Formation and Functioning of Bimetallic Nanocatalysts: The Power of X‐ray Probes

Abstract Bimetallic nanocatalysts are key enablers of current chemical technologies, including car exhaust converters and fuel cells, and play a crucial role in industry to promote a wide range of chemical reactions. However, owing to significant characterization challenges, insights in the dynamic phenomena that shape and change the working state of the catalyst await further refinement. Herein, we discuss the atomic‐scale processes leading to mono‐ and bimetallic nanoparticle formation and highlight the dynamics and kinetics of lifetime changes in bimetallic catalysts with showcase examples for Pt‐based systems. We discuss how in situ and operando X‐ray spectroscopy, scattering, and diffraction can be used as a complementary toolbox to interrogate the working principles of today's and tomorrow's bimetallic nanocatalysts.


Introduction
Bimetallic nanoparticle (NP) catalysts display extraordinary physicochemical properties compared to their monometallic counterparts. [1][2][3][4] When initially introduced by Sinfelt et al.,N i-Cu, Ru-Cu, and Os-Cu were found to reduce undesired CÀCa ctivation for hydrogenolysis compared to monometallic Ni, Ru, and Os,w hile maintaining their CÀH activation abilities for dehydrogenation. [5] This early discovery triggered the exploration and widespread use of bimetallic nanocatalysts,i ncluding Pt-Pd, Pt-Ru, Pt-Ir,P t-Re alloys,t o induce selectivity shifts towards desired reaction products. [6][7][8][9] Today,bimetallic NPs possess recognized abilities to promote reforming,hydrogenolysis,(de)hydrogenation, and oxidation reactions. [10] Recently,a ttention has been invested in the development of bimetallic catalysts for upgrading biomass to fuels and chemicals, [10,11] the production of hydrogen, [12] as well as in the use of earth-abundant transition (bi)metallic nanocatalysts. [13] Bimetallic NP electrocatalysis [14] is another application area, which however lies beyond the scope of this Minireview.
Theo rigin of the performance shift in bimetallic compared to monometallic nanocatalysts is mainly attributed to electronic (ligand) and geometric (ensemble) effects. [15] By alloying at ransition metal with ad onor metal, the d-band center shifts to lower energies,t hereby reducing the adsorbate-metal bond strength, which alters its selectivity.T he geometric effect results from the decrease of the active metal ensemble size at the surface of ab imetallic NP owing to the presence of alloying metals.S uch decrease can result in (partial) inhibition of structure sensitive reactions for which active metal islands are required, and hence forms at ool to steer the catalyst selectivity.
Ideally,anatomically tailored bimetallic nanocatalyst can be fabricated with meticulously selected NP shape,s ize, composition, and stability to yield maximal activity and full selectivity towards target reaction products.W hilst increasingly realistic catalyst screening methods have been developed by computational approaches, [16] experimental characterization of the bimetallic nanocatalyst is equally required.
Particularly,charting all possible active sites in as ingle bimetallic NP and probing their abundance and chemical and structural nature during catalyst formation and functioning will provide input for the pursued rational design. [17] X-ray-based characterization is ideal for monitoring the electronic and structural state of (bi)metallic nanocatalysts.I nv iew of the decadeslong development of X-ray tools and the advent of X-ray sources with unprecedented opportunities,t his Minireview provides ab rief overview of the current X-ray toolbox for nanocatalyst characterization and its application under relevant conditions.

Bimetallic Nanocatalyst and its Complexity
(Bi)metallic catalysts were initially regarded as static entities.However, the past decades have brought firm understanding that (supported) metal catalysts are dynamic nanomaterials with evolving properties over the catalyst lifetime. Fore xample,m etal nanocatalysts in action can reversibly switch between work and sleep mode,orcommunicate within and between single NPs,reminiscent of living organisms. [18,19] Understanding the intricate nanoscale phenomena which underlie catalyst formation, functioning,and aging is aformi-Bimetallic nanocatalysts are key enablers of current chemical technologies,i ncluding car exhaust converters and fuel cells,a nd play acrucial role in industry to promote aw ide range of chemical reactions.H owever,o wing to significant characterization challenges, insights in the dynamic phenomena that shape and change the working state of the catalyst await further refinement. Herein, we discuss the atomic-scale processes leading to mono-and bimetallic nanoparticle formation and highlight the dynamics and kinetics of lifetime changes in bimetallic catalysts with showcase examples for Pt-based systems. We discuss howi nsitu and operando X-rays pectroscopy, scattering, and diffraction can be used as acomplementary toolboxtointerrogate the working principles of todaysa nd tomorrowsb imetallic nanocatalysts.
dable challenge owing to their complexity.T od ecipher the observed phenomena in greater detail, defining four layers of complexity aids in deconvoluting the structural complexity of bimetallic nanocatalysts.T hese four layers are shown in Figure 1.

Layer 1: Active Site Heterogeneity
ANPisasolid ensemble of atoms with finite dimensions, typically exposing low-indexed facets to minimize the Gibbs free energy.T he NP surface contains low-coordinated atoms (the active sites) situated at corners,edges,kinks,and terraces with different reactivity. [20] Such heterogeneity in the nature of the active sites introduces the first layer of complexity in structure-performance relationships,e ven for the most simple case of am onometallic NP.C uenya et al. showed that differently-shaped 1nmPtNPs have lower onset temperature for 2-propanol partial oxidation owing to ah igher degree of Pt undercoordination. [21]

Layer 2: Second Metal
Thesecond layer of complexity arises owing to the change in the active site number and nature upon introduction of as econd metal to the NP.T he active site distribution on the bimetallic NP surface depends on its mixing configuration, ranging from core-shell to ordered and randomly homogeneous alloys. [11] Them ixing enthalpy and entropy determine the intimacyo fm ixing and degree of alloy ordering, respectively,a nd thereby the active site number and nature. Ty pically,the metal with lowest sublimation temperature will segregate to the NP surface (under specific conditions), affecting the active site number. [5] Notably,n anoalloy formation can lead to NP shape and size changes compared to their monometallic analogues.

Layer 3: External Ligands:Support and Environment
Thet hird layer of complexity arises from the structural and chemical transformation of the bimetallic NP owing to the reaction environment, either gas or liquid, and the support. These interactions provide external ligands to the bare bimetallic NP,w hich can strongly alter its shape,s ize, composition, strain, and electronic properties,a nd thereby the catalytic performance.F or example,M ager-Maury et al. [22] and Taoe tal. [23] demonstrated that H 2 and CO change the shape and size of Pt NPs,r espectively.F renkel et al. [24] provided evidence of gas-and support-induced lattice strain in Pt NPs.T he degree of charge transfer between the support and the NP (Schwabb effect) can affect the NP electron density,a ss hown by Lykhach et al. [25] for Pt/CeO 2 .

Layer 4: Active-Site Modification and Poisoning
Supported bimetallic NPs under reaction conditions are prone to reaction-induced active site modification and poisoning.T he former changes the nature of the metal site and includes the incorporation of subsurface Ha nd Ci nP d hydrogenation catalysts or oxidation state changes at the NP surface. [26,27] These modifications can improve or inhibit the catalytic activity.I nc ontrast, poisoning hinders further reaction and reduces the number of available active sites. Most common are active site blockage by hydrocarbonderived CH x fragments or irreversible adsorption by CO or S. [28,29] NP sintering can be classified under poisoning as well, considering it blocks the original metal site by the active metal itself,causing an irreversible decrease in the number of available sites.

The X-ray Toolbox
X-ray sources have revolutionized materials characterization at an unprecedented pace.T oday,t hird-generation synchrotrons emit collimated X-ray beams with variable energy and spectral brightness 10 orders of magnitude higher than X-ray tubes. [30][31][32] Typically,hard X-rays (> 2000 eV) are used for characterization of (bi)metal catalysts owing to their i) high penetrating power, required for in situ and operando metrology,i i) high scattering cross-section for metal atoms compared to lighter elements contained in hydrocarbons or the support, and iii)element specificity originating from element-dependent X-ray absorption edges.I nc ontrast, soft X-rays (< 2000 eV) have limited penetrative power but show high sensitivity and element specificity towards lighter elements,f or example,c ontained in reaction products (for example,C x H y ), and metal valence-band properties.F igure 2 provides ab rief overview of X-ray tools to study supported metal catalysts,while Figure 3shows which X-ray tools can be used to yield structural or electronic information at specific length scales.Inwhat follows,the advantages and limitations of these X-ray methods are discussed for the study of bimetallic NPs.

X-ray Absorption, Emission, and Photoelectron Spectroscopy
In X-ray absorption spectroscopy (XAS), ac ore-level electron is excited to unoccupied valence states or the continuum. [33] By varying the incident X-ray energy across an absorption edge,the X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure  (EXAFS) are recorded. XANES yields information on the unoccupied valence states and the local geometry,w hile EXAFS provides structural details on the local environment around the X-ray absorber.The main advantage of XAS over other X-ray techniques is that it does not require long-range order and the technique is element-specific.
Thecore hole generated during X-ray absorption is filled by electrons from higher energy levels,l eading to X-ray and Auger emission. Fluorescence detection of the emitted X-rays is advantageous compared to transmission XAS as it allows to i) detect trace amounts with high sensitivity,i i) collect sitespecific XAS based on different spin states,i ii)obtain high energy resolution fluorescence detected (HERFD) XAS owing to reduced core-hole lifetime broadening of fluorescent decay channels,a nd iv) probe the (un)occupied density of states and differentiate between different low-Z ligands by (non-)resonant X-ray emission spectroscopy (XES). [34,35] When the incident X-ray energy exceeds the binding energy of core electrons,t he energy of outgoing photoelectrons can be measured by X-ray photoelectron spectroscopy (XPS), yielding the composition, oxidation state,a nd electronic structure of the X-ray absorbers.The advantage of XPS relative to other X-ray tools is its surface sensitivity, which allows for depth profiling in as mall number of monolayers.A lthough initially the technique required UHV conditions,m ajor progress has been made to allow nearambient pressure (NAP) XPS. [36] 3.2. X-ray Scattering and Diffraction X-ray scattering results from diffuse and/or coherent scattering,t he latter known as Bragg X-ray diffraction (XRD). [37] While intense XRD features mainly appear in the wide-angle X-ray scattering (WAXS, > 58 8)region, diffuse scattering is manifested in both the small-angle X-ray scattering (SAXS, < 58 8)and WAXS region. [38] Diffuse scattering originates from (sub-)nm electronic density contrast, providing i) nm-scale structural information on the NP shape, size,a nd spacing at small angles,a nd ii) -scale structural information on the interatomic distances of X-ray scattering atomic pairs at wide angles.From the latter,apair distribution function (PDF) can be obtained after inverse Fourier transformation. PDF does not require long-range order (similar to EXAFS) but displays sensitivity over longer length scales without element specificity.
To reduce undesired bulk scattering from the catalyst support, NP-decorated planar supports can be probed by grazing-incidence SAXS and WAXS,r espectively GISAXS and GIWAXS. [39,40] By setting the incident angle close to the critical angle of the support, typically below 18 8,n ear-surface scattering is enhanced, which provides high sensitivity to (bi)metal NPs.

In Situ and Operando Metrology
To probe all layers of complexity in bimetallic nanocatalysts,insitu and operando characterization is essential to capture their structural and electronic properties under relevant conditions.O wing to their penetrative nature,h ard X-rays are powerful tools to monitor nanocatalysts under harsh conditions,w hich resulted in the design of dedicated high-temperature and high-pressure reaction cells. [41] Soft Xray tools favor UHV,though major progress has been made to allow gas pressures and elevated temperatures to mimic realistic conditions.F or example,t ime-resolved NAP-XPS gained prominence in operando characterization, [42] taking advantage of small dead-volume reaction cells [43] and fast delay line detectors.X RS can probe soft X-ray edges using Raman scattered hard X-rays,a llowing in situ and operando conditions.

Time-and Spatially Resolved X-ray Characterization
In the past decades,t ime-resolved in situ or operando Xray metrology has become possible owing to the development of i) beamline optics for time-resolved experimentation, ii)high-temperature and high-pressure reaction cells (Section 3.3) and product monitoring,a nd iii)improved detector sensitivities and data transfer.N otable examples include Quick-XAS, [44] Dispersive-XAS,a nd high-energy resolution off-resonant spectroscopy (HEROS), [45] allowing (sub-)second XAS data acquisition for monitoring the temporal changes in catalysts.S cattering and diffraction tools mainly benefited from improved detectors with larger 2D area to capture wider 2q ranges in one snapshot, smaller pixel sizes for better pattern resolution, and higher maximum count rates for the use of more intense X-ray beams.
Current state-of-the-art X-ray nanoscopes have spatial resolutions at best approaching 15 nm, which are still away from the nm length scale required for imaging catalytically relevant single NPs. [30] Even though tools like microfocus XRD [46] and coherent diffraction imaging (CDI) [47] are promising,t od ate (bi)metallic nanocatalyst characterization still mainly uses bulk X-ray tools,m ore and more in tandem with aposteriori data treatment. Forasummary of the latest Figure 3. Overview of structurala nd electronic information content that X-ray tools provide on functionaln anomaterials over different length scales.

Angewandte
Chemie developments in spatiotemporal imaging of heterogeneous catalysts,w er efer to ar ecent review. [30]

APosteriori Data Treatment
As size and information content of datasets increase, smart, often (semi-)automated aposteriori data treatment methods can replace classic time-expensive and comparatively inaccurate analysis. [48] In X-ray spectroscopy,f or example,m odulation excitation spectroscopy coupled with phase sensitive detection (MES-PSD) increases the signal sensitivity towards the active part of the catalyst, for example, the NP surface in reaction, while filtering out the spectator contribution. [49] Wavelet-transformed (WT) XAS can simultaneously determine the atom type (k-space) and location (Rspace) of the X-ray absorbersn eighbors,i nc ontrast to kspace-blind Fourier transformed (FT) EXAFS. [50,51] ForX-ray scattering techniques,c orrelation methods are applied to extract scattering fluctuations over the detector pattern angles,t ermed X-ray cross-correlation analysis (XCCA), [52] and time,n amed X-ray photon correlation spectroscopy (XPCS), [53] yielding structural information and insight in the system dynamics,r espectively.Arelatively recent approach, applicable to ab road set of X-ray methods,i sm achine learning, which can empirically link signal features to material properties if trained thoroughly. [54]

Bimetallic Nanocatalyst Formation
To illustrate todayss tatus and future potential of X-ray tools,h erein, we present showcase examples which mainly focus on Pt-based bimetallic nanocatalysts.T his catalyst class can serve as archetypal example as it has been at the center of decades-long catalysis research, especially for the development and application of X-ray tools.

Birth of aNanoparticle
In industry,i mpregnation methods are widely used to deposit metal precursors on porous supports,f ollowed by drying, calcination, and reduction to yield supported NPs. [57] Filez et al. [50] used WT XAS to study decomposition of aP t(acac) 2 precursor impregnated on aM g(In)(Al)O x support. Prior to calcination, Pt À Obonds are observed typical of the square planar Pt(acac) 2 geometry (Figure 4a). The double-scattering Pt À C À O À Pt foothill to the Pt À Op eak originates from the precursor ligand geometry.After calcination at 650 8 8C, Pt-Mg and/or Pt-Al support peaks appear besides PtÀOb onds,s howing al igand change around Pt owing to acac decomposition and strong Pt-support binding during calcination (Figure 4b). No PtO 2 or Pt NPs are formed as their peaks are absent, which is rapidly checked by WT XAS (Figure 4c,d), suggesting atomic Pt on the support.
Szlachetko et al. [45] used HEROS to monitor the decomposition of Pt(acac) 2 in H 2 at 150 8 8C. By tuning the incident Xray energy below the Pt edge (off-resonant), single-shot XES can be collected by ahigh-resolution dispersive spectrometer with sub-second time resolution. XAS can then be obtained from off-resonant XES by the Kramers-Heisenberg formalism. Self-absorption does not occur in HEROS,o pposed to fluorescence XAS.Atwo-step Pt(acac) 2 decomposition mechanism is seen in H 2 :i )a decrease of the white line (WL) height owing to adecreasing density of unoccupied 5d states,f ollowed by ii)a peak shift showing reduction to metallic NPs (Figure 4e,f). Notably,Saha et al. [58] used in situ total X-ray scattering (1 sr esolution) to study Pt and Pt 3 Gd formation from their precursor states.
An emerging method to deposit metal NPs with (sub-)nm control is atomic layer deposition (ALD). [59] By combining O 2 and N 2 plasma (N 2 *) Pt ALD,Dendooven et al. [56] independently tuned the Pt NP size and center-to-center distance on ap lanar support (Figure 5a). GISAXS showed that after depositing as elected number of Pt NPs by MeCpPtMe 3 -O 2 ALD,subsequent MeCpPtMe 3 -N 2 *ALD only increased their size,w hile keeping the center-to-center distance constant: i) the constant q y -values of the lobe maxima evidence the constant NP spacing, while ii)the shift of the lobe minima in q y and q z to lower values and the appearance of as econdary lobe is characteristic for an increasing NP width and height, respectively.N otably,aQXAS study showed that the Pt NP

Angewandte
Chemie size and spacing can also be tuned after synthesis by redox cycling, using ap re-selected temperature and reduction gas, for example,H 2 or CO. [60] This is in line with an early Tu rbo-XAS study of Nagai et al. [61] showing temperature-dependent in situ redispersion in Pt/CeZrYO x automotive catalysts upon redox cycling.
While the Pt NP size and shape can be extracted from XAS [62] and (GI)SAXS [63] modeling,the electronic properties can be interrogated for example by RIXS. [35] Glatzel et al. [55] used RIXS to map the occupied density of 5d valence states of Pt 6 NPs (by means of the energy transfer) in bare and COadsorbed state (Figure 5b). CO adsorption atop of Pt 6 lowered the 5d band center relative to the Fermi level. The more the 5d band center moves below Fermi level, the weaker the interaction with new adsorbates; [15] thus Pt adsorbates can change catalytic activity.

Alloy Formation:H ydrogen Spillover and Metal Mobility
Ramachandran et al. [64] used XAS,i nsitu XRD and GISAXS to study Pt-In nanoalloy formation starting from Pt NPs on an In 2 O 3 support. In situ XRD during H 2 -TPR showed ag radual Bragg shift from Pt to Pt 3 In due to Inincorporation in face-centered cubic (fcc) Pt (Figure 6a). Pt 3 In alloying is confirmed by aXAS edge blue-shift relative to bulk Pt and an In-contribution in its fcc-type EXAFS fingerprint (Figure 6b). In situ GISAXS showed NP growth during Pt-In alloying (lobe shift to smaller q y and q z values, Figure 6c,d), with subsequent NP height increase upon further heating (secondary lobe shift to smaller q z ). Filez et al. [65,66] further detailed the alloying process as follows: i) dissociation of H 2 on Pt, ii)Hspillover to and iii)long-range transport across the support, iv) (partial) reduction of indium oxide by H, v) short-range transport to Pt, followed by vi)full reduction to In 0 and (vii)Pt-In alloying.
Once formed, determining the size,s hape,a nd composition of bimetallic NPs is crucial to link structural features to catalytic performance.T ao et al. [67] used AP-XPS to study the extent of metal mixing and segregation in bimetallic NPs under different gas atmospheres.B ym easuring at different incident X-ray energies,d epth profiling is achieved, showing that Rh 0.5 Pt 0.5 alloyed NPs have good mixing properties but expose more Rh at the NP surface in vacuum (Figure 7a). Under 100 mTorr reducing H 2 or oxidizing NO gas,t he NP surface composition changes,r esulting in increased Rh segregation (Figure 7b). Timoshenko et al. [68] recently developed apowerful approach to reconstruct 3D atomic models of mono-and bimetallic NPs by combining WT XAS,supervised machine learning,a nd molecular dynamics simulations (Figure 7c). Deriving the 3D NP structure in real-time from in situ or operando XAS could revolutionize nanocatalysis.

Active State
X-ray microscopy shows potential to map the nanocatalyst active state under reaction or relevant conditions.U sing microfocus scanning XRF combined with computed tomography (m-XRF-CT), Price et al. [69] mapped the Pt and Mo composition across aM o-promoted Pt/C catalyst particle during liquid-phase hydrogenation of nitrobenzene with 5 5 mm 2 pixel size.S trong intraparticle heterogeneities were observed, namely Mo residing mostly in the particle core while Pt is strongly abundant at the edge (Figure 8a,b). In contrast to the mmr esolution of m-XRF-CT,B ragg coherent diffraction imaging (CDI) can map in situ the lattice displacement in amodel Pt NP during methane oxidation with 14 nm resolution. [47] Such displacements originate from reactant adsorption and thus allow for active site localization. Strong distortions are observed at the NP surface corner and edge regions owing to methane oxidation, which are restored after reaction (Figure 8c,d).

Dynamic Restructuring
Redekop et al. [70] investigated the dynamic phase changes in activated Mg(Ga)(Al)O x -supported Pt-Ga nanoalloys during 600 8 8CH 2 -O 2 redox cycling (Figure 9a). Such conditions mimic reaction-regeneration cycles in the industrial propane dehydrogenation processes (for example,O leflex) or rapid redox cycles in car exhaust convertors.After one redox cycle, H 2 gas leads to Ga 2 O 3 /Pt reduction causing Pt-Ga nanoalloy formation into aP t-rich intermetallic or solid solution, and aG a-rich phase which disappears over longer timescales, presumably at ransient NP surface alloy,w hich might affect the catalyst performance.
Recently,F ilez et al. [71] refined the steps involved in segregation alloying of Pt 13 In 9 into In 2 O 3 /PtO x and back to Pt 13 In 9 during high-temperature O 2 -H 2 redox cycling (Figure 9b). By kinetic modeling of QXANES data, partial reaction orders,r ate constants,a nd Arrhenius parameters were estimated, allowing to construct the kinetic reaction cycle that steers the dynamic restructuring of Pt 13 In 9 nanoalloys ( Figure 9c). In O 2 ,Pt 13 In 9 decomposition and Pt surface oxidation simultaneously take place with high activation energy in which Pt oxidation is rate-determining (Figure 9d). In contrast, the reverse processes in H 2 steer the equilibrium state from In 2 O 3 /PtO x back to Pt 13 In 9 through In 2 O 3 and PtO x reduction followed by Pt-In alloying,b oth showing low apparent activation energies.K inetic modeling of QXAS can thus identify the reaction steps governing the dynamic restructuring of the nanocatalyst.
Beyond model redox cycling,V irnovskaia et al. [72] found with AP-XPS that the surface of Pt-Sn nanocatalysts is Snenriched during high-temperature alkane dehydrogenation and Sn remains partially oxidized even in ah ighly reducing hydrocarbon environment. Barbosa et al. [73] showed by AP-XPS during methanol steam reforming that the surface of Pt-In NPs is In-enriched, suppressing CO formation during reaction.

Deactivation by NP Sintering
During NP sintering,the number of active sites decreases by NP growth. Iglesias-Juez et al. [74] simulated possible Pt-Sn NP structures based on Pt-Pt and Pt-Sn EXAFS coordination numbers [75] for different dehydrogenation-regeneration C 3 H 8 -O 2 -H 2 cycle(s), showing:1 )NPs ize increase,2 )progressive Sn-enrichment of the NP,a nd 3) increased mixing from core-shell-to random-type alloys (Figure 10 a). A gradual NP size increase is also observed by Filez et al. [71] via adecreased XANES WL height under O 2 during 60 H 2 -O 2 redox cycles (Figure 10 b). In O 2 ,a nI n 2 O 3 /PtO x composite is formed in which PtO x consists of ametallic core with oxidized surface,t he latter yielding increased WL heights.W ith decreasing dispersion, which is due to sintering during redox cycling, the fraction of oxidized surface decreases,l eading to aW Ld ecrease.N otably,S olano et al. [76] also studied Pt NP sintering with in situ GISAXS under different O 2 partial pressures to monitor the NP size and spacing in real time.The observed behavior indicates the key role of the outer PtO 2 shell, stable at low temperature,a nd its thermal reduction creating mobile species that trigger particle sintering.

Conclusions and Outlook
Ac omplementary X-ray toolbox is currently available to extract the size,s hape,c omposition, spacing,a nd electronic properties of bimetallic NPs under in situ or operando conditions.T hese tools include X-ray absorption, emission, and photoelectron spectroscopy along with scattering and diffraction, often in combination with advanced aposteriori data treatment. This toolbox currently provides mechanistic insights into the formation, functioning,a nd deactivation of nanocatalysts in real time,l eading to am ore detailed understanding of bimetallic NPs.
Future trends will favor increased spatiotemporal resolution of the X-ray experiment. On the one hand, X-ray techniques allowing milli-to femtosecond time resolution have strong potential to uncover currently scarce kinetic descriptions of transient changes in catalytic solids and molecular adsorption-desorption phenomena under reaction conditions,r espectively.O nt he other hand, the avenue of spatially resolved X-ray micro-and nanoscopy will provide   (1)(2)(3)(4)(5) simultaneous Pt 13 In 9 segregation, Pt surface oxidation, (5-8,1) consecutive Pt reduction and Pt-In alloying. c) reaction mechanism includingArrhenius parameters as derived from QXAS kinetic modelling. d) E a Àln(k)p lot for (red, top) Pt oxidation, (green, top) Pt-In segregation in O 2 and (green, bottom) Pt-In alloying and (red, bottom) PtO x reduction in H 2 .The rectanglewidth represents the range to which ln (k)varies over the measured temperaturer ange; its height is the 66 %confidence interval of E a. Adapted from Refs. [70,71].

Angewandte
Chemie the means to map the composition and structure of metal catalyst particles and single crystals under real conditions,for example by (S)TXM, [77,78] m-XRD-CT, [46] or CDI, [47] uncovering particle heterogeneities commonly present in heterogeneous catalysts. [79,80] As the spatial resolution of state-of-theart X-ray nanoscopes (ca. 15 nm) still exceeds the dimensions of catalytically relevant NPs,m achine learning methods for obtaining the 3D NP structure show disruptive potential to revolutionize this field of research. [30,68] Methods which increase the sensitivity of at echnique,s uch as wavelet analysis [50] or MES, [49] can be used to efficiently extract apparently hidden information.
Future advances greatly rely on new X-ray sources.T he advent of X-ray free-electron lasers (XFELs) announces highly coherent, intense,a nd short-pulsed X-ray beams, facilitating the investigation of matter dynamics on atomic length scales with femtosecond time resolution. [81] Furthermore,l aboratory-based XAS,X ES,a nd (GI)SAXS become increasingly popular, allowing everyday access to traditional synchrotron techniques. [82] These evolutions hold bright prospects for X-ray tools in nanocatalysis,a nd more general in functional materials research.