Preparation of catalysts involves mainly two steps as described below:

Step 1: Commercial TiO₂ (anatase, Powder, Kronos, BET-SA—315 m²/g) was impregnated with aq. solution of SbCl₃ (Alfa Produkte, Karlsruhe, Germany) by keeping 8 wt% Sb with respect to the total amount of the catalyst and kept it aside for 1 hour followed by precipitation with (NH₄)₂SO₄ and kept at 70˚C for 1 hour on a water bath. After cooling to room temperature the solution was neutralized with 25% ammonia (adjusted to pH of 7) and heated on the water bath for another hour. Afterwards the slurry was filtered and dried on a rota-vapor to remove excess water; the resulting solid mass was further dried in an oven at 120˚C for 16 h, followed by calcination at 400˚C for 3 h in air (50 ml/min).

Step 2: The above mentioned Sb impregnated TiO₂ sample was again impregnated with desired amount of acidified aqueous solution of PdCl₂ (99.8%, Alfa Produkte, Karlsruhe, Germany) to get the desired amount of Pd (10 wt%) the excess solvent was removed by rota vapor, and the sample was dried in an oven at 120˚C for 16 h. In a similar fashion, all of the catalysts were prepared using four different oxide supports such as ZrO₂, SiO₂, and γ-Al₂O₃. More details on catalyst preparation have been reported elsewhere [36]. The denotation and composition of different samples prepared are presented in Table 1. The composition of the materials was checked by inductively coupled plasma-optical emission spectrometry (ICP-OES), and the values were found to be close to the theoretical ones.

Methods: TPR/H₂ up to 250˚C (Low Temperature Reduction) to prepare Pd particles.

TPR/H₂ up to 350˚C (High Temperature Reduction) to check a SMSI effect by antimony.

CO pulse chemisorption at r.t. after TPR and subsequent elution with a He flows at final temperature.

The BET surface areas and pore size distribution of the catalysts were determined using a Micromeritics Gemini III 2375 instrument by N₂ physisorption at −196˚C. Before the measurements, the known amount of catalyst was evacuated for 2 h at 150˚C to remove physically adsorbed water.

Samples for transmission electron microscopy (TEM) were prepared by depositing the sample on a copper grid (300 mesh) coated with lacy carbon film. TEM analysis was performed using a CM-20 microscope (twin) (Philips, The Netherlands) at 200 kV with EDAX PV9900.

The catalysts were characterized using (AMI-1 (Zeton/Altamira)), TCD.

Catalyst pretreatment: calcination in air at 300˚C for 2 h, catalyst sample weight 0.12 to 0.14 g.

5.1% H₂/Ar, flow rate 50 ml/min, 10 K/min, calibrated by 500 µl Ar-Pulses into 5.1% H₂/Ar.

500 µl pulses of 19.7% CO/He into He flow 50 ml/min.

BET surface areas and pore volumes of the fresh catalysts prepared with four different supports are presented in Table 1. Of the four supports used, the surface area of the TiO₂ (anatase)-supported catalyst decreased drastically (from 315 to 78 m²/g) after impregnation of both Sb (8 wt%) and Pd (10 wt%). However, no such drastic decrease in surface area was seen in the catalysts prepared with other supports. These decreases in surface area and pore volume after the deposition of Sb and Pd is due mainly to penetration of dispersed Pd and Sb species into the pores of the support, as well as to solid-state reactions between the dispersed components and the support. But the extent of decrease in surface areas depends on the nature of the support applied. Table 1 shows that of all of the supports studied in this series, SiO₂-supported Pd catalyst exhibited the highest surface area (193 m²/g) and pore volume (0.542 cm³/g); ZrO₂ had the lowest surface area (47 m²/g) and pore volume (0.107 cm³/g).

Pore size distributions of fresh catalysts are portrayed in Figure 1. These figures clearly show that all of the catalysts exhibited unimodal pore volume distribution. Although all of these catalysts had a unimodal distribution, they had some differences, especially in terms of pore size and distribution (Figure 1). Such differences in pore size are due to differences in the type of oxide supports applied and their varying surface areas. The catalysts prepared with reducible supports (TiO₂, ZrO₂) with low surface areas. showed a narrow distribution in the range of ca. 20 - 50 Å with a dominant pore diameter around 35 Å. On the other hand, the catalysts with high surface areas (SiO₂, γ-Al₂O₃) displayed a considerably broader range of pore size distribution, from ca. 25 - 150 Å, with dominant pore diameters around 85 Å (SiO₂) and 60 Å (γ-Al₂O₃), respectively.

Electron micrographs of the fresh (activated) 10%Pd8%Sb catalysts with different supports are depicted in Figure 2. All of these catalysts showed a narrow particle size distribution of Pd, but with varying sizes, depend on the type of support used. It can be seen from Figure 2(a) that the fresh TiO₂-supported catalyst exhibited relatively larger Pd particles compared with the other supports.

The Pd particles on TiO₂ support were spherical shaped and ranged in size from 1 to 10 nm. Shen et al. [38] reported from their TEM analysis on TiO₂-supported Pd catalysts with different Pd loadings that Pd particles are hemispherical in shape with a narrow size distribution of 4 - 20 nm, which are fixed in the well-developed planes of TiO₂. As a result of such matching, these authors suggested that the support structure could stabilize Pd particles and favour strong metal-support interaction. In the case of fresh SiO₂-supported catalyst (Figure 2(c)), the Pd particles are uniformly distributed on the support with size ranging from 2 to 4 nm and shape again nearly spherical These results are in agreement with those reported by Lyubovsky et al. [39]. ZrO₂- and γ-Al₂O₃-supported catalysts (Figure 2(b) and Figure 2(d)) showed Pd morphology similar to that of TiO₂- and SiO₂-supported ones but with smaller Pd particles. Among the four fresh solids investigated, the ZrO₂-supported catalyst (Figure 2(b)) exhibited very small Pd particles (1.5 - 2 nm) with a narrow size distribution, more or less comparable to the γ-Al₂O₃-supported solid. The decreasing order of Pd particle size in these fresh 10%Pd8%Sb catalysts over the different supports is as follows:

TiO₂ > SiO₂ > ZrO₂ ∼ Al₂O₃

TPR of Sb₂O₃: The TPR plot shown in Figure 3 revealed that the reduction of unsupported Sb₂O₃ only starts at about 480˚C. The stepwise reduction to Sb(0) is not finished at 800˚C and the shoulder at the high temperature flank is not a real peak but it is due to the end of the temperature ramp. Presumably, the reduction Sb₂O₃ to Sb occurs stepwise, e.g. Sb₆O₁₃ of could be formed as intermediate.

TPR of Sb₂O₃/SiO₂: the reduction of supported Sb₂O₃ did not start at lower temperature but rather at slightly higher temperature (ca. 500˚C) as can be seen in Figure 4.

High temperature reduction (HTR): Figure 5 shows the plot obtained with Pd/Sb₂O₃/Al₂O₃. The first negative peak is due to H₂ desorption by decomposition β-Pd hydride. That means that a portion of Pd(II) is already reduced to metallic Pd at room temperature (r.t.) during the isothermal segment of the TPR procedure which is not monitored. It is well known that metallic Pd absorbs H₂ into its bulk forming β-Pd hydride. The decomposition peak is superimposed by the further reduction of Pd(II). 3 peaks in the temperature range up to ca. 250˚C point to the reduction of different Pd(II) species or Pd(II) species of different size. The additional peak with maximum at the final temperature should be attributed to a reduction of Sb₂O₃ since reduction of supported PdO or PdCl₂ is usually finished at temperatures below 300˚C. Probably, hydrogen activated on Pd reduces Sb(III) at lower temperature than in case of the Pd free sample. The peak maximum at final temperature is an apparent maximum again, because the applied isothermal holds in the TPR procedure.

A similar HTR-plot was obtained with the Pd/Sb₂O₃/ZrO₂ sample (Figure 6). Pd hydride decomposition was very distinctive. Again a reduction of Sb₂O₃ besides Pd(II) was observed.

The β-Pd hydride decomposition is completely superimposed by the H₂ consumption in the case of the Pd/Sb₂O₃/TiO₂ sample (Figure 7).

From the literature relevant for the SMSI effect it can be concluded that ZrO₂ and TiO₂ are still not reduced at 350˚C even in the presence of noble metals.

All three samples showed no CO consumption in the subsequent pulse chemisorption measurement. Obviously, CO chemisorption on Pd is completely suppressed by SbO_x species by a typical SMSI effect as known for other reducible supports such as TiO₂, ZrO₂, CeO₂ and Nb₂O₅ [40].

This conclusion was supported by an experiment to check the reversibility of the SMSI state. After the CO pulse measurement sample Pd/Sb₂O₃/ZrO₂ was oxidised by a TPO treatment up to 400˚C (1 h hold at final temperature). Subsequently, it was reduced by a low temperature reduction (LTR) procedure up to 250˚C only. Now, a significant CO adsorption could be stated.

In summary, the TPR studies revealed a close contact between the subsequently supported catalyst components Sb and Pd, which result in an easier reducibility of Sb₂O₃. To prevent a reduction and/or migration of SbO_x species onto the Pd particles (SMSI see below) we suggested limiting the reduction temperature to 250˚C. This temperature has to be also applied at least for elution of the TPR gas to obtain a hydrogen free Pd surface for a subsequent CO pulse chemisorption measurement.

TPR Plots for all catalysts are shown in Figures 8-10 and Figure 11. A partial reduction of Sb₂O₃ already at 250˚C cannot be excluded because a clear separation between complete Pd(II) reduction and beginning Sb₂O₃ reduction which was not possible. Moreover, the long tailed profile during the isothermal segment points to reduction of a low amount of Sb₂O₃. However, we believe that SbO_x species cannot migrate onto the Pd particles at 250˚C because we obtained reasonable values for CO chemisorption.

At least the different CO chemisorption can be compared or related to properties of the various supports and to the catalytic activity. The results of the CO chemisorption measurements after LTR are summarised in Table 2. Most surprisingly, a very low CO chemisorption was reproducibly obtained with Pd/Sb₂O₃/TiO₂.

However, in case of large Pd particles also bridged bound CO (CO: Pd = 1/2) is present. Consequently dispersity and surface area could be up to twice and particle size up to a half of the values given in Table 2.

*Assumption 10% Pd, CO:Pd = 1 and using 1.27 × 10¹⁵ Pd/cm² (average face density); **Assuming spherical Pd particles.

S_Pd(cm²/g) = CO(µmol/g) × 6 × 10¹⁷ µmol⁻¹/1.27 × 10¹⁵ Pd∙cm⁻²

D(%) = (Pd^s/Pd_t) × 100% = (CO/Pd_t) × 100%

V_Pd = m_pd/ρ_Pd = n × 4πd³, S_Pd = n × πd², V/S = d/6, d = 6 × m_pd/ (ρ_Pd × S)

n: number of Pd-particles; m_pd(mg_Pd/g), ρ_Pd = 12 g/cm³ volumic mass density Currently, the real Pd content (r%Pd) of the samples is determined by ICP-OES. The result has to be considered to correct the values of the dispersity and particle size.

D_corr = D × 10%Pd/r%Pd

d_corr = d × r% Pd/10%Pd

Table 3 demonstrates the reversibility of the blocking of the CO chemisorption after HTR and subsequent TPO/LTR, respectively. This behaviour is written in the literature as typical for a SMSI effect [41]. TPO was performed up to 400˚C, subsequent reduction only up to 250˚C.