficulty comes from the fact that the catalytic reaction is orchestrated by a number of factors and from that Au has unique properties. In many cases, it is difficult to phase out the individual role of these factors which are highly sensitive to the experimental conditions and preparation history.13 For example, when a Au/Fe2O3 catalyst is prepared by the coprecipitation (CP) method,12b the dried catalyst usually contains a lot of cationic Au species, but these revert to the metallic state after calcination above 300 °C. To identify the range of Au oxidation states involved, X-ray photoelectron spectroscopy (XPS) is often applied. Unfortunately, compared to many other noble metals such as Pt, Pd, and Rh, oxidic Au species are very sensitive to ambient atmosphere and unstable at elevated temperatures. As a result, techniques such as XPS may yield (11) (a) Chen, M. S.; Goodman, D. W. Science 2004, 306, 252. (b) Santra, A. K.; Goodman, D. W. J. Phys.: Condens. Matter 2002, 14, 31. (12) (a) Guzman, J.; Gates, B. C. J. Am. Chem. Soc. 2004, 126, 2672. (b) Hutchings, G. J.; Hall, M. S.; Carley, A. F.; Landon, P.; Solsona, B. E.; Kiely, C. J.; Herzing, A.; Makkee, M.; Moulijim, J. A.; Overweg, A.; Fierro-Gonzalez, J. C.; Guzman, J.; Gates, B. C. J. Catal. 2006, 242, 71. (13) Kung, M. C.; Davis, R. J.; Kung, H. H. J. Phys. Chem. C 2007, 111, 11767.

10.1021/la800395k CCC: $40.75  2008 American Chemical Society Published on Web 07/08/2008

CO Oxidation/Decomposition on Au/R-Fe2O3, R-Fe2O3

results that are only meaningful under analysis conditions, that is, in vacuo, and potentially irrelevant to the catalyst under reaction conditions. Furthermore, besides changes in the properties of Au itself, the catalyst support may also experience a series of textural and even phase changes during pretreatment.14 Thermogravimetry coupled to on-line FTIR spectroscopy (TGFTIR) is established as a convenient and versatile technique to investigate in situ catalyst activation and activity profiling under flow conditions,15 and it has not yet been exploited significantly in the investigation of Au catalysts. The thermobalance provides accurate quantitative analysis through weight change, while FTIR provides evolved gas analysis. Together, from different aspects, TG-FTIR can provide insight into thermally activated processes involving the surface and/or bulk of the solid sample under inert or reactant gas flow. Herein, we report the use of TG-FTIR to study the interaction between CO and Au/R-Fe2O3, with the aim of rationalizing the unexpected detection of carbon deposits in transmission electron microscopy (TEM) micrographs after relatively low temperature exposure to CO (vide infra). Precise weight loss data coupled with relative changes in evolved gas composition have been compared for 3% Au/R-Fe2O3 and the R-Fe2O3 control, subjected to various prereduction treatments: directly in CO or prereduction in H2 followed by carburizing treatment in CO. Key differences are attributable to the presence of Au nanoparticles, which accelerate the catalysis of CO oxidation (nondissociative adsorption) in two temperature ranges. The first range is below 100 °C where absorbed oxygen species and/or OH groups are present, and the second one is at the high-temperature range (above 400 °C) where the CO decomposition (dissociated) reaction has been initiated after metallic Fe is formed that can supply the oxygen for oxidation of CO. In the medium temperature range, where the transition from R-Fe2O3 to R-Fe3O4 takes place, Au nanoparticles do not accelerate the CO oxidation, as it is controlled by the supply of lattice oxygen. These results clearly indicate that the Au particles can promote the oxidation of CO with oxygen in the whole temperature range (from RT to 700 °C), can activate the absorbed oxygen species, but cannot activate the lattice oxygen in the catalyst supports, such as R-Fe2O3, γ-Al2O3, and ZrO2, directly.

Experimental Section All chemicals were purchased from Aldrich except for γ-Al2O3 (Puralox, UR-160, with a surface area of ca. 160 m2/g) which was a gift of SASOL. One dimensional (1D) FeO(OH) nanorods were synthesized by a hydrothermal method using tetraethylammonium hydroxide (TEAOH) as the structure director, and they were further converted to a 1D and porous R-Fe2O3 support by calcination at 300 °C for 1 h.16 For supported Au catalyst preparation (3% Au/RFe2O3, 3% Au/γ-Al2O3, and 3% Au/ZrO2), we employed our recently invented Au colloid based method.17 In this technique, very small Au colloids were first produced by the reduction of HAuCl4 with NaBH4 in the presence of a capping reagent (lysine) and the catalyst support in a pH range of 5-8.5. At the same time, ultrasonic irradiation (sonication) was applied during the reduction to facilitate the deposition of the Au colloids on the catalyst support efficiently. The catalysts were washed several times with DI H2O, dried at 60 °C, and calcined at 300 °C to remove the capped lysine molecules (14) Hodge, N. A.; Kiely, C. J.; Whyman, R.; Siddiqui, M. R. H.; Hutchings, G. J.; Pankhurst, Q. A.; Wagner, F. E.; Rajaram, R. R.; Golunski, S. E. Catal. Today 2002, 72, 133–144. (15) Highfield, J.; Loo, L. S.; Zhong, Z.; Grushko, B. Carbon 2007, 45, 2597. (16) Zhong, Z.; Ho, J.; Teo, J.; Shen, S. C.; Gedanken, A. Chem. Mater. 2007, 19, 4776. (17) (a) Zhong, Z.; Lin, J.; Teh, S. P.; Teo, J.; Dautzenberg, F. M. AdV. Funct. Mater. 2007, 17, 1402. (b) Zhong, Z.; Teo, J.; Lin. M.; Ho, J. Top. Catal. 2008, in press.

Langmuir, Vol. 24, No. 16, 2008 8577

Figure 1. TEM images of 3% Au/R-Fe2O3 calcined at 300 °C in air for 1 h: (A) low magnification and (B) high-resolution TEM (HRTEM) image showing the detailed structure of one Au particle on R-Fe2O3 taken along the [011] zone axis. (The scale bar is 20 nm in (A).)

before the carbonization reaction, the temperature-programmed reduction (TPR) experiments, and the in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) study. The measurement of the catalytic activity for CO oxidation was carried out in a fixed-bed microreactor. Prior to the test, the catalyst (10 mg) mixed with 20 mg of R-Al2O3 was activated in air at 300 °C for 1 h. After cooling to room temperature, a reactant gas containing 1% CO in air was passed through the catalyst bed. The outlet gas was analyzed on-line with gas chromatography (GC) (Shimadzu-14B). During the measurement, the gas hourly space velocity (GHSV) was varied from 40 000 to 400 000 h-1 depending on various catalysts, so as to obtain a CO conversion of ∼10% for calculation of the specific activity. The carbonization reactions on the three supported Au catalysts and on the catalyst support (1D R-Fe2O3) were carried out in a quartz fixed-bed reactor which was vertically set in an electronic furnace using premixed 5% CO in helium as the reactant gas (National Oxygen Company, Singapore). The temperature-programmed reduction (TPR), using 2.5% CO in helium and 5% H2 in helium for the prepared catalyst support (1D R-Fe2O3) and the Au/R-Fe2O3 catalyst, respectively, was conducted on a Setaram Setsys 1200 thermobalance, coupled through a heated transfer line/gas cell to a Digilab Excalibur FTS-3000 FTIR spectrometer, operating in the “kinetics” mode of the Resolutions Pro software package (the setup was described in our previous report15). An FTIR absorbance spectrum at 4 cm-1 resolution in the range of 4000-400 cm-1 (midrange MCT detector) was stored every minute (120 co-added scans), from which recovery of CO2 was monitored by the area under the bending vibration centered at ∼670 cm-1. Conversion of CO was monitored by diminution in the area under the fundamental stretching envelope centered at ∼2140 cm-1. The DRIFTS study for CO adsorption was conducted on a modified Harricks model HVDR2 reaction cell to allow CO gas (5% CO in helium) to flow through the catalyst bed (200 mg) during spectra acquisition. The in situ spectra were recorded on a Bio-Rad FTIR 3000MX spectrometer with a resolution of 4 cm-1. The morphology and size of the support and the Au catalyst before and after the carbonization were observed employing a transmission electron microscope (Tecai TF20 Super Twin, 200 kV). Powder XRD analysis was conducted on a Bruker D8 Advance X-ray diffractometer with Cu KR1 radiation.

Results and Discussion The TEM images of the calcined 3% Au/R-Fe2O3 catalyst at 300 °C in air are shown in Figure 1. The Au particles (black dots) were well dispersed on the support and had a size distribution in the range of 2-5 nm. From the high-resolution TEM (HRTEM) image shown in Figure 1B, the lattice distances of the Au{111} and Au{200} planes can be resolved, while the measured lattice spacing of R-Fe2O3 is 0.25 nm, corresponding to the {011} plane. Due to the large lattice mismatch between the Au particle and the R-Fe2O3 surface, the interaction resulted in the formation of

8578 Langmuir, Vol. 24, No. 16, 2008

Zhong et al.

Figure 3. XRD patterns for (a) R-Fe2O3 and (b-d) 3% Au/R-Fe2O3 discharged after exposure to 5% CO at (a,c) 500 °C for 1 h and at (b) 400°C and (d) 600°C for 1 h, respectively. For comparison, the standard XRD patterns of Fe3O4, Fe3C, and metallic Fe are shown. Figure 2. 3% Au/R-Fe2O3 catalyst calcined at 300 °C and exposed to 5% CO in He at (A) 300 °C, (B) 400 °C, and (C) 600 °C for 1 h. For comparison, a detail from the corresponding R-Fe2O3 support, calcined at 300 °C and exposed to 5% CO in He at 600 °C for 1 h, is shown in (D).

many atomic steps on the Au surface beside the exposed {111} and {200} planes. From the junction of the HRTEM image and the angle between step planes and {111}, {200} planes, stepped surfaces are identified as {122} and {322} planes. The {122} plane contains a (111) step and (111) terrace, while the {322} plane has a (111) step and (100) terrace. The high density of Au atoms on steps in such an arrangement would most probably contribute to the high catalytic performance of the Au particles. In the 3% Au/γ-Al2O3 and 3% Au/ZrO2 catalysts, the Au particle size was 2-5 and 2-8 nm, respectively, after the calcination at 300 °C in air. The larger size of the Au particles in the 3% Au/ZrO2 catalyst is due to the relative low isoelectric point (IEP) of ZrO2, which has significant influence on the final Au particles in the catalysts.17a In a basic CO oxidation catalytic analysis, the specific rates of 3% Au/R-Fe2O3, 3% Au/ZrO2, and 3% Au/γ-Al2O3 in a reactant gas of 1% CO in air at 30 °C were measured to be 4.33 × 10-4, 2.38 × 10-5, and 5.5 × 10-6 molCO · gAu-1 · s-1, respectively. The above 3% Au/R-Fe2O3 catalyst was further carbonized in 5% CO in helium in the range of 300-700 °C (Figure 2). After exposure to 5% CO in helium for 1 h at 300 °C, there was no carbon deposited on the catalyst (Figure 2A); however, after the reaction temperature was increased to 400 °C, layers of carbon were observed on the Fe or on its oxide particles, with thicknesses of ∼3-4 nm. Further increase of the reaction temperature to 500, 600, and 700 °C led to the formation of more carbon deposits evidenced by a growth in thickness of the carbon layers (ca. 10 nm in Figure 2C). Similar results were observed on R-Fe2O3 after exposure to 5% CO (Figure 2D). It should be pointed out that even after screening many areas during the TEM observations, no carbon deposits having structures such as carbon nanotubes, nanofibers, and layered carbon graphene sheets were seen to be growing out directly from the Au particles, indicating that these carbon deposits were grown only on Fe and not on Au particles. This is quite different from the cases of supported Fe, Ni, and

Co catalysts18 and the cases of supported Au catalysts on Si/SiO2 and on Al2O3-SiO2 using ethylene and acetylene as reactant gases, respectively.19 The XRD results after carburization in 5% CO at 400 and 500 °C for 1 h show that only Fe3O4 and Fe3C were present, with more carbide present in the Au catalyst (Figure 3). No free Fe or intermediate FeO phases were detected, suggesting that these were rapidly reduced and carburized autocatalytically, with metallic Fe being active in CO dissociation. After carburization at 600 °C in 5% CO, all the peaks corresponding to Fe3O4 and Fe3C, except that of metallic Fe, had disappeared (Figure 3d), indicating Fe3C was an intermediate phase during the reaction. To provide more definitive details of the CO interaction with the 3% Au/R-Fe2O3 catalyst and the R-Fe2O3 support, these were further investigated under a flow of 2.5% CO in helium by TGFTIR. Gravimetric temperature-programmed reaction curves are shown in Figure 4A, while the evolved gas profiles for the same experiments are given in Figure 4B. A striking feature, absent in the case of R-Fe2O3, was the rapid loss of ∼2 wt % for 3% Au/R-Fe2O3 at ∼50 °C (red curve in Figure 4A). This was confirmed as primarily genuine reaction, and not simply desorption, by the associated consumption of CO and evolution of CO2, as seen in Figure 4B. Since at this temperature it is impossible to activate the lattice oxygen in R-Fe2O3 for the oxidation reaction, it is clear that it represents the reaction of CO with labile surface OH species and/or preadsorbed O2 species, whose formation and reduction are intimately linked to the presence of Au. The next weight loss (close to 3%) and associated CO2 evolution, beginning at around 200 °C and completing by ∼320 °C, were very similar for both samples and correspond to the reduction of R-Fe2O3 to Fe3O4. It is quite clear that Au does not catalyze this process, despite a claim to the contrary in the literature.20 Beyond 320 °C, the Fe3O4 phase was further reduced to FeO and/or Fe, showing a maximum rate (and corresponding CO2 evolution peak) at ∼530 °C. Although the degree of reduction in CO appears to be far from complete, as seen by comparing (18) Zhong, Z.; Chen, F.; Xiong, X.; Soon, H. C.; Lin, J.; Tan, K. L. J. Nanosci. Nanotechnol. 2004, 4(1/2), 183. (19) (a) Lee, S. E.; Yamada, M.; Miyake, M. Carbon 2005, 43, 2654. (b) Bhaviripudi, S.; Mile, E.; Steiner, S. A., III; Zare, A. T.; Dressehaus, M. S.; Belcher, A. M.; Kong, A. M. B. J. Am. Chem. Soc. 2007, 129, 1516. (20) Ilieva, L. I.; Andreeva, D. H.; Andreev, A. A. Thermochim. Acta 1997, 292, 169.

CO Oxidation/Decomposition on Au/R-Fe2O3, R-Fe2O3

Langmuir, Vol. 24, No. 16, 2008 8579

Figure 5. On-line FTIR analysis of gases evolved during the reduction of R-Fe2O3 (black curve) and 3% Au/R-Fe2O3 (red curve) in 2.5% CO in He and the subsequent reoxidization in 10% O2 in helium at 350 °C. Reduction/carburization was made with a ramping rate of 10 °C/min to 350 °C and a hold at 350 °C for 1.5 h. After purging with N2 for 20 min, 10% O2 in N2 was introduced at 350 °C to oxidize the carbon deposit (evolution peak of CO2 from oxidation of precarburized R-Fe2O3 is displaced for clarity).

Figure 4. Recorded (A) TG and (B) FTIR profiles for R-Fe2O3 (black curve) and 3% Au/R-Fe2O3 (red curve) in 2.5% CO/He [TG reduction profile of R-Fe2O3 (blue) in pure H2 shown for comparison in (A)]. Samples (20 mg) were exposed to 2.5% CO (or pure H2) at 30 mL/min, and then the temperature was ramped to 700 °C at a ramping rate of 10 °C per min.

the weight loss curve for Fe2O3 in H2 (blue curve, Figure 4), this may be an exaggerated difference, since any Fe formed would be expected to carburize, such that the overall weight change is the product of partially canceling factors. Both the TG and FTIR data show that the final reduction stage is accelerated by the presence of Au. Their correspondence also suggests that most of the CO2 was derived from genuine reduction, that is, extraction of residual lattice oxygen, and not from the Boudart (disproportionation) reaction: 2CO f CO2 + C. This further illustrates the role of lattice oxygen in CO oxidation at this temperature range. By 700 °C, the overall weight losses for 3% Au/R-Fe2O3 and R-Fe2O3 were 17.5% and 13.1%, respectively, as compared to a theoretical weight loss for the reduction to FeO and metallic Fe of 10% and 30%, respectively (assuming Au is present as Au0). To verify the occurrence of carbon deposition at low temperature, prereduction in 2.5% CO in He (ramp to 350 °C at a rate of 10°C/min and hold for 1.5 h) was followed by intermediate purging in N2 and introduction of 10% O2 in N2 at 350 °C. FTIR profiles for gas compositional changes during the full experiments are shown in Figure 5. The two distinct lowtemperature reduction features described previously (Figure 4) for the Au catalyst were again observed, with CO consumption and peak CO2 evolution at ∼3 min (∼60 °C) and ∼27 min (∼300 °C), with the latter also evident and of similar intensity to that in the study of R-Fe2O3 support. During the subsequent temperature hold at 350 °C, the intensity of the CO2 peak from the Au catalyst, although decaying gradually, nevertheless

Figure 6. TG profiles for R-Fe2O3 (black) and 3% Au/R-Fe2O3 (red) for prereduction in H2, followed by a N2 flush and exposure to CO. In detail, samples were first reduced in pure H2 at a flow rate of 90 mL/min at 350 °C. After purging with N2 at 350 °C for 20 min, they were exposed to 2.5% CO/He for 40 min at 350 °C (Fe3C formation). A ramp to 450 °C (10 °C/min) and hold for 1 h was then applied (coke formation).

generally exceeded the level evolved from the support. This confirmed the promoting effect of Au in the reduction by CO of intermediate Fe3O4 to FeO and/or Fe even at 350 °C. In reoxidation, the CO2 peaks derived from deposited carbon and/ or iron carbide(s) were virtually instantaneous and of similar intensity in both samples (trace for R-Fe2O3 offset for clarity in Figure 5). Carburization evidently occurs contemporaneously with reduction under CO. To give more information on the nature of the associated carbon, experiments were performed in which the samples were prereduced in 5% H2/N2 at 350 °C and, after intermediate flushing, exposed to 2.5% CO/He between 350 and 450 °C. As seen in Figure 6, both the Au catalyst and the support lost close to 30 wt % relative in H2, equal to the theoretical loss for the reduction of Fe2O3 to the metallic state. The low level of Au made it difficult for any definitive conclusion as to its initial oxidation state. Upon exposure to CO at 350 °C, both samples showed

8580 Langmuir, Vol. 24, No. 16, 2008

Zhong et al.

because lattice oxygen mobility is rate-limiting in the reduction process:

g250 °C,

R-Fe2O3 f Fe3O4 + O

(2)

Above 350°C (see Figure 4), Fe3O4 is further reduced to FeO and metallic Fe, leading to CO dissociation, carbon deposition (see Figure 5) and/or Fe3C formation (see Figure 6) depending on the temperature:

g350 °C,

Figure 7. TEM micrographs of 3% Au/γ-Al2O3 after 1 h exposure to 5% CO/He at (A) 300 °C, (B) 400 °C, (C) 500 °C, and (D) 700 °C.

very similar profiles with a distinct initial weight gain, relative to the starting weight (Fe2O3) of 4.7% for 3% Au/R-Fe2O3 and 4.9% for R-Fe2O3. These were very close to the theoretical mass gain for the reaction 3Fe + C ) Fe3C of 7.0% against Fe or 4.9% against Fe2O3. Thus, any metal is converted to Fe3C under relatively mild conditions, in agreement with the XRD result (Figure 3). The ramp to 450 °C resulted in an acceleration of the weight gain rate, until it reached a relatively constant value that was maintained for most of the final hold. This must be attributed to carbon deposition. The current understanding is that Fe3C must decompose back into metallic iron and phase-separated carbon.15 Unfortunately, this decomposition results in no weight change and so is not directly visible in TG. Disintegrated Fe metal then sustains continuous carbon buildup until the particle is fully encapsulated and coking stops. This provides a rationale for the carbon layer buildup seen in HRTEM (Figure 2). As a comparison, the highly dispersed 3% Au/γ-Al2O3 and 3% Au/ZrO2 catalysts were also exposed to 5% CO in He in the fixed-bed reactor in the range of 300-700 °C. In contrast to Fe2O3-supported Au, no carbon deposition was observed in either of these catalysts (Figure 7). Since neither γ-Al2O3 nor ZrO2 is reducible under the applied conditions, this was to be expected. However, these experiments also verified that Au itself is inactive in CtO dissociation and carbon deposition. The key factor underpinning the role of Au in CO oxidation, coupled with reduction of R-Fe2O3, is the availability (lability) of oxygen species:

e100 °C,

CO + O f CO2

(1)

Well below 100 °C (see Figure 4), the presence of highly active oxygen species on the surface of 3% Au/R-Fe2O3 caused rapid oxidation of CO. They may be, or are derived from, surface OH groups/adsorbed H2O,21 absorbed oxygen species, or a combination of these. In the temperature range up to 350°C (see Figure 4), the Au nanoparticles do not display any catalytic function, probably (21) Gavril, D.; Georgaka, A.; Loukopoulos, V.; Karaiskakis, G.; Nieuwenhuys, B. E. Gold Bull. 2006, 39(4), 192.

CO + 3Fe f Fe3C + O

(3)

While carbon derived from metastable Fe3C may further reduce the surrounding oxide, the coproduced surface O atom (from CO dissociation) is labile, and the catalytic function of Au nanoparticles for CO oxidation becomes obvious again. From these results, it is clear that the CO oxidation on the supported Au catalyst can be divided into two stages. One is the interaction of CO with surface oxygen species (below 100 °C), and the other is the CO interaction with the bulk oxygen species or the lattice oxygen (above 100 °C). Meanwhile, the significance of the catalytic power of the Au nanoparticles is dependent on the availability and on the types of oxygen species. In the first stage (below 100 °C), because of the existence of highly active surface oxygen species, the catalytic function of the Au particles is fully expressed, while in the second stage and in the middle temperature range (100-350 °C), due to the lattice oxygen mobility being low, the Au nanoparticles cannot accelerate the reaction of CO oxidation. As one reaches the high-temperature range (above 350 °C), the formation of some metallic Fe sites results in a direct reaction with CO, thus releasing labile oxygen, and the catalytic function of the Au particles is revealed again. The coupled TG-FTIR technique can monitor the reaction progress of CO oxidation by analyzing the evolved CO and CO2 concentration and by measuring the weight changes from low to high temperatures, but it cannot provide detailed information on the status of the adsorbed CO molecules on the supported Au catalysts. CO adsorption on Au surfaces has been characterized by a number of spectroscopic techniques including in situ CODRIFTS for determining the Au oxidation states and the status of the absorbed CO molecules.22 Generally, the electronic property of Au surface is significantly affected by supporters. Schubert et al.23 have classified the supports into “active” and “inert” material according to their role for oxygen supply during reaction. Among active supporters including R-Fe2O3 and CeO2, the vibrational stretching of the linear absorbed CO on Au bonds was observed in two regions: 2130-2140 and 2000-2100 cm-1, corresponding to CO-(Au+)n and CO-(Au0) species, respectively. The former usually appears on the untreated catalysts prepared by the deposition-precipitation and coprecipitation methods and disappears after reduction in H2. More interestingly, those Au particles could be reoxidized by the oxygen from the supporters under mild reaction conditions. In particular, the oxidation-reduction of Au nanparticles on R-Fe2O3 is recognized to associate its remarkable activity in CO oxidation. Our previous DRIFTS results24 showed that the CO absorbance peak was located at 2100 cm-1 for Au/γ-Al2O3 and a similar peak at 2107 cm-1 was observed on a Au/R-Fe2O3 catalyst prepared by our method (not shown for the sake of brevity), indicating that the CO molecules were linearly absorbed on metallic Au species.25 (22) Gates, B. C. Chem. ReV. 1995, 95, 511. (23) Schubert, M. M.; Hackenberg, S.; van Veen, A. C.; Muhler, M.; Plzak, V.; Behm, R. J. J. Catal. 2001, 197(1), 113. (24) Han, Y. F.; Zhong, Z.; Ramesh, K.; Chen, F.; Chen, L. J. Phys. Chem. C 2007, 111, 3163. (25) Schubert, M. M.; Haring, T. P.; Brath, G.; Gasteiger, H. A.; Behm, R. J. Appl. Spectrosc. 2001, 55(11), 1537.

CO Oxidation/Decomposition on Au/R-Fe2O3, R-Fe2O3

This is quite different from the supported Au catalysts prepared by the deposition-precipitation and the coprecipitation methods, in which usually CO absorbed on oxidic Au species could be observed.26–28 The existence of the metallic Au species in our samples is also in agreement with our previous XPS measurements.16 This result is logical for supported Au catalysts prepared by our method, as NaBH4 was used as the reducing agent and the catalyst was calcined at 300 °C for 1 h. Both experimental techniques will lead to the formation of the metallic Au species. However, the possibility that the cationic Au species acts as an active site still cannot be excluded, as the trace amount of Au cations may just locate at the interface between the Au particles and the catalyst support that favors CO oxidation,29 or the Au cations may be associated with the active oxygen species. After being purged with CO gas in the DRIFTS study, the active oxygen might be removed and thus the Au cations were reduced to metallic Au again. Gates and Guzman30 observed that the CO oxidation activity increased with the cationic Au fraction on a Au/MgO catalyst, and Hutchings et al.31 reported that in a series of Au/ R-Fe2O3 catalysts, the higher the activity, the higher the concentration of cationic Au the catalyst had. However, in order not to diverse the focus in this study, we have no intention to discuss this issue deeply, as more experimental evidence and discussions on this issue can be found in a recent review article by Gates and Fierro-Gonzalez.32 Although the above studies have revealed good insights into the interactions between CO, the Au catalyst, and the pure support by TG-FTIR, and the adsorbed status of CO molecules on catalyst surface by DRIFTS, a chief difficulty remains, namely, the identification of the active surface oxygen species that is most relevant to the catalytic properties of Au or the route to activate the molecular oxygen for CO oxidation in the low temperature range.13 There is no final conclusion on this issue as of today, but important progress has been made in the last several years. For instance, Fu et al. observed a low reduction peak near 100 °C in a H2-TPR experiment on supported Au/CeO2 catalyst and confirmed that Au facilitated the activation of surface oxygen species on ceria.33 Ilieva et al. confirmed the existence of F centers (electrons in cation vacancies) in Au/ZrO2 by electron spin resonance (ESR)34 and observed enhanced dissociative adsorption of water on Au/R-Fe2O3.20 Furthermore, first-principle calculations show that O2 and H2O may coadsorb on Au/MgO, leading to the formation of a hydroperoxyl-like complex.35 Anderson et al reported that, on Aun/TiO2(110) catalyst, the CO oxidation activity is closely related to the ability of Au clusters to bind molecular O2.36 On the same catalyst, Hu et al.37 showed by density functional theory that molecular O2 could be adsorbed in the presence of surface OH groups. Recently, Bokhoven et (26) Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Andreeva, D.; Tabakova, T. J. Catal. 1999, 188, 176. (27) Boccuzzi, F.; Chiorino, A.; Manzoli, M.; Lu, P.; Akita, T.; Ichikawa, S.; Haruta, M. J. Catal. 2001, 202, 256. (28) Venkov, T.; Klimev, H.; Centeno, M. A.; Odriozola, J. A.; Hadjiivanov, K. Catal. Commun. 2006, 7, 308. (29) Haruta, M. CATTECH 2002, 6, 102. (30) Guzman, J.; Gates, B. C. J. Am. Chem. Soc. 2004, 126, 2672. (31) Hutchings, G. J.; Hall, M. S.; Carley, A. F.; Landon, P.; Solsona, B. E.; Kiely, A.; Herzing, A.; Makkee, M.; Moulijin, J. A.; Overweg, A.; Fierro-Gonzalez, J. C.; Guzman, J.; Gates, B. C. J. Catal. 2006, 242, 71. (32) Fierro-Gonzalez, J. C.; Gates, B. C. Catal. Today 2007, 122, 201. (33) Fu, Q.; Kudriavtseva, S.; Saltsburg, H.; Flytzani-Seephanopoulos, M. Chem. Eng. J. 2003, 93, 41. (34) Ilieva, L.; Sobczak, J. W.; Manzoli, M.; Su, B. L.; Andreeva, D. Appl. Catal., A 2005, 291, 85. (35) Bongiorno, A.; Landman, U. Phys. ReV. Lett. 2005, 95, 106102. (36) Lee, S.; Fan, C.; Wu, T.; Anderson, S. L. J. Chem. Phys. 2005, 123, 124710. (37) Liu, L. M.; McAllister, B.; Ye, H. Q.; Hu, P. J. Am. Chem. Soc. 2006, 128, 4017.

Langmuir, Vol. 24, No. 16, 2008 8581 Scheme 1. Reaction Steps of CO with r-Fe2O3 and the Role of Nanosized Au Particlesa

a In each step, the theoretical weight loss (Wt) is indicated. The catalytic function of Au nanoparticles for CO oxidation is not evident in step 2, probably due to restricted mobility of lattice oxygen below 300 °C.

al.38 reported that, in Au/Al2O3 catalyst, metallic Au0 could activate O2 molecules directly at 298 K and form AuxOy species, as determined by in situ time-resolved high energy resolution fluorescence detected (HERFD) X-ray spectroscopy. This oxidized Au species is short-lived under the reaction conditions but can react with CO directly to form CO2.38 Therefore, in the Au/R-Fe2O3 catalyst, the metallic Au particles probably can activate molecular O2 in the presence of adsorbed H2O or surface OH groups to form AuxOy-like species or Au+-O2--like species that are highly active for CO oxidation. In a very recent DRIFTS study on Au/TiO2 catalyst done by Date´ et al., the role of moisture was proposed to be in the activation of molecular oxygen and in the decomposition of carbonate species, which have accumulated on the catalyst surface, thereby blocking the active sites.39 In this work, TG-FTIR has verified the existence of these active oxygen species at low reaction temperatures (below 100 °C) and given a rough idea of their concentration in a single preparation. However, this method and the DRIFTS results cannot give more structural information about these active oxygen species. More focused investigations of the “quantitative reactive” type explored here, together with very sensitive surface spectroscopic characterization to the surface oxygen species under in situ conditions, are clearly necessary but beyond the scope of the present article.

Conclusions In summary (Scheme 1), nanosized and highly dispersed Au particles on R-Fe2O3, γ-Al2O3, and ZrO2 have been prepared by a sonication-assisted colloid-based method, and various interactions between CO and the surface and bulk of oxygen species in the catalyst have been evaluated by TG-FTIR. At very low reaction temperatures, the reaction of CO with active surface oxygen species is catalyzed by Au nanoparticles. Au may have also been responsible for creating these active surface oxygen species in the first instance, although it is not yet clear if these are present on Au itself or have spilled over to the adjacent support surface (step 1). Above 250 °C, CO reacts directly with lattice oxygen in R-Fe2O3 and Fe3O4 (steps 2 and 3). In the latter case, Au has no influence, because at that temperature the ratedetermining step is the mobility of lattice oxygen. In step 4, the promoting effect of Au is once again expressed, because once the metallic Fe is available, the CO dissociation reaction is initiated and Fe3C is formed. Labile surface oxygen is available once (38) van Bokhoven, J. A.; Louis, C.; Miller, J. T.; Tromp, M.; Safonova, O. V.; Glatzel, P. Angew. Chem., Int. Ed. 2006, 45, 4651. (39) Date´, M.; Imai, H.; Tsubota, S.; Haruta, M. Catal. Today 2007, 122, 222.

8582 Langmuir, Vol. 24, No. 16, 2008

more for CO oxidation, thereby initiating the Au particles to become active again (step 4). Although steps 3 and 4 are distinguished in Scheme 1, no evidence exists from either TG or XRD for the formation of bulk wu¨stite (FeO). Reduction of Fe3O4 to metallic Fe may proceed in a concerted manner, with initially formed Fe promoting further reduction of the surrounding oxide by carbon. In this carbonization process, the Fe3C phase is identified by XRD and TG analysis to be the intermediate product. Differing from the case of the Au/R-Fe2O3 catalyst, no carbon deposit is observed on the 3% Au/γ-Al2O3 and Au/ZrO2 catalysts after exposure to CO in the temperature range from 300 to 700 °C. In the latter two catalysts, since the catalyst supports are not reducible in the temperature range, no metallic Al or Zr is produced except for the metallic Au particles. These different behaviors for CO decomposition on the three catalysts indicate that, although the Au nanoparticles can activate the absorbed oxygen molecules, which are highly active for CO oxidation, they cannot activate the lattice oxygen in the three catalyst supports and cannot dissociate the CO molecules directly. In the case of the Au/R-Fe2O3 catalyst, the appearance of the metallic Fe is the

Zhong et al.

turning point where the CO dissociation reaction can take place on the metallic Fe sites, thus releasing O2 and accelerating the CO oxidation. It should be pointed out that, though this TG-FTIR technique can monitor the interaction of CO molecules with the supported Au catalysts in a wide temperature range, it is not able to identify the status of the absorbed CO molecules and the low-temperature active oxygen species. Combining DRIFTS and TG-FTIR with some other surface spectroscopic techniques that are able to determine the status of surface active oxygen species under reaction conditions should be an effective route to understand the nature of Au catalysis and some other gas-phase catalytic reactions. Acknowledgment. This research was supported by the Agency for Science, Technology and Research in Singapore. Z.Z. thanks Dr. Frits M. Dautzenberg for his helpful advice, Ms. Z. Wang for technical assistance, and Drs. P. K. Wong, Armando Borgna, and Keith Carpenter for their kind support of this project. LA800395K