Design of efficient Pt-based electrocatalysts through characterization by X-ray absorption spectroscopy

Abstract

A method is described to determine the internal structure of electrocatalyst nanoparticles by in situ X-ray absorption spectroscopy (XAS). The nondestructive spectroscopic technique typically utilizing synchrotron radiation as the source measures changes in the X-ray absorption coefficient as a function of energy. The bulk technique has found its use for materials characterization in all scientific areas, including nanomaterials. The analysis of the internal structure of nanoparticles reveals interatomic distances and coordination numbers for each element, and their values and mutual relations indicate whether the elements form a homogeneous or heterogeneous mixture. The core-shell heterogeneous structure in which certain elements are predominantly located in the core, and others form the encapsulating shell is of particular importance in catalysis and electrocatalysis because it may reduce the amount of precious metals in nanoparticles by replacing the atoms in the core of nanoparticles with more abundant and cheaper alternatives. The examples of nanoparticle structures designed in the laboratory and the approach to model efficient catalysts through systematic analysis of XAS data in electrochemical systems consisting of two and three metals are also demonstrated.

Keywords

X-ray absorption spectroscopy / EXAFS / XANES / nanocatalysts / core shell

Introduction

Limited number of techniques can probe sample catalysts in situ, for example during varying the electrochemical potential. Vacuum techniques, such as X-ray photoelectron spectroscopy (XPS), Auger electron spectroscopy (AES) and low-energy electron diffraction (LEED), although primarily surface sensitive, cannot be combined with aqueous electrochemical systems due to water evaporation and, therefore, provide only information about initial and final states of the catalyst surface, i.e., before and after the electrochemical reaction.In situ infrared reflection-absorption spectroscopy (IRRAS) and differential electrochemical mass spectroscopy (DEMS) give information on the nature of particles and their changes in the solution and at surface-solution boundary during the potential excursion, but provide little insight in the changes of the internal catalyst structure. For the latter studies, the X-ray absorption spectroscopy (XAS) is the ideal technique because the X-rays have sufficient energy to penetrate through the solution layer and into the catalyst, providing information on the structure and changes in the tridimensional environment surrounding the elements comprising the nanoparticle. The technique can be applied to both crystal surfaces and amorphous materials in any physical state, as well as to thin films with the thicknesses of one atomic layer or less.

The major attention in the last years was devoted in developing nanomaterials that successfully catalyze an electrochemical reaction, yet contain minimal amount of rare and expensive precious metals. Despite the extensive research, the best catalysts for both anodic and cathodic reactions in fuel cells still contain certain amount of platinum. Minimizing its content is accomplished either by alloying platinum with other more abundant metals, or by forming a Pt thin layer (often a monolayer) on the surface of the nanoparticle made by other metals. In the latter systems all Pt atoms are exposed to the solution. The inner structure of the nanoparticle of two or three elements is easily determined by XAS spectroscopy. The technique determines the spatial orientation of atoms surrounding the atom which absorbs the X-rays; because it is element-sensitive, it returns information on the three-dimensional atomic network, including the number of bonds (i.e., coordination numbers) of one element to others comprising the nanoparticle. The relationship between the coordination numbers determines the inner structure of the nanoparticle, i.e., whether it is a homogeneous solid solution of randomly distributed atoms, or heterogeneous solution in which atoms of different kind are segregated either within one nanoparticle (so called core-shell structure) or form different particles so that each type of atoms is preferentially located in its own particle. The coordination numbers are also related to the nanoparticle diameter, and XAS is able to detect nanoparticles of any size larger than individual atoms.

Operating principles

Matter_interacts with the X-ray beam by the statistical effect. An atom in ground state absorbs X-rays and ejects an electron (so-called photoelectron) from its inner shell. Depending on the shell it is ejected from, the K, L, M, and so on electronic transitions are distinguished, i.e., K-, L- or M- edges, where K-edge corresponds to eviction of electrons from 1ssubshell. There are three L-edges, labeled such that the L1 edge corresponds to an electronic transition from 2ssubshell, while the L2 and L3 edges denote the transition from 2p subshell. The latter two transitions have different energies due to the spin-orbital coupling, so that l + s corresponds to L2, and l – s corresponds to L3 transition. Similarly, there are five M transitions (one 3s, three 3p and two 3d), but they are generally of little use in the XAS technique [1].

The change in absorption coefficient (m) around the edge is followed in XAS as the function of X-ray energy E; graphically represented, the spectrum is in principle similar to the one shown in Fig. 1(a). The XAS spectrum is divided into two parts: the spectral part withinca. 50 eV near the edge is called X-ray absorption near-edge spectroscopy (XANES), whereas the part of the spectrum further away from the edge is termed extended X-ray absorption fine structure (EXAFS). The two parts of the spectrum offer related, yet different information on the analyzed system.

X-ray absorption coefficient decreases with energy by the relation m ~ 1/E⁴, but near the edge it sharply increases [1,2]. If the atom that absorbs the monochromatic X-rays (the so-called photoabsorber) were completely isolated,m would continue to decline by the above m-E relation following the sharp rise at the edge. However, in the condensed system the electron wave traveling from the photoabsorber is reflected by the electron cloud of a neighboring atom (scatterer) and travels back. The reflected wave interferes with the new electronic wave coming out of the photoabsorber, producing a dumping sinusoidal function superposed to them ~ 1/E⁴ curve. These interference oscillations depend on nature and the three-dimensional arrangement of scattering atoms, the so-called fine structure in EXAFS. By subtracting the spectrum of an isolated atomm₀(E), dividing it with the edge jump Dm₀ and normalizing the oscillations to unit absorption, one obtains the graph similar to that in Fig. 1(b). By default, the abscise is given in the wave numberk, which depends on the difference of X-ray energy E and the energy of the edge E₀ as k = 0.512 (E−E₀)^1/2 and is expressed in inverse Angstroms, Å⁻¹. To enlarge the amplitude of the oscillations especially at higher values of k, the normalized absorption c(k) is multiplied by kⁿ (n = 1, 2, 3 …). The sine wave in Fig. 1(b) is called the X-ray absorption spectrum ink-space. By Fourier transformation of this spectrum, one obtains the Fourier transform magnitude as a function of the distance from the photoabsorber, the so-called spectrum inR-space (see below). Distances in R-space spectrum are related to the actual distance between the atoms of the photoabsorber and scatterer, but are usually shorter by 0.3 to 0.5 Å due to phase shift [1,2]. The true distances between the atoms are revealed by subsequent fitting the experimentally obtained spectrum to the theoretical one, based on the predicted structure of the nanoparticle (vide infra).

Because the XANES part of the XAS spectrum corresponds to the multielectron transitions that are too complicated to express mathematically, no equation describes this spectral region. Hence, the XANES spectrum is modeled qualitatively (and in certain instances also quantitatively) by linear combination or principle component (LCA or PCA) analyses. The edge shift toward higher or lower energies is the consequence of the oxidation state of the photo-absorber and can also be modeled. Furthermore, the XANES spectrum often shows unique “pre-edge” features that are easily identified and can be used in fingerprinting. If the sample is a mixture of two or more components, one can carry on the LCA technique in an attempt to make a match of the spectrum of the sample by adding together fractions of the spectra of various standards. The computer technique works well as long as the standards for all constituents are known. On the other hand, if one has a limited knowledge of the sample but plenty related spectra, as in a series of samples with various compositions or a single sample under different conditions, the PCA computer routine can be used to identify the set of components that account for the spectra in the decreasing order of importance. The technique provides information whether individual constituents are present and how many of them are in the sample. It is most often used when the spectra are taken “in-operando” while conditions like temperature, pressure, voltage, or concentration cause the change in the XANES spectra.

The spectrum in the EXAFS region can be relatively easily described mathematically. The expression for the normalized absorptionc(k) contains a pre-exponential part (amplitude) proportional to the nature and number of identical scatterers and inversely proportional to the photoabsorber-scatterer distance [3]. This amplitude is multiplied by a sinusoidal function that includes the phase shift, and two dampening terms that depend on various factors, including losses caused by the multielectron effect, the probability of elastic scattering, and the variance in the photoabsorber-scatterer distance. EXAFS spectrum processing is based on varying the parameters in EXAFS equation so that the experimental spectrum best matches the theoretical one. The theoretical spectrum is constructed from the crystal lattice of a selected model featuring scatterer atoms, their coordination number, and the distance to the absorber. At the end of the fitting routine, the computer program lists the resulting parameters including the coordination number of the elements and the interatomic distances. The values and relations between them are in direct connection with the internal structure of the sample. For example, for the homogeneous distribution of two alloyed metals in the nanoparticle (Fig. 2), it has been shown that their coordination numbers are directly related to their mole fractions [4]. It is further found that the following conditions can be assumed for the alloy system: ① both metals must have the same number of neighbors, so the total coordination number of the metal M1 must be equal to that of the other metal M2; ② the distance between the atoms of the two metals M1-M2 must be greater than the distance M1-M1 but smaller than M2-M2, because of different atomic diameters of the two metalsd_M1<d_M2. Similarly, for the other two nanoparticle systems shown in Fig. 2, a set of appropriate coordination numbers and interatomic distances that describe the system can be derived [5]. Since the EXAFS technique is element-specific, the processing of EXAFS spectra allows complete analysis of the particles, including its size and composition, which is of great importance for the electrochemistry and electrocatalysis.

Experiment

A good EXAFS spectrum must have a signal-to-noise ratio greater than 1000. To obtain such a spectrum in a reasonable time (few minutes), the flux of incident X-rays must be of the order of 10¹⁰ photons per second or greater, and the width of the monochromatic beam about 1 eV or smaller. These requirements can be only found in the X-rays produced by a synchrotron. Fortunately, there are about a hundred synchrotron sources in the world, many of which have more than one experimental station (beam line) for XAS experiments.

The spectro-electrochemical cell for in situ measurements used in the studies presented below allows for spectral collection in both transmission and fluorescent mode (Fig. 3). It consists of an electrocatalyst deposited on carbon cloth (or carbon paper), a Nafion membrane that separates the catalyst (working electrode) from the auxiliary electrode (usually Pt foil) placed outside of the X-ray path, and Teflon rings that seal the working-auxiliary electrode sandwich pair placed in a grove of the body of the cell. The groove is filled with electrolyte in contact with the third (reference) electrode, also placed away from the X-rays path. The EXAFS spectrum is measured at the constant potential once the current reaches the steady-state [6].

Results and discussion

Although great research effort has been devoted to development of better and cheaper catalysts for fuel cells in recent decades, platinum is still unrivaled catalyst for both anodic and cathodic electrochemical reactions. It appears that any successful catalyst must contain some Pt in it. However, pure platinum gradually loses its catalytic activity due to a number of reasons, including surface poisoning. For the commercial utilization, the catalyst must be durable, the Pt content should be as small as possible, and yet the catalyst must reach or overtake the activity of pure Pt. Several successful catalysts have been developed in the laboratory [5]; here a few examples are shown which best illustrate the power ofin situ XAS techniques in resolving the inner structure of fuel cell catalyst for ethanol oxidation and oxygen reduction.

Oxidation of ethanol

Due to its potential energy density, available production from renewable sources, and ease of storage and transport, ethanol is considered to be the perfect fuel for use in fuel cells in vehicles [7]. Unfortunately, ethanol oxidation pathway is complex, and on pure Pt surface the partial oxidation pathway prevails, which is the main reason that this reaction is not yet commercially utilized. A complete oxidation of the alcohol produces 12 electrons per molecule and yields water and carbon dioxide as the products, but on a pure platinum surface the ethanol oxidation products are mainly acetaldehyde and acetic acid, yielding 2 or 4 electrons per molecule in the practical potential range of fuel cells<0.6 V. Thus, the ethanol C-C bond remains intact; the partial products can be further oxidized only at potentials higher than 0.6 V, when Pt surface becomes covered by oxygen-containing particles.

The complete oxidation pathway can be increased somewhat in binary catalytic systems by alloying platinum with other metals that are covered with oxides at potentials lower than 0.6 V, for instance with tin, but partial oxidation products still predominate. Studies have shown that a successful catalyst for ethanol oxidation should contain three metals [8]. Thus far, the best catalyst for the reaction is the ternary Pt-Rh-SnO₂nanoparticle catalyst that successfully splits the C-C bond at low potentials, and oxidizes ethanol to CO₂with minor quantity of partial oxidation products [9]. The catalytic process is the result of the interplay of all components, so that rhodium adsorbs ethanol and cleaves the C-C bond, Pt-modifies Rhd-orbitals for easy removal of the oxidation products, and SnO₂ transfers oxygen particles resulting from water dehydrogenation at the oxide onto the Pt-Rh surface where they are oxidizing ethanol. The transmission electron microscope (TEM) results show that the SnO₂particles are about 10 nm in size, while the Pt-Rh particles are within 1–3 nm. Measurements of X-ray diffraction (XRD) and energy-dispersive scanning transmission electron microscopy (STEM-EDS) show that the metal particles are in one phase, which means that the Pt and Rh atoms are homogeneously distributed throughout the particle and there are no particles in which either metal prevails. The mole fractions of Pt and Rh are approximately 0.6 and 0.4 forx (Pt) and x (Rh), respectively, as established by inductively coupled plasma atomic emission spectroscopy (ICP-AES).

The results of the above techniques have been confirmed by analysis of the in situ EXAFS measurements. The XAS spectra of SnO₂ at various potentials have been found to be potential-independent, ie., the oxide surface composition does not change during the potential excursion [5,9]. To confirm the assumption that Pt and Rh make solid solution, selected EXAFS spectra obtained at Pt and Rh edges have been analyzed by fitting them to the theoretical spectra of Pt-Rh alloy. These theoretical spectra have been constructed by choosing the space group in which the first metal (for instance, Pt) forms the crystal lattice and by the replacement of one atom in its first coordination sphere by the other metal (e.g. Rh). Each of the two theoretical spectra (Pt-Rh and Rh-Pt) has been used in the fit: the former has been used to fit the experimental spectrum of Pt L3-edge, and the latter for that of Rh K-edge [5]. The resulting fits are shown in Fig. 4. The obtained sets of coordination numbers for each metal have been used to establish the inner structure of the particle and to calculate the mole fraction of the Pt-Rh nanoparticle.

As shown above, the mole fraction of the two constituents in a binary homogeneous mixture is directly related to the ratio of their coordination numbers, i.e.,x(Pt)/x(Rh) = N(Rh-Pt)/N(Pt-Rh). The obtained ratio of coordination numbers, N(Rh-Pt)/N(Pt-Rh) = 2.1±0.3, agrees within the experimental error with the ratio of mole fractions obtained by ICP-AES technique, 1.6±0.2. Further, a homogeneous solid solution should have the sum of all coordination numbers for one metal equal to that of the other. The total coordination numbers for platinum and rhodium obtained by EXAFS analysis wereN(Pt-M) = N(Pt-Pt) + N(Pt-Rh) = 9.5±0.8, and N(Rh-M) = N(Rh-Rh) + N(Rh-Pt) = 10.8±0.8. The close values of the sums confirm the homogeneous distribution of the Pt and Rh atoms within the Pt-Rh nanoparticles. This is strengthened further by the bond lengths; as obtained by the fit, the Pt-Rh bond length (2.725±0.004 Å) is shorter than that of Pt-Pt but larger than that of Rh-Rh (2.743±0.003 and 2.705±0.005 Å, respectively), which is expected for a solid solution. Finally, the average total coordination number for the two metals, 10.0±0.8, indicates that the Pt-Rh particle size is between 1.5 and 3.5 nm, which agrees well with the value obtained by TEM analysis. As can be seen from the foregoing, EXAFS spectroscopy replaces the data of four independent experimental techniques, which best describes the strength of this spectroscopic method.

The content of the PtRhSnO₂/C catalyst has been further optimized by varying the mole fractions of the three constituents and measurements of their activity. Thein situ infrared spectroscopy shows that the best catalyst, i.e., the one that produces the highest relative amount of CO₂ in comparison to the partial oxidation products, has the mole fraction of Pt:Rh:Sn= 3:1:4 [10]. The XANES analysis confirms the optimal content of the constituents. It also further demonstrates that the best catalyst should contain Pt and Rh in the metallic (zero-valent) state in the whole potential region of practical interest for fuel cells, and that the mole fraction of Sn (in the form of SnO₂) must be carefully chosen, as too large a content may partially oxidize the noble metals when the oxygen particles from SnO₂ are not used completely in ethanol oxidation reaction [11].

Reduction of oxygen

While there are a number of potential fuels for the anodic reaction in fuel cells, the fuel for the cathode side is almost exclusively oxygen, usually from air. A new type of catalyst has been developed for the oxygen reduction reaction: Pt-monolayer catalyst, which has several unique properties in addition to the low amount of Pt in the particle [12–14]. The Pt monolayer on nanoparticles made of a suitable substrate (e.g. metal or metal alloy) is formed by Pt-Cu galvanic displacement in a previously-deposited Cu-monolayer by under potential deposition [12]. Here, the XAS characterization of the following catalysts are described: Pt monolayer at palladium nanoparticles, surrounded by carbon for electric contact, Pt_ML/Pd/C; Pt monolayer on Pd-Au alloy, Pt_ML/PdAu/C; and Pt monolayer on nanoparticles composed of nickel and iridium, wherein Ni atoms are predominantly in the center while iridium atoms are predominantly located in the shell surrounding the center, i.e., Pt_ML/IrNi/C [12–14]. All of the above catalysts are found to be more active than the pure platinum. Furthermore, the latter catalyst is of special importance for commercial use since it contains the least amount of precious metals.

The XAS spectra of Pt_ML/Pd/C catalyst substantially differ from that of Pt and Pd metal foils (Fig. 5). In addition to the differences in the amplitudes, observed at both spectra, the spectrum of Pt L3 ink-space also shows differences in the phase of the oscillations [6,12]. The smaller amplitudes of the catalyst reflect the smaller cluster sizes around atom-absorber in the nanoparticle vs. that in the bulk. However, the variance in phase indicates dissimilar spatial environment of platinum in the Pt-monolayers from that in the bulk Pt.

Coordination numbers have been obtained by fitting of the experimental spectrum with the theoretical one in which one atom in the first coordination sphere is replaced by another kind of atom, in the manner described previously (vide supra). The obtained coordination numbers, viz.N(Pt-Pt) = 5.8±0.8, and N(Pt-Pd) = 2.7±0.7 are close to that expected for a monolayer of Pt on a flat semi-infinite Pd (111) surface, (6 and 3). For a core-shell nanoparticle, it is expected that the coordination numberN(Pt-Pt) should be smaller, to some extent, due the edges which have a smaller coordination number. For a cuboctahedron particle of 4.2 nm in diameter (as measured by X-ray diffraction), the coordination numbers should be between 5 and 6 forN(Pt-Pt), and about 3 for N(Pt-Pd), which is consistent with the values obtained by fitting. Furthermore, the bond lengths for Pt-Pt and Pt-Pd obtained by fitting are lower than those of Pt-Pt in the bulk. It is, therefore, concluded that the nanoparticles are partly compacted. The compression of the nanoparticle is likely the cause for the three times higher activity of Pt_ML/Pd/C catalyst for oxygen reduction vs. that of pure Pt [15]. In addition, the reduction in the Pt-Pt distance increases the protection of the interior palladium core from corrosion and dissolution in acidic electrolyte during the potential cycling. It should be noted that Pd atoms are dissolving slowly, as after a large number of potential cycles (>100000) between 0.6 and 1.4 V, the Pd XAS signal could no longer be detected, and the activity of the catalyst decreases by about 37% of the original value. However, even such nanoparticles with a hollow core still shows activity which exceeds that of the nanoparticle with the same diameter and made of Pt atoms only [12].

It is envisaged that the addition of larger atoms in the core would diminish the core dissolution, as such larger atoms may clog the imperfections in the Pt monolayer through which Pd atoms escape. Indeed, the addition of 10% gold forms the Pd₉Au alloy core; upon covering it with a monolayer of Pt, an active catalyst for oxygen reduction is obtained that shows an excellent stability, as after 100000 cycles the catalyst exhibits only 8% decreased activity compared to the original value. Furthermore, the Pd EXAFS spectra inR-space at different potentials shows no noticeable peaks corresponding to the Pd-O, except at the highest potential of 1.12 V [13].

Further progress in the discovery of an efficient catalyst for oxygen reduction with the least amount of precious metal is achieved by Pt monolayer deposition on a core consisting of iridium and nickel [14]. The internal structure of the catalyst has been determined by fitting the XAS spectra to the theoretical ones, constructed in the manner described above. Simultaneous fitting of Ni K-edge and Pd L3-edge has been performed with reasonable constrains that reduce the number of the fitting parameters and produce better agreement with the data. As expected, the Ir-Ir bond length in the nanoparticle is shorter than that of bulk iridium and larger than that in the bulk nickel.

Interpretation of coordination numbers for this type of nanoparticles is somewhat more complicated, because the relationship between coordination numbers obtained by the fit does not correspond to the ratio of mole fractions expected for homogeneous distribution of the two metals. Namely, if the atoms Ir and Ni form solid solution, the ratio of mole fractionsx(Ni)/x(Ir) should be equal to the ratio of coordination numbers N(Ni-Ni)/N(Ni-Ir). However, the ratio of coordination numbers obtained by the fit is almost twice larger than the ratio of mole fractions (1.2 and 0.56, respectively), indicating that Ir and Ni form heterogeneous nanoparticle. The large value ofN(Ni-Ni) indicates that nickel is preferentially in the core of the nanoparticle, whereas iridium is in the shell. However, the coordination number values do not match those expected for a core-shell particle with a monolayer shell [5]. With the help of an internet program [16], it is possible to distinguish the particle as Pt_ML/ Ir_2ML/ IrNi, i.e., having two layers of Ir between a Pt monolayer and IrNi alloy. This internal structure has been proven by the X-ray diffraction and scanning electron microscopy [14]. This catalyst has worked well in the tests, with the mass activity more than three times greater than the commercially available Pt catalyst of approximately equal nanoparticle size, and the stability significantly better than the commercial catalyst after 50000 cycles.

Oxidation and Pt nanoparticles

The electrochemical cell (Fig.3) makes it possible to measure both in situ XAS and XRD of the same sample. In situ XRD patterns from the same specimen measured immediately after the XAS at each electrochemical potential shows fairly small broadening of the peaks with increasing potentials; the particle size changes from 2.6 to 2.1 nm by applying potentials from 0.41 to 1.51 V, indicating that the thickness of the oxide formed is approximately monolayer thick or less. Although the potential to 1.51 V significantly enhances the Pt white line due to oxidation, the oxide is confined only on the top surface layer of the nanoparticles [17].

The delta mu (Dm) technique has been developed as a surface-sensitive (surface-sensitive) method to identify surface/adsorbate interactions [18,19]. By subtracting spectra of the same sample at two different potentials, the method isolates the surface-sensitive interactions since those of bulk metal–metal are eliminated by the subtraction. The obtainedDm spectra are interpreted by the comparison with theoretical curves made based on crystallographic models. Figure 6 depicts theDm spectra of the Pt/C catalyst at 0.71, 0.91, and 1.11 V, calculated by the equation (Dm(V) = m(V) −m(0.41 V)), where m(0.41 V) is the reference signal of Pt/C catalyst that is free from adsorbate species on the Pt surface. TheDm (0.71 V) shows a small positive peak at a few eV above the E₀. With a further increase in potential, its intensity increases significantly, and the peak position shifts slightly toward higher energy values. The observation is interpreted to O or OH adsorption in a-top sites of Pt at 0.71 V, followed by n-fold bonded configuration at higher potentials. Another marked feature is that a negative peak slightly below theE₀ starts to appear at potentials above 0.91 V, which could be attributed to the formation of subsurface O due to a “place- exchange” process, as indicated by comparison to the theoretically calculatedDm spectral signatures [18–20]. The interfacial place exchange of the adsorbed O (OH) and platinum atoms then forms a Pt–O quasi-3D lattice at higher potentials.

To determine the fraction of metallic Pt and PtO₂ components at different potentials, the linear composition analysis (LCA) technique has been employed. By fitting analysis of EXAFS spectra it is found that the molar contents of Pt metal and PtO₂ at the highest potential used in the study (1.51 V) were 74% and 26%, respectively [17]. As in situ XRD measurements revealed, the oxide formation at 1.51 V is confined only to the top surface layer on the nanoparticles. Calculation of the ratio of surface atoms to total atoms as a function of particle size givesN_s/N_t= 0.50 for icosahedron and cuboctahedron nanoparticles with a diameter of 2.6 nm, indicating that 76% of surface atoms are oxidized at 1.51 V, assuming the oxidation takes place only on the top layer of the nanoparticles.

Conclusions

The in situ XAS has become an irreplaceable technique for studies in the field of electrocatalysis. The strength of the technique to characterization of the internal structure of nanoparticles is illustrated through selected examples of fuel cell electrocatalysts for the ethanol oxidation and oxygen reduction reactions. By the spectral analysis of XAS spectra, details on oxidation states, electronic states, coordination numbers, and bond length, as well as their changes during the electrochemical reaction can be obtained. Thus, a direct insight in the relationship between the structure and activity of the catalyst may be found. In addition, thein situ XAS technique gives specifics comparable to that of many experimental techniques often used in laboratory practice. EXAFS discloses the molar composition (like ICP-AES), determines particle size (as TEM), internal structure (like STEM-EDS), and the bond length between atoms (like XRD), and can be applied to nanoparticles smaller in size than the practical limit of most aforementioned techniques. Given that synchrotrons are now generally available throughout the world and that their use is often free, researchers are able to obtain time on an XAS beam line and use the acquired data to understand better the reaction mechanism on the catalyst surface and to formulate the approach to the construction of new, more stable and active catalysts.

References

[1] Calvin  S. XAFS for Everyone. Boca Raton, FL: CRC Press-Taylor and Francis Group, 2013

[2] Bunker  G. Introduction to XAFS: A Practical Guide to X-ray Absorption Fine Structure Spectroscopy. Cambridge: Cambridge University Press, 2010

[3] Stern  E A . Theory of the extended X-ray absorption fine structure. Physical Review B: Condensed Matter and Materials Physics, 1974, 10(8): 3027–3037

[4] Frenkel  A. Solving the 3D structure of metal nanoparticles. Zeitschrift fur Kristallographie, 2007, 222(11): 605–611

[5] Sasaski  K, Marinkovic   N S. X-ray absorption spectroscopic characterization of nanomaterial catalysts in electrochemistry and fuel cells. In: Kumar C S S R. (Ed.) X-ray and Neutron Techniques for Nanomaterials Characterization, Chapter 6, and the literature cited therein. New York: Springer, 2016

[6] Sasaki  K, Wang   J X, Naohara   H, Marinkovic  N ,  More  K ,  Inada  H ,  Adzic  R R . Recent advances in platinum monolayer electrocatalysts for oxygen reduction reaction: scale-up synthesis, structure and activity of Pt shells on Pd cores. Electrochimica Acta, 2010, 55(8): 2645–2652

[7] Lamy  C, Coutanceau   C, Leger  J M . The direct ethanol fuel cell: a challenge to convert bioethanol cleanly into electric energy. In: Barbaro P, Bianchini C. (Eds.) Catalysis for sustainable energy production. Weinheim, Germany: WILEY-VCH Verlag GmbH & Co. kGaA, 2009

[8] Demirci  U B . Theoretical means for searching bimetallic alloys as anode electrocatalysts for direct liquid-feed fuel cell. Journal of Power Sources, 2007, 173(1): 11–18

[9] Kowal  A, Li   M, Shao  M ,  Sasaki  K ,  Vukmirovic  M B ,  Zhang  J ,  Marinkovic  N S ,  Liu  P ,  Frenkel  A I ,  Adzic  R R . Ternary Pt/Rh/SnO₂electrocatalysts for oxidizing ethanol to CO₂. Nature Materials, 2009, 8(4): 325–330

[10] Li  M, Kowal   A, Sasaki  K ,  Marinkovic  N S ,  Su  D ,  Korach  E ,  Liu  P ,  Adzic  R R . Ethanol oxidation on the ternary Pt–Rh–SnO₂/C electrocatalysts with varied Pt:Rh:Sn ratios. Electrochimica Acta, 2010, 55(14): 4331–4338

[11] Li  M, Marinkovic   N, Sasaki  K . In situ characterization of ternary Pt–Rh–SnO₂/C catalysts for ethanol electrooxidation. Electrocatalysis, 2012, 3(3–4): 376–385

[12] Sasaki  K, Naohara   H, Cai  Y ,  Choi  Y M ,  Liu  P ,  Vukmirovic  M B ,  Wang  J X ,  Adzic  R R . Coreprotected platinum monolayer shell high-stability electrocatalysts for fuel cell cathodes. Angewandte Chemie International Edition, 2010, 55(8): 2645–2652

[13] Sasaki  K, Naohara   H, Choi  Y M ,  Cai  Y ,  Chen  W H ,  Liu  P ,  Adzic  R R . Highly stable Pt monolayer on PdAu nanoparticle electrocatalysts for the oxygen reduction reaction. Nature Communications, 2012, 3: 1115

[14] Sasaki  K, Kuttiyiel   K A, Barrio   L, Su  D ,  Frenkel  A ,  Marinkovic  N ,  Mahajan  D ,  Adzic  R R . Carbon supported IrNi core-shell nanoparticles: synthesis, characterization, and catalytic activity. Journal of Physical Chemistry C, 2011, 115(20): 9894–9902

[15] Zhang  J L ,  Vukmirovic  M B ,  Xu  Y ,  Mavrikakis  M ,  Adzic  R R . Controlling the catalytic activity of platinum-monolayer  electrocatalysts for oxygen reduction with different substrates. Angewandte Chemie International Edition, 2005, 44(14): 2132–2135

[16] Glasner  D, Frenkel   A I. Geometrical characteristics of regular polyhedra: application to EXAFS studies of nanoclusters. AIP Conference Proceedings, 2007, 882(1): 746–748

[17] Sasaki  K, Marinkovic   N S, Isaacs   H S, Adzic   R R. Synchrotron-based in situ characterization of carbon-supported platinum and platinum monolayer electrocatalysts. ACS Catalysis, 2016, 6(1): 69–76

[18] Arruda  T M ,  Shyam  B ,  Ziegelbauer  J M ,  Mukerjee  S ,  Ramaker  D E . Investigation into the competitive and site-specific nature of anion adsorption on Pt using in situ X-ray absorption spectroscopy. Journal of Physical Chemistry C, 2008, 112(46): 18087–18097

[19] Teliska  M, O’Grady   W E, Ramaker   D E. Determination of O and OH adsorption sites and coverage in situ on Pt electrodes from Pt L23 X-ray absorption spectroscopy. Journal of Physical Chemistry B, 2005, 109(16): 8076–8084

[20] Conway  B E . Electrochemical oxide film formation at noble metals as a surface-chemical process. Progress in Surface Science, 1995, 49(4): 331–452

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg