1.4 Miniaturization of Metal Catalysts: From Supported Catalysts to Single‐Atom Sites
A core criterion in the design of catalysts is to maximize the active metal dispersion (the ratio of surface atom to bulk), such that the highest reactivity per amount of metal loading on the catalyst can be achieved. This is especially relevant when precious metals are used, which is indeed the case for a large number of catalytic reactions. In the abovementioned historical overview, the strategies for maximizing dispersions include making metal sheets into metal sponges and gauzes of fine wires, as well as depositing active metals onto high‐surface‐area supports to make very fine deposits or thin atomic layers.
The synthesis of supported catalysts capitalizes on the strong interfacial interactions between the active metal and the (usually oxide) support to allow the former to exist as stable and small size deposits. Without the strong interfacial interactions, the initially small deposits tend to diffuse on the support surface and coalesce with another deposit until its surface energy (i.e., a function of surface area) decreases to that of the interfacial energy. Incipient wetness impregnation is by far the most common procedure for the preparation of supported catalysts, where metal precursor solution is drawn into the pores of the support by means of capillary effect. To prevent overflowing of the solution to the external surface of the support, the solution volume introduced in each impregnation step should not exceed that of the pore volume (typically maintained at 80–90%). More liquid solution can be introduced repeatedly upon complete drying of the liquid solvent, leaving behind more metal salt within the pore during each repetition. The advantage of the incipient wetness impregnation is that it does not require very strong interactions between the oxide support surface and the coordinated metal cation from the precursor to reach the desired loading amount. On the contrary, wet impregnation is when the porous support is immersed in the metal precursor solution and the amount that penetrates the pores depends on the metal precursor–support interactions. If the interaction is strong, the impregnated concentration would be higher than that of the bulk, and vice versa. Further drying to remove the solvent from the pores and calcination yield the supported catalysts in both cases of impregnation. To minimize coalescence between the metal deposits during the calcination step, it is essential to remove moisture and oxygen by flowing inert gas and introducing a small amount of NO, respectively [53]. Other techniques such as deposition–precipitation, chemical vapor deposition, and the one‐step flame synthesis (see Chapter 10) have also become popular alternatives for producing supported metal catalysts. The ability to produce small Pt deposits on carbon support has been one of the major breakthroughs that led to popularity of low‐temperature H2‐polymer electrolyte membrane (PEM) fuel cell. In fact, the amount of the ∼3.5 nm Pt used is so small (0.2 mg/cm2 of fuel cell, compared with 28 mg/cm2 in the early days) that it significantly reduced the device cost and thus popularising the H2‐PEM fuel cell [54]. Chapter 32 introduces the design of electrocatalysts for PEM fuel cell applications, while Chapter 4 complements nicely the strategies of using carbon supports for such purpose.
Besides maximizing the metal dispersions, further miniaturization of metal deposits to or approaching the quantum‐related level can result in altered electronic properties not otherwise seen in larger particles. Gold catalysis is an intriguing example of such a phenomenon, which was led notably by the independent efforts of Graham J. Hutchings and Masatake Haruta since the mid‐1980s. They showed that gold, which was classically believed to be almost inactive, can be made extremely active in the hydrochlorination of acetylene [55] and the oxidation of carbon monoxide (at −77 °C!) [56], respectively, when made less than 25 nm. The latter, which gold size was 4.5 ± 1.6 nm, was first prepared by the coprecipitation technique but was later superseded by the deposition–precipitation technique in which dissolved gold precursor was precipitated by raising the pH of the medium in the presence of suspended oxide support. Over time, the commercial flame‐synthesized P25 TiO2 became the preferred support. Many new reactions by gold catalysis followed in the next three decades, ranging from the oxidation of aqueous polyalcohols to carboxylic acids, selective oxidation of cyclohexane to cyclohexanol and cyclohexanone, epoxidation of propylene, water‐gas shift, to the selective hydrogenation of 3‐nitrobenzene and the hydrogenation of alkynes to alkenes. Size‐dependent turnover frequencies (i.e., conversion rate per active site) is typically observed due in part to the variation of electronic interactions, with the optimum gold deposit size for CO oxidation in the range of 2–4 nm [57, 58]. The size‐dependent activity is a general phenomenon as observed readily on different metal deposits including cobalt for FTS [59], palladium for Suzuki coupling [60], and platinum for propane dehydrogenation [61].
Synthesizing ultrasmall size deposits of less than 2 nm (<100 atoms), or so‐called metal clusters (or nanoclusters as a more appealing terminology), can be quite challenging because of their high surface energies. At such a size, the surface energy can become so overwhelming that even when deposited onto high‐surface‐area supports, the metal deposits prefer to exist as larger sizes so as to minimize the exposed surface area (and hence the total energy). In such cases, stabilizing ligands such as glutathione, cetyl trimethyl ammonium bromide (CTAB), and poly(vinylpyrrolidone) (PVP) that bind to the surface of the small deposits can be added during the synthesis procedure. With the ligands being exposed and having lower surface energy than the bare metals, they protect the metal clusters from coalescing or dissolving. Metal clusters are so called not just to distinguish them from the larger nanosized particles, but importantly they reach a state where they no longer behave like metals. As a result of the size quantization effect that gives rise to the discrete orbitals and formation of an energy gap, they essentially behave more as semiconductors. The effect is not unlike the size quantization phenomenon commonly observed for semiconductor photocatalysts with the diameter smaller than the Bohr excitonic radius. Although the term quantum dot (commonly abbreviated as Q‐dot) refers exclusively to such semiconductor particles, by the same definition, metal clusters should also be termed quantum metals! [62]. In that respect, Chapter 5 readily lays out the physics as well as the design principles of different metal clusters catalysts. Metal clusters, with or without ligand bound, have been shown to exhibit catalytic properties different from that of larger nanoparticles, for example, the highly selective oxidation of cyclohexane to cyclohexanone over Ag6/graphene oxide, 100% selectivity of 4‐nitrobenzaldehyde to 4‐nitrobenzyl alcohol over Au99(SPh)42/CeO2, and electrocatalytic reduction of carbon dioxide to carboxylic acid Cu32H20L12 (L = dithiophosphate ligand). Because of the semiconductor nature of metal clusters, they can even function as photocatalysts, for example, in the photocatalytic degradation of aqueous organic micropollutants over glutathione‐protected gold clusters [63, 64].
Single‐atom catalysts (SACs) represent the ultimate extreme end of catalyst miniaturization. First demonstrated by John Meurig Thomas of the Davy‐Faraday Research