Intermetallic: A Pseudoelement for Catalysis - Accounts of Chemical

Dec 8, 2017 - The interval between the Fermi level and the top of the d band is closely related to the selectivity of CO2 for the SRM: the larger the ...
1 downloads 12 Views 2MB Size
Article pubs.acs.org/accounts

Cite This: Acc. Chem. Res. 2017, 50, 2879−2885

Intermetallic: A Pseudoelement for Catalysis Published as part of the Accounts of Chemical Research special issue “Advancing Chemistry through Intermetallic Compounds”. A. P. Tsai,*,†,‡ S. Kameoka,† K. Nozawa,§ M. Shimoda,‡ and Y. Ishii∥ †

Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Sendai 980-8577, Japan National Institute for Materials Science, Tsukuba 305-0047, Japan § Department of Physics and Astronomy, Kagoshima University, Kagoshima 890-0065, Japan ∥ Department of Physics, Chuo University, Tokyo 112-8551, Japan ‡

CONSPECTUS: A clear correlation between electronic structure and CO2 selectivity for steam reforming of methanol (SRM) was obtained with PdZn, PtZn, NiZn, and PdCd intermetallics on the basis of experiments and calculations. In order to rule out the effects of oxide supports, the intermetallic powders were simply prepared by alloying in an arc furnace followed by crushing in a mortar. PdZn and PdCd exhibit valence electronic densities of states similar to that of Cu and significant chemical shifts (larger than 1 eV) of Pd 3d states with respect to pure Pd, as verified by high-resolution hard X-ray photoelectron spectroscopy (HXPS) measurements and density functional theory (DFT) calculations. Consequently, they show the similar high selectivity of CO2 for the SRM reaction. However, this is not the case for PtZn and NiZn because of the slight differences in their valence electronic structures from that of PdZn. The interval between the Fermi level and the top of the d band is closely related to the selectivity of CO2 for the SRM: the larger the interval is, the higher is the selectivity of CO2. According to DFT calculations for bulk PdZn performed by Chen et al. (Phys. Rev. B 2003, 68, 075417), the (111) and (100) surfaces exposing Zn and Pd in an equimolar ratio are more stable than the (001) or (110) surfaces terminated by alternative Zn or Pd layers. First-principles slab calculations for PdZn, PtZn, and NiZn show that bond breaking on the surface leads to a reduction in the d bandwidth but that the d band for stable (111) or (100) surfaces remains essentially unchanged from that of the bulk. It is intriguing that PdZn and PdCd do not contain Cu but show similar valence electronic structure and catalytic selectivity, and hence, a concept is proposed where PdZn and PdCd are regarded as pseudoelements of Cu. The basis of this concept is like electronic structure, like catalysis, which has been demonstrated by experiments and calculations. This is a logical way to enable us to look for new catalysts in which precious metals are partially or completely replaced by base metals. We do not expect that this concept can be applied to all catalytic reactions, but this approach is one of most promising ways to derive a better understanding of the origin of catalytic mechanisms and eventually allow us to design useful catalysts intentionally in the future. This Account reviews the authors’ published works on this topic. The potential of a given intermetallic for catalysis was first indicated by Schwab in the 1940s, when a γ phase with limited compositional range revealed especially higher activity than any other alloys in the Cu−Sn1 and Cd−Au2 systems for dehydrogenation of formic acid. The γ phase is a typical intermetallic and is also an “electron compound” within the definition given by Hume-Rothery,3 where the structure is stable in a narrow compositional range with a definite value of the valence concentration or electron/atom ratio (e/a). It is well-known that alloying can improve the catalytic properties of the parent catalyst, and an intermetallic has often been treated as a kind of alloy.4,5 Therefore, many approaches have been developed to use a given intermetallic to improve the catalytic properties of an element that is a constituent of the parent intermetallic.6,7 However, if one looks at intermetallics in terms

1. INTRODUCTION Recently, catalysis studied in terms of metallurgy is of increasing interest and represents a rising stream of publications in the community. Several new forms of catalysts have been developed, and some new insights have been derived within this framework. In many cases, an intermetallic compound has been treated as either a “pseudoelement” or a “precursor”. In the former case, the intermetallic compound shows similar electronic structure and catalytic selectivity as another single element. Here, the keywords are “electronic structure” and “catalytic selectivity”. In the latter case, the intermetallic compound exhibits high activity due to a unique microstructure derived by leaching or oxidation treatments. In this Account, we review how an intermetallic can be viewed as a pseudoelement in terms of catalytic selectivity and electronic structure. As the best examples, we focus on the intermetallics NiZn, PdZn, and PtZn since they have been well-studied with various techniques and theoretical calculations. © 2017 American Chemical Society

Received: September 27, 2017 Published: December 8, 2017 2879

DOI: 10.1021/acs.accounts.7b00476 Acc. Chem. Res. 2017, 50, 2879−2885

Article

Accounts of Chemical Research of electronic structure, any given intermetallic can be viewed as a “new element” since its electronic structure could be completely different from those of the constituents. Consequently, we may expect an intermetallic to have totally new catalytic properties, different from those of its constituents. From this point of view, intermetallics possess immeasurable potential for catalysis. In this Account, the following two catalytic reactions are mentioned frequently: CH3OH + H 2O → CO2 + 3H 2

(1)

CH3OH → CO + 2H 2

(2)

The first reaction is stream reforming of methanol (SRM), and the second is simply methanol decomposition (MD). Palladium is known to be an efficient and thermally stable catalyst for various hydrogenation and dehydrogenation processes. Unfortunately, Pd catalysts are not efficient for SRM since they are almost 100% CO selective via MD (eq 2). Nevertheless, Pd dispersed on a ZnO support reveals high selectivity of CO2 and maintains high activity and thermal stability for SRM.8,9 The drastic change in selectivity has been attributed to the formation of the intermetallic PdZn upon heating under a reducing atmosphere or the reaction conditions.8 It seems that the Pd−Zn bonding in the intermetallic is more favorable than the Zn−O bonding in the oxide under certain conditions. Strong bonding between the constituents is a characteristic feature of intermetallics. In addition, Pd/ZnO (PdZn) also shows high activity for dehydrogenation of methanol10 and methanol partial oxidation11 (CH3OH + 1/2O2 → 2H2 + CO2). Interestingly, Cu catalysts are known to be good catalysts for these reactions. This raised a very intriguing but very basic question: how can an intermetallic (PdZn) show similar catalytic properties as another element (Cu)? If its origin is clarified, this would permit the design of new catalysts based on intermetallics. In order to gain insights into this, we have been investigating this issue12,13 over the past decade by means of density functional theory (DFT) calculations, SRM reaction studies, and hard Xray photoelectron spectroscopy (HXPS) characterization of the catalysts, and we are quite convinced that in some cases an intermetallic can be viewed as a pseudoelement in terms of catalysis. An important feature that deserves mention in our work is the use of samples without oxide supports for all of the experiments in order to rule out the possible contributions made by the supports. In this Account, we review mainly our work on this topic. Details of the experiments and calculations are given in refs 12 and 13.

Figure 1. Illustration of the L10 structure of PdZn with a bodycentered tetragonal lattice. Circles in blue and red represent Pd and Zn atoms, respectively. Pd atoms occupy eight vertices and the face centers of the top and bottom planes, and Zn atoms occupy the remaining four face centers. The Pd and Zn layers stack alternately along [001]. The (001) atomic planes contain only either Pd or Zn atoms, while the (100) and (111) planes have the same atomic composition as in the bulk (50 atom % Pd and 50 atom % Zn). Adapted from ref 12.

same L10 structure but different degrees of distortion along the c axis. If the atoms of the L10 structure are all replaced by Cu atoms, then the lattice becomes a simple fcc Cu lattice, and hence, the comparison between the intermetallics and Cu should be on the same structural basis. Calculated band structures for pure Cu and a series of L10 structures are shown in Figure 2. Band structure calculations were performed with the tight-binding linear muffin-tin orbital

2. DENSITY OF STATES AND CATALYTIC PROPERTIES It is interesting to note that Cu has the same electron/atom ratio as PdZn (e/a = 1). Intuitively, one might expect that the electronic structure plays a dominant role in these two materials. PdZn has an L10 structure (CuAu-type) with a bodycentered tetragonal (bct) lattice in which Zn and Pd atoms occupy vertices and body-center sites, respectively. The L10 structure also can be viewed as a tetragonally deformed B2 (CsCl) structure. Conventionally, the bct structure can be alternatively described as a face-centered tetragonal (fct) lattice. The relations between the lattice parameters of the two descriptions are afct = abct√2 and cfct = cbct = c. An illustration of the PdZn structure is shown in Figure 1. In addition to PdZn, three related intermetallics, PtZn, NiZn, and PdCd, have the

Figure 2. Valence band spectra obtained from calculations for fcc Cu, fcc Pd, and four intermetallics. It should be noted that the distributions, widths, and positions of the d states are similar for Cu, PdZn, and PdCd. The strong peaks around E = 8 eV are Zn 3d states, and those at E = 9 eV are Cd 4d states. The dotted line is a guideline for comparing the positions of the d-band tops. Adapted from ref 12. 2880

DOI: 10.1021/acs.accounts.7b00476 Acc. Chem. Res. 2017, 50, 2879−2885

Article

Accounts of Chemical Research (TB-LMTO) method in the atomic-sphere approximation (ASA). First, a significant difference in density of states (DOS) between pure Cu and Pd can be easily recognized, where the tops of the 3d bands of the former are located around 1.5 eV below the Fermi level (EF) whereas the 4d bands of the latter show a much broader distribution and their tops cross EF. In terms of electronic structure, pure Cu and pure Pd are two extreme cases because of the different distributions of the d bands; consequently, the former is a typical diamagnetic element whereas the latter is paramagnetic with considerable magnetization.13 Surprisingly, similar features are observed for the intermetallic PdZn, but they arise from a different origin than for pure Cu. The most drastic change in the DOS of Pd due to formation of the intermetallic with Zn is the apparent depression of the 4d states of Pd near EF, which makes the DOS similar to that of Cu. The similar DOS is also observed for the intermetallic PdCd. Obviously, the formation of the intermetallics with Zn would depress the d states close to EF, and this also occurs in PtZn and NiZn, but the degree of the depression is not so drastic as in PdZn and PdCd. The peaks around E = 8 eV and E = 9 eV are attributed to Zn 3d and Cd 4d states, respectively. They are the fingerprints of alloying. Like pure Cu, PdZn and PdCd exhibit a diamagnetic nature in magnetic measurements.13 The origin of the significant depression of d states around EF is revealed in a series of calculated band dispersions, which can be divided into two parts. The first contribution is the formation of the bonding between Pd and Zn, where the bonding−antibonding splitting of dyz and dzx disappears in PdZn at the Γ and M points. The dyz and dzx states extend in the nearest-neighbor Pd−Zn directions. Because the Zn 3d band is far below the Pd 4d band, the d−d mixing between Zn and Pd is supposed to be much smaller than that between Pd and Pd. Thus, the dyz and dzx bands at approximately −2 eV have a nonbonding nature in PdZn. The second contribution is the tetragonal lattice distortion in the Pd−Zn structure. The PdZn structure has lattice constants with c/a (fct) ∼ 0.82, and the lattice distortion reduces the degree of band dispersion in the ab plane. The splitting of the dispersionless dx2−y2 branches on the Γ−Z and A−M lines is dominated by hopping in the ab plane and is reduced in the intermetallic PdZn. On the other hand, the tetragonal lattice distortion is smaller for PtZn (c/ aPtZn = 0.87) and NiZn (c/aNiZn = 0.85), and hence, the smaller depression of the d states is not in conflict with the calculation shown in Figure 2. The catalytic properties for the SRM reaction were evaluated and are shown in Figure 3. The four intermetallics were prepared simply by sintering in vacuum-sealed quartz tubes at elevated temperatures. The Cu catalyst was Raney Cu prepared by leaching of Al2Cu in NaOH solution, and the Pd catalyst was a commercial Pd black. Although the surface area was small for all of the samples, they revealed more than 5% conversion of methanol. This ensured the reliability of the selectivity of CO2 (i.e., CO2/(CO2 + CO)) in the measurements. Details are given in refs 12 and 13. Raney Cu representing the Cu catalyst is known as the best catalyst for the SRM reaction, and it shows almost 100% selectivity of CO2 over the temperature range measured. On the contrary, Pd black is at the other extreme, showing almost zero selectivity of CO2. This is not surprising since Pd is known as a catalyst for DM (eq 2), which produces CO rather than CO2. The extreme difference in catalytic selectivity for Cu and Pd is well-reflected in the differences in their calculated DOSs, as shown in Figure 1.

Figure 3. Selectivities of CO2 at elevated temperatures under SRM for Cu, PdZn, PtZn, NiZn, and PdCd powders. The selectivity of CO2 is close to 100% over the measured temperature range for Cu, PdZn, and PdCd but zero for Pd black. Adapted from ref 12.

However, the selectivity of the catalyst no longer resembles that of Pd after Pd is alloyed with Zn to form the intermetallic PdZn. The nearly zero selectivity of CO2 over the Pd catalyst jumps to ∼100% over PdZn. Such a drastic change in selectivity is not merely a modification of the catalytic properties of Pd but is totally a replacement of the catalytic function of Cu. This is consistent with the calculated DOS shown in Figure 2. A similar result was also observed for another intermetallic, PdCd, indicating the universality of the parallel correlation between DOS and catalytic function, at least for SRM. It should be noted that Cd is in the same column as Zn in the periodic table and forms the same L10 structure in its intermetallic with Pd. Therefore, in terms of DOS and catalytic function, we may speculate that PdZn ≈ Cu ≈ PdCd and that the intermetallic PdZn (or PdCd) can be regarded as a pseudoelement of Cu. We also observed a very clear correlation between the location of the d band and the selectivity of catalysts for SRM. In PdZn, the top of the d band is located ∼1.5 eV below EF, as in Cu. In PtZn and NiZn, the d bands are shifted to slightly lower binding energies, corresponding to decreased CO2 selectivity. A close correlation between d-band position and adsorption energy has been reported for many molecules.14 A lower binding energy of the s states promotes the donation of electrons to empty antibonding molecular eigenstates.

3. BULK ELECTRONIC STRUCTURES STUDIED BY HARD X-RAY PHOTOELECTRON SPECTROSCOPY Electronic structures of PdZn have been measured on the Pd/ ZnO catalyst8 and PdZn/Ru thin films.15 However, these measurements all suffered from interference by either the oxide support or the substrate, and thus, the results were not precise enough for discussion of the bulk electronic structures in detail. We conducted hard X-ray photoelectron spectroscopy measurements on samples used for the catalytic reaction using highbrilliance synchrotron radiation with hν = 5.951 keV at beamline 15XU at the SPring-8 facility in Japan. The valence band structures for three intermetallics along with that of Cu are shown in Figure 4. Although the measurements were performed at room temperature, the locations of the d-band tops relative to EF are very close to those calculated at 0 K for the three intermetallics, and their order (PdZn > PtZn > NiZn) is the same as that in the calculations. This order also 2881

DOI: 10.1021/acs.accounts.7b00476 Acc. Chem. Res. 2017, 50, 2879−2885

Article

Accounts of Chemical Research

Figure 4. Valence band structures of PdZn, PtZn, and NiZn obtained by high-resolution HXPS. Those of their constituent elements and Cu are also shown for comparison. Alloying with Zn led to a depression of d states near EF for PdZn and PtZn but not for NiZn. It should be noted that the distributions and positions of d states are very similar for PdZn and Cu.

corresponds to the degree of depression of the d band of each element generated by alloying with Zn. When the valence bands of Cu and PdZn are superimposed, it is clear that the locations of the d-band tops and the widths of the d bands are very similar. Basically, the valence band structures obtained by XPS measurements are consistent with the calculated ones (Figure 2). The core-level shifts of Pd 3d, Pt 4f, Ni2p, and Zn 2p generated by alloying are shown in Figure 5. Surprisingly, the chemical shift of Pd 3d to higher energy in PdZn is as large as 1.3 eV, and the degrees of shift are in the order PdZn > PtZn > NiZn. HXPS measurements for PdCd (Figure 6) showed results similar to those for PdZn. On the other hand, the shift of Zn 2p is to lower energy, and the degree of the shift is almost the same for the three intermetallics. In terms of charge transfer, it seems that Zn serves as an acceptor receiving electrons from Pd and Pd serves as a donor donating electrons to Zn. It should be noted that Pd is paramagnetic with certain high magnetization, whereas it becomes diamagnetic like Cu in PdZn and PdCd. This behavior is parallel evidence for the charge transfer. It is interesting to indicate that the charge transfer from Zn to Pd showed up in the analysis of chemical bonding by the electron localizability approach.16 Consequently, the bonding of intermetallic PdZn should have strong ionic character, which becomes much weaker for PtZn and is almost vanishing for NiZn. Similar results have also been reported for supported catalysts by Rodriguez.15 However, a strong correlation between valence electronic structure and core-level shift is observed in these measurements for the three intermetallics. A shoulder originating from pure Zn is observed for the Zn 2p peaks of NiZn, indicating the existence of the Zn component and the lower stability of NiZn among the three intermetallics. Recently, Friedrich et al.17 reported that unsupported NiZn catalysts are unstable under SRM reaction conditions and decompose, forming ZnO and a Ni-enriched NiZn alloy, which is consistent with the HXPS measurements.

Figure 5. Core-level HXPS spectra of Pd 3d, Pt 4f, Ni 2p, and Zn 2p for PdZn, PtZn, NiZn, Pd, Pt, and Ni. The core-level shifts are in the order PdZn > PtZn > NiZn.

Figure 6. (top) Valence HXPS spectra for PdCd, Cu, and PdZn and (bottom) core-level HXPS spectra for Pd and PdCd. It should be noted that the d states are similar for PdCd, Cu and PdZn.

2882

DOI: 10.1021/acs.accounts.7b00476 Acc. Chem. Res. 2017, 50, 2879−2885

Article

Accounts of Chemical Research

4. SURFACE ELECTRONIC STRUCTURES Thus far we have established a parallel relationship between bulk electronic structure and catalytic selectivity. However, since the selectivity of CO2 is a surface property, surface states of the intermetallics must be examined as well in order to gain insights into the role of electronic structure. Regarding the surface stability, Chen et al.18 had calculated the stability of various planes for PdZn and PtZn. It was found that the (100) and (111) planes reveal lower surface energies than the (110) planes for fcc Cu and that the trend is maintained but becomes stronger in PdZn. For example, the surface energies γ(111), γ(110), and γ(100) are 1.34, 1.58, and 1.53 J/m2, respectively, for Cu and 1.17, 1.57, and 1.23 J/m2, respectively, for PdZn. Interestingly, the (100) and (111) planes have the composition Pd 50Zn 50 (i.e., the bulk composition), whereas the (001) and (110) planes are either 100% Pd or 100% Zn. In the case of fcc Cu, the surface energy is considered to be dominated by the atomic density of the surface, while in the case of intermetallic PdZn, the bonding between a similar or dissimilar pair of elements is more critical. It should be noted that γ(100) is much smaller for PdZn than for fcc Cu. It is most likely that (100) and (111) would be present at the surface in the intermetallic PdZn. According to the HXPS measurements, if one assumes the bonding nature between Pd and Zn atoms where Pd−Zn > Pd−Pd and Pd−Zn > ZnZn, it is reasonable that the surface of PdZn would show energetically stable surfaces with 50% Pd and 50% Zn. Experimentally, both the (111) and (100) surfaces have been demonstrated by lowenergy electron diffraction patterns to be preferentially exposed even at elevated temperatures.19 Therefore, we should be able to consider only the (111) and (100) planes when we deal with the catalytic properties of the intermetallic PdZn. The calculated valence electronic DOSs at the (111) and (100) surfaces for the three intermetallics are shown in Figures 7 and 8, and the bulk valence DOSs (black lines) are also present for comparison. The surface DOSs in both cases are those for the topmost layer. The positions of the d-band top in the (100) surface DOSs are almost the same as those in the bulk for all three intermetallics. This is almost true as well for the (111) surface DOSs, except that the surface d-band top shows a slight shift toward the Fermi level. This slight shift of the d-band top in the (111) surface is caused by splitting of the degenerate dyz and dzx states due to the lower symmetry of the (111) surface. Ganduglia-Pirovano et al.20 have discussed the relation between the shift and the depression of the surface d band. According to that discussion, the d-band center shifts toward the Fermi level for late transition metals such as Pd to compensate for the increasing d-band filling caused by the bandwidth reduction at the surface. A review with a similar point of view describing the potential energy for the MD reaction on the Cu(111) and Pd(111) surfaces is very intriguing.21 Although the neighboring Pd−Pd bonds are missing in the (100) and (111) surfaces, one may expect a similar scenario to the pure transition metal surfaces for the dband shift in Pd−Zn. This is a plausible explanation for the same position of the d-band top for the surface layer and the bulk. Very recently, Krajči et al.22 investigated the selectivity of SRM catalyzed on the three isostructural surfaces PdZn(111), PtZn(111), and NiZn(111) using DFT calculations. According to these calculations, formaldehyde (CH2O) seems to be an important intermediate of the SRM process, serving as a

Figure 7. Calculated valence electronic DOSs at the (111) surfaces of PdZn, PtZn, and NiZn. Black and colored lines represent the bulk and surface DOSs, respectively. Adapted from ref 13.

Figure 8. Calculated valence electronic DOSs at the (100) surfaces of PdZn, PtZn, and NiZn. Black and colored lines represent the bulk and surface DOSs, respectively. Adapted from ref 13.

branching point for different product channels. There is a clear correlation between high CO2 selectivities and weak adsorption energies of CH2O. The adsorption energy of CH2O becomes more exothermic in the sequence PdZn (Eb = −21 kJ/mol) < PtZn (Eb = −35 kJ/mol) < NiZn (Eb = −46 kJ/mol). This 2883

DOI: 10.1021/acs.accounts.7b00476 Acc. Chem. Res. 2017, 50, 2879−2885

Article

Accounts of Chemical Research

more energy than breaking bonds between similar elements by surface formation. To understand the trend in surface selection, further analysis of the bonding nature will be helpful. We leave these issues as open questions for better understanding of the valence electron structure and its effects on catalytic function.

sequence corresponds to the decrease in the energy of the dband top of the constituent transition metal shown in Figure 2 from −1.8 to −1.2 to −0.9 eV, respectively. This explains why PdZn can be a pseudoelement of Cu with similar CO2 selectivity whereas this is not the case for PtZn or NiZn.



5. CONCLUDING REMARKS We have reviewed a number of examples on the basis of a series of experiments and calculations, and we conclude that in terms of electronic structure and catalytic selectivity, an intermetallic can be treated as a pseudoelement. This conclusion is supported by clear correlations between bulk and surface electronic structure and CO2 selectivity for the SRM reaction obtained in PdZn, PtZn, NiZn, and PdCd intermetallics. Some comparisons with other studies of intermetallic and alloy catalysts are relevant. The intermetallic Al13Fe4, which has a complex structure, was shown to be an active and selective semihydrogenation catalyst like Pd.23 However, it seems that the valence electronic structure of Al13Fe4 is not similar to that of Pd, and hence, it cannot be interpreted as a pseudoelement with respect to Pd. The high catalytic activity of the Al13Fe4 seems to have another origin. Apart from intermetallics, a CuNi solid solution has been studied. Its valence electronic structure obtained by both calculation and XPS experiments was similar to that of Pd. It also presented a similar selectivity for SRM and methanol decomposition as Pd.24 In other work, solid-solution nanoparticles containing precious metals have been studied on the basis of valence DOS.25 However, in this case the structure could be a pure solid solution or a core−shell structure, depending strongly on the effects of temperature on surface tension and particle size. Hence, the correlation between valence DOS and catalytic properties cannot be simply derived. The major issue described in this Account is the concept of a pseudoelement, which is based on the fact of like valence electronic structure, like catalytic f unction, as demonstrated by experiments and calculations. This is a logical way to enable us to look for new catalysts in which precious metals are partially or completely replaced by base metals. Of course, this concept cannot be applied to all catalytic reactions, but this approach is one of most promising ways to derive a better understanding of the origin of catalytic mechanisms and eventually allow us to design useful catalysts intentionally in the future. We have focused on the valence electronic DOS and the dband position as indicators of catalytic selectivity. Concerning the electronic structures of intermetallics, however, more careful analysis and arguments are needed. For example, we speculate that Pd serves as a donor whereas Zn serves as an acceptor on the basis of HXPS measurements. However, corelevel shifts may be determined not only by the valence electron distribution but also other effects such as relaxation of the valence electrons caused by the core hole. The electron configurations of isolated Pd and Zn atoms are d10s0 and d10s2, respectively. If one expects naively that an average electron configuration for PdZn is similar to that of Cu, which is d10s1, Pd should be an acceptor instead of a donor. More theoretical studies on core-level shifts in intermetallics are desirable. Further analysis of surface stability is also desirable. According to analysis of the calculated band dispersion, the bonding− antibonding splitting in pure Pd disappears in PdZn because of the nearest-neighbor bonds between dissimilar elements. This seems to give a nonbonding nature for Pd−Zn bonds in the intermetallic PdZn. However, comparison of the surface energies leads to arguments that breaking Pd−Zn bonds costs

AUTHOR INFORMATION

ORCID

A. P. Tsai: 0000-0002-7905-9915 Notes

The authors declare no competing financial interest. Biographies An-Pang Tsai is a professor at the Institute of Multidisciplinary Research for Advanced Materials (IMRAM), Tohoku University. His research interests include the formation, structure, and application of quasicrystalline alloys. He also studies catalysis in terms of metallurgy by controlling their electronic structures and tailoring the microstructures of intermetallics and different types of alloys. Satoshi Kameoka is an associate professor at IMRAM. His research interests are catalysis by metals and alloys. He collaborates with Prof. Tsai in studies of metallurgy for advanced catalysis materials. Kazuki Nozawa is an associate professor in the Department of Physics and Astronomy at Kagoshima University. His current research interests include the correlation between catalytic properties and electronic structures of intermetallic compounds and also the electronic structure of single-element quasicrystalline ultrathin films. Masahiko Shimoda is a former chief researcher at the National Institute for Materials Science, Tsukuba, Japan. His research interests include surface structure and film growth on intermetallics. Yasushi Ishii is a professor of physics at Chuo University. His main research focuses on theoretical studies of electronic structures and properties of solids, in particular intermetallic compounds. He is also interested in dynamical aspects of quasicrystals.



ACKNOWLEDGMENTS The authors are grateful to Drs. Ueda and Y. Yamashita at the National Institute for Materials Science for assistance during the HXPS measurements at beamline 15XU at Spring-8 (2017A4908, 2016B4907, 2016A4905, 2015B4900). This work was supported in part by Grants-in-Aid for Scientific Research ((A) 15H02299 and (C) 17K05059) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan. Support from the “Dynamic Alliance for Open Innovation Bridging Human, Environment and Materials” is also highly appreciated.



REFERENCES

(1) Schwab, G. M. Metal electrons and Catalysis. Trans. Faraday Soc. 1946, 42, 689−697. (2) Schwab, G. M.; Pesmatjoglou, S. Metal electrons and alloy catalysis. J. Phys. Colloid Chem. 1948, 52, 1046−1053. (3) Hume-Rothery, W. J. Nature, properties and formation of intermetallic compounds. Inst. Met. 1926, 35, 295−361. (4) Sinfelt, J. H. Bimetallic Catalysts: Discoveries, Concepts, and Applications; John Wiley & Sons: New York, 1983. (5) Ponec, V.; Bond, G. C. Catalysis by Metals and Alloys, 1st ed.; Studies in Surface Science and Catalysis, Vol. 95; Elsevier Science: Amsterdam, 1995. (6) Llorca, J.; Delapiscina, P. R.; Fierro, J. L. G.; Sales, J.; Homs, N. Influence of metallic precursors on the preparation of silica-supported 2884

DOI: 10.1021/acs.accounts.7b00476 Acc. Chem. Res. 2017, 50, 2879−2885

Article

Accounts of Chemical Research PtSn alloy - characterization and reactivity in the catalytic activation of CO2. J. Catal. 1995, 156, 139−146. (7) Komatsu, T.; Satoh, D.; Onda, A. Ti−Pt intermetallic compound catalysts more active than Pt for hydrogen dissociation and ethylene hydrogenation. Chem. Commun. 2001, 1080−1081. (8) Iwasa, N.; Masuda, S.; Ogawa, N.; Takezawa, N. Steam-reforming of methanol over Pd/ZnO: effect of formation of PdZn alloys upon the reaction. Appl. Catal., A 1995, 125, 145−157. (9) Takezawa, N.; Iwasa, N. Steam-reforming and dehydrogenation of methanol: difference in the catalytic function of Cu and group VIII metals. Catal. Today 1997, 36, 45−56. (10) Iwasa, N.; Takezawa, N. New supported Pd and Pt alloy catalysts for stream reforming and dehydrogenation of methanol. Top. Catal. 2003, 22, 215−224. (11) Cubeiro, M. L.; Fierro, J. L. C. Selective production of hydrogen by partial oxidation of methanol over ZnO-supported Pd catalysts. J. Catal. 1998, 179, 150−162. (12) Tsai, A. P.; Kameoka, S.; Ishii, Y. PdZn = Cu: Can an intermetallic compound replace an element. J. Phys. Soc. Jpn. 2004, 73, 3270−3273. (13) Nozawa, K.; Endo, N.; Kameoka, S.; Tsai, A. P.; Ishii, Y. Catalytic properties dominated by electronic structures in PdZn, NiZn, and PtZn intermetallic compounds. J. Phys. Soc. Jpn. 2011, 80, 064801. (14) Nilsson, A.; Pettersson, L. G. B.; Hammer, B.; Bligaard, T.; Christensen, C. H.; Norskov, J. K. The electronic structure effect in heterogeneous catalysis. Catal. Lett. 2005, 100, 111. (15) Rodriguez, J. A. Interactions in bimetallic bonding: electronic and chemical properties of PdZn surfaces. J. Phys. Chem. 1994, 98, 5758−5764. (16) Friedrich, M.; Ormeci, A.; Grin, J.; Armbrüster, M. PdZn or ZnPd: Charge transfer and Pd-Pd bonding as the driving force for the tetragonal distortion of the cubic crystal structure. Z. Anorg. Allg. Chem. 2010, 636, 1735−1739. (17) Friedrich, M.; Teschner, D.; Knop-Gericke, A.; Armbrüster, M. Surface and subsurface dynamics of the intermetallic compound ZnNi in methanol steam reforming. J. Phys. Chem. C 2012, 116, 14930. (18) Chen, Z. X.; Neyman, K. M.; Gordienko, A. B.; Rösch, N. Surface structure and stability of PdZn and PtZn alloys: Density functional slab model studies. Phys. Rev. B: Condens. Matter Mater. Phys. 2003, 68, 075417. (19) Gabasch, A.; Knop-Gericke, R.; Schlogl, R.; Penner, S.; Jenewein, B.; Hayek, K.; Klotzer, B. Zn Adsorption on Pd (111): ZnO and PdZn alloy formation. J. Phys. Chem. B 2006, 110, 11391. (20) Ganduglia-Pirovano, M. V.; Natoli, V.; Cohen, M. H.; Kudrnovský, J.; Turek, I. Potential, core-level, and d band shifts at transition-metal surfaces. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 8892. (21) Greeley, J.; Nørskov, J. K.; Mavrikakis, M. Electronic structure and catalysis on metal surfaces. Annu. Rev. Phys. Chem. 2002, 53, 319− 48. (22) Krajči, M.; Tsai, A. P.; Hafner, J. Understanding the selectivity of methanol steam reforming on the (111) surfaces of NiZn, PdZn and PtZn: insights from DFT. J. Catal. 2015, 330, 6−18. (23) Armbrüster, M.; Kovnir, K.; Friedrich, M.; Teschner, D.; Wowsnick, G.; Hahne, M.; Gille, P.; Szentmiklosi, L.; Feuerbacher, M.; Heggen, M.; Girgsdies, F.; Rosenthal, D.; Schlögl, R.; Grin, J. Al13Fe4 as a low-cost alternative for palladiumin heterogeneous hydrogenation. Nat. Mater. 2012, 11, 690−693. (24) Tsai, A. P.; Kimura, T.; Suzuki, Y.; Kameoka, S.; Shimoda, M.; Ishii, Y. Effect of electronic structures on catalytic properties of CuNi alloy and Pd in MeOH-related reactions. J. Chem. Phys. 2013, 138, 144701. (25) Kobayashi, H.; Kusada, K.; Kitagawa, H. Creation of novel solidsolution alloy nanoparticles on the basis of density-of-state engineering interelement fusion. Acc. Chem. Res. 2015, 48, 1551−1559.

2885

DOI: 10.1021/acs.accounts.7b00476 Acc. Chem. Res. 2017, 50, 2879−2885