Catalytic Properties of Heusler Alloys for Steam Reforming of Methanol

Intermetallic compounds have attracted research attention in catalysis because of their unique catalytic properties. Recently, a group of intermetallic compounds, referred to as Heusler alloys (X2YZ), has been investigated as new catalysts. In this study, catalytic properties of 14 Heusler alloys with X = Fe, Co, Ni, or Cu; Y = Ti, Mn, or Fe; Z = Al, Si, Ga, Ge, or Sn for the steam reforming of methanol were examined. Co2TiAl and Ni2TiAl alloys exhibited relatively high H2 production rates because of the formation of fine particles via the selective oxidation of Ti. X2MnZ alloys exhibited high CO2 selectivity because of a water-gas shift reaction catalyzed by using MnO that was formed during the reaction. Crystal phases, surface microstructures, and surface compositions of most alloys were changed because of the reaction, and the formation of fine particles possibly assisted in the observation of catalytic activity. Heusler alloys can be beneficial as catalyst precursors by the selection of appropriate elemental sets depending on target reactions.


INTRODUCTION
Intermetallic compounds exhibit unique atomic ordered structures and electronic structures, even at the surface, rendering unique catalytic properties. Currently, they are gaining increasing popularity as new catalysts. 1−3 Almost all of the reported intermetallic catalysts are binary systems. However, ternary intermetallic compounds can render synergistic effects among three elements, which serve as novel catalysts different from ordinary binary catalysts. Heusler alloys (X 2 YZ) constitute a group of ternary intermetallic compounds with an L2 1 structure (other types are also present). 4 Typically, the main element X belongs to groups 8− 12, while subelements Y and Z belong to groups 3−8 and 13− 15, respectively, which is similar to most of the intermetallic catalysts reported thus far. Various possible elemental sets of X, Y, and Z can afford new catalysts, and catalysis can be tuned by utilizing in-between compositions (e.g., X 2 YZ 0.5 Z 0.5 ′ ). However, Heusler alloys have not been identified as catalysts until only recently. Typically, catalysts are prepared by a liquid process in a form of supported nanoparticles, which requires considerable trials and errors to establish preparation conditions for each elemental system. This fact makes it difficult to screen ternary intermetallic compounds, including Heusler alloys, as catalysts.
Most of the stable intermetallic compounds, even in ternary (or higher) systems, can be easily prepared by metallurgical methods such as arc-melting. By this method, Heusler catalysts in a form of unsupported powders have been screened by our group. 4−6 Previously, the oxidation of carbon monoxide over Heusler catalysts revealed that X mainly exhibits catalytic activity and Y and Z modulate the activity and durability. 5 Our group has also investigated the selective hydrogenation of alkynes and reported that unlike practical Pd-based catalysts, non-noble metal catalysts such as Co 2 (Mn or Fe)Ge exhibit high alkene selectivity, and catalysis can be systematically controlled by elemental substitution (Co 2 Mn x Fe 1−x Ga y Ge 1−y ). 6 It would be desirable and beneficial to screen these emerging ternary intermetallic catalysts for other reactions to develop new catalysts and gain fundamental knowledge. Hence, the steam reforming of methanol (SRM) is conducted herein, for which intermetallic catalysts are typically applied. 1−3,7−11 The reaction CH 3 OH + H 2 O → 3H 2 + CO 2 is one possibility for producing hydrogen from safe storage for use as a clean energy source in society. 12,13 The catalytic properties of 14 Heusler alloys with X = Fe, Co, Ni, or Cu; Y = Ti, Mn, or Fe; Z = Al, Si, Ga, Ge, or Sn, which were easily prepared equilibrium phases, and structural changes occurring during the reaction were examined in this report.

RESULTS AND DISCUSSION
2.1. SRM with the Reactant CH 3 OH/H 2 O = 2:3. Figure 1 shows production rates of H 2 , CO, CO 2 , and CH 4 for the SRM with the reactant CH 3 OH/H 2 O = 2:3 during heating and cooling processes. For Co 2 TiAl (Figure 1a), the H 2 production rate started to significantly increase at 500°C with the heating, but it decreased at 600°C. A higher H 2 rate was observed during cooling than during heating. Similarly, for Ni 2 TiAl, the H 2 rate suddenly started to increase at 450°C. Furthermore, the H 2 production rate at ≤400°C was greater during cooling than during heating. Studies reported that Ni 3 Al and Ni 3 (Si,Ti) form Ni fine particles on the surface because of the selective oxidation of Al and Si, respectively, during methanol decomposition, thereby exhibiting high activity. 14−16 Similarly, after SRM, Ni 2 TiAl exhibited Ni fine particles, as can be observed by X-ray diffraction (XRD) and scanning electron microscopy (SEM) analyses (Figure 2a,b, respectively); these fine particles were probably formed because of the selective oxidation of Ti as revealed by X-ray photoelectron spectroscopy (XPS) analysis summarized in Table 1. The increased H 2 production rate at ≥450°C and the higher H 2 production rate at 400°C during cooling than during heating were attributed to the Ni fine particle formation. SEM images of Co 2 TiAl after SRM mainly revealed a number of carbon nanofibers (CNF) with fine particles, while occasionally observed fine particles on the surface (Figure 2b). The crystal structure and composition of these fine particles were not observed by XRD (Figure 2a) and energy-dispersive X-ray spectrometry (EDX) analyses. However, the fine particles in Figure 2b (region 2) were probably Co precipitated via the selective oxidation of Ti; this precipitated Co could serve as active centers, which produced an increased amount of H 2 during cooling than during heating. This mechanism was deduced from Table 1 and the subsequent experiments discussed in the next section.
The decrease in the H 2 production rate at 600°C for Co 2 TiAl was mainly related to the decrease in the methanol Figure 1. Production rates of H 2 , CO, CO 2 , and CH 4 in SRM with the reactant of CH 3 OH/H 2 O = 2:3 during heating and cooling processes for (a−c) X 2 TiAl with X = Co, Ni, Cu; (d−f) X 2 TiSn with X = Fe, Co, Ni; (g,h) X 2 MnSn with X = Co, Ni; (i−k) Co 2 MnZ with Z = Ge, Ga, Si; (l) Co 2 TiGe; and (m,n) Co 2 FeZ with Z = Ge, Ga. Circles, triangles, squares, and crosses represent H 2 , CO, CO 2 , and CH 4 , respectively. The reaction was conducted with an LHSV of ∼20 h −1 for powders sieved to 20−63 μm. Each data point was obtained after maintaining constant for 30 min at each reaction temperature. 3, and (c) feeding rate of the reactant mixture and experimental and ideal maximum H 2 production rates. In (a), asterisks represent X 2 TiAl peaks. In (b), two representative regions are shown for Co 2 TiAl. In (c), red (heating) and pink (cooling) circles represent the feeding rate (left axis) and the ideal H 2 production rate assuming methanol decomposition (CH 3 OH → 2H 2 + CO) (right axis), and blue (heating), and pale blue (cooling) represent the experimentally obtained H 2 production rate.

ACS Omega
Article feeding because of the blocking of the extensive growth of CNF ( Figure 2b, region 1). Compared to Ni 2 TiAl, Co 2 TiAl exhibited a larger number of CNF, as indicated by the SEM images and by the higher surface area of Co 2 TiAl (3.26 m 2 g −1 ) than that of Ni 2 TiAl (0.32 m 2 g −1 ) after the reaction. Compared to Co 2 TiAl, Ni 2 TiAl exhibited a smaller decrease in methanol feeding ( Figure 2c); therefore, the smaller H 2 production rate at 500°C in the cooling phase than in the heating phase for Ni 2 TiAl (Figure 1b) is related to the coarsening of Ni particles at ≥550°C and not to the decrease in methanol feeding. Besides H 2 , CO was the main product over these alloys, which was typical for pure Co and Ni. 13 Thus, H 2 is predominantly produced by methanol decomposition (CH 3 OH → 2H 2 + CO), and most of the methanol is consumed at ≥550°C for both alloys as the H 2 rates almost reach their ideal maximum values (assuming decomposition) ( Figure 2c). The increase in the CO 2 production rate at high temperatures confirms the growth of CNF via the disproportionation of CO (2CO → CO 2 + C).
Conversely, Cu 2 TiAl did not catalyze the reaction ( Figure  1c). SEM images revealed homogeneously precipitated fine particles ( Figure 3). However, XPS analysis revealed a very small surface composition of Cu (Table 1), indicating that active fine particles are not formed by selective oxidation. Co and Ni exhibited similar Gibbs energies of oxidation, but Cu exhibited a considerably more positive Gibbs energy ( Table  2). 17 Fine particle formation by selective oxidation possibly depended on a balance of component oxidizability; thus, Cu is probably difficult to precipitate on the surface of alloy with Ti and Al.
In case of X 2 TiSn, Fe 2 TiSn exhibited extremely low activity ( Figure 1d) and formed several phases after the reaction ( Figure 4a). Co 2 TiSn and Ni 2 TiSn basically exhibited similar production rates although Ni 2 TiSn was slightly active compared to Co 2 TiSn (Figure 1e,f). In addition, Co 2 TiSn and Ni 2 TiSn exhibited similar structural changes, affording marginal X 3 Sn 2 (Figure 4a), as well as similar SEM images revealing fine particles (Figure 4b) after the reaction. EDX analysis indicated that their alloy compositions are stoichiometric in most areas after the reaction. Similar to X 2 TiAl, oxidizability (Fe > Co ≅ Ni; Table 2) likely played a role in the structural change through the reaction, consequently dominating the catalytic properties.
Similar to X 2 TiSn, Co 2 MnSn and Ni 2 MnSn exhibited similar catalytic properties (Figure 1g,h). In addition, these alloys exhibited similar structural changes, affording marginal X 3 Sn 2 and MnO ( Figure 5a). SEM images after the reaction revealed fine particles in some parts of Co 2 MnSn and all parts of Ni 2 MnSn (Figure 5b). Even in the case of X 2 MnSn, X = Co

ACS Omega
Article and Ni exhibited similar results, as well as for X 2 TiAl and X 2 TiSn. However, Co 2 MnSn and Ni 2 MnSn mainly produced CO 2 besides H 2 , unlike (Co or Ni) 2 TiAl and (Co or Ni) 2 TiSn. In addition, the main byproduct of Co 2 MnGe, Co 2 MnGa, and   21 Actually, the CO 2 selectivity (molar ratio of CO 2 in all byproducts) decreased with the increase in the temperature for Mn-containing alloys (Table  3), supporting the proposed mechanism.
Co 2 TiGe did not catalyze the reaction (Figure 1l). Fine particle formation was not observed by SEM (Figure 7b). Only this alloy clearly exhibited TiO 2 peaks in the XRD pattern ( Figure 7a). The surface Ti composition was considerably greater than the stoichiometry even before the reaction (Table  1). These facts indicated that active sites are not formed on the surface covered by inert TiO 2 .
Co 2 FeGe exhibited an unusual change in the H 2 production rate: it increased with heating but decreased at 550°C ( Figure  1m). SEM revealed fine particles with diameters of ∼400 nm (Figure 8a: region 1), and grains with diameters less than 100 nm are apparently precursors of the 400 nm particles ( Figure  8a: region 2). The precipitation of small grains possibly increased the H 2 production rate at 450°C. XRD and XPS analyses revealed that most of the surfaces are covered by Fe 3 O 4 (Figure 8b and Table 1). Thus, the coarsening of the    In (a), asterisks represent X 2 TiSn peaks, "CoGe?" peaks were likely CoGe with the C2/m space group, and "Un" peak was unknown, but possibly hcp-Co.

ACS Omega
Article grains and the formation of Fe 3 O 4 possibly decrease the H 2 production rate at 550°C. All of the alloy powders changed their colors after the catalytic reaction, but the metallic gray color of Co 2 FeGa did not change, although the color of the bottommost region (ca. 10 vol %) of the catalyst bed changed to black. XRD ( Figure  9a), SEM, and EDX results did not reveal changes in structures and compositions of the metallic gray powders during the reaction, except for the presence of minute CNFs in some areas of SEM images (Figure 9b). However, XPS analysis revealed that most of the surfaces are covered by Ga oxides (Table 1). Although studies on the SRM and methanol decomposition over Ga oxides at high temperatures of greater than 500°C have not been reported, studies revealed that Ga oxides strongly interact with methanol molecules at a lower temperature. 22 Thus, the H 2 production over Co 2 FeGa is likely related to the presence of surface Ga oxides. The SEM image of the bottommost black powders revealed a significant growth of CNF (Figure 9b). As these black powders were obtained at the bottommost region and the H 2 production rate only exhibited a marginal difference between the heating and cooling processes, the fine particles on CNFs did not contribute to the H 2 production.
2.2. Further Investigation on Co 2 TiAl, Ni 2 TiAl, Co 2 FeGe, and Co 2 FeGa. Co 2 TiAl and Ni 2 TiAl apparently exhibited high activity for methanol decomposition, which corresponded to the first step of SRM, because of the fine particle formation. Pure Co and Ni powders sieved to 20−63 μm exhibited low activity (Figure 10), possibly related to the large particle size for methanol decomposition that is structuresensitive. 23 To validate the importance of fine particle formation, effects of steam and oxygen were investigated. Figure 11 shows the results of methanol decomposition using CH 3 OH as the reactant without H 2 O, the SRM using CH 3 OH/H 2 O = 2:1 as the reactant, and the SRM using CH 3 OH/H 2 O = 2:3 as the reactant after the oxygen pretreatment at 500°C. For Co 2 TiAl, a H 2 production rate for 2:1 SRM was greater than that for methanol decomposition (Figure 11a,b), but it was less than that for 2:3 SRM (Figure  1a), indicating that active fine particles are formed via oxidation by H 2 O. The highest H 2 production rate was observed for 2:3 SRM after the oxygen pretreatment ( Figure  11c), indicating that fine particle formation is considerably enhanced by O 2 instead of H 2 O. Ni 2 TiAl exhibited a similar tendency with Co 2 TiAl (Figure 11d−f), although similar results were obtained for 2:1 and 2:3 SRM reactions (Figures  11e and 1b). Figure 12 shows SEM images after these reactions. Co 2 TiAl exhibited significant CNF growth in all reactions, where almost all areas were covered by CNF in case of the methanol decomposition and 2:1 SRM. In case of the surface after

ACS Omega
Article oxygen-pretreated SRM, fine particle formation was occasionally observed, which was likely enhanced by O 2 pretreatment (Figure 12d), although the structure of fine particles was not observed by XRD analysis (Figure 13a). As indicated by SEM and XPS analyses, significant CNF growth likely increased in the order of methanol decomposition >2:1 SRM >2:3 SRM without and with oxygen pretreatment (Table 1), but the H 2 production rate followed the reverse order. Thus, the active centers of Co 2 TiAl are likely fine particles formed by the selective oxidation of Ti, not the fine particles on CNF. For Ni 2 TiAl, only methanol decomposition afforded an excess growth of CNF (Figure 12e). After 2:1 and 2:3 SRM, similar surfaces (Figures 12f and 2b) as well as catalytic properties (Figures 11e and 1b) were observed. After oxygen-pretreated SRM, the surface clearly exhibited large numbers of fine particles (Figure 12g), which was verified by XRD analysis, with the strongest intensity peaks observed for Ni in the series (Figure 13b). The H 2 production rate clearly depended on the number of fine particles, indicating that the active centers are fine particles formed by the selective oxidation of Ti as well as in Co 2 TiAl. Note that the topmost region (ca. 20 vol %) of the catalyst bed changed the color to black after SRM with oxygen pretreatment and revealed significant CNF growth ( Figure  12h, Table 1). However, our conclusion was not altered as

ACS Omega
Article CNFs were apparently not grown at 400°C during heating, while a higher H 2 production rate was observed during heating than during cooling at this temperature. Jang et al. reported that the activation of Ni 3 Al catalysts by the selective oxidation of Al is enhanced by the addition of some steam (CH 3 OH/ H 2 O = 10:1) in methanol decomposition, but it is significantly suppressed by the addition of more steam (CH 3 OH/H 2 O = 2:1 and 1:1). 24 The activation of Ni 2 TiAl was more enhanced by the addition of more steam and using more severe oxidant (O 2 ). As Ni fine particle formation required the selective oxidation of Ti in the presence of Al, more severe oxidation conditions were likely preferred.
In terms of application, the activity of oxygen-pretreated Ni 2 TiAl is herein compared to those of reported catalysts. The turnover frequency (TOF) is often evaluated to compare the activity. TOF values have been reported for many transitionmetal-supported catalysts, however, strongly depending on surface areas and dispersion of metals. 25,26 Thus, activities cannot be compared by TOF between supported small catalysts and our unsupported large catalysts. A H 2 production rate per catalyst weight is crucial for applications. Supported Cu catalysts exhibited H 2 production rates of ∼200 μmol g −1 s −1 at ∼250°C and ∼300 μmol g −1 s −1 at ∼300°C. 13,27 In our previous study, Raney Pd exhibit a H 2 production rate of 87 μmol g −1 s −1 at 300°C. 28 However, herein, Ni 2 TiAl exhibited a H 2 production rate of 8.6 μmol g −1 s −1 at 300°C. Considering that the size of Ni fine particles was up to 400 nm and that their precursors were powders sieved to 20−63 μm, a considerably higher H 2 production rate is expected by using, for example, precursor powders with a diameter of ∼2 μm or foils with a thickness of ∼2 μm.
Co 2 FeGe and Co 2 FeGa exhibited relatively high H 2 production rates for 2:3 SRM (Figure 1m,n). As the main byproduct was CO, H 2 was supposedly produced by methanol decomposition. Actually, these catalysts were active for the methanol decomposition without steam ( Figure 14). For Co 2 FeGe, rather than fine particle formation, only surface roughening was observed in the SEM images ( Figure 15a). XPS analysis indicated that the chemical composition among metals is not changed ( Table 1). As H 2 was produced in the low-temperature region of ≤400°C, metallic Co 2 FeGe was considered to be active for methanol decomposition. Lower H 2 production rates observed at ≤400 and ≥550°C for 2:3 SRM than those for methanol decomposition (Figures 1m and 14a) were likely related to the oxidation by H 2 O at ≤400°C and to the significant formation of Fe 3 O 4 at ≥550°C. Metallic Co 2 FeGe with a diameter of 20−63 μm was active for the structure-sensitive reaction possibly due to the specific electronic structure or a specific reaction field constructed by atomic ordered structures. For Co 2 FeGa, the H 2 production rate was slightly less for the methanol decomposition than that for 2:3 SRM. SEM only revealed surface roughening, not fine particle formation (Figure 15b). XPS analysis revealed an increase in the composition of Ga among metals (Table 1). However, it was less significant than that observed for 2:3 SRM. Thus, Ga oxides are considered to play a key role in catalysis of methanol decomposition. By the comparison of all results, secondary phases that might be FeO for Co 2 FeGe and Co 2 C for Co 2 FeGa (Figure 15c) were not likely to contribute to the methanol decomposition without steam.

METHODS
Heusler alloy ingots were prepared by arc-melting and subsequent annealing, the conditions of which were described previously. 5,6 First, the ingots were crushed using a mortar and pestle. Second, the obtained powders were sieved to 20−63 μm for catalytic measurements. The alloys were (almost) a single phase with a highly ordered L2 1 Heusler structure, 5,6 which was examined by XRD for powders with a size of less than 20 μm post-annealed at 600°C for removing strains.
A catalytic reaction was conducted in the gas phase using a fixed-bed flow reactor. A powder catalyst was held on quartz wool fixed in a quartz tube with an internal diameter of ∼7.5 mm. Considering formula weights, catalyst amounts were 0.60, 0.75, and 0.90 g for alloys with Z = Al, Si; Z = Ga, Ge; Z = Sn, respectively, so that the catalyst layer height was ∼5 mm. The catalyst was preheated at 600°C for 1 h under H 2 flow for removing surface oxides and strains introduced because of crushing. A reactant mixture of CH 3 OH and H 2 O with a molar ratio of 2:3 was fed using a plunger pump at a constant rate of 0.1 mL min −1 , where the actual measured rate was typically 0.07 g min −1 corresponding to 0.08 mL min −1 and a liquid hourly space velocity (LHSV) of ∼20 h −1 . In addition to the reactant, 30 mL min −1 of N 2 carrier gas was introduced during the reaction. The reaction temperature was controlled using an electric furnace surrounding the reactor. Gaseous products (H 2 , CO, CO 2 , and CH 4 ) were analyzed by gas chromatog-

ACS Omega
Article raphy (Shimadzu GC 14A). 29 Their production rates were estimated from the analyzed concentrations and downstream flow rate measured at room temperature using a flow meter. These analyses were performed 30 min after achieving the measurement temperatures from every 50°C from 300°C when heating up to 600°C and every 100°C when cooling down to 300°C.
For selected samples, the effect of steam was investigated by the SRM using the reactant CH 3 OH/H 2 O = 2:1 and 1:0 (methanol decomposition), with a methanol feeding rate of 0.04 g min −1 , corresponding to that in the SRM with the reactant CH 3 OH/H 2 O = 2:3. In addition, the effect of oxygen pretreatment was investigated by the addition of a heating process at 500°C for 1 h under a 100 mL min −1 flow of [1% O 2 /He balance] gas before the reaction.

CONCLUSIONS
Fourteen Heusler alloy catalysts were examined for SRM. Depending on elemental sets, various catalytic properties were observed because of structural changes occurring during the reaction; these structural changes apparently depended on a balance of component oxidizability. The H 2 production rates over Co 2 TiAl and Ni 2 TiAl significantly increased because of the formation of fine particles via the selective oxidation of Ti during the reaction, although most of H 2 was produced by methanol decomposition. H 2 production was enhanced by oxygen pretreatment, promoting fine particle formation. All of the Mn-containing alloys formed MnO, which catalyzed the water-gas shift reaction and produced CO 2 from CO generated by methanol decomposition. Co 2 FeGa apparently exhibited activity for methanol decomposition because of the formation of Ga oxides. By choosing appropriate elemental sets, Heusler alloys can be good precursors of catalysts that exhibit high activity and desirable selectivity. On the other hand, Co 2 FeGe apparently catalyzed methanol decomposition in the initial metallic states, possibly related to its unique electronic structure or a unique atomic ordered reaction field.