Diamond Annual Review 2021/22

66 67 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 1 / 2 2 D I A M O N D L I G H T S O U R C E A N N U A L R E V I E W 2 0 2 1 / 2 2 X-ray ptychography investigates coking of solid catalysts in 3D Related publication: Weber, S., Batey, D., Cipiccia, S., Stehle, M., Abel, K. L., Gläser, R., & Sheppard, T. L. Hard X-ray nanotomography for 3D analysis of coking in nickel-based catalysts. Angewandte Chemie International Edition 60 , 21772–21777 (2021). DOI: 10.1002/anie.202106380 Publication keywords: Carbon; Methanation of CO 2 ; Nickel; Raman Spectroscopy; X-ray Ptychography C atalysts speed up chemical reactions and are widely used in industrial processes. Chemical reactions involving carbon-containing species can form carbon deposits (coke) on the catalyst surface. As catalysts are typically porous materials, extensive coke formation can restrict accessibility of reactants and products through the pores, leading to so-called mass transport limitations, and can physically block the active sites of the catalyst that are responsible for facilitating the chemical reaction. While diagnosing the presence of coke is relatively simple, obtaining information on the location, chemical nature, and severity of coke formation is more complicated. This study aimed toexamine coke formationon several catalysts usingX-rayptychography. It neededaveryhigh image resolution tovisualise the catalyst sample, the pores, and possible coke deposits, all of which can occur on the scale of micrometres to nanometres. The imaging method used also had to be sensitive enough to differentiate which parts of the catalyst are coked. The Coherence Branchline (I13-1) is specially designed for high-resolution imaging with X-ray in 2D and 3D. X-ray ptychography also allows scientists to discover the electron density of the sample, which can be used to examine where coking has occurred. The results can be applied to study industrial catalyst samples during their life cycle, to understand the location and nature of coke deposits. An improved fundamental understanding of catalyst deactivation is also important for catalyst design and synthesis, potentially tailoring the pore structure to withstand coking. Catalyst coking is a major deactivation pathway of solid catalysts applied in the chemical industry. In principle, any process involving carbonaceous species might suffer from this type of deactivation, where a variety of solid carbon products may deposit on the catalyst surface. This in turn can block the catalytic active sites or fill the pores of the catalyst, inducing mass transport limitations. In the case of mass transport limitations, tailoring the catalyst pore structure by introducing hierarchical pore systems on different length scales ( e.g ., meso-/ macroporous catalysts) is one possible strategy to mitigate the effects of coking, therefore improving catalyst performance and stability 1 . Catalyst coking is often diagnosed and analysed by bulk characterisation techniques ( e.g ., elemental analysis, IR/Raman spectroscopy, thermal analysis methods), which can provide information about the amount of coke formed or the type of coke ( e.g ., graphitic, amorphous, activated carbon). However, conventional methods provide little information about the location of the carbon species formed, which is relevant in the context of mass transport limitations and pore blocking. A spatially resolved investigation of coke deposition should ideally be carried out on whole catalyst particles, which typically range from tens of µm ( i.e the size of a single human hair) all the way up to the mm or cm scale in industrial applications. Considering the vast difference in length scales which are relevant for catalysis, hard X-ray imaging methods are required to investigate the whole sample without imposing limitations on beam transmission, which would be the case for soft X-ray techniques or electron microscopy. Among the currently available hard X-ray methods, X-ray Ptychography combined with tomography can retrieve quantitative 3D information about the electron density (N e ) distribution of an investigated sample 2,3 . Additionally, this method allows for high-resolution imaging (routinely below 100 nm in 3D) of extended sample volumes (tens of µm in diameter) due to the use of hard X-ray as the probe. To explore the potential of Ptychographic X-ray computed tomography (PXCT) to detect and reveal the location of coking in 3D, a hierarchically porous Ni/Al 2 O 3 catalyst was selected as a case study 4 . This catalyst has applications in CO 2 methanation or CH 4 dry reforming, both of which are relevant in the context of reducing CO 2 emissions, and in storage of energy from intermittent power sources ( e.g ., wind, solar) in chemical form. The catalyst itself exhibits a hierarchical pore structure of meso- (2-50 nm) and macropores (>50 nm). During the experiments, the catalyst was first activated using a microcapillary setup equipped with a hot-air blower, custom-made gas dosing system, and online mass spectrometer for gaseous product analysis. The activation procedure was carried out in a gas flow of 25% H 2 /He (20 mL.min -1 ) at about 700 °C followed by reaction conditions of 20% H 2 /5% CO 2 / He at 400 °C to obtain an activated catalyst sample (Ni/Al 2 O 3 -ha). An artificially coked sample (Ni/Al 2 O 3 -hc) was obtained by a subsequent coking step in 4% CH 4 /He at 400 °C after reaction conditions, while the formation of coke during the latter step was confirmed by operando Raman spectroscopy. Selected particles of the Ni/Al 2 O 3 -ha and Ni/Al 2 O 3 -hc samples (around 30 to 50 µmdiameter) were thenmounted on Al tomography pins using a dual-beam focused ion beam scanning electron microscope (FIB-SEM) with Pt deposition. The Ni/Al 2 O 3 -ha and Ni/Al 2 O 3 -hc samples were then studied by PXCT at beamline I13-1 of Diamond Light Source, which is specialised in coherent hard X-ray imaging methods such as ptychography. For each sample, 1,000 2D-projections were acquired over an angular range of 180°, followed by image reconstruction to obtain tomograms (3D digital volumes) of the complex refractive index (δ). From the δ tomograms, the electron density (N e ) tomograms of the catalysts can be directly calculated 2,3 . The N e tomograms (Fig. 1) in the present work have an isotropic voxel size of ca. 37 nm and an estimated resolution of 74 to 83 nm based on Fourier shell correlation. The tomograms readily show the connected macropore network of Ni/Al 2 O 3 -ha and Ni/Al 2 O 3 -hc, but the resolution is not sufficient to directly resolve the mesopores. For further analysis, the tomograms were segmented and labelled based on the N e distribution into regions containing pores (orange), nanoporous catalyst body (grey) and contamination (green) (Fig. 1), following which the N e distribution of the whole catalyst particles was determined (Fig. 2a). The contaminationmainly originates fromPt used to mount the sample particles on the pins during FIB-SEM preparation, which can be selected and discarded from the 3D volumes before further analysis. Comparing the N e distribution for the whole catalyst particles, a shoulder with increased N e can be observed on the nanoporous catalyst body peak at N e >0.5 e – Å –3 for the Ni/Al 2 O 3 -hc sample. This is additionally illustrated by plotting the mean N e with the standard deviation (σ) of the nanoporous catalyst body label depending on the tomography slice number (Fig. 2b). Over thewhole slice range, themean N e of Ni/Al 2 O 3 -hc is larger than for Ni/Al 2 O 3 -ha, while the σ value also highlights the increased contribution from regions of higher N e . The normalised N e distribution of the whole nanoporous catalyst labels of Ni/Al 2 O 3 -ha and Ni/Al 2 O 3 -hc were used to critically select a threshold for segmentation of the Ni/Al 2 O 3 -hc into coked and uncoked labels (Fig. 2c). No obvious coke formation was detected inside the macropores of the Ni/Al 2 O 3 -hc catalyst. Based on the distribution of the coked label (Fig. 2c), it is postulated that coke forms in mesopores within the catalyst body, with a pronounced coke distribution at the exterior of the particle. Similar observations were recently reported by X-ray holotomography for fluid catalytic cracking particles with a 3D spatial resolution of ca. 200 nm 5 . However, in the present case further studies at different coking stages should be performed to validate this conclusion. In principle, the electron density distribution and thus the results of PXCT are sensitive to any changes of the atomic configuration in 3D space, requiring careful interpretation of the results. In the present study, it is reasonable to attribute the observed N e increase to coking. Other possible factors such as Ni nanoparticle sintering can be neglected, due to the comparison of similar treatment and activation conditions at high temperatures for both samples, with only a short coking time (30 min) at the reaction temperature. In conclusion, X-ray Ptychography allows routine analysis of catalyst particles up to tens of µm in diameter with sub-100 nm resolution. At this level of performance however the method is currently only feasible at dedicated synchrotron beamlines for coherent hard X-ray microscopy, such as I13-1 of Diamond Light Source. This is due to the high coherent photon flux required for such measurements. On the other hand, the quantitative electron density contrast available from PXCT can not only be used to analyse coking of catalysts, but also to directly visualise the location of coke deposits. Based on this knowledge, the methodology presented here has great potential to provide fundamental knowledge on catalyst deactivation for various use cases. Obtaining such information at the nanoscale on extended sample volumes is hardly possible with any other currently established method. Understanding andmitigating catalyst deactivation is a crucial component in future knowledge- based tailoring of catalyst synthesis methods as well as structural properties ( e.g ., pore networks) to increase the catalyst stability, improve the efficiency and lifespan particularly for industrial applications. With the current shift towards next generation diffraction limited synchrotron light sources, such as Diamond- II, it is anticipated that the resolution, performance, and applications of hard X-ray imaging will stand to gain quite dramatically in future. References: 1. Wang, G. et al. Rational design of hierarchically structured porous catalysts for autothermal reforming of methane. Chemical Engineering Science 65 , 2344–2351 (2010). DOI: 10.1016/j.ces.2009.09.079 2. Dierolf, M. et al . Ptychographic X-ray computed tomography at the nanoscale. Nature 467 , 436–439 (2010). DOI: 10.1038/nature09419 3. Diaz, A. et al . Quantitative x-ray phase nanotomography. Physical Review B 85 , 020104 (2012). DOI: 10.1103/PhysRevB.85.020104 4. Weber, S. et al . Porosity and structure of hierarchically porous Ni/Al 2 O 3 Catalysts for CO 2 Methanation. Catalysts 10 , 1471 (2020). DOI: 10.3390/ catal10121471 5. Veselý, M. et al. 3-D X-ray nanotomography reveals different carbon deposition mechanisms in a single catalyst particle. ChemCatChem 13 , 2494–2507 (2021). DOI: 10.1002/cctc.202100276 Funding acknowledgement: Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – 406914011. This work was partly carried out with the support of the Karlsruhe Nano Micro Facility (KNMF), a Helmholtz Research Infrastructure at Karlsruhe Institute of Technology (KIT), which provided access to FIB instruments via proposal 2019-022-026980. This work was carried out with the support of Diamond Light Source, instrument I13-1 (proposal MG24079-1). We acknowledge KIT and DFG for financing the Raman spectrometer system (INST121384/73-1). Open access funding enabled and organised by Projekt DEAL. Corresponding authors: SebastianWeber, Karlsruhe Institute of Technology (KIT), sebastian.weber@kit.edu Dr. Thomas L. Sheppard, Karlsruhe Institute of Technology (KIT), thomas.sheppard@kit.edu . Imaging andMicroscopy Group Beamline I13-1 Figure 1: PXCT of the investigated activated catalyst Ni/Al 2 O 3 -ha (a) and artificially coked catalyst Ni/Al 2 O 3 -hc (b) particles. Each reconstructed volume (grey) is shown with a cut through the middle, illustrating the segmented and labelled tomograms (grey = nanoporous catalyst body, orange = pores, green = contamination) and a grayscale electron density (N e ) image of a typical slice through the volume (colour bar in N e /e – Å –3 , N e offset to 0 with respect to air close to the sample). Figure 2: Global electron density (N e ) distribution from PXCT of the activated and artificially coked catalyst particle labels of Ni/Al 2 O 3 -h (a). Mean, standard deviation (σ) and variance (σ 2 ) in N e of the segmented catalyst body label in a selected slice range (b). Segmented tomogram slice of Ni/Al 2 O 3 -hc (c) showing less electron dense (grey, assigned to uncoked catalyst body) and higher density (red, assigned to coked catalyst body) regions by binary thresholding of the normalised N e distribution.