Learn about the tests, in situ XAS, XRD and Raman used to characterise catalyst phase morphology.

Over the past decade, many synchrotron accelerator technologies have been developed to improve the temporal, energetic, and spatial resolution for determining the structure, dynamics, and kinetics of catalytic active sites. With the development of individual techniques, new and improved experimental methods aimed at studying in situ catalytic mechanisms are also evolving. Today, the most advanced approach combines complementary techniques in a single experiment for catalytic studies conducted on-site or in operation. As the scope of in situ research expands from model catalysts to real systems, new challenges arise. One of these challenges is the presence of many competitive factors influencing the catalytic process: for example, the heterogeneity of particle size and shape, temperature and pressure gradients, and the influence of the support and adsorbates. Another complex issue is the existence of multiple length scales defining a truly catalytic system. Therefore, researchers such as S. N. Ehrlich from the University of Delaware, A. I. Frenkel from Yale University, and E. Stavitski from Brookhaven National Laboratory combine spectroscopic and scattering techniques to elucidate processes occurring simultaneously at different length scales (from tens of picometers to micrometers).

To demonstrate the feasibility of this approach, a inherently complex catalytic system was selected: a chromium-oxide-modified iron oxide used for the high-temperature water-gas shift (WGS) reaction. The WGS reaction is an important industrial process in which carbon monoxide reacts with steam to produce carbon dioxide and hydrogen molecules. Several in situ studies involving different types of WGS catalysts (3% Cr2O3/Fe2O3 and Fe2O3) have been conducted using a combination of different techniques, focusing on issues related to structural changes, catalytic active sites, the role of catalyst promoters or stabilizers, and reaction mechanisms. By combining various techniques for studying catalyst performance, monitoring, and online real-time product analysis, further progress can be made in mechanism studies under operating conditions. The authors combined XAS/XRD/Raman measurements at the X18A beamline of the National Synchrotron Light Source (NSLS) at Brookhaven National Laboratory. These studies provide new information on the oxidation state of the catalyst, local atomic structure, bulk and surface crystal structures before, during, and after the WGS reaction. This work opens up new opportunities for similar in situ studies of a wide range of heterogeneous catalytic reactions.

The first set of experiments involved a series of Fe-K edge XANES spectra of pure Fe2O3 exposed to O2 at room temperature, 400 ℃, during the WGS reaction at 400 ℃, and after the WGS reaction at room temperature. The Fe-K edge positions were found to be the same when exposed to O2 flow at room temperature and 400 ℃, indicating no change in the charge of Fe. Compared to the oxidation state of Fe before the WGS reaction, the Fe-K edge shifted to lower energy at 400 ℃ under WGS conditions, a shift similar to the partial reduction of Fe in Fe2O3 to Fe3O4. Under WGS flow conditions, the Fe-K edge shifted to higher binding energy, which is explained by the oxidation effect of steam at temperatures below the WGS reaction temperature (below 200 ℃) and the subsequent decrease in CO activity. However, the XANES technique alone cannot identify or quantitatively analyze the different iron oxide phases (e.g., α-Fe2O3, γ-Fe2O3, or Fe3O4) that may coexist during the decomposition process at the same temperature before, during, and after the reaction.

Figure 2 shows the κ space (a) and γ space (b, c) Fe-K edge EXAFS spectra sequences of 3% Cr2O3/Fe2O3 and Fe2O3 catalysts before and after the WGS reaction. The data indicate significant changes in the local structure environment of Fe, although the changes differ significantly between the two systems. After the WGS reaction, the decrease in intensity of the first peak corresponding to Fe-O between 1-2.0 Å can be explained by a reduction in the Fe-O coordination number or an increase in the disorder of bond lengths, or both. However, the decrease in Fe coordination number when different iron oxide mixtures are present is too small to be accurately detected by EXAFS analysis. The disorder of bond lengths reduces the intensity of EXAFS oscillations, which is due to partial decoherence of the connecting X-ray absorption atoms and their nearest neighboring photoelectron paths caused by bond length disorder due to configuration, thermal disorder, or both. Disorder can be quantified by σ, the standard deviation of bond length R, usually defined as σ^2 = (R-R)^2. For small to moderate disorder, the radial distribution function can be approximated by a Gaussian distribution, and the relevant EXAFS equation is exp(-2k^2σ^2), where the decrease in intensity of the EXAFS due to bond length disorder is caused by this term. A common method is to extract the disorder parameter by fitting the EXAFS equation; however, this method fails in cases of strong, non-Gaussian disorder, leading to incorrect results from Gaussian fitting methods. Yevick and Frenkel demonstrated that in an asymmetrically disordered system (where even the third cumulant is insufficient to make the cumulant expansion converge), the distances, coordination numbers, and disorder parameters obtained from theory fitting of EXAFS data based on Gaussian estimates will be incorrect. In the system described by the authors, the origin of this enhanced disorder is the coexistence of different iron oxide phases, each with its own Fe-O and Fe-Fe distance distributions, which will be discussed in more detail below.

In figures 2b and 2c, the second peak corresponds to the contribution of Fe-Fe bonds to the Fe-K edge EXAFS. A decrease in the intensity of the second peak in the Fourier-transformed EXAFS signal indicates the presence of an uneven mixture of iron oxides with different Fe-Fe distances in the 3% Cr2O3/Fe2O3 sample compared to the Fe2O3 sample. Different forms of iron oxide can be distinguished through quantitative EXAFS analysis, but it is essential to satisfy the condition of phase homogeneity to ensure reliable results. In this work, XRD measurements showed the coexistence of different Fe phases at different temperatures, suggesting that the fitting method used for the EXAFS data analysis was inaccurate.

Another result from the EXAFS data is the comparison of data between O2 and WGS conditions at 400°C. The most significant effect is the reduction in the intensity of the second peak under O2 conditions compared to WGS conditions at the same temperature (Figure 2b). This reduction is consistent with significant disordering of the Fe environment under O2 before the start of the WGS reaction. This interpretation of the EXAFS data is consistent with in situ XRD results. Figure 3 shows the XRD spectra at different stages of the WGS reaction. The black lines in figures 3a and 3b correspond to the XRD spectra of Fe2O3 and 3% Cr2O3/Fe2O3 at room temperature before the WGS reaction, indicating the presence of pure γ-Fe2O3. The Bragg peaks of the 3% Cr2O3/Fe2O3 sample are noticeably broader and weaker compared to the corresponding Fe2O3, indicating that the average particle size of the 3% Cr2O3/Fe2O3 catalyst is smaller than that of the Fe2O3 catalyst. However, no Bragg peaks of Cr were detected, suggesting that Cr is incorporated into the Fe2O3 lattice. The red line in figure 3b corresponds to the XRD spectrum of the 3% Cr2O3/Fe2O3 sample collected in an O2 flow at 400°C before the WGS reaction, indicating the presence of the α-Fe2O3 phase. This change is not significant in the XRD of the Fe2O3 catalyst (red line, figure 3a), indicating that the incorporation of chromium affects the high-temperature reducibility of the Fe2O3 bulk. This effect may also be the reason for the presence of two types of iron oxide crystal forms (α- and γ-Fe2O3) observed in the Cr-doped samples at 400°C.

Figure 4 shows the XRD diffraction patterns in the range of 2θ from 48° to 51°. During the WGS reaction at 400°C, the XRD patterns of Fe2O3 and 3% Cr2O3/Fe2O3 (green lines, Figures 4a, b) show a shift of the Bragg peaks towards lower angles. This is consistent with the transformation of γ-Fe2O3, which has a larger lattice parameter, to Fe3O4. This observation is also consistent with the reduction of iron oxides observed in XANES (Figure 1). After cooling, the process of some Fe3O4 oxidizing to Fe2O3 is represented by a shift of the Bragg peaks towards higher angles (blue line, Figure 4a). It is important to note that the Bragg peaks after cooling (blue line) do not return to the same positions as before heating (black line), indicating that the phase transition to Fe3O4 in the WGS flow reaction is not fully reversible. The main difference is that the 3% Cr2O3/Fe2O3 catalyst system undergoes a transformation from γ-Fe2O3 to α-Fe2O3 to Fe3O4, whereas in the Fe2O3 catalyst, the transformation seems to occur directly from γ-Fe2O3 to Fe3O4 without the formation of α-Fe2O3 in between.

However, analyzing the phase morphology of the two catalysts based solely on XRD data would be overly simplistic. Combining the results of XRD and EXAFS provides a deeper understanding of the unique iron oxide phase transformations. The XRD data (Figure 3b) indicate the presence of α-Fe2O3 in the 3% Cr2O3/Fe2O3 sample at the start of the temperature ramp, while no α-Fe2O3 is present in the Fe2O3 sample (Figure 3a). Although the XRD peak of α-Fe2O3 is not present during the WGS reaction intermediate and after, the EXAFS data suggest that the low-dimensional component of α-Fe2O3 or other octahedral Fe-coordinated compounds (e.g., γ-FeOOH) may be present in the Cr2O3/Fe2O3 system throughout the temperature cycle. The coexistence of these heterogeneous Fe compounds in the Cr2O3/Fe2O3 system leads to greater disorder in the Fe-Fe distances and a decrease in the intensity of the second peak in the EXAFS data after the WGS reaction (Figures 2b, c). The presence of multiple Fe phases compared to a single Fe phase leads to a larger spread in Fe-Fe distances in the Cr2O3/Fe2O3 system. A similar conclusion can be drawn for the Fe2O3 sample in O2 flow at 400°C: although the XRD peak of α-Fe2O3 is hardly visible in Figure 3, the EXAFS data (Figure 2b) indicate greater disorder in the second peak region under O2 conditions at the same temperature compared to WGS conditions, consistent with the presence of strong disorder or low-dimensional phases such as α-Fe2O3. It has recently been observed that XAFS is more sensitive to these disordered phases when both techniques are used in the same experiment, allowing their presence to be detected before XRD. As shown below, Raman measurements also confirm the presence of α-Fe2O3 at room temperature, supporting this model.

When the authors interpreted the measurement results, a new and more complex picture of the phases and their transformations emerged (Table 1). Since the analysis capability of XRD methods is limited in the presence of strong dispersion or disordered species, Raman spectroscopy was used to obtain additional information about the mixed phases in the samples. Raman spectra of Fe2O3 and 3% Cr2O3/Fe2O3 catalysts were collected at room temperature under the same conditions as XAFS and XRD data. The Raman spectra of both catalysts before the WGS reaction are shown in Figure 5a. Several peaks were observed in the Fe2O3 catalyst before hydration pretreatment at 226, 244, 261, 292, 378, 496, and 635 cm-1. The peaks at 226 cm-1 and 292 cm-1 are likely attributed to the α-Fe2O3 phase, while the broad peak at 378 cm-1 and the peaks at 244, 261, 496, and 635 cm-1 are likely attributed to the γ-Fe2O3 phase. Additionally, the peak at 261 cm-1 may correspond to a γ-FeOOH phase. Raman spectra of the 3% Cr2O3/Fe2O3 catalyst collected at room temperature before the WGS reaction showed peaks at 378, 492, 632, and 830 cm-1. The peaks at 378, 492, and 632 cm-1 are most likely from the γ-Fe2O3 phase, as described above. The peak at 830 cm-1 is attributed to the hydrated CrO42- oxyanion.

Raman spectroscopy is more sensitive than XRD spectroscopy to oxide phases with sizes of less than 4 nm. Since the α-Fe2O3 phase was not detected by XRD in the Fe2O3 catalyst (Figure 3a(i)), but was detected by Raman spectroscopy (Figure 5a), the α-Fe2O3 phase is most likely present in the form of dispersed regions on the surface of the catalyst particles. This conclusion is also consistent with the behavior observed in the EXAFS data, indicating an increase in disorder of the Fe-Fe distances at 400 °C under O2 flow. However, the Raman spectrum of 3% Cr2O3/Fe2O3 did not show a Raman peak for α-Fe2O3, suggesting that the addition of Cr helps stabilize γ-Fe2O3 as the sole phase at room temperature before the WGS reaction. XRD and EXAFS indicate that at higher temperatures, both samples exhibit significant formation of bulk α-Fe2O3 phase. The dispersed α-Fe2O3 phase is present and only exists in the post-treated Fe2O3 sample, which is a new and surprising result that could only be obtained through in situ Raman spectroscopy experiments. Figure 5b shows the Raman spectra of 3% Cr2O3/Fe2O3 before and after the WGS reaction. The significant disappearance of the CrO42- Raman peak at 830 cm-1 suggests that the doped CrOx is almost completely irreversibly reduced during the WGS reaction, indicating that Cr dissolves into the main oxide support during the WGS reaction, as proposed in the literature. This process, along with other irreversible processes (such as incomplete reoxidation of the Fe2O3 catalyst (Figure 1) and stabilization of the α-Fe2O3 phase at the end of the cycle (Figure 3)), may be the cause of catalyst deactivation. Therefore, the combination of various in situ experimental techniques can successfully monitor the changes between different phases of the catalyst during the reaction cycle, revealing the relationships between different processes. This study also demonstrates the analytical capabilities of combining complementary methods in studying the catalytic activity, selectivity, and deactivation mechanisms of actual catalytic systems with component heterogeneity and large spatial dimensions.

Reference Patlolla, A., Carino, E. V., Ehrlich, S. N., Stavitski, E., & Frenkel, A. I. (2012). Application of operando XAS, XRD, and Raman spectroscopy for phase speciation in water gas shift reaction catalysts. ACS Catalysis, 2(11), 2216-2223.