2017 Volume 58 Issue 5 Pages 776-781
We investigated the catalytic properties of Fe1−xNix nanoparticles prepared by a polyol-based method that allowed the precise control of their composition and structure. In the hydrogenation of propyne, the reaction rate decreased with decreasing x, which was in good agreement with the scenario that inactive Fe atoms dilute active Ni ensembles in particles at 0.55 ≤ x ≤ 1. For particles with x ≤ 0.46, it was suggested that slightly active Fe-Ni ensembles mainly contributed to the activity due to small population of active Ni ensembles. Since the change in catalytic properties with x was well explained by the assumption that the surface composition corresponded to x, it was concluded that the surface composition dominating the catalytic properties can be tuned by adjusting the entire composition of nanoparticles.
Noble metals have been very widely used as catalysts, such as Pt/Pd/Rh in three-way catalytic converters and Pd for the hydrogenation of hydrocarbons, because of their high activity and high stability.1–4) However, as noble metals are very expensive and their resources are localized and limited, alternative catalysts consisting of base metals are required. To develop high-performance catalysts, alloying is often used to tune the properties or to create novel properties.5–12) For industrial use, catalysts must be nanoparticulate to ensure high surface area. In base metal alloys, however, it is difficult to synthesize nanoparticles via a liquid-phase route because the oxidation-reduction potentials of their constituent elements are low. Hence, the development of an effective preparation process for such nanoparticles is desired.
Very recently, we have developed a novel process to synthesize Fe-Ni alloy nanoparticles with sizes less than ca. 50 nm by hydrogen reduction of Fe-Ni hydroxides precipitated from polyol solution.13) X-ray diffraction (XRD) and inductively coupled plasma-atomic emission spectrometry (ICP-AES) revealed that their crystal structures and chemical compositions were precisely controlled in good accordance with the phase diagram of the system. Thus, this method can be applied to the preparation of a variety of nanoparticles for magnetic materials, catalytic materials, and so on. However, the catalytic properties of nanoparticles prepared in this way have not been investigated, though their structure and composition were shown to be similar to those of the bulk materials.
The structure and composition at the surface of nanoparticles are often different from those in the bulk. For example, surface orientations can be controlled for elemental metals and even alloys using appropriate preparation conditions,14,15) which implies that inappropriate preparation conditions can lead to undesirable surfaces. In alloys, large differences in the surface energies of the constituents causes surface segregation, such as in Ni-Cu.16) Besides the surface energy, the surface composition depends on the conditions used in liquid-phase synthesis owing to the different reduction processes of the constituents. For example, in noble-metal core-shell nanoparticles, the tendency to form shells follows the trend Rh > Pd > Pt > Au,17) but the surface energies follow the trend Rh > Pt > Pd > Au.18)
In this study, we investigated the catalytic properties of composition-controlled Fe-Ni nanoparticles prepared by the above-mentioned method, and verified that their surface chemical states were as well controlled as their bulk states by analyzing the composition dependence of their catalytic properties.
Details on sample preparation are given in Ref. 13). Fe-Ni hydroxides with a layered double hydroxide structure were obtained by precipitation from an ethylene glycol solution containing FeCl2·4H2O, Ni(CH3COO)2·4H2O, and NaOH. By heating under H2 gas flow at 400℃ (Fe-Ni) or 300℃ (Ni) for 1 h, the hydroxides were reduced to Fe1−xNix alloy nanoparticles with x = 0.33, 0.46, 0.55, 0.70, and 1, as measured by ICP-AES. XRD revealed that the nanoparticles were single phase with an fcc structure (except that with x = 0.33, which consisted of bcc-Ni-poor and fcc-Ni-rich phases). Their surface areas (ABET) were estimated to be 1.6–7.2 m2 g−1 (Table 1) using the Brunauer-Emmett-Teller method with Kr adsorption (MicrotracBEL, BELSORP-max volumetric adsorption instrument).
| x | ABET/m2 g−1 | Ea/kJ mol−1 | Reaction order | |
|---|---|---|---|---|
| C3H4 | H2 | |||
| 1 | 3.2 | 49 | −0.3 | 0.9 |
| 0.70 | 5.1 | 45 | −0.2 | 0.8 |
| 0.55 | 5.0 | 41 | – | – |
| 0.46 | 1.6 | 36 | – | – |
| 0.33 | 7.2 | 42 | −0.3 | 1.0 |
Catalytic measurements were conducted in a conventional fixed-bed flow reactor. The hydrogenation of propyne (aC3H4 + bH2 → cC3H6 + dC3H8) was used as a model catalytic reaction. The gaseous reactant mixture used was 1% C3H4/55% H2/He-balance, except in the measurements used to estimate reaction orders (see below). The products and unreacted reactants were monitored using an Agilent 490 Micro gas chromatography apparatus equipped with a thermal conductivity detector and a PoraPLOT Q column, and the conversion of C3H4 and selectivity for products (i.e., the ratio of C3H6 and C3H8) were estimated. The reaction rate of C3H4 per total surface area (r (mL (STP) min−1 m−2)) was used to compare the catalytic activities of the samples.
An appropriate amount of catalyst (7–33 mg) was supported on quartz wool in a quartz tube with a 4 mm internal diameter surrounded by an electric furnace. Before measurement, heating the catalyst under a H2 gas flow at 400℃ (Fe-Ni) or 300℃ (Ni) for 1 h was performed to remove the surface oxides formed in air. After cooling to room temperature, the reactant was introduced into the catalyst channel using mass flow controllers at a typical flow rate of 30 mL (STP) min−1 for 1 h. The gas passed through the catalyst was analyzed during heating from room temperature to 150℃ at a rate of 2.5℃ min−1. A decrease in catalytic activity during the measurement was not significant, as confirmed by measurement during the cooling process. Products resulting from the decomposition and oligomerization of hydrocarbons were not detected below 150℃.
Reaction orders with respect to C3H4 and H2 were estimated by measurement of the reaction rate at 40℃ (x = 1 and 0.70) or 120℃ (x = 0.33) using reactant mixtures of n% C3H4 (n = 0.5–1.7)/20% H2/He-balance and 1% C3H4/m% H2 (m = 5–55)/He-balance, respectively.
Figures 1(a)–(e) show the conversion of C3H4 and the selectivities for C3H6 and C3H8 vs. temperature for a series of Fe1−xNix samples. Note that the C3H4 conversion curves cannot be compared between each sample because the weights and surface areas were different. For pure Ni (Fig. 1(a)), the C3H8 selectivity is low in the low-conversion region, but drastically increases as the conversion increases (> 50%). This behavior is in good agreement with results in the literature.19,20) The reaction process on Ni surface is schematically described in Fig. 2. The hydrogenation proceeds on the surface by addition of hydrogen to adsorbed hydrocarbons ((a) and (b) → (c) → (f) in Fig. 2). C3H4 molecules strongly adsorb (Fig. 2(b)) and hardly desorb, while C3H6 molecules can desorb (Fig. 2(d)) but hardly readsorb (Fig. 2(e)) in the presence of strongly adsorbed C3H4, corresponding to the situation at low C3H4 conversion, which reduces the opportunity of formation of C3H8 (Fig. 2(f)). On the other hand, C3H6 becomes able to readsorb when more C3H4 molecules leave the catalyst surface at high conversion, which leads to the production of C3H8 (Fig. 2(f)). This is the mechanism of the drastic change in selectivities.19,20) Such the change in selectivity is smaller but also observed for x = 0.70 (Fig. 1(b)). Whereas, no drastic change in selectivity is observed for x ≤ 0.55 (Figs. 1(c)–(e)). These results indicate that Fe atoms affect selectivity.
Conversion of C3H4 (filled circles) and selectivity (open circles, left: C3H8, right: C3H6) for Fe1−xNix alloy catalysts with x = (a) 1, (b) 0.70, (c) 0.55, (d) 0.46, and (e) 0.33; (f) reaction rate of C3H4 (r); and production rates (rp) of (g) C3H6 and (h) C3H8. Legend in (f) shows data symbols in (f)–(h); filled circles: x = 1, triangles: x = 0.70, squares: x = 0.55, diamonds: x = 0.46, and open circles: x = 0.33.
Schematic of the hydrogenation of propyne on Ni surface. (a) Dissociative adsorption of H2, (b) adsorption of C3H4, (c) addition of hydrogen to the propyne intermediate, (d) desorption of C3H6, (e) readsorption of C3H6, (f) addition of hydrogen to the propene intermediate, and (g) desorption of C3H8. The step shown in (e) occurs with difficulty in the presence of strongly adsorbed propyne.
Figures 1(f)–(g) show the reaction rate of C3H4 (r) and the production rates (rp) of C3H6 and C3H8 in mL min−1 m−2. Note that, in the temperature region where the C3H4 conversion is 100%, the data cannot be discussed between samples. The C3H4 reaction rate in Fig. 1(f) is higher for higher x, and thus, so too with both productions. From Figs. 1(g) and (h), the pure Ni seems to be the best catalyst to obtain C3H8, and to obtain C3H6 when unreacted C3H4 is allowed to remain (low C3H4 conversion region). When large fraction of C3H6 is required without unreacted C3H4, it is better to use the Fe-Ni alloy catalysts at higher temperature.
Figure 3 shows r vs. x at 40℃. The measurement for x = 1 was performed four times using four different samples including the sample used in Fig. 1 because the error was large due to the small amount of samples.21) The average is shown at x = 1 in Fig. 3 with the error corresponding to the maximum and minimum values. The rate decreases with a decrease in x. Since commercial Fe powder shows no catalytic activity at 40℃, it is considered that inactive Fe atoms dilute the active Ni ensembles.
Reaction rate of C3H4 (r) at 40℃ for Fe1−xNix alloy catalysts. Inset is a magnification of x ≤ 0.8. The rate for x = 0 was assumed to be 0 by measurement using commercial Fe powder. Although the error bar for x = 1 seems to be large in linear scale, it is acceptable since the error depends on the r value multiplicatively.21) Dependence of the r value on x will be essentially discussed in log scale using Fig. 6 in which the error bar is obviously acceptable.
Figure 4 shows the Arrhenius plots of r. The activation energies (Ea) listed in Table 1 were estimated from Fig. 4 by fitting the data below the temperature, above which the drastic change in selectivity occurs. The estimated Ea for x = 1 is in good agreement with the literature (51 kJ mol−1), whereas reported Ea for pure Fe is about 31 kJ mol−1.19) The Ea values in Table 1 decrease with a decrease in x, indicating that alloying with Fe reduces the apparent Ea.
Arrhenius plot of the reaction rate of C3H4 (ln r vs. 1/T) for Fe1−xNix alloy catalysts with x = 1 (filled circles), 0.70 (triangles), 0.55 (squares), 0.46 (diamonds), and 0.33 (open circles). Typical temperatures in ℃ are shown at the top of the graph as a guide.
Figures 5(a) and (b) show double logarithmic plots of r vs. the C3H4 and H2 concentrations in the reactants, respectively. The reaction orders with respect to C3H4 and H2 were estimated from Fig. 5 and are listed in Table 1. The values are similar for the pure Ni and Fe-Ni samples, and roughly correspond to the values in the literature (C3H4: −0.2–0 (Ni); 0 (Fe), H2: 1.0–1.2 (Ni); 1.0 (Fe)).19,20) It is clear that C3H4 strongly adsorbs and the step determining reaction rate is the dissociative adsorption of hydrogen (Fig. 2(a)) or the addition of hydrogen to an intermediate of the adsorbed propyne (Fig. 2(c)).22–25)
Double logarithmic plots of the dependences of C3H4 reaction rate (r) on concentrations of (a) C3H4 and (b) H2 in the reactants for Fe1−xNix alloy catalysts with x = 1 (filled circles), 0.70 (triangles), and 0.33 (open circles). Data were recorded at 40℃ for x = 1 and 0.70 and at 120℃ for x = 0.33. The data shown are the averages of several measurements and the error bars correspond to the maximum and minimum values. Data are shown in arbitrary units because they are offset for clarity. Concentrations used are shown at the top of the graph as a guide.
Based on the scenario that inactive Fe atoms dilute active Ni ensembles in Fe-Ni alloy catalysts, the reaction rate (r) is proportional to the existence probability (population) of active Ni ensembles,26,27) according to the following equations:
| \[r = r_{\rm Ni} x^N,\] | (1) |
| \[\log r = \log r_{\rm Ni} + N \log x,\] | (2) |
Double logarithmic plots of C3H4 reaction rates (r) at 40℃ (circles) and 70℃ (triangles) vs. Ni composition (x) for Fe1−xNix alloy catalysts. The x values of the data points are shown at the top of the graph as a guide. The solid line is a fitting of eq. (2) with N = 7 to data at 40℃ with 0.55 ≤ x ≤ 1. Open symbols indicate values for x = 0.43 estimated from the values for x = 0.33 consisting of two phases.
The sample with x = 0.33 is a eutectic of Ni-poor and Ni-rich phases, whose interface could contribute the catalytic properties similar to the case of the interface between a catalyst and a support. Assuming that the two phases have x = 0.055 and 0.43, as indicated in the phase diagram at 400℃ (reduction temperature),31) their mole fraction is estimated to be 34:66 from XRD by comparing experimental and calculated intensities, which roughly correspond to the value expected from phase diagram (27:73). Since the phase with x = 0.055 is considered to be completely inactive due to its very low Ni content, the r value of the phase with x = 0.43 was calculated as rx=0.33/0.66 and plotted in Fig. 6. The r value for x = 0.43 is similar to that for x = 0.46. Thus, we rule out additional effects raised from the eutectic structure. The small difference in r may be attributed to the difference in surface areas (particle sizes) since this reaction is structure sensitive.32)
The deviation of data for x ≤ 0.46 from the extrapolated line in Fig. 6 can be explained by the presence of slightly active ensembles consisting of both Fe and Ni atoms. At higher x, the contribution of the Fe-Ni mixed ensembles to the reaction is negligible compared with that of the much more active Ni ensembles, while the Fe-Ni ensembles become dominant at lower x because the population of active Ni ensembles with N ≥ 7 become vanishing. This hypothesis is supported by the decrease in Ea with decreasing x (Table 1), which reflects the contribution of Fe atoms to the catalytic reaction because Ea for Fe is smaller than that for Ni, as discussed above.
The origin of the apparent Ea may be understood in terms of the reaction orders, which indicate that the rate determining step is the dissociative adsorption of hydrogen (Fig. 2(a)) or the addition of hydrogen to the intermediate of adsorbed propyne (Fig. 2(c)).22–25) Even if the latter step is rate determining, the apparent Ea probably depends on the H2 dissociation, because enhancement of the dissociation provides more H atoms, which increases the probability of reaction between the H atom and the intermediate. Experiments and calculations clearly indicate that the dissociative adsorption energy of H2 is negatively larger for Fe than for Ni, which predicts a smaller Ea for the adsorption on Fe than on Ni through the Brønsted-Evans-Polanyi relationship.33,34) Thus, the decrease in Ea with a decrease in x is owing to the increase in the contribution of Fe-Ni ensembles.
In addition, the dependence of the product selectivities on composition (Fig. 1) also supports the above hypothesis. Since the readsorption of C3H6 is impossible at low C3H4 conversion as discussed above, C3H8 can be produced only by consecutive reactions on the surface from C3H4 (Figs. 2(c) and (f)). Thus, the negatively larger adsorption energy of C3H6 is considered to bring about a higher C3H8 selectivity. In general, adsorption energies of hydrocarbons are negatively larger on Fe than on Ni,35) which predicts a higher C3H8 selectivity at low conversion for the Fe-Ni alloys than for pure Ni. Such a tendency is actually observed; for example, the C3H8 selectivities at 40℃ are 19.5% for Ni (Fig. 1(a)) and 38.0% for Fe0.45Ni0.55, which indicates that the reaction occurs on not only the Ni ensembles but also on the Fe-Ni ensembles. The moderate decrease in C3H8 selectivity with temperature for all the samples in Fig. 1 is probably due to an increase in the desorption of C3H6.
As discussed above, the catalytic properties appear to change in accordance with the bulk composition (x). This indicates that the surface composition is in accordance with the bulk composition without significant segregation and heterogeneity owing to the reduction behavior of the hydroxides. It is also important to indicate that catalytic reaction is a very sensitive probe for surface state of alloy.
The catalytic properties of composition-controlled Fe1−xNix alloy nanoparticles have been studied with propyne hydrogenation. The reaction rate decreased with a decrease in x, which was attributed to the dilution of active Ni ensembles by the inactive Fe atoms. This scenario was available to interpret the catalytic properties of the alloy particles with compositions of 0.55 ≤ x ≤ 1, whereas alloy particles with x ≤ 0.46 deviated from the scenario. This is considered that active Ni ensembles dominated the activity at larger x while less active Fe-Ni ensembles mainly contributed to the activity at smaller x. This hypothesis was supported by the change in apparent activation energy with x, which was revealed to depend on the dissociative adsorption of H2 from the reaction orders. The present study indicated that the catalytic properties were well dominated by the bulk composition in the Fe-Ni nanoparticles. Thus, this synthesis process of alloy nanoparticles is promising for the development of new catalysts. We believe that these Fe-Ni alloys can be used for appropriate catalytic reactions such as the Fischer-Tropsch synthesis,36,37) and that the method can be applied to the synthesis of other base metal alloy catalysts.
The work was supported by IMRAM Research Project in Tohoku University (FY2015) and JSPS KAKENHI Grant Number 25709055. We also would like to thank the Frontier Research Institute for Interdisciplinary Sciences of Tohoku University for their financial support.
