1 Introduction

The three-way catalysis (TWC) system is the current exhaust purification technology used for gasoline-fueled automobiles, which requires platinum group metals (PGMs) such as Pt, Pd and Rh as active catalyst components [1–5]. As automobile usage in emerging countries is rapidly increasing, automotive emission regulations are more severely restricted. As such, the demand for PGMs is increasing and thus the development of catalysts using lower amounts of PGMs is required. The catalyst design strategy for this purpose is based on various metal oxide support interactions [6–17]. Thus, further development of novel support materials that can stabilise active PGM elements is challenging.

Among PGM elements, palladium (Pd) has high catalytic activities especially for oxidation reactions in TWC. The Pd catalyst deactivation during thermal ageing is caused by the sintering of metallic Pd, which is especially pronounced under fuel-rich conditions compared to lean environment [18, 19]. Although γ-Al2O3 has been widely used as support for Pd catalysts, we are interested in the use of aluminium oxide borate (10Al2O3 · 2B2O3, 10A2B). The first catalytic application of 10A2B was reported by Garbowski et al., who studied characteristics as a support for high-temperature methane combustion [20]. More recently, we reported that 10A2B is a potential support material suitable for Pd, having a higher specific surface area and thermal stability than conventional γ-Al2O3 [21, 22]. The most beneficial feature of 10A2B is its stability under high-temperature water vapour, which is an important property for TWC support materials. Another interesting feature is its cation-substitution capability; among various transition metal cations, Fe3+ is most easily accommodated in the 10A2B structure. This structural modification significantly enhanced the NO reduction efficiency under fuel-rich conditions with an air-to-fuel ratio (A/F) of <14.6 [22]. Moreover, the redox reaction between Fe3+ and Fe2+ can contribute to the oxygen storage capacity (OSC) of the catalyst [22]. According to the fundamental results of 10A2B powders, we have extended our research to honeycomb-shaped catalysts for practical applications in the TWC system. In this study, we have prepared two types of monolithic honeycomb catalysts using 10A2B and Fe-10A2B as Pd supports. The TWC performance of Rh/Pd bilayer-coated honeycomb catalysts in simulated exhaust comprising NO–CO–C3H6–O2–CO2–H2O gas mixtures was evaluated as a function of the Rh:Pd loading ratio. Furthermore, the Pd/Fe-10A2B monolayer catalyst was studied under rich conditions (A/F < 14.6) to induce higher NO conversion compared to a reference catalyst (Pd/CeO2-ZrO2).

2 Experimental

2.1 Catalyst Preparation

The powders of 10A2B and Fe-10A2B were prepared via a wet process. H3BO3 (Wako Pure Chemicals) was dissolved in water and boehmite (Disperal, Sasol) was added with a molar ratio of Al:B = 20:4.8 under vigorous stirring. A 20 % excess of H3BO3 was used as it is volatilised during subsequent drying and calcination processes. The as-obtained slurry was dried at 120 °C for 12 h, ground and heated at 300 °C for 1 h. Finally, the as-obtained powder was calcined at 1000 °C for 5 h in air to complete solid-state reactions to a 10A2B phase. γ-Al2O3 was prepared from boehmite as a reference by drying and calcination at 600 °C. Lanthanum (3 mol%) was impregnated in 10A2B and γ-Al2O3 using an aqueous solution of La(NO3)3 (Wako Pure Chemicals), followed by drying and calcining at 600 °C for 3 h in air. The La addition is known to stabilise γ-Al2O3 against phase transformation to α-Al2O3 [23–25]. Fe-10A2B was prepared in a manner similar to 10A2B by using H3BO3, Fe(NO3)3 (Wako Pure Chemicals) and boehmite. Solid solutions of CeO2-ZrO2 of 1:9 and 4:6 M ratios were prepared as reference supports via co-precipitation method. To aqueous solutions containing Ce(NO3)3 and ZrO(NO3)2 (Wako Pure Chemicals), a 12 % NH3 solution was added under vigorous stirring until the pH reached 10. The precipitates were washed with water, dried and calcined at 700 °C for 3 h to form single-phase solid solutions of tetragonal fluorite-type structure (see Fig. S1 in Electronic supplementary material (ESM)). As-prepared compounds with CeO2:ZrO2 M ratios of 1:9 and 4:6 were used for the top and bottom layers, respectively, of bilayer-coated honeycomb catalysts.

As shown in Fig. 1, three types of honeycomb-coated catalysts were prepared in the present study. For preparing honeycomb catalysts, various slurries were prepared by ball-milling the catalyst powders, inorganic binders and water. Monolayer-coated honeycombs (type A) were prepared by dipping a cordierite honeycomb (25.4 mmφ × 30 mm, 600 cells in−2, NGK Insulator) into the slurry containing either 10A2B or γ-Al2O3, which was followed by impregnation with an aqueous solution of Pd(NO3)2 (Tanaka Kikinzoku), drying and calcining at 600 °C for 3 h in air. The Pd loading of the as-calcined honeycomb catalysts corresponded to 1.3 g L−1. Bilayer-coated honeycombs (type B) were prepared by a stepwise dipping process. First, the cordierite honeycomb was dipped into the slurry containing either 10A2B or γ-Al2O3 and CeO2-ZrO2 (CeO2:ZrO2 = 4:6) and was subsequently impregnated with an aqueous solution of Pd(NO3)2. The mass ratio of 10A2B or γ-Al2O3 and CeO2-ZrO2 was 1:1. After calcining at 600 °C for 3 h in air, a second dipping was conducted using a slurry containing CeO2-ZrO2 (CeO2:ZrO2 = 1:9) and γ-Al2O3. The mass ratio of CeO2-ZrO2 and γ-Al2O3 was 3:1. This was followed by impregnation with an aqueous solution of Rh(NO3)3 (Tanaka Kikinzoku), drying and calcining at 600 °C for 3 h in air. As-calcined bilayer-coated honeycomb catalysts with a top Rh layer and a bottom Pd layer contained 1.0 g L−1 of Rh + Pd with different weight ratios of Pd:Rh (3:1, 5:1, 10:1 and 19:1). Different monolayer-coated honeycombs (type C) were prepared in a similar manner using the same cordierite honeycombs and a slurry containing either 10A2B, γ-Al2O3, CeO2-ZrO2 (CeO2:ZrO2 = 4:6) or Fe-10A2B. After subsequent impregnation with an aqueous solution of Pd(NO3)2 and drying, they were calcined at 600 °C for 3 h in air. The Pd loading of as-calcined honeycomb catalysts corresponded to 0.7 g L−1.

Fig. 1
figure 1

Schematic illustrations of three types of honeycomb catalysts. CZ means CeO2–ZrO2

Full size image

2.2 Catalyst Characterisation

X-ray diffraction (XRD) patterns of 10A2B and Fe-10A2B powders were obtained using a Rigaku RINT-TTR III diffractometer with Cu Kα radiation (50 kV, 30 mA). The metal loading was analysed by the ICP analysis, which agreed with the calculated values based on the synthesis procedure described above within an error of 10 %. Brunauer–Emmett–Teller (BET) surface areas (S BET) were calculated using N2 adsorption isotherms obtained at −196 °C (Belsorp-mini, Bel Japan). The Pd metal dispersion was evaluated using pulsed CO chemisorption at 50 °C (Belcat, Bel Japan) after sample reduction with H2 at 400 °C. The metal dispersion is expressed as the molar ratio of chemisorbed CO per loaded Pd (CO/Pd). The microstructure of the materials was observed using a field emission scanning electron microscope (XL30, FEI) and a scanning electron microscope (SEM : Miniscope TM3000, Hitachi). The spatial distribution of Rh and Pd in bilayer-coated honeycombs was analysed by X-ray line analysis technique (Shimadzu EPMA-1720H).

2.3 Catalyst Performance Testing

Prior to the catalytic tests, all as-prepared honeycomb catalysts were thermally aged at 1000 °C for 25 h by continuously switching between two gas feeds, i.e. stoichiometric (50 s) and lean conditions (50 s). The stoichiometric gas composition (A/F = 14.6) was C3H6 (1670 ppm), O2 (0.75 %) and N2 (balance), while the lean gas was composed of air with 10 % H2O. Catalytic activity tests on the honeycomb catalysts were performed in a flow reactor at atmospheric pressure. Catalytic light-off tests of the stoichiometric simulated gas stream of CO (0.73 %), C3H6 (400 ppm), NO (500 ppm), O2 (0.50 %), CO2 (14 %), H2O (10 %) and N2 (balance) over the honeycomb catalysts (types A and B) were performed at a heating rate of 20 °C min−1 and space velocity (SV) of 98,000 h−1. The gas composition corresponded to the stoichiometric A/F of 14.6. The concentrations of CO, C3H6 and NO in the effluent gas were analysed online using a motor exhaust gas analysis system (MEXA 9100, Horiba) equipped with non-dispersive infrared detectors and a flame ionisation detector.

The steady-state catalyst performance for honeycomb catalysts (type C) was evaluated at 400 °C using gas mixtures containing CO, C3H6, H2, NO, O2, CO2, H2O and N2 balance with various A/F values (14.2–14.8) at SV = 98,000 h−1 (See Table S1 in ESM). The concentrations of CO, C3H6 and NO in the effluent gas were analysed online using an exhaust gas analyser (MEXA 9100). Catalytic light-off tests on the honeycomb catalysts (type C) were performed in three different modes (I, II and III) under rich conditions (A/F = 14.2). Mode I used gas mixtures containing NO, CO, C3H6, H2, O2, H2O and N2 balance, whereas modes II and III excluded C3H6 and C3H6/H2, respectively (See Table S2 in ESM). The effluent gas composition was analysed online using an exhaust gas analyser (Best Sokki, Sesam3-N and Bex-5200C) equipped with an FT-IR spectrometer and a flame ionisation detector.

3 Results and Discussion

3.1 Type A Monolayer Catalysts

Figure 2 shows SEM images of γ-Al2O3 and 10A2B after calcination at 1000 °C. Contrary to the amorphous texture of γ-Al2O3, 10A2B was composed of larger particles (∼50–150 nm) with an elliptical shape. The S BET value of 10A2B (70 m2 g−1) was less than that of γ-Al2O3 (81 m2 g−1), whereas the Pd metal dispersions for their Pd-supported catalysts were nearly identical (CO/Pd values close to 50 %). Table 1 shows the influence of thermal ageing under dry and humidified atmospheres on the S BET value of each support material. The S BET value of 10A2B decreased with increasing temperature and was less than that of γ-Al2O3 when calcined under dry air. However, the result was different when the calcined sample was further heated under a stream of 10 % H2O/air; 10A2B had a greater S BET at 1200 °C, while γ-Al2O3 showed more steep sintering because the phase transformation to α-Al2O3 was significantly accelerated (see Fig. S2 in ESM). Similar effects of water vapour on the phase transformation of transition alumina are reported in the literature [26–28]. By contrast, water vapour had no detectable influence on the structure and the crystallinity of 10A2B (see Fig. S3 in ESM). Thus, stability under hydrothermal conditions is a beneficial feature of 10A2B as a support material for automotive catalysts.

Fig. 2
figure 2

SEM images of γ-Al2O3 (top) and 10A2B (bottom) after calcination at 1000 °C

Full size image
Table 1 BET surface area (S BET/m2 g−1) of support materials after thermal ageing
Full size table

The effect of the support material on TWC performance was studied using monolayer-coated honeycomb catalysts (type A). Figure 3 shows the catalytic light-off curves of the stoichiometric simulated gas mixture (A/F = 14.6) for honeycomb catalysts containing Pd/10A2B or Pd/γ-Al2O3. Both catalysts displayed very similar light-off curves for CO, C3H6 and NO. To compare their differences, Table 2 lists the light-off temperatures (T 50), which was defined as the temperature at which the conversion of each gas species reaches 50 %, as well as the apparent conversion at 400 °C (η 400), which was used to evaluate the performance after the light-off temperature. The T 50 values for Pd/10A2B were 6–8 °C higher than that of Pd/γ-Al2O3 for each gas species. Nevertheless, the former achieved 3–4 % higher η400 values than the latter, with the values for CO and C3H6 being especially close to complete removal. One may point out that the differences of η 400 values are too small to correlate with the catalytic activity. However, the difference in the η 400 values between Pd/10A2B and Pd/Al2O3 is reproducible and is in accordance with the activity trend of bilayer-coated honeycomb catalysts as described in the following section. Because the η400 value is regarded as an activity measure after the light-off, which is a very important feature considering the stringent automotive emission regulations.

Fig. 3
figure 3

Catalytic light-off curves of CO, C3H6 and NO over monolayer honeycomb catalysts (type A) under stoichiometric conditions (A/F = 14.6). Temperature ramp 20 °C min−1

Full size image
Table 2 Catalytic activities of Pd/10A2B and Pd/γ-Al2O3
Full size table

3.2 Type B Bilayer Catalysts

Figure 4 shows a SEM image of the fracture surface of an as-prepared bilayer honeycomb catalyst with a top Rh catalyst layer and a bottom Pd catalyst layer. The top layer is approximately 40-μm thick at the corner of the square cells and contains Rh loaded on CeO2-ZrO2 and γ-Al2O3, whereas the bottom layer is approximately 90-μm thick at the corner of the square cells and contains Pd loaded on 10A2B and CeO2-ZrO2. The spatial distribution of Rh and Pd in each layer was demonstrated by SEM/EDX analysis (Fig. S4 in ESM). These layers were in close contact with the surface of the cordierite honeycomb substrate. Similar microstructure was observed for another bilayer honeycomb catalyst composed of a top Rh/CeO2-ZrO2/Al2O3 layer (CeO2:ZrO2 = 1:9) and a bottom Pd/Al2O3/CeO2-ZrO2 layer (CeO2:ZrO2 = 4:6). Therefore, the only difference between these two bilayer honeycomb catalysts is their support materials (10A2B or Al2O3) in the bottom layer as shown in Fig. 1.

Fig. 4
figure 4

SEM image of the fracture surface of the bilayer-coated honeycomb catalyst (type B) with a Pd/10A2B + CeO2-ZrO2 bottom layer and Rh/CeO2 − ZrO2 + γ-Al2O3 top layer

Full size image

These two types of bilayer honeycomb catalysts (type B) with different Rh:Pd ratios were used to simulate a TWC reaction. The total loading of Rh + Pd (1.0 g L−1) was maintained unchanged. Figure 5 shows the catalytic light-off curves for the stoichiometric simulated gas mixture (A/F = 14.6). The light-off temperature (T 50) was found to be dependent on the support materials for the bottom Pd layer; catalysts containing Pd/10A2B were inferior to Pd/γ-Al2O3. Conversely, catalysts containing Pd/10A2B demonstrated higher conversion efficiencies after the light-off (η 400). These trends are in accordance with those of the aforementioned monolayer honeycomb catalysts (type A, Fig. 3). To elucidate the effect of Rh:Pd ratios on the catalytic performance, these two activity measurements (T 50 and η 400) were plotted as a function of the Rh:Pd ratio, as shown in Fig. 6. To confirm the reproducibility of the plots in Fig. 6, the light-off reaction test was performed eight times using several batches of honeycombs prepared using the same procedure. The observed relative errors for T 50 values in Fig. 6a were 2.1 %(CO), 1.2 %(HC) and 2.9 %(NOx), and those for η 400 values in Fig. 6b were 0.4 %(CO), 0.5 %(HC) and 4.1 %(NOx). At the highest Rh content (Pd:Rh = 3:1, Rh = 0.25 g L−1), the catalyst with 10A2B showed higher T 50 values than that with γ-Al2O3. However, the difference in T 50 was not obvious when the Rh content was less than 0.2 g L−1. As the Rh loading was decreased, η 400 continuously decreased for the honeycomb comprising a Pd/γ-Al2O3 bottom layer. In contrast, the η 400 value was maintained almost unchanged (>96 % conversion) when Pd/10A2B was used rather than Pd/γ-Al2O3 as the bottom catalyst layer. It was suggested that the Pd catalyst on the bottom layer showed higher catalytic performance after the light-off temperatures (≥400 °C) regardless of the amount of Rh. This is consistent with the results shown in Fig. 3 because when the Pd-only catalyst (type A) was loaded on 10A2B, higher conversions were achieved after the light-off temperature compared with Pd/γ-Al2O3. Although Rh is an indispensable active component of the TWC system, particularly for efficient conversion of NO to N2 [2, 29, 30], it is one of the rarest platinum group metals for this purpose. The present results on bilayer honeycomb catalysts demonstrated that the Rh content can be decreased without destroying the catalytic performance by using Pd/10A2B.

Fig. 5
figure 5

Catalytic conversion of CO, C3H6 and NO over bilayer-coated honeycomb catalysts (type B) under stoichiometric conditions (A/F = 14.6, 20 °C min−1)

Full size image
Fig. 6
figure 6

a Light-off temperature (T 50). b Conversion at 400 °C (η 400) as a function of Rh loading for Rh/Pd bilayer-coated honeycomb catalysts (type B) containing Pd/10A2B or Pd/γ-Al2O3 bottom layers. Total metal loading (Rh + Pd) is constant at 1.0 g L−1

Full size image

3.3 Type C Monolayer Catalysts

The effect of Fe-substitution in 10A2B on the catalytic performance was studied using monolayer-coated honeycombs (type C). Substitution of Fe for the Al site in 10A2B was demonstrated using XRD (See Fig. S5 in ESM). Although Fe-substitution did not affect the particle morphology, the S BET value of Fe-10A2B (44 m2 g−1) was smaller than that of neat 10A2B (70 m2 g−1). Figure 7 shows the A/F dependence on conversion of CO, C3H6 and NO at 400 °C. In the lean region (A/F > 14.6), CO and C3H6 were completely converted to CO2/H2O with all catalysts, while NO conversion was strongly inhibited because of excess O2 in the gas phase. In the rich region (A/F < 14.6), Pd/Fe-10A2B showed higher conversion of C3H6 and NO compared to Pd/γ-Al2O3 and Pd/10A2B. According to our previous study on powdered catalysts [22], Fe-substitution improves the catalytic NO conversion in the rich region (A/F < 14.6) because of enhanced activity for NO–C3H6, CO–H2O and NO–H2 reactions. Since the higher conversion of C3H6 and NO was also observed for Pd/CeO2-ZrO2, the following studies were performed using Pd/CeO2-ZrO2 as a reference catalyst. Our previous study also demonstrated that the redox reaction between Fe3+ and Fe2+ in Fe-10A2B generates an OSC that causes an efficient buffering effect on dynamic A/F fluctuation, which is a well-known function of CeO2-ZrO2 [31–35]. However, its contribution to the present catalytic performance under a static rich condition should be negligible.

Fig. 7
figure 7

Catalytic conversion of CO, C3H6 and NO at 400 °C over monolayer honeycomb catalysts (type C) in the range of 14.2 ≤ A/F ≤ 14.8

Full size image

Figure 8 shows the catalytic light-off curves for CO, C3H6 and NO in the CO–C3H6–H2–NO–O2–H2O reaction (mode I, A/F = 14.2, see Table S2 in EMS for detailed gas compositions) over monolayer honeycomb catalysts (type C) comprising Pd/CeO2-ZrO2 and Pd/Fe-10A2B. Evidently, higher conversion of NO occurred with Pd/Fe-10A2B compared to Pd/CeO2-ZrO2, although their conversions of C3H6 and CO were comparable. On NO reduction, N2O was negligible while nearly stoichiometric NH3 was yielded over both catalysts. Note that NO conversion to NH3 began at ∼200 °C and conversion of CO and C3H6 were negligible at this temperature, suggesting the occurrence of an NO–H2 reaction. However, the NO conversion levelled off and then decreased at ≥300 °C. The irregular shape of NO conversion is possibly associated with the increased rate of the H2–O2 reaction at this temperature. However, further increase in temperature led to an increase in the NO–H2 reaction because CO–H2O and C3H6–H2O reactions began to produce H2.

Fig. 8
figure 8

Catalytic light-off curves of CO–C3H6–H2–NO–O2–H2O–N2 and NH3/N2O formation over monolayer honeycomb catalysts (type C, mode I) at A/F = 14. Solid lines Pd/Fe-10A2B. Dotted lines Pd/CeO2-ZrO2. Temperature ramp 20 °C min−1

Full size image

To demonstrate this hypothesis, catalytic light-off tests of CO and NO in the CO–NO–H2–O2–H2O reaction (mode II) were performed at A/F = 14.2 (Fig. 9). Upon removal of C3H6, NO reduction to NH3 was further accelerated at lower temperatures (<200 °C) compared with Fig. 8. In addition, the irregular shape of both NO conversion and NH3 concentration were no longer observed because of the higher H2/O2 ratio in the gas feed. Moreover, the suppression of CO–O2 reaction is also beneficial for H2 production via the water-gas shift reaction. Even under these conditions, Pd/Fe-10A2B was superior to Pd/CeO2-ZrO2 for the NO–H2 reaction. Figure 10 shows the results of the CO–NO–O2–H2O reaction (mode III). Because H2 was absent in this case, a higher light-off temperature was required for NO (≥240 °C) compared to Figs. 8 and 9. Thus, NO conversion was driven by the CO–H2O reaction followed by the NO–H2 reaction. In addition, the NO–CO reaction may contribute to this conversion, as can be assumed from the higher concentrations of N2O compared to Figs. 8 and 9. Importantly, Pd/Fe-10A2B exhibited higher NO conversion than Pd/CeO2-ZrO2 over the entire temperature range. Under rich conditions, Pd/Fe-10A2B can contribute to NO conversion, which is superior to Pd/γ-Al2O3 and Pd/10A2B, because of its higher catalytic activity for the NO–H2 reaction to NH3.

Fig. 9
figure 9

Catalytic light-off curves of CO–NO–H2–O2–H2O–N2 and NH3/N2O formation over monolayer honeycomb catalysts (type C, mode II) at A/F = 14.2. Solid lines Pd/Fe-10A2B. Dotted lines Pd/CeO2-ZrO2. Temperature ramp: 20 °C min−1

Full size image
Fig. 10
figure 10

Catalytic light-off curves of CO–NO–O2–H2O–N2 and NH3/N2O formation over monolayer honeycomb catalysts (type C, mode III) at A/F = 14.2. Solid lines Pd/Fe-10A2B. Dotted lines Pd/CeO2-ZrO2. Temperature ramp: 20 °C min−1

Full size image

4 Conclusions

Three different honeycomb-coated catalysts containing 10A2B and Fe-10A2B as supports for Pd were prepared and their catalytic performances under simulated TWC conditions were compared to those of conventional Pd/γ-Al2O3 and/or Pd/CeO2-ZrO2 catalysts. The monolayer-coated Pd/10A2B catalyst (type A) achieved higher conversion of CO, C3H6 and NO after the light-off temperature (400 °C), compared to conventional Pd/γ-Al2O3. The Rh/Pd bilayer-coated honeycomb catalysts (type B) using Pd/10A2B were also able to preserve higher levels of CO, HC and NO conversion at 400 °C compared with Pd/γ-Al2O3 when Rh loading in the top layer was decreased. These results imply that Pd/10A2B is a promising catalyst support for the development of automotive catalysts with decreased Rh contents. The monolayer-coated catalyst containing Pd/Fe-10A2B demonstrated characteristic features under a fuel-rich atmosphere, in which CO–H2O and subsequent NO–H2 reactions significantly accelerated NO conversion.