A model high surface area alumina-supported palladium catalyst

A catalyst preparative procedure is described that produces a high surface area alumina-supported palladium catalyst that yields an atypical chemisorbed carbon monoxide infrared spectrum.

This inherently residue-free substrate provides a useful reference for evaluation of catalyst crystallite morphology and its effect on reactivity profiles.

Continuing development of structure–activity relationships relevant to applied heterogeneous catalysis1,2 requires benchmark catalysts in which crystallite morphology is well defined and in which residues, resulting from the synthesis, which can complicate reactivity profiles3 and the vibrational spectra of probe molecules,4 are absent.

Alumina-supported palladium catalysts for this purpose are usually prepared by surface science methods5,6 but this limits their ability to be tested in conventional microreactors.

This disadvantage has been overcome by the synthesis of a benchmark alumina supported palladium material, which is an active catalyst for alkyne hydrogenation.

The novel synthetic feature is the use of thermally unstable tetramminepalladium(ii) tetraazidopalladate(ii)7 as the precursor compound rather than conventional precursors such as PdCl28 or Pd(NO3)2,9 which can potentially leave residues on the activated catalyst that can affect catalytic performance.10,11

The catalyst is prepared according to eqns. (1) and (2) via a controlled explosion of the precursor.7

Tertramminepalladium(ii) tetraazidopalladate(ii) was synthesized by reaction of Pd(NH3)4Cl2 with sodium azide in water.7

It was mixed with alumina (Degussa Oxid C, BET surface area 109 m2 g−1) in a Pyrex evacuable reaction vessel to produce a Pd loading of ca. 10% w/w.

The reactor was connected to a gas handling facility equipped with an evacuated reservoir of sufficient volume to contain the complete release of ammonia and nitrogen gas from the thermal decomposition of the tertramminepalladium(ii) tetraazidopalladate(ii).

This material decomposes at 419 K 7 and so the reaction mixture was progressively warmed from room temperature to 473 K to effect the decomposition process.

This resulted in the rapid breakdown of the starting material that permitted the remaining solid product to mix intimately with the support material.

To ensure no unstable residues remained within the mixture, the solid was then further warmed to 873 K in vacuum.

This combination of procedures produced a free-flowing powder of uniform grey appearance.

The sample was sealed in a glass sample container and was stable on exposure to air, where it was sieved to a grain size of 250–500 μm for subsequent analysis.

CAUTION! Even though the resulting Pd/Al2O3 catalyst is stable, palladium azide compounds are explosive, see .ref. 7

Analysis by atomic absorption spectroscopy showed the palladium loading to be 7.3%.

BET analysis indicated a surface area of 94 m2 g−1, indicating the new preparative method has no detrimental effect on total surface area.

Pulsed chemisorption of CO showed the catalyst to saturate at 45 μmol CO g−1, which corresponds to 5.42 × 1019 surface Pd atoms g−1, assuming a Pd : CO surface stoichiometry of 2 : 18 and represents a metal dispersion of 13.1%.

Assuming the metal particles are spheres of equal diameter, the chemisorption result equates to a mean particle size of 8.3 nm.

Elemental analysis performed using a CE-440 elemental analyser prior to the reduction process8,9 was unable to detect the presence of nitrogen in either the support material or the Pd (azide)/Al2O3 catalyst.

A representative TEM micrograph for the catalyst is presented in Fig. 1; metal crystallites are clearly discernible.

Some of the particles exhibit a distinct hexagonal structure, consistent with cubo-octahedra, as described by Hardeveld and Hartog12 and observed elsewhere.1,13

Recognition of the general shape of the particles indicates that, to a first approximation, the particle surface can be described in terms of a small number of low index planes, i.e. (111) and (100), as identified by STM for the surface science model catalysts.5,6

Overall, the catalyst exhibits mostly regular particles, with diameters clustered about ca. 5 nm and 15 nm.

This range is generally consistent with the mean particle size predicted from CO chemisorption.

High resolution TEM images show the Pd particles to be often composed of crystalline sub-units separated by grain boundaries, twin planes, etc.

(insert in Fig. 1).

Thus, the polyhedral nanoparticles are not always perfect single crystals but, nevertheless, they are still dominated by well developed low index facets with a low level of surface defects (one should note that distortions by the microcrystalline alumina support typically prevent high resolution imaging of particle facets).13

The background subtracted diffuse reflectance spectrum in the region 1600–2300 cm−1 for a saturation coverage of CO on the Pd(azide) catalyst, achieved using a pulse-flow arrangement at 293 K, is presented in Fig. 2(a).

This spectrum is atypical for CO chemisorption on a supported palladium catalyst, in that there is effectively no contribution from linear CO at ca. 2100–2050 cm−1.14–16

To illustrate this point the DRIFTS spectrum for a saturation CO coverage for a 1% Pd/Al2O3 catalyst prepared via wet impregnation and using Pd(NO3)2 as the precursor compound is presented in Fig. 2(b).

Elemental analysis of this catalyst prior to reduction produced a nitrogen content of 0.42%.

CO chemisorption showed this catalyst to have a metal dispersion of 78%, which corresponds to a mean paricle size of ca. 1.4 nm.

The spectrum more closely resembles that presented for numerous Pd/Al2O3 catalysts, e.grefs. 14–16., with a distinct signal due to linear CO clearly apparent.

The Pd(azide)/Al2O3 spectrum is characterised by two intense, symmetrical features: a broad band at 1923 cm−1 and a sharp feature at 1984 cm−1 (FWHM = 21 cm−1).

On the basis of CO adsorption studies on metal single crystal and nanoparticle model catalysts16–19 the 1923 cm−1 band is assigned to μ3 hollow bonded CO and μ2 bridge bonded CO on the (111) planes of the Pd particles.

The origin of the sharp 1984 cm−1 feature is more complex.

According to combined vibrational and density functional studies on Pd surfaces of varying roughness, it may originate from CO adsorbed on (100) facets and from CO bridge bonded to particle edges and steps.20

However, even these contributions cannot explain the high intensity of the 1984 cm−1 peak and “intensity borrowing” from bridging CO on (111) facets (1923 cm−1) may therefore increase the intensity of the 1984 cm−1 peak.21

In the pulse chemisorption arrangement employed here, only high purity helium is present over the catalyst during spectral acquisition.

This situation equates to CO spectra recorded post-evacuation of CO for the more conventional equilibrium measurements, which typically produce CO bands in the 2100 cm−1 region,14 characteristic of linear (on-top) CO. Studies on model Pd–Al2O3 catalysts revealed that more intense linear bands were observed for rougher surfaces.5,6,19

The effective absence of on-top CO in Fig. 2 is therefore interpreted as indicating that the present catalyst has a low concentration of relatively open sites of low metal coordination number.

In this respect, the infrared data are broadly consistent with the electron microscopy.

Recently published CO spectra for well-annealed Pd/MgO catalysts also exhibit a minimal linear CO contribution under comparable conditions; in contrast the behaviour of a Pd/γ-Al2O3 catalyst (prepared from PdCl2) is less well defined and a significant contribution from linear CO is retained upon evacuation.22

Furthermore, since surface impurities such as carbon increase the amount of linear bonded CO,23 the diffuse reflectance spectrum indicates that the Pd particles derived from the azide precursor exhibit clean facets.

The synthetic approach described in this work has the advantage of producing structured metal particles on a high surface area alumina support, that is ubiquitous in heterogeneous catalysis.

In fact, alumina is thought to be the most widely used commercial carrier (support) material.24

Given this wide application, future studies could examine the sensitivity of particle morphology to alumina structure by investigating structurally pure aluminas such as γ- and η-aluminas.

To demonstrate that the Pd(azide)/Al2O3 catalyst exhibits some effectiveness as a catalytic substrate, Fig. 3 presents the results for the hydrogenation of propyne at 293 K in a continuous flow reactor.

Under the selected reaction conditions, full conversion and complete hydrogenation of the propyne are observed, with no loss of activity apparent over the duration studied.

It is acknowledged that operation at full conversion prevents a more complete analysis of the capability of the Pd(azide)/Al2O3 catalyst to function as an alkyne hydrogenation catalyst, nevertheless, Fig. 3 clearly indicates this substrate to constitute an active catalyst, worthy of further investigation.

It is worthwhile considering the value of this material.

It is readily acknowledged that such an exotic (and potentially dangerous) preparative procedure could never be used to prepare industrial catalysts.

Furthermore, the catalyst described here offers modest dispersion.

However, this latter attribute could be improved by moderating the anneal temperature.25

Additionally, the preparative technique described inherently leads to a residue-free material and the chemisorbed CO infrared spectrum indicates well ordered metal crystallites.

Bertarione et al. note that the linear CO signal reflects the defect concentration of supported Pd catalysts.22

Following this approach, Fig. 2(b) shows, in agreement with Bertarione et al., that conventional catalyst preparations of alumina-supported Pd catalysts invariably lead to defect rich substrates.

In contrast, Fig. 2(a) clearly shows the azide derived catalyst to exhibit a low concentration of defect sites.

Thus, this catalyst then provides a reference material that could be used to evaluate the role that defect sites play in the surface chemistry/catalysis of supported palladium catalysts and is therefore ideally suited to in situ studies.26

Although other preparative procedures could, in principle, also prepare such useful highly ordered catalysts, e.g. carefully controlled impregnation methods or MOCVD-based techniques,27 the azide based preparative route has the added advantage that it yields a minimal residue catalyst.

In summary, a novel preparative approach is described that inherently leads to a residue-free high surface area alumina-supported Pd catalyst, that presents a well defined crystallite morphology.

A schematic representation of the adsorption sites populated at saturation CO coverage is shown in Fig. 4.

This catalyst can be used to evaluate the role of low index planes in selective hydrogenation reactions and represents a useful benchmark against which preparative procedures, particle roughness, and catalytic activity can be evaluated.

Work is in progress to explore these exciting possibilities.