ZnO is a CO(2)-selective steam reforming catalyst.

ZnO was tested as possible methanol and - since formaldehyde is one of the key intermediates in methanol conversion reactions - also as formaldehyde steam reforming catalyst. Catalytic experiments in a batch as well as a flow reactor resulted in highly selective steam reforming, though at low specific activities, of formaldehyde and methanol over ZnO toward CO(2) (selectivity of 95-99.6%). Comparison of the behavior of ZnPd near-surface intermetallic phases, unsupported intermetallic ZnPd and supported ZnPd/ZnO catalysts reveals that formaldehyde is formed from methanol in parallel with CO(2) on the former, while on unsupported intermetallic ZnPd and ZnO-supported ZnPd, it is efficiently reacted toward CO(2), thus, a beneficial role of ZnO in oxidizing formaldehyde-derived intermediates toward CO(2) is evident.

Introduction

CO2-selective methanol steam reforming (MSR) has been extensively studied in the past decades with special focus on Cu/ZnO [1] and a novel class of oxide-supported Pd catalysts [2-5]. The activity and selectivity of the latter was assigned to the intermetallic compounds (IMCs) ZnPd, GaPd2, and InPd, by investigations into either bulk materials or supported systems based on the respective oxides ZnO, Ga2O3 and In2O3 [2-4,6]. Usually, formation of the intermetallic compound was obtained by reduction in the catalysts in hydrogen at temperatures between 523 and 773 K [2-4]. As increasing knowledge is gathered on the methanol steam reforming capability of the ZnO-supported intermetallic compound ZnPd, results suggest a crucial synergism between the intermetallic compound and the oxide support regarding water activation for efficient CO2 formation [3,4,6,7]. Thus, the catalytic properties of the intermetallic compound are only part of the story, and a more complex picture including the catalytic activity and selectivity of the corresponding support must be assumed to be prevalent [5,8]. This is especially important when water activation, one of the most crucial steps in CO2-selective methanol steam reforming, is discussed. The importance of a bifunctional synergism has been highlighted, especially for the ZnPd/ZnO catalysts as well as for the unsupported intermetallic compound ZnPd, where the presence of oxidized Zn species has been linked to a high CO2-selectivity [5,6]. A first step toward a deeper understanding of the active phases in MSR catalysts is to elucidate the catalytic performance of the single constituents of the catalysts, which also includes the supporting oxides in conventional catalysts. A fair amount of data already exists for Ga2O3 and In2O3 [9-12], wherein the latter has been identified as a highly CO2-selective MSR catalyst [11,12]. To our knowledge, MSR experiments on pure ZnO, namely ZnO nanorods, have been mentioned only once, without stating of selectivity [13]. Furthermore, a beneficial role in methanol conversion and synthesis has been suspected for ZnO, and quite some evidence is presented that its addition improves the CO2-selectivity of Cu [14] and Pd-containing catalysts [5,6,15]. Recently, quantum chemical calculations suggested a possible role of ZnO in methanol steam reforming [7]. Nevertheless, the methanol steam reforming performance and ability of single-phase ZnO have been addressed only partially yet [14] and has been even explicitly denied in some studies [2,16]. We therefore present an unambiguous clarification of the behavior of ZnO in both methanol and formaldehyde steam reforming (FSR). The latter is of special importance, since formaldehyde not only represents a key intermediate in all methanol conversion reactions, but also provides a crucial link to a thorough understanding of the more complex catalytic systems based on intermetallic compounds.

Experimental methods

For all measurements, ZnO powder (Alfa Aesar, 99.99%) was used. The surface area of the powder was determined by N2-BET measurements as approximately 14 m2/g. Catalytic batch measurements in methanol and formaldehyde steam reforming as well as catalyst activation treatments were performed in a dedicated NI Labview-automatized recirculating batch reactor of about 8 ml volume. The system allows for automated pretreatment cycles (oxidative and reductive) and reaction sequences. Details are given elsewhere [11]. A quadrupole mass spectrometer (Balzers QMG 311) attached to the recirculating batch Duran glass reactor via a capillary leak was used for detection of the reaction educts and products. All MSR reactions were conducted with methanol/water mixtures of a 1:9 volume composition of the liquid phase at room temperature, corresponding to a gas-phase ratio of 1:2 at room temperature. All methanol/water mixtures were degassed by repeated freeze-and-thaw cycles. For each catalytic MSR experiment, about 50 mbar methanol/water was mixed with 7.5 mbar Ar (to be measured at m/z = 40) as internal standard to account for the decrease of the mass spectrometer signal due to the continuous gas withdrawal through the leak. Finally, He was added to yield 1 bar total pressure. Standard mass spectrometer calibration was applied as well as correction for fragmentation [11]. In a typical experiment, the catalyst was exposed to the reaction mixture, and the temperature was ramped with 5 K/min to the final value. For data evaluation, the relative intensities of the mass spectrometer signals were converted into partial pressures via external calibration using gas mixtures of defined partial pressures. To simulate formaldehyde steam reforming conditions, a solution of 30% formaldehyde in water was heated to 350 K to result in a defined and constant 50 mbar formaldehyde/water equilibrium mixture in the gas phase after a certain equilibration period (composition approximately 1:1). Catalytic flow experiments in methanol steam reforming were carried out in a flow reactor system (Microactivity Reference, PID Eng&Tech). 6 g of ZnO powder was placed inside a silica-coated stainless steel tube (inner diameter 7.9 mm), which was mounted inside a heated box to prevent condensation of liquids. The reactive feed consisted of 0.0325 mL/min liquid (50 mol.% methanol (Sigma–Aldrich, ⩾99.9%), 50 mol.% deionized water), which is evaporated before being mixed with 13.2 mL/min N2 and 1.6 mL/min He (both Praxair, 99.999%). N2 was used as carrier gas, while He was applied as inert tracer gas to calibrate the gas volumes. The gas composition in the product stream was determined by a gas chromatograph (Varian Micro GC CP4900), allowing quantitative CO determination down to 20 ppm. Amounts of unconverted methanol and water as well as the potential product formaldehyde were not determined by GC because they were separated from the product gas by a cooling trap and a subsequent Nafion® membrane before being injected into the GC. The CO2-selectivity is calculated by dividing the concentration of CO2 by the sum of the concentration of all carbon containing products. The CO level is calculated by dividing the CO concentration by the sum of the concentrations of all gases in the product mixture, disregarding the inert gases N2 and He. A typical MSR flow reactor experiment consisted of heating in N2/He until 433 K (5 K/min) before the methanol/water mixture was added. Heating was continued with 2 K/min. Each data point was collected after 2 h onstream at the respective temperature. By varying catalyst amounts, mass transport limitation could be excluded for the flow reactor setup.

Results and discussion

Fig. 1 summarizes a typical MSR batch experiment over ZnO. As shown, CO2 formation starts at rather low temperatures shortly above 473 K. Strong acceleration of CO2 formation occurs only above 600 K. Most important, CO formation is suppressed over the entire reaction temperature region monitored, at least up to 673 K. The hydrogen trace increases accordingly, matching the theoretical product composition of MSR. The total conversion of methanol was determined to about 53%. Flow reactor experiments also proved ZnO to be highly selective to CO2 in MSR (Fig. 2), reaching a selectivity of 99.6% at 573 K (MeOH conversion: 3.6%), while the theoretical MSR product ratio was fulfilled very accurately. The detected CO levels are similar to those reported on the reference system CuO/ZnO/Al2O3 [17] and on unsupported ZnPd [6], being in the range of 1000–2000 ppm. Due to the low activity of the powder at temperatures below 573 K, no reasonable selectivity determination was possible. The observed activity at 613 K corresponds to a methanol conversion of 26%; thus, ZnO is capable of forming CO2 selectively even at relevant conversions. Fig. 3 shows an Arrhenius plot that is based on the values obtained during MSR on ZnO in the flow setup. The slope obtained from the linear regression equation allowed calculating the apparent activation energy for methanol steam reforming as 144 kJ mol−1. Ex situ characterization by XRD of ZnO, ZnPd, and ZnPd/ZnO after the MSR flow experiments did neither show any changes of the present phases nor any additional phases.

Fig. 1

Temperature-programmed methanol steam reforming over single-phase ZnO in a recirculating batch reactor. The sample was pre-oxidized at 673 K in 1 bar O2 before the catalytic measurement. 20 mg ZnO powder was used for the experiments.

Fig. 2

Temperature-programmed methanol steam reforming over single-phase ZnO (no pretreatment) in a flow reactor. Every data point was acquired after 2 h time onstream at the respective temperature in a steady state regime. Error bars retrieved from the applied GC method are smaller than the depicted symbols that represent the respective data points.

Fig. 3

Arrhenius plot of ZnO in methanol steam reforming (MeOH:H2O = 1:1) according to the data shown in Fig. 2. The equation for the linear regression is shown.

Both batch and flow MSR reactor studies unambiguously show that ZnO itself is a highly CO2-selective (95–99.6%), yet not highly active, methanol steam reforming catalyst.

In a further reaction, the formaldehyde steam reforming (FSR) capability of ZnO was tested in the recirculating batch reactor (Fig. 4). In general, the overall reaction temperature profile and selectivity pattern appear to be very similar to the one observed in MSR (cf. Fig. 1), but with the difference of a lower onset temperature of H2/CO2 formation in FSR. CO2 formation starts at roughly 480 K in FSR as compared to ∼540 K in MSR (compare Figs. 1 and 4, cf. Fig. 5). CO formation is again suppressed up to the highest reaction temperature measured. Hydrogen formation is observed according to the theoretical product ratio of formaldehyde steam reforming. The total conversion of formaldehyde was determined to be 60%. Hence, ZnO also selectively catalyzes the reaction of formaldehyde to CO2. Putting these results into perspective with other oxides typically used in combination with Pd-based intermetallic compounds as selective methanol steam reforming catalysts, we stress similarities especially with In2O3, which essentially follows the same selectivity trend [10,11]. The formation of reaction products in methanol steam reforming over oxidic catalysts has shown to be strongly dependent on a delicate balance of acidic and basic surface sites, thereby influencing the pathway of the adsorbed methoxy group either to condensed products (e.g., dimethylether) or formaldehyde/formiate species and subsequently carbon oxides. Generally, basic oxides (as ZnO and In2O3) tend to favor carbon oxide formation due to the efficient H abstraction by neighboring basic oxygen species [18].

Fig. 4

Temperature-programmed formaldehyde steam reforming over single-phase ZnO in a recirculating batch reactor. The sample was pre-oxidized at 673 K in 1 bar O2 before the catalytic measurement. 22 mg ZnO powder was used for the experiments.

Fig. 5

Turnover frequencies for H2 as a function of reaction temperature for both the formaldehyde (open circles) and the methanol steam reforming reaction (closed circles). Both experimental curves have been fitted by an Arrhenius equation to determine the apparent activation energy.

To localize the potential rate-limiting step of MSR, we found it useful to directly compare the activities of ZnO in methanol and formaldehyde steam reforming. Fig. 5 highlights the dependence of the H2-TOF (per Zn–O site and second) of the reaction temperature as measured in the batch reactor. Note that the TOF numbers represent the lowest possible estimate, since the average surface density of Zn–O entities was derived from the volume density of ZnO, that is, basically, the full BET area was used for normalization of the rate. Both experimental curves have been fitted by the Arrhenius equation to extract the respective apparent activation energies. As indicated, an activation energy of ∼130 kJ mol−1 for the methanol steam reforming reaction has been determined, corroborating the results from the flow measurements (144 kJ mol−1) shown in Fig. 3. In contrast, the activation energy for formaldehyde steam reforming was determined to be around 110 kJ mol−1. This provides an additional hint that initial methanol activation may be the rate-determining step on ZnO. The value of 130–144 kJ mol−1 for MSR does approach 160 kJ mol−1 reported for ZnO nanorods by Flytzani-Stephanopoulos et al. [13]. In addition, the lower value of ∼110 kJ mol−1 for FSR corresponds well to the MSR activation barrier reported for Au/ZnO catalysts [19]. This indicates that our activation energy for FSR over ZnO is in the range of Au/ZnO for MSR, suggesting that Au may play a vital role in the first step of the reaction, namely methanol activation/selective dehydrogenation to formaldehyde. The same catalytic function was assigned to Pd–Zn near-surface intermetallic phases as well as unsupported intermetallic bulk ZnPd in our previous studies [5,6], but also in theory work by Neyman et al. [20].

A possible bifunctional synergism of ZnO and ZnPd in the ZnPd/ZnO catalyst is supported by the comparison of the apparent activation energies of the three respective systems in MSR. The EA of ZnO was determined as 130–144 kJ mol−1 from our batch and flow reactor experiments, being significantly larger than the reported value for ZnPd/ZnO (93 kJ mol−1, [21]). The apparent activation energy of the unsupported, ZnO-free, intermetallic compound ZnPd as judged by XPS (Pd = 50.2 at.%) was calculated as EA = 120 kJ mol−1 [6], being also larger than the value for ZnPd/ZnO. ZnO-modified ZnPd yielded much lower activation energies of 69 kJ mol−1.

To correlate the activity of ZnO to literature-reported values of ZnPd and ZnPd/ZnO, the TOF numbers were estimated to be ∼3 × 10−4 s−1 for FSR (500 K) and ∼5 × 10−6 s−1 for MSR (500 K) in the batch reactor (cf. Fig. 5) and ∼6 × 10−5 s−1 for MSR (513 K) in the flow reactor. This is much less than reported both for MSR on ZnPd/ZnO (0.497 s−1 at 500 K [2], 0.39 s−1 at 523 K [22]) as well as on the Pd–Zn near-surface intermetallic phase (0.04 s−1 at 540 K [5]). The comparison of the EA and the TOF for the different materials shows clearly a synergistic interaction of ZnO and ZnPd in MSR.

In summary, the results reveal a surprisingly high CO2-selectivity (95–99.6%) at low specific activity of single-phase ZnO in MSR, which helps gaining an enhanced understanding of the catalytic performance of the scrutinized intermetallic compound ZnPd [6]. Reviewing the catalytic activity and selectivity of the relevant literature-reported highly CO2-selective ZnPd/ZnO system [2,7,15,16,22,23], we obtain the following picture, which was earlier suggested both experimentally [5] and theoretically [20], for the single constituents and their contribution to the overall catalytic activity and selectivity: On ZnO, formaldehyde is steam-reformed with high selectivity into CO2 (∼95%), but at a much higher rate/lower activation barrier than methanol. Furthermore, ZnO is an essential part in the ZnPd/ZnO catalyst, being also responsible for improved CO2-activity in MSR, most presumably by forming a highly active interface with ZnPd. In this respect, it is also worth noting that over pure ZnO, CO2 formation is also observed upon reaction of pure methanol. This indicates a certain reducibility of ZnO in methanol (directly at the surface or at the phase boundary ZnO–Pd) and the involvement of lattice oxygen of ZnO [23,24]. Hence, in supported systems, the potential role of ZnO primarily maybe the further oxidation of formaldehyde-derived species potentially delivered by the intermetallic compound ZnPd by diffusion to/across the bimetal/oxide phase boundary. This is corroborated by the fact that also over the supported ZnPd/ZnO system formaldehyde is rapidly reacted toward CO2 [2]. Putting also recent experiments on ZnPd near-surface intermetallic phases (NSIP) [5] and unsupported intermetallic bulk ZnPd [6] into perspective, we note that the low-temperature CO2-activity and selectivity crucially depends on the presence of oxidized Zn species in contact to ZnPd during methanol steam reforming, consequently resulting in an increase in both activity and selectivity due to a bifunctional improvement of methanol activation by the intermetallic compound and by enhanced CO2-selective conversion of formaldehyde-containing species over the ZnO support or under participation of the oxidized Zn species.