Heusler Alloys: A Group of Novel Catalysts.

In this study, we investigated the catalytic properties of various Heusler alloys for the hydrogenation of propyne and the oxidation of carbon monoxide. For propyne hydrogenation, Co2FeGe alloy showed a higher activity than that of elemental Co, where neither Fe nor Ge showed any activity. This clearly indicates an alloying effect. For the oxidation of carbon monoxide, although most alloys showed a significant change in catalytic activity during measurement due to an irreversible oxidation of the alloy, Co2TiSn alloy showed a very small change. The results indicate that the catalytic activity and stability of a Heusler alloy can be tuned by employing an appropriate set of elements.

Introduction

A Heusler alloy is an intermetallic compound (ordered alloy) described as X2YZ with an L21 structure (Figure ).[1] It is called a full-Heusler alloy in a narrow sense, whereas XYZ with a C1b structure is called a half-Heusler alloy. Typically, X and Y are transition metals in groups 3–8 and 8–12, respectively, and Z is in group 13–15.[2] Because many Heusler alloys are ferromagnetic, they have been mainly studied as magnetic materials since the discovery of Cu2MnAl in 1903.[3] In particular, they have recently attracted much attention in the field of spintronics.[4] There are also of paramount importance in other fields, such as Ni2MnGa for a ferromagnetic shape memory alloy[5] and Fe2VAl for a thermoelectric material.[6] One interesting feature of Heusler alloys is that their electronic structure can be tailored by elemental substitution in accordance with the rigid-band approximation. This enables tuning of the functional properties.

Figure 1

Crystal structure of (full-) Heusler alloy (X2YZ): L21 structure. Drawing was done using VESTA.[1]

Transition metals exhibit various catalytic functions, and their catalytic properties are dominated by the surface and electronic structures and defects. Alloying modifies the atomic configuration at the surface and the electronic structure of any given element, resulting in variation in the catalytic properties. An intermetallic compound is an extreme alloy that exhibits a specific atomic arrangement at the surface and an electronic structure completely different from that of its constituent elements. Thus, intermetallic compounds have recently attracted attention as new catalysts.[7−9]

Heusler alloys are potentially valuable catalysts because there are so many possible sets of elements and electronic tuning can be done by elemental substitution. To our knowledge, there have been no experimental studies on the catalytic properties of Heusler alloys; however, only one theoretical study has appeared very recently.[10] In this article, we describe our experimental investigation of the catalytic properties of 12 Heusler alloys for the hydrogenation of propyne and the oxidation of carbon monoxide as a first step toward exploring new catalysts.

Results and Discussion

Structural Characterization

Powder samples of Heusler alloys were prepared by arc melting and annealing, followed by crushing. X-ray diffraction (XRD) was performed to characterize the ordered structure. The diffraction pattern of L21-phase was sufficiently observed for all of the samples, as shown in Figure . Although several samples showed unknown extra peaks, their intensities were negligible (Figure S2 and Table S1). Thus, samples were almost of single phase.

Figure 2

XRD patterns of (a, b) Co2TiSn and (c, d) Co2FeGe; (b) and (d) are calculated patterns. Inset in (c) is magnified image around 111 and 200 peaks along with calculated peaks (blue bars).

Many of Heusler alloys have two disordered phases: B2-phase (Y–Z disorder) and A1-phase (X–Y–Z disorder, i.e., completely random) (Figure S1). Considering these disorders, the degree of chemical ordering in Heusler alloys is typically evaluated by Webster’s model using factors S and α.[11] The S-factor corresponds to the long-range order parameter in the binary alloy.[12]S = 0 means A1-phase, and S = 1 means no disorder of X atoms. The α-factor describes a disorder between Y and Z atoms. α = 0 means no Y–Z disorder, and α = 0.5 means no Y–Z order. Thus, S = 1 and α = 0 indicate L21-phase, and S = 1 and α = 0.5 indicate B2-phase. These factors are obtained by the following equationswhere I200, I111, and Ifund are the integrated intensities of 200 and 111 superlattice peaks and a fundamental peak, respectively, the numerator is an experimentally obtained value, and the denominator is the calculated value in the perfect-order case. In our study, to avoid the influence of a preferred orientation, the S value was estimated using a 400 peak for Ifund in eq , and the α value used was the average of the ones estimated using Ifund for the 220, 400, 422, and 440 fundamental peaks in eq . Table shows the obtained S-factors and α-factors. A high S value and a low α value were obtained for all of the samples, indicating that the alloys sufficiently ordered in the L21-structure.

Table 1

Prepared Heusler Alloys and Their Ordering Factors, S and α, with Standard Deviation (SD)a

material	S	SD	α	SD
Fe2TiSn	1.0		0.05
Co2TiAl	0.97	0.04	0.14	0.02
Co2TiGe	1		0.01
Co2TiSn	0.99	0.01	0.01	<0.01
Co2MnSi	1.00	0.04	0.06	<0.01
Co2MnGe	1.0		0.05
Co2MnSn	0.97	0.01	0.05	<0.01
Co2FeGe	0.97	0.02	0.04	<0.01
Ni2TiAl	0.98	0.01	0.01	<0.01
Ni2TiSn	1.0		0.00
Ni2MnSn	1.0		0.0
Cu2TiAl	0.95	<0.01	0.00	<0.01

The SD value for S was estimated from the deviations between experimental data and fitting curve using Voigt function. The SD for α includes an SD among α values estimated using Ifund for the 220, 400, 422, and 440 peaks, in addition to the fitting errors. S = 1 and 1.0 mean that S values were somewhat larger than 1.0 and 1.00, respectively. α = 0.0 means that the α value was smaller than 0.00 (negative value).

Hydrogenation of Propyne

A catalytic reaction involving 1% C3H4 (propyne)/55% H2/He-balance mixture was investigated in a fixed-bed flow reactor using a catalyst sample pretreated with H2 at 600 °C for 1 h to remove surface oxides. The reaction was monitored during heating the catalyst from 40 to 200 °C. To check the change in the catalytic performance during measurement, the reaction was also monitored during a cooling process and the second run of the heating and cooling cycle. The catalytic activity in the first heating process was higher than that in the rest of the processes, whereas the temporal change was well settled in the second cycle (Figure S3). Thus, we discuss the results upon cooling in the second cycle (Figure ). The Heusler alloy catalysts except Co2FeGe and Co2MnGe showed only a few percent C3H4 conversion even at 200 °C, as shown by the open squares (Co2TiSn as a representative) in Figure a. The Co2FeGe and Co2MnGe showed catalytic activity with C3H4 conversions of 63 and 30%, respectively, at 200 °C.

Figure 3

Catalytic properties for hydrogenation of propyne (C3H4): (a) conversion of C3H4 and (b) selectivity of products (upper: C3H6, lower: C3H8). C3H4 conversions for the other Heusler alloys were similar to those for Co2TiSn, which is shown as a representative alloy in (a). Data of elemental Co were obtained using a commercial Co powder.

In preliminary experiments, commercial powders of Fe, Mn, and Ge were verified to show no catalytic activity for propyne hydrogenation, whereas Co powder showed a certain level of activity, as shown in Figure a,b. Thus, Co atoms were considered to mainly contribute to the catalytic activities of Co2FeGe and Co2MnGe, so the product selectivities were similar for these three catalysts. However, the Co2FeGe showed a higher C3H4 conversion than the commercial Co powder even though the population of Co atoms in the Co2FeGe is half of that in the Co powder (total surface area was the same for each sample, as described in Section ). This clearly indicates an alloying effect on the catalytic activity through changes in the atomic arrangement and/or electronic structure. On the other hand, the Co2MnGe showed about half the C3H4 conversion of Co2FeGe. This difference was brought about by the difference in constituent elements (Mn vs Fe), whereas both the Fe and Mn powders did not show any activity probably due to too strong adsorption of hydrocarbons. In general, the adsorption energy of hydrocarbons on transition metals is higher for a lower group number, as expected from the electronic structure derived by calculation.[13] Thus, propyne adsorption sites consisting of Co and Mn atoms are considered to be less active than those consisting of Co and Fe. The difference in the dissociation yield of H2 is another possible reason. According to first-principles calculations, in the minority spin state, the Fermi level lies in the band gap for Co2MnGe,[14,15] whereas it is at the bottom of the conduction band for Co2FeGe.[16,17] Because a high electron density at the Fermi level induces dissociation of H2 through electron transfer from the catalyst to molecules,[7] Co2FeGe with a larger number of electrons is considered to have a higher activity for propyne hydrogenation.

As discussed above, the tested Heusler alloys except Co2FeGe and Co2MnGe showed very low activity. Here, we discuss why their activities were much lower than those of Co2FeGe and Co2MnGe. The first factor is the difference in the electronic structure. The density of states for Co2MnSi, Co2MnGe, and Co2MnSn are very similar,[15] whereas only the Co2MnGe showed a relatively high activity. Thus, the electronic structure is unlikely the sole reason. The second factor is the formation of inactive surface oxides. Many of the samples included base metals that have very stable oxides, as can be seen from the standard Gibbs energies of formation (ΔGf°, Table ).[18,19] If the presence of surface oxides was the cause, as the Co2MnGe showed activity in catalytic reaction, the surface oxides of Mn and the other elements in Table with ΔGf° smaller than Mn were probably reduced upon heating under H2 flow. In contrast, the surface oxides of Al, Si, and Ti were hardly removed by the H2 reduction. Because Fe–Ti–O natural surface oxides decompose into Fe and TiO2 when heated under a H2 atmosphere,[20] it is likely that such residual surface oxides substantially decreased the active ensembles in the alloys containing Al, Si, and/or Ti. In contrast, for the Ni2MnSn and Co2MnSn alloys, which showed very low activity, the surface oxides were likely reduced because the ΔGf° values for NiO, CoO, and SnO2 are larger than those for MnO. A plausible reason for the low level of activity is the surface segregation of Sn, which inhibits H2 dissociation,[21] because its surface energy is much lower than that of Ni, Co, and Mn.[22] In addition, according to the theoretical calculation for a stable surface termination on Co2MnZ(100) (Z = Si, Ge, Sn), the Co termination becomes unstable with an increasing number of core electrons of Z atoms; Mn, Z, and MnZ-mixed terminations become stable; and Co termination should be impossible, especially for Z = Sn.[23] It is thus presumed that catalytically active Ni and Co atoms were embedded underneath the surface, resulting in the observation of low catalytic activity.

Table 2

Standard Gibbs Energies of Oxide Formation for Elements Constituting Heusler Alloys[18,19]a

element	oxide	ΔGf° [kJ (O2 1 mol)−1]
Fe	Fe3O4	–508
Co	CoO	–428
Ni	NiO	–423
Cu	Cu2O	–292
Ti	TiO2 (rutile)	–890
Mn	MnO	–726
Al	Al2O3 (a)	–1055
Si	SiO2 (quartz)	–857
Ge	GeO2 (tetragonal)	–521
Sn	SnO2	–516

Values were normalized to 1 mol of O2 (2m/nM + O2 = 2/nMO).

Oxidation of Carbon Monoxide

As with the propyne hydrogenation, the measurement was conducted in the fixed-bed flow reactor during two cycles of heating and cooling in the temperature range of 80–600 °C. A CO-rich mixture (1.2% CO/0.4% O2/He-balance) was used as a reactant to suppress the irreversible oxidation of catalysts. While selected results are discussed here, all of the results are provided in the Supporting Information (Figures S4 and S5). Figure shows the conversion of CO in the carbon monoxide oxidation upon heating in the first cycle. It is important to note that the ideal CO conversion was 66.7% when only CO + (1/2)O2 → CO2 occurred without any other reactions, such as the irreversible oxidation of catalyst, as the concentration ratio in the reactant was CO/O2 = 3:1. For X2TiSn, X2TiAl, and X2MnSn (Figure a–c), X = Co showed a higher activity than other alloys. This tendency is similar to that for the elemental catalysts (commercial powders) (Figure d). Thus, element X probably played the main role in the catalytic reaction for the carbon monoxide oxidation.

Figure 4

Conversion of CO in oxidation of carbon monoxide by (a) X2TiSn, (b) X2TiAl, (c) X2MnSn, and (d) elemental X catalysts (commercial powders) under CO-rich condition (CO/O2 = 3:1). Color symbols of each X in (a)–(c) correspond to those represented in (d). Data are for the first heating process. CO conversion beyond 66.7% by elemental Cu at higher temperatures in (d) indicates reduction of Cu oxide formed at lower temperatures.

Figure a1,b1,c1 shows the CO conversion in all of the thermal processes for the X2TiSn alloys. The Fe2TiSn and Ni2TiSn exhibited hysteresis between each thermal process, as shown in Figure a1,c1, which indicated a change in the surface states of catalysts during the reaction. Such hysteresis was also observed for most alloys (Figure S4). The behavior of hysteresis was different for each alloy; for example, Figure a1 shows that only the second heating datum is substantially different from that of the others, whereas Figure c1 shows three types of curves: the first heating, the second heating, and two coolings. Thus, it is difficult to fully reveal the origin of such complex behaviors. In contrast, the Co2TiSn alloy did not exhibit significant hysteresis even though the elemental Co (commercial powder) showed a certain level of hysteresis, as shown in Figure b1,d1. The component of the O2 conversion consumed for the irreversible oxidation of these catalysts was estimated using (O2 conversion) – 3/2 (CO conversion), as shown in Figure a2,b2,c2,d2. The irreversible oxidations of the Fe2TiSn and Ni2TiSn were much greater than those of the Co2TiSn and elemental Co. Most of the other alloys also showed significant irreversible oxidation (Figure S5). These findings indicate that the hysteresis in the CO conversion curves was due to a complex oxidation behavior resulting from the combination of individual oxidations of each element in the ternary alloys.

Figure 5

Conversion curves for oxidation of carbon monoxide in (a1, a2) Fe2TiSn, (b1, b2) Co2TiSn, (c1, c2) Ni2TiSn, and (d1, d2) elemental Co under CO-rich condition (CO/O2 = 3:1) and (e1, e2) Co2TiSn and (f1, f2) elemental Co under O2-rich condition (CO/O2) = (1:1). Upper graphs for each material show conversion of CO. Lower graphs show conversion of O2 consumed for irreversible oxidation of catalyst. Data points were divided among thermal processes, as are shown by legends in (a1) and (a2). Heating: filled symbols; cooling: open symbols; first cycle: deep and pale green squares (CO) and red and pink circles (O2); and second cycle: deep and pale orange diamonds (CO) and deep and pale blue triangles (O2).

To get insight into the stability of Co2TiSn, a catalytic measurement with an O2-rich reactant (CO/O2 = 1:1, each concentration: 0.67%) was performed for the Co2TiSn and elemental Co, as shown in Figure e1,e2,f1,f2 (irreversible oxidation: (O2 conversion) – 1/2 (CO conversion)). Although the hysteresis was enhanced for both samples compared to that under the CO-rich condition, the irreversible oxidation was obviously smaller for the Co2TiSn than that for the elemental Co.

Figure shows the XRD patterns obtained after O2-rich measurement. For the elemental Co, the peak intensities of the oxides were higher than those of metallic Co (hexagonal close-packed (hcp) and face-centered cubic (fcc)). In contrast, for the Co2TiSn, the oxide peak was much lower than the metallic Co2TiSn peak. The appearance of Co3Sn2 in Figure b is probably due to phase separation resulting from the preferential oxidation of Ti. Although the origin of the oxidation resistance of Co2TiSn is still unclear, this resistance is considered to be the main reason for the stability of the catalytic performance. However, the Co2TiAl showed a large hysteresis (Figure S4b) even though the irreversible oxidation was relatively small (Figure S5b). Thus, other factors must have affected the stability besides the oxidation resistance, which is just one of the origins.

Figure 6

XRD patterns of (a) elemental Co and (b) Co2TiSn after catalytic measurement for oxidation of carbon monoxide under O2-rich condition (CO/O2 = 1:1) and simulated patterns of (c) hcp-Co, (d) fcc-Co, (e) Co3O4 (space group: Fd3̅m), (f) CoO (Fm3̅m), (g) L21-Co2TiSn, and (h) Co3Sn2 (Pnma). Symbols h, f, and * indicate peaks of hcp-Co, fcc-Co, and L21-Co2TiSn, respectively. Reference XRD patterns can be downloaded from AtomWork (http://crystdb.nims.go.jp) provided by the National Institute of Materials Science.[24] Patterns were simulated with the data published in refs (25) (hcp-Co, CoO (space group Fm3̅m)), (26) (in French, fcc-Co), (27) (Co3O4(Fd3̅m)), (28) (L21-Co2TiSn), and (29) (Co3Sn2 (Pnma)).

Methods

Preparation and Characterization of Catalysts

Twelve Heusler alloys listed in Table were prepared from pure metallic sources (purity: >99.9%) by arc melting, followed by annealing (homogenization and then ordering: conditions are listed in Table S1) under an Ar atmosphere. Annealed ingots were crushed using a hammer and a mortar and pestle, and the powder obtained was sieved to a particle size of 20–63 μm for catalytic measurement and <20 μm for XRD measurement (the commercial Co powder was also sieved to 20–63 μm as a reference catalyst). The XRD measurement was done using the Bragg–Brentano geometry with Cu Kα radiation (Rigaku, Ultima IV Diffractometer) to determine whether the sample was of single phase or not and the degree of chemical ordering. The powder sample was annealed at 600 °C for 1 h under a H2 atmosphere (the same condition as that of the pretreatment of the catalyst) before the XRD measurement to remove the strain and defects introduced due to the crushing. The degree of order was evaluated using factors S and α estimated from the intensity ratio of the superlattice and fundamental peaks using eqs and 2, respectively, as described in Section . Theoretical intensity ratios in the perfect-order case in eqs and 2 were calculated from the atomic scattering factors, including the anomalous dispersion terms[30] the multiplicity factor, the Lorentz–polarization factor, and the Debye–Waller factors.[31,32]

The surface area of the powder catalyst after H2 pretreatment (Table S1) was estimated using the Brunauer–Emmett–Teller method with Kr adsorption (MicrotracBEL, BELSORP-max volumetric adsorption instrument).

Evaluation of Catalytic Properties

Catalytic measurements were conducted in a standard fixed-bed flow reactor. The hydrogenation of propyne (aC3H4 + bH2 → cC3H6 + dC3H8) and the oxidation of carbon monoxide (CO + (1/2)O2 → CO2) were used as typical catalytic reactions. The reason for using propyne (C3H4: HC≡C–CH3) is that the simplest alkyne, acetylene, easily produces oligomers, and further oils in a catalytic reaction then poison the catalyst. The reactants used were gaseous mixtures: 1% C3H4/55% H2/He-balance for the propyne hydrogenation and 1.2% CO/0.4% O2/He-balance for the carbon monoxide oxidation. The carbon monoxide oxidation was conducted under a CO-rich condition (CO/O2 = 3:1) to suppress the irreversible oxidation of the catalyst. The carbon monoxide oxidation with 0.67% CO/0.67% O2/He-balance (O2-rich condition) was also performed for the Co2TiSn and the commercial Co powder catalysts.

The catalyst was supported on quartz wool in a quartz tube with a 4 mm internal diameter and surrounded by an electric furnace. The amount of catalyst was employed such that the total surface area (At) was 0.027 m2, and the flow rate of reactants (F) was fixed to 30 mL min–1, meaning that At/F was the same for all measurements. Therefore, the data obtained can be compared without normalization (the reaction yield is proportional to At/F (=Avtc = Av/SV, Av: surface area per volume; tc: contact time; SV: space velocity)).

Before measurement, heating of the catalyst under a H2 gas flow at 600 °C for 1 h was carried out to remove the surface oxides. After cooling to room temperature, the reactant mixture was introduced through mass flow controllers into the catalyst channel. The products and unreacted reactants were monitored using a gas chromatograph (Agilent, 490 Micro GC equipped with a thermal conductivity detector and two columns: Molesieve 5A and PoraPLOT Q) during the continuous heating and cooling of the catalyst. Temperature range, heating and cooling rates, and sampling step were 40–200 °C, 2.5 °C min–1, and 5 °C, respectively, for the propyne hydrogenation, and 80–600 °C, 5 °C min–1, and 10 °C, respectively, for the carbon monoxide oxidation, although the actual cooling rates were lower than the set rates, especially below 200 °C. The heating and cooling cycle was run twice to check the change in the catalytic performance during measurement.

The catalytic properties for the propyne hydrogenation were investigated using the conversion of C3H4 ((Crea – Cpro)/Crea) and a selectivity of products (CC3H6 or CC3H8/(CC3H6 + CC3H8)), where Crea and Cpro are the concentrations of C3H4 in the reactant and the product, respectively, and CC3H6 and CC3H8 are the concentrations of C3H6 and C3H8 in the product, respectively. Products resulting from the decomposition and oligomerization of hydrocarbons were not detected below 200 °C. For the carbon monoxide oxidation, the catalytic properties were investigated using the conversions of CO and O2 in a manner analogous to that for the propyne hydrogenation.

Conclusions

Twelve kinds of Heusler alloys with highly ordered L21 single phases were investigated in terms of catalysis. For propyne hydrogenation, the Co2FeGe alloy showed a higher activity than the elemental Co even though the elemental Fe and Ge did not show activity, which indicates an alloying effect. The Co2MnGe showed a lower activity than the Co2FeGe, which is attributed to the difference in the electronic structure. However, the other alloys showed very low activity. One possible reason for the low activity is the embedding of catalytically active atoms under the surface with residual oxides and/or nonactive elements. For the carbon monoxide oxidation, the alloys with X = Co in X2YZ showed a higher activity than the other alloys with X = Fe, Ni, and Cu, which was similar to the tendency in elemental X powders. Thus, element X in X2YZ apparently played the main role in the catalytic reaction. While most of the other alloys showed hysteresis in the activity for the cycles of heating and cooling due to the irreversible oxidation of the alloy, the Co2TiSn alloy showed a very small hysteresis. The stability of the Co2TiSn is attributed to the high resistance to irreversible oxidation. Our findings indicate that a suitable catalyst can be developed for a variety of target reactions by using Heusler alloys consisting of appropriate elements.