An analytical scanning transmission electron microscopy study of the support effects on Mn-promoted Co Fischer–Tropsch catalysts

Abstract

The metal–support interaction in supported heterogeneous catalyst systems has been long known to affect the activity and selectivity of the catalyst. In this paper, we will combine scanning transmission electron microscopy with electron energy-loss spectroscopy to investigate the interaction between the Mn promoter, the Co nano-catalysts, and the SiO₂ support. By comparing these results with similar catalysts on TiO₂ supports, we will discuss the effect of metal–support interaction and interfacial oxygen vacancies on the catalyst morphology.

---

Introduction

The demand for alternative energy sources to crude-oil based fuels has been increasing over the last few decades, in particular at times of high petroleum prices, decreasing petroleum reserves and soaring petroleum consumption. Fischer–Tropsch (FT) synthesis could be one path to alleviate this looming energy crisis, by utilizing natural gas and converting it into synthetic diesel fuelvia heterogeneous catalysts. For this reason, extensive investigations have been conducted on the Co-based FT catalysts in order to improve the performance of the reaction by steering selectivity to longer chain hydrocarbons for use as diesel fuel, by using different supports or adding promoters. The roles of the TiO₂, SiO₂ and Al₂O₃ supports have been studied in previous work.^1–8 Moreover, it is found that Mn is an effective promoter for increasing the selectivity of the Co-based catalysts on both TiO₂ and SiO₂ supports.^9,10

Yet, the fundamental, atomic-scale mechanism of the promoter-element on the catalysts' activity and selectivity remains still unclear. Morales et al. proposed that the improved performance of Mn-promoted Co catalysts was due to a Mn–Co compound formed on the surface of the catalyst.^1,11 As these Mn-oxide species cannot be completely reduced at the temperatures regularly used for the FT-reaction, it has been suggested that the generation of mixed Co–Mn oxides could potentially create active sites with better selectivity.^12,13

It was previously proposed that the catalyst synthesis (i.e. dry impregnation or strong electrostatic adsorption) can play an important role in its final performance since the promoter should be closely associated with the active metal to be effective. Dry Impregnation (DI) is universally used to distribute the precursor onto the support.¹⁴ It is considered as a simple method for randomly dispersing metals onto the support and strong precursor–support interaction is not guaranteed. On the other hand, strong electrostatic adsorption (SEA) provides a strong interaction between the precursor and the support, which is based on the relative charges of the support and the metal precursor. The SEA method was shown to be very effective in selectively depositing precursors onto a specific site on the substrate.¹⁵

In particular, for the Mn promoted Co-based catalysts on TiO₂ supports, we have previously shown that Mn only adsorbs onto the Co particles using SEA, while a random distribution of Mn was found in systems synthesized using DI.¹⁶ However, it was shown that after reduction the Mn promoter had migrated to the Co/TiO₂ interfaces for both synthesis methods.^16,17 The strong metal–support interaction (SMSI)^18–20 was shown to be responsible for the formation of oxygen vacancies at the Co/TiO₂ interface, which then act as anchoring sites for the mobile Mn atoms during the reduction process.¹⁷

In contrast, SiO₂ is considered as an inert support and no SMSI is expected between the SiO₂ supports and the supported catalyst particles. As a result, we do not expect to find oxygen vacancies at the Co/SiO₂ interface, and therefore we do not expect preferential migration of Mn towards the Co/SiO₂ interface. Mn-promoted Co/SiO₂ synthesized using either DI or SEA will thus serve as important benchmark systems to examine the effect of metal–support interaction and the effect of oxygen vacancies on the elemental distribution and morphology of the Mn-promoted Co catalyst.

In this paper, we will utilize high-resolution Z-contrast imaging in a scanning transmission electron microscope (STEM) in combination with spatially resolved electron energy-loss spectroscopy (EELS) to study the migration of Mn in promoted Co-based catalysts on SiO₂ supports after calcination and reduction. We will then compare the differences in particle morphology and elemental distribution between catalysts on SiO₂ and TiO₂ supports; in particular the presence of oxygen vacancies will be examined. The mechanism of Mn migration and its effect on the FT catalyst performance will be discussed.

Sample preparation and experimental methods

A Co₃O₄/SiO₂ (∼5 wt% Co) supported material was prepared by DI of a Co(NO₃)₂·6H₂O (Aldrich, ≥98%) precursor onto Aerosil 300 SiO₂. Briefly, Co(NO₃)₂·6H₂O was dissolved in an appropriate amount of deionized H₂O to fill the SiO₂ pore volume and added drop-wise onto the SiO₂ support. This material was then dried at room temperature overnight and subsequently calcined at 400 °C for 4 h (5 °C/min ramp). The final Co₃O₄/SiO₂ material was then divided and loaded with manganese (∼1.5 wt% Mn). The first promoted sample (Mn^SEA/Co/SiO₂) was prepared via SEA of a KMnO₄ (Aldrich, ≥99%) precursor from a solution adjusted to pH 1 following the method described previously.¹³ The second promoted sample (Mn^DI/Co/SiO₂) was prepared by DI of a Mn(NO₃)₂·4H₂O (Arcos, p.a.) salt with no pH adjustments. Following Mn addition, the samples were dried at room temperature overnight and calcined in air at 350 °C for 4 h (5 °C/min ramp). The reduced samples analyzed were treated in flowing H₂ at 350 °C for 2 h (5 °C/min ramp).

Temperature Programmed Reduction (TPR) experiments of the calcined catalyst materials were performed on a Micromeritics Autochem 2920 using a Thermal conductivity detector (TCD). After stabilization of the baseline, the reduction was carried out at a heating rate of 10 °C min⁻¹ in a 10% H₂/Ar flow.

To characterize the catalyst samples, we utilize the UIC-JEOL2010F TEM/STEM,^21,22 equipped with a Schottky field-emission electron gun operated at 200 kV, an ultra-high-resolution objective lens pole-piece (URP), and a post-column Gatan Imaging Filter (GIF). It has been set up for chemical analysis with high spatial resolution, resulting in an electron probe size of 1.8 Å at a convergence angle of 15 mrad and a spectrometer collection angle of 28 mrad. The core-loss and low-loss spectrum images were both acquired at the same location of each sample, with the acquisition time of 0.7 s pxl⁻¹ (for core-loss) and 0.1 s pxl⁻¹ (for low-loss) respectively. The dispersion for the spectra was set up to be 0.5 eV pxl⁻¹ (using a 1k × 1k camera). The core-loss EELS spectrum includes the O K-edge at 532 eV, the MnL-edge at 640 eV and the CoL-edge at 779 eV. The low-loss EELS spectrum includes the zero-loss peak and the SiL-edge at 99 eV. The Co and MnL-edges were calibrated by using the O K-edge onset, while the SiL-edge was calibrated by the zero-loss peak.

Results

Fig. 1 shows the TPR profiles of the Mn^DI/Co/SiO₂ and Mn^SEA/Co/SiO₂ samples, as well as an unpromoted Co/SiO₂ sample. In the profile of the unpromoted Co/SiO₂ sample, two peaks at 241 °C and 284 °C can be seen. These two peaks represent the reductions of Co₃O₄ to CoO, and CoO to metallic Co.^4,23,24 There is a small peak at 550 °C attributed to the reduction of Co silicate.⁶ For Mn^DI/Co/SiO₂, a shoulder at 241 °C and a peak at 284 °C still can be found, which means a portion of the Co particles still contain Co-oxide. For Mn^SEA/Co/SiO₂, the profile shape at temperatures less than 300 °C is similar to that of Co/SiO₂. However, the first reduction peak shifts to lower energy (from 241 °C to 216 °C in Mn^SEA/Co/SiO₂), which might be due to the addition of Mn to the Co/SiO₂ sample leading to an increase in the number of oxygen vacancies and the oxygen ion mobility.^10,25,26 The shift of the second peak to 300 °C is attributable to the fact that the Mn suppresses the reducibility of CoO to metallic Co.^9,11

Fig. 1 The TPR profiles of the Co/SiO₂, Mn^DI/Co/SiO₂ and Mn^SEA/Co/SiO₂ samples.

Above 300 °C, both promoted Co/SiO₂ samples show a peak and a broad feature extending up to 600 °C. More specifically, in Mn^DI/Co/SiO₂, we find a peak at 328 °C and a subsequent broad shoulder extending to 600 °C. Liang et al have previously shown that the reduction of spinel type Co–Mn mixed oxide results in a H₂ consumption peak in the temperature range between 300 °C and 600 °C.²⁷ We therefore ascribe our measured peak extending to 600 °C with a maximum at 328 °C to the reduction of the spinel type Co–Mn mixed oxide formed between the Mn promoters and Co catalysts. Moreover, Morales et al. have found a H₂ consumption peak at 550 °C due to the reduction of mixed Co–Mn spinel oxide materials.¹¹ The Co silicate reduction peak is also located at 550 eV. Based solely on our TPR results, we therefore assign the shoulder at 550 °C in the Mn^DI/Co/SiO₂ profile as arising from both Co–Mn spinel phase and Co silicate reduction. Similarly, for the Mn^SEA/Co/SiO₂ sample, we attribute the broad shoulder from 328 °C to 600 °C to the reduction of Co–Mn oxides.

It is interesting to note here that the TPR profiles for both Mn^DI/Co/SiO₂ and Mn^SEA/Co/SiO₂ samples are very similar, although the Mn distribution should be significantly different. For the Mn^DI/Co/SiO₂ sample, Mn should be randomly distributed over the Co/SiO₂ support. The higher dispersion of Mn should result in less interaction between the Mn and the Co particles, and thus a smaller concentration of the Co–Mn spinel phase. On the other hand, the SEA method guarantees that all of the Mn is deposited onto the Co particles in the Mn^SEA/Co/SiO₂ sample, resulting in a much larger concentration of the Co–Mn mixed oxide. However, the TPR profiles show that both methods generate comparable amounts of Co–Mn phase, indicating that Mn migrates to the Co particles and forms a compound with Co after deposition by DI.

In order to better understand the TPR results, we now use high-resolution Z-contrast imaging and EELS imaging to characterize the local chemical composition and electronic structures of the different catalyst samples. Around 20 images were acquired for each sample taken from regions that contain both Co and Mn. Fig. 2(a) shows the EELS spectrum image map of the Co, Mn and Si elemental distribution for the Mn^DI/Co/SiO₂ sample after calcination. For each element (Co, Mn and Si), the integrated spectrum intensity after background subtraction over an energy range of 20 eV is plotted as a function of position. As a result, it can be clearly seen that the Mn particles do not associate with the Co particles using dry impregnation. In particular, we see that a Mn particle (approximately 30 nm in diameter) is separated about 30 nm from a Co particle (around 20 nm in diameter). However, Fig. 2(b) shows another elemental distribution for a different location in the same sample. Here, there are two Co particles of around 15 nm and 10 nm in diameters on the surface SiO₂ support, while a 25 nm Mn particle covers both the Co particles and SiO₂ support. In particular, we find that the Mn particle covers the entire larger Co particle and only parts of the smaller Co particle.

Fig. 2 The elemental distribution of the Mn^DI/Co/SiO₂ sample (a) and (b) after calcination; (c) and (d) after reduction. Red represents Co, green represents Mn, and blue represents Si.

These results highlight two important features of the Mn^DI/Co/SiO₂ sample: (1) the initial distribution of Mn on the Co/SiO₂ surface is random; (2) although the initial location of Mn after deposition should be separated from Co, some Mn can still move to the vicinity of the Co particles during the calcination.

Fig. 2(c) and (d) show the elemental distribution for the Mn^DI/Co/SiO₂ sample after reduction. In both locations shown here, we can see that the Mn is in close proximity to the Co particles. Fig. 2(c) shows a smaller Co particle (around 8 nm in diameter) without any Mn, while parts of the larger Mn particle (around 10 nm × 20 nm) overlap with two Co particles around 10 nm in diameter. In Fig. 2(d), a Co particle of 10 nm in diameter sits on the edge of the SiO₂ support, and a 15 nm Mn particle overlaps the position of the Co particle. Although the images shown here imply that Mn is always associated with Co after reduction, we can still find single Mn particles without any Co (not shown here). These spectrum images suggest that the catalyst reduction treatment does not lead to a fundamental change in the Co and Mn elemental distributions compared to the calcined sample. Yet, for Mn that is in close proximity to Co particles in the calcined sample, it appears that Mn migrates towards Co during calcination, and forms a Co–Mn oxide, which then leads to the high peak corresponding to a Co–Mn compound in the TPR profile of Mn^DI/Co/SiO₂. Another interesting phenomenon shown in the spectrum images is that the Mn particles are larger than the Co particles, considering that the Mn is only 1.5 wt% compared to the 5 wt% Co. This could be due to the lack of reactivity of the SiO₂ support such that Mn and SiO₂ do not interact, facilitating Mn sintering on the SiO₂ surface.

The elemental distribution for the Mn^SEA/Co/SiO₂ sample after calcination is shown in Fig. 3(a) and (b), taken from different locations within the same sample. Fig. 3(a) shows a lower magnification spectrum image of a Co and Mn particle sitting at the edge of the SiO₂ support. It does not appear that these are two individual particles, but rather a Co core covered by a Mn shell. Moreover, we do not find any evidence of isolated Mn on the surface of SiO₂. Our analysis of different locations within the Mn^SEA/Co/SiO₂ sample confirms that Mn is selectively adsorbed on Co and not on SiO₂. In Fig. 3(b), a higher magnification spectrum image clearly shows that a Mn shell is formed on the surface of a 15 nm Co particle. The selective adsorption of Mn on Co creates a large Co/Mn interface, which leads to the large peak between 300 °C and 600 °C in our TPR profiles (Fig. 1).

Fig. 3 The elemental distribution of the Mn^SEA/Co/SiO₂ sample (a) and (b) after calcination; (c) after reduction. Red represents Co, green represents Mn, and blue represents Si.

Fig. 3(c) shows the elemental distribution for the Mn^SEA/Co/SiO₂ sample after reduction. Similar to the sample after calcination (Fig. 3(a) and (b)), Mn is only found on the Co particle; no Mn is on the SiO₂ support after the reduction treatment. In the particular location shown in Fig. 3(c), a Mn-covered Co particle (15 nm diameter) sits on the edge of the SiO₂ support. Although the relative intensity of Mn at the bottom of the Co particle is higher than at the top, a Mn signal could be detected over the entire surface of the Co particle. That means the Mn shell still covers the whole Co particle surface after the reduction treatment. These results indicate that if the Mn is initially on the Co particles, it prefers to stay on the Co surface, and does not migrate towards the Co/SiO₂ interface during either the calcination or reduction treatment.

Next, we will examine the elemental composition and electronic structure of the Co/SiO₂ interface by looking at the near-edge fine-structure of Si and CoL-edges.²⁸Fig. 4 shows the SiL-edges acquired from the amorphous SiO₂ at the Co/SiO₂ interface for our four samples. For each spectrum, SiL₃, L₂ and L₁ edges are located at 108 eV, 116 eV and 160 eV, respectively. The broad peak at 130 eV is due to the plural scattering events of the incident electrons, which usually occurs at larger sample thicknesses. The edge-onset of the SiL-edges is found to be 105.5 eV, using the second derivative method in Digital Micrograph. We cannot find any sign of the Si⁰ edge signal, which should have an edge onset of 99 eV. Therefore, we find that the fine structures of all SiL-edges indicate the presence of stoichiometric SiO₂²⁹ at the Co/SiO₂ interface. The absence of shift in the SiL-edge or the Si⁰ intensity indicates that there is no oxygen vacancy at the Co/SiO₂ interface.

Fig. 4 Si L-edges acquired at the Co/Si interface from Mn^DI/Co/SiO₂ and Mn^SEA/Co/SiO₂ samples after calcination and after reduction. The dash line indicates the edge onset at 105.5 eV.

The near-edge fine structures of CoL_2,3-edges were acquired at the same locations as those of the SiL-edges (Fig. 5). The change in the Co valence can be quantified using the change in the CoL_2,3-edge onset, as well as the Co L₃/L₂-ratio.³⁰ While it is customary to align the CoL_2,3-edge onset with respect to the O K-edge onset at 532 eV, this will not be possible in our current experiment. Due to the presence of SiO₂, the O K-edge shows a significant shift towards lower energies that is solely attributed to the SiO₂ and not a shift in the Co valance. Therefore, we will only use the changes in the Co L₃/L₂-ratio to quantify the Co valence. In Fig. 5, the CoL-edge is aligned to the position of the maximum intensity of the Co L₃ peak, in order to compare the changes in the L₃/L₂ ratio. Fig. 5(a) shows the CoL_2,3-edges acquired from the calcined and reduced Mn^DI/Co/SiO₂ samples. By using the second derivative method,³¹ the Co L₃/L₂ ratios of the calcined sample and reduced sample are found to be (2.2 ± 0.1) and (4.1 ± 0.1), respectively. According to our previous analysis of the Co L₃/L₂ ratio for different Co-oxide reference spectra,³⁰ we find that the Co-oxide particles are Co₃O₄ after calcination, and CoO after the reduction treatment. Similarly, the analysis of the CoL-edges in the Mn^SEA/Co/SiO₂ samples reveals a CoL₃/L₂ ratio in the calcined sample and reduced sample (2.1 ± 0.1) and (4.3 ± 0.1), respectively (Fig. 5(b)). Again, this indicates the presence of Co₃O₄ particles after calcination and CoO particles after reduction. It is important to note that all the particles analyzed are still oxides, even after reduction. This implies that either the samples were re-oxidized when exposed to air after reduction treatment, or formed Co–Mn or Co silicate compounds, as shown in Fig. 1, resulting in a reduction temperature higher than 350 °C used here.

Fig. 5 Co L_2,3-edges acquired at the Co/Si interface after calcination and reduction from (a) Mn^DI/Co/SiO₂ sample and (b) Mn^SEA/Co/SiO₂ sample.

Discussion

Our experimental results show that even in the Mn^DI/Co/SiO₂ sample, there appears to be a driving force for the Mn and Co particles to agglomerate after the calcination treatment. Moreover, we have shown that the Mn/Co particle morphology (in both the Mn^DI/Co/SiO₂ and Mn^SEA/Co/SiO₂ samples) does not change during the reduction treatment, as opposed to the Mn/Co/TiO₂ system, where a migration of Mn towards the Co/TiO₂ interface was previously reported.¹⁷Table 1 shows a summary of the Mn location in Mn promoted Co catalysts on SiO₂ and TiO₂ supports.

Table 1 The Mn location in Mn promoted Co catalysts on SiO₂ and TiO₂ supports

SiO2 support			TiO2 support
	Mn^DI/Co/SiO2	Mn^SEA/Co/SiO2	Mn^DI/Co/TiO2	Mn^SEA/Co/TiO2
Calcined	Large Mn particles form on SiO2. Mn particles also form on Co surface.	Mn forms shell covering the Co particle surfaces.	Mn is randomly distributed on Co and TiO2 surfaces.	Mn forms shell covering the Co particle surfaces.
Reduced	No significant change is observed compared to calcined sample.	No significant change is observed compared to calcined sample.	Mn moves to the Co/TiO2 interface.	Mn moves to the Co/TiO2 interface.

---

We propose the surface energies of the different oxides to be one reason to explain the agglomeration of the Mn–Co particles in the Mn^DI/Co/SiO₂ sample. Although it is hard to determine the surface energies of all the possible exposed surface orientations for all the oxides present in our samples, we will utilize the following trends to support our claim: the Ramsdellite MnO₂ (101) surface energy is 1.52 J m⁻²,³² that of Co₃O₄ (111) is 1.779 J m⁻²,³³ and the SiO₂ surface energy is 0.605 J m⁻².³⁴ During calcination of the Mn^DI/Co/SiO₂ sample, MnO₂ prefers to move onto the Co₃O₄ particle surface in order to decrease the surface energy of the system. In the Mn^SEA/Co/SiO₂ sample, where Mn is already selectively deposited on the Co₃O₄ surface, MnO₂ will remain there and stay away from the SiO₂ surface in order to maintain the lower surface energy of the system. On the other hand, the formation of a Co–Mn spinel phase will result in an increased interaction energy at the Mn/Co interface³⁵ that would act as an enthalpic driving force for the migration of Mn over the surface of the Co particles. Fig. 2(b) shows that the Co particle nearly completely overlaps with the Mn after calcination, and potentially covers the entire surface of the Co particles. Fig. 3 shows that for the Mn^SEA/Co/SiO₂ sample, the Mn shell appears to cover the entire surface of the Co particles, after both the calcination and the reduction treatments.

The migration of Mn in the Co-based catalysts on TiO₂ supports during reduction treatment compared to the calcined sample has been reported previously,^1,3 resulting in an enhanced catalyst performance of the catalysts due to the so-called promotion effect. However, the mechanism for the Mn migration during reduction is still unclear. In our previous study using TiO₂ supports, we had shown that by using the SEA method, Mn could form a homogeneous shell over the Co particle after calcination. The Mn shell ruptured after reduction treatment and the Mn migrated to the Co/TiO₂ interface.¹⁶ By using DI as the synthesis method to deposit Mn on the Co/TiO₂ sample, we found that the Mn also moved from the TiO₂ support to the Co/TiO₂ interface after reduction. Moreover, since the oxygen vacancies present on the TiO₂ surface are known to act as anchoring sites for mobile the metal atoms,³⁶ we speculated that the oxygen vacancies found at the Co/TiO₂ interface promote the migration of Mn towards the Co/TiO₂ interface.¹⁷

Although SiO₂ does not partially reduce, the formation of interfacial compounds such as Co silicate in the SiO₂ supported Co catalyst can still occur. It has been previously reported that the decomposition of Co(NO₃)₂·6H₂O forms cobalt silicate on the silica support,^37,38 and Ernst et al. reported that Co₂SiO₄ forms at reduction temperatures as low as 350 °C.⁵ It has also been shown that the Co silicate reduces above 500 °C.^5,6,39Fig. 1 shows the TPR profile for a Co/SiO₂ catalyst, and the peak at 550 °C is assigned to the reduction of Co₂SiO₄. Since the calcination and reduction treatments for the Mn^DI/Co/SiO₂ and Mn^SEA/Co/SiO₂ samples are all carried out at 350 °C, it is entirely possible that an interfacial Co₂SiO₄ phase is still present. Unfortunately, our TPR measurements cannot distinguish between the reduction peaks of the Co–Mn spinel phase and Co₂SiO₄, since they both overlap in the temperature range between 300 °C and 600 °C. Moreover, our EELS analysis would not be able to distinguish the SiO₂ support from an interfacial Co₂SiO₄ or CoSiO₃ phase,^5,39,40 since the valence state for Si is 4+ in either phase, as measured by the SiL-edge EELS spectrum in Fig. 4. However, regardless of the formation of an interfacial Co silicate, we do not find any electron transfer between the silica support and the Co catalyst, and no oxygen vacancies are measured at the Co/SiO₂ interface.

In order to verify that the oxygen vacancies provide the driving force for the formation of an interfacial Co–Mn catalyst, we can study the evolution of the Mn promoter in Co/SiO₂ catalysts. The oxygen vacancy formation energy on SiO₂ is nearly twice as high as that on TiO₂ surfaces. More specifically, the oxygen vacancy formation energy on the SiO₂ surface was calculated to be 8.5 eV,⁴¹ while that for TiO₂ rutile and anatase surfaces was found at 3.77 eV and 4.94 eV, respectively.³⁶ Therefore, the formation of oxygen vacancies is expected to be much less likely in the Co/SiO₂ compared to the Co/TiO₂ interface under similar reduction conditions. Thus, the different Co/Mn particle morphology changes during reduction on SiO₂ supports (as shown in Fig. 2 and 3) compared to that on TiO₂ supports can be attributed to the lack of interfacial oxygen vacancies, and anchoring sites for the mobile Mn atoms. As a result, we do not see any migration of the Mn on the Co particle towards the Co/SiO₂ interface, which happens in the Co/TiO₂ system as previously reported.^13,16,17 Therefore, by comparing the results presented here with previous studies of similar catalysts on TiO₂, where a strong metal–support interaction exists, we can show that the presence of interfacial oxygen vacancies is important for the formation of an interfacial Mn–Co catalyst, and thus a critical factor to the promotional effect of Mn/Co FT catalysts.

Conclusion

Temperature programmed reduction and Z-contrast imaging combined with electron energy-loss spectrum imaging is used to study the elemental distribution and morphology of the Mn-promoted Co nanoparticles on SiO₂. By comparing the results with those of Mn-promoted Co particles on TiO₂,^13,16,17 we find that oxygen vacancies at the metal–support interface are important for understanding not only the migration effects of Mn on the Co particle surface, but also the promoter effects on the Mn promoted Co-based FT catalysts. More specifically, we find that the oxygen vacancies mediate formation of an interfacial Mn–Co phase with active sites that contribute to the higher selectivity found in Mn-promoted FT catalysts.

Acknowledgements

The authors Y.Z and R.F.K would like to acknowledge support for this research by the American Chemical Society-Petroleum Research Fund (grant number 47307-G). T.E.F, R.J.M and J.R.R would like to acknowledge support from the National Science Foundation (CBET-#0626505). The authors would acknowledge the Research Resource Center of UIC for providing instrumentation for the experiments.

References

1. F. Morales, F. M. F. de Groot and O. L. J. Gijzeman, et al. , J. Catal., 2005, 230, 301.
2. F. Morales, F. M. F. de Groot and P. Glatzel, et al. , J. Phys. Chem. B, 2004, 108, 16201.
3. F. Morales, D. Grandjean and F. M. F. de Groot, et al. , Phys. Chem. Chem. Phys., 2005, 7, 568.
4. R. Riva, H. Miessner and R. Vitali, et al. , Appl. Catal., A, 2000, 196, 111.
5. B. Ernst, S. Libs and P. Chaumette, et al. , Appl. Catal., A, 1999, 186, 145.
6. A. Kogelbauer, J. C. Weber and J. G. Goodwin, Catal. Lett., 1995, 34, 259.
7. O. Borg, S. Eri and E. A. Blekkan, et al. , J. Catal., 2007, 248, 89.
8. M. Vo, D. Borgmann and G. Wedler, J. Catal., 2002, 212, 10.
9. F. Morales, E. de Smit and F. M. F. de Groot, et al. , J. Catal., 2007, 246, 91.
10. D. Das, G. Ravichandran and D. K. Chakrabarty, Appl. Catal., A, 1995, 131, 335.
11. F. Morales, D. Grandjean and A. Mens, et al. , J. Phys. Chem. B, 2006, 110, 8626.
12. M. J. Keyser, R. C. Everson and R. L. Espinoza, Appl. Catal., A, 1998, 171, 99.
13. T. E. Feltes, L. Espinosa-Alonso and E. d. Smit, et al. , J. Catal., 2010, 270, 95.
14. Catalysis: an integrated approach to homogeneous, heterogeneous and industrial catalysis, Elsevier, 1993.
15. J. R. Regalbuto, Catalyst Preparation: Science and Engineering, Taylor and Francis/CRC Press, 2007.
16. T. E. Feltes, Y. Zhao and R. F. Klie, et al. , ChemCatChem, 1065, 2, 2010.
17. Y. Zhao, T. E. Feltes and J. R. Regalbuto, et al. , Catal. Lett., 2011, 141, 641.
18. S. J. Tauster, S. C. Fung and R. T. K. Baker, et al. , Science, 1121, 211, 1981.
19. S. J. Tauster, S. C. Fung and R. L. Garten, J. Am. Chem. Soc., 1978, 100, 170.
20. S. J. Tauster, Acc. Chem. Res., 1987, 20, 389.
21. E. M. James and N. D. Browning, Ultramicroscopy, 1999, 78, 125.
22. E. M. James, N. D. Browning and A. W. Nicholls, et al. , J. Electron. Microsc., 1998, 47, 561.
23. S. Bessell, Appl. Catal., A, 1993, 96, 253.
24. H. Ming, B. G. Baker and M. Jasieniak, Appl. Catal., A, 2010, 381, 216.
25. G. Ravichandran, D. Das and D. K. Chakrabarty, J. Chem. Soc., Faraday Trans., 1993, 90, 1994.
26. S. Varma, B. N. Wani and N. M. Gupta, Appl. Catal., A, 2001, 205, 295.
27. Q. Liang, K. Chen and W. Hou, et al. , Appl. Catal., A, 1998, 166, 191.
28. R. F. Klie, M. M. Disko and N. D. Browning, J. Catal., 2002, 205, 1.
29. C. C. Ahn and O. L. Krivanek, EELS Atlas, Arizona State University, 1983.
30. Y. Zhao, T. E. Feltes and J. R. Regalbuto, et al. , J. Appl. Phys., 2010, 108, 063704.
31. G. A. Botton, C. C. Appel and A. Horsewell, et al. , J. Microsc., 1995, 180, 211.
32. R. R. Maphanga, S. C. Parker and P. E. Ngoepe, Surf. Sci., 3184, 603, 2009.
33. X.-L. Xu, Z.-H. Chen and Y. Li, et al. , Surf. Sci., 2009, 603, 653.
34. S. H. Overbury, P. A. Bertrand and G. A. Somorjai, Chem. Rev., 1975, 75, 547.
35. S. A. Stevenson, J. A. Dumesic and R. T. K. Baker, et al., Metal–support interactions in catalysis, sintering, and redispersion, Van Nostrand Reinhold company, NY, 1987.
36. H. Iddir, V. Skavysh and S. Ogut, et al. , Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 73, 041403.
37. M. P. Rosynek and C. A. Polansky, Appl. Catal., 1991, 73, 97.
38. K. E. Coulter and A. G. Sault, J. Catal., 1995, 154, 56.
39. I. Puskas, T. H. Fleisch and J. B. Hall, et al. , J. Catal., 1992, 134, 615.
40. H. Ming and B. G. Baker, Appl. Catal., A, 1995, 123, 23.
41. T. Bredow and G. Pacchioni, Chem. Phys. Lett., 2002, 355, 417.

---