1. Hydrosilylation
The hydrosilylation reaction is widely utilized in preparing silicone-based materials such as aerogels and surfactants. Among various catalysts, homogeneous Pt catalysts are frequently used in industrial processes. In general, the reaction mechanisms of alkene hydrosilylation over Pt (0) catalysts include the Chalk–Harrod or modified Chalk–Harrod mechanism
[1]. The Chalk–Harrod mechanism is divided into three steps: (1) Si-H oxidative addition to Pt, (2) alkene insertion into the Pt-H bond, and (3) Si-C reductive elimination. In the modified Chalk–Harrod mechanism, the second step is replaced by alkene insertion into the Pt-Si bond, followed by the C-H reductive elimination. Therefore, a typical Karstedt’s catalyst, with the zero oxidation state at the Pt center, could exhibit superior catalytic performance for anti-Markovnikov addition hydrosilylation reactions even under mild working temperatures
[2]. However, certain kinds of problems, especially concerning the separation and recycling of homogeneous catalysts, will always persist. Therefore, Pt-based SACs are likely the alternatives to replace traditional Pt-complex catalysts.
In this context, one of the most important issues lies in the fabrication of stable Pt SACs. In 2017, Beller’s group presented the first heterogeneous single-atom Pt catalyst for hydrosilylation reactions
[3]. They anchored isolated Pt species on alumina nanorods and thus prepared Pt/NR-Al
2O
3-IP with high reactivity comparable to Karstedt’s catalyst, demonstrating that SACs can be as effective as homogeneous catalysts. Moreover, various functional groups on the alkenes can be tolerated by Pt/NR-Al
2O
3-IP SAC, which can be separated and reused with good stability.
With the first successful case of single Pt atoms catalyzed hydrosilylation in hand, more efforts have been devoted to further improve catalytic performance by regulating the coordination structure of Pt active centers. Carbon-based supports are widely used in industrial processes due to their low cost, high surface areas, and tunable surface properties since the heteroatoms (O, N, P, and S) can be easily doped on/in the carbon materials. Therefore, some carbon-supported Pt SACs are investigated here in detail regarding their potential industrial applications. For example, up to 5.3 wt% loading of Pt is atomically dispersed on the N-doped graphene (Pt-ISA/NG) with the Na
2CO
3-assisted one-pot pyrolysis synthetic strategy
[4]. In the benchmark reaction of hydrosilylation of 1-octene, Pt-ISA/NG with a stable Pt-N
4 site exhibited high selectivity. The reactivity of Pt-ISA/NG was 4 times higher than that of commercial Pt/C with the same Pt loadings, suggesting that the completely exposed metal centers are pivotal for achieving high catalytic performance
[4]. In addition, P was added along with N to tune the electronic state of Pt centers with a similar structure. The PN-doped carbon nanofibers were fabricated based on the reaction between P
2O
5 and
N-Methyl-2-pyrrolidone. Pt single-atoms were anchored by the -PN- units on the support and denoted as Pt
1@C
NP-600, which displayed high TOF for hydrosilylation (9.2 × 10
6 h
−1) and 99% selectivity
[5]. Unlike the traditional (modified) Chalk–Harrod mechanism, the reaction mechanism of Pt
1@C
NP-600 showed the in situ formation of cyclization of ketene oligomers and subsequent polymerization at elevated temperatures, as evidenced by the FTIR and NMR results. Moreover, doping of O and Cl with Pt atoms resulted in electron-deficient Pt centers. For example, Pt SAC supported on humic matter (Pt
1@AHA_U_400) showed the best activity thus far in hydrosilylation of 1-octene (TOF = 3.0 × 10
7 h
−1) with >99% selectivity
[6]. DFT calculations revealed that this high performance can be attributed to the atomic dispersion and the electron-deficiency state of Pt species since Pt was connected with two Cl and two O ions. According to DFT calculations, Pt
1@AHA_U_400 prefers to undergo the modified Chalk–Harrod mechanism regarding hydrosilylation reactions, owing to its low energy barrier.
Besides heteroatom doping, the carbon support could also offer a versatile platform to add another metal species alloying with Pt. In this case, Fe was added to form a PtFe
3 catalytic center on N-doped carbon spheres (PtFe
3/CN)
[7]. The electrons could transfer from neighboring Fe species to Pt single atoms in a PtFe
3/CN catalyst. Consequently, the electron-rich isolated PtFe
3 sites showed higher catalytic performance (99% selectivity and ca. 740,000 of TON) than the Pt
1/CN SAC and K
2PtCl
6 homogeneous catalysts. The hydrosilylation mechanism of the PtFe
3/CN catalyst was investigated using DFT calculations, proposing the Chalk–Harrod mechanism: Si-H bond oxidative addition to isolated Pt center and alkene insertion into the Pt-H bond. These results show that the electron-rich catalytic centers may further improve the hydrosilylation reactivity; however, these sites may be more vulnerable to aggregate than the electron-deficient Pt sites. Therefore, a porous confinement strategy is developed to locate the Pt sites within a nanocage and maintain the valance of Pt between 0 and +2. The confined Pt SAC (COP1-T-Pt) was ten times more active than the Karstedt’s catalyst
[8]. Besides, remarkable site-selectivity as well as good recyclability of COP1-T-Pt was achieved. Both quantitative EXAFS fitting and FDMNES simulation supported that Pt was anchored by the surrounding Pt-C and Pt-Si bonds, forming a familiar structure as the key intermediate in the Chalk–Harrod mechanism.
The above cases illustrate that isolated Pt centers may be suitable replacements for the conventional homogeneous catalysts in hydrosilylation reactions. Inspired by these results, other noble metals have also been employed. A recent example showed that ZrO
2-supported Ru SAC was more reactive than the RuCl
3 homogeneous catalyst in ethylene hydrosilylation
[9]. The reaction mechanism on the Ru
δ+/ZrO
2 SAC is proposed to follow the Chalk–Harrod mechanism according to DFT simulation. Based on the mechanism insights obtained from Pt-based SACs, we can anticipate that the electronic state of other metal species may be further tuned by the coordinating species, such as heteroatoms, to achieve better catalytic performance than homogeneous catalysts.
2. Hydroformylation
Currently, the hydroformylation reaction is homogeneously catalyzed in industrial processes, producing aldehydes from olefins and syngas (CO/H
2). Compared to the above hydrosilylation reaction, achieving high chemo-selectivity in hydroformylation reactions is quite difficult as the side-reaction of olefin hydrogenation simultaneously occurs under H
2 atmospheres. Consequently, the homogeneous Wilkinson’s catalyst (RhCl(PPh
3)
3) is frequently used for its high reactivity and chemo-selectivity, while in the heterogenous catalysts, compared with Rh NPs, the singly dispersed Rh species are more capable to prevent hydrogenation pathways due to the relatively weak bonding effect with H
2 [10]. However, isolated metal ions may be reduced with CO and/or H
2 gas, resulting in the deactivation of catalysts via aggregation. Therefore, the active structure of SACs should achieve a balance between strong stabilization effects to prevent sintering and flexible coordination with substrates
[11].
Regarding oxide supports, three synthetic strategies are developed to design stable and reactive Rh SACs, including creating oxygen vacancies on the supports, fabricating dual-metal sites, and forming metal alloy species. A systematic study on some representative oxide-supported Rh SACs was performed, correlating the densities of oxygen vacancies with catalytic performances
[12]. It is obvious that the existence of oxygen vacancies on the oxide surface leads to the dominant hydroformylation pathway rather than the hydrogenation reaction. Moreover, the oxygen-defective SnO
2-supported Rh SAC showed hundreds of times higher reactivity than that on ZrO
2 support in the gas-phase hydroformylation of ethylene. These results revealed that the delicate design of the surface defects, such as vacancies and step sites, are helpful for achieving excellent catalytic activity on the reducible-oxide-supported Rh SACs. The Heck–Breslow mechanism is proposed for the Rh catalysts on the oxygen-defective support. After CO and H adsorption, insertion of ethylene into the Rh-H bond provided an ethyl group. Following this, an additional CO molecule inserted into the Rh-ethyl bond, and the subsequent reductive elimination formed the final product. However, oxygen vacancies are rarely found on inert oxides such as Al
2O
3 and SiO
2, so dual metal sites are concisely fabricated to increase reactivity. In the case of Al
2O
3-supported Rh-WO
x catalysts, increasing the loading of WO
x could lead to different structures of catalytic sites, and the atomically dispersed Rh-W pair sites achieved the highest rate of ethylene hydroformylation to propanal
[13]. Unlike traditional Rh SACs, the synergetic effects of Rh and W species was suggested by unique mechanism calculations involving Rh-assisted WO
x reduction, ethylene transfer from WO
x to Rh atoms, and H
2 dissociation at the Rh-WO
x interface. In the above two cases, the Rh centers are both in the positive valence state, while the electron-rich Rh single atoms are seldom reported. Researchers fabricated ZnO-nanowire-supported Rh SACs and used the in situ EXANES and XPS techniques to verify that after mild H
2 reduction at 200 °C, the isolated Rh species were at the approximately Rh
0 state
[14]. A plausible explanation for this is that Zn
2+ was simultaneously reduced to Zn
0, transforming electrons to lower the oxidation state of the nearby Rh species. The Rh/ZnO SACs were even slightly more reactive and chemo-selective than the homogeneous Wilkinson’s catalyst in styrene hydroformylation, demonstrating that regulating the valence state of Rh sites is pivotal for achieving excellent catalytic performance.
Although very high reactivity is fulfilled by the oxide-supported Rh SACs, regio-selectivity in hydroformylation is another concern. For the SACs on the open oxide supports, a similar number of linear and branched aldehyde products always generate at the same time
[14]. Therefore, the spatial confinement strategy is thus developed to precisely control the regioselectivity. In this context, the zeolite matrix
[15] and the porous polymer
[16] are identified as suitable supports to locate single Rh atoms. The Rh@Y zeolite catalyst was prepared with the in situ hydrothermal method, achieving 100% chemo-selectivity and good regioselectivity toward heptanal (linear/branched = 2.2) in the hydroformylation reaction of 1-hexene
[15]. Inspired by the homogeneous catalytic systems, wherein high regio-selectivity can be obtained by adding excess amount of phosphine ligands, Rh@POP-PTBA-HA-50 was designed to encapsulate atomically dispersed Rh species within porous monophosphine polymers, and the selectivity of linear aldehyde was remarkably increased to 92% in 1-octene hydroformylation
[16]. Detailed mechanism studies revealed that abundant phosphine ligands within polymer substrates endow the isolated Rh sites with high activity and stability. The similar coordinating effect of N to the single metal centers might also promote regioselectivity. For example, a Ru@NC SAC exhibited very good activity (TOF = 12,000 h
−1) and remarkable regioselectivity (linear/branched = 93:7) in the hydroformylation of 1-hexene, owing to the presence of Ru(II)-N
x-active species on the N-doped carbon support
[17].
The heteroatom-doped strategy can even be applied to the non-noble metal SACs on oxide materials. The ZrP-supported Co SAC showed 91.3% aldehyde selectivity in 1-octene hydroformylation
[18]. The strong bonding effect between Co sites and the phosphate groups on ZrP are the origins responsible for suppression of Co ions regarding leaching, thus improving the catalytic performance and stability of the Co SAC. In situ FT-IR analysis was used to clarify the reaction mechanism. The proposed reaction pathway involved the following steps such as alkene coordination, hydride and CO migratory insertion forming the acyl-like species, and reductive elimination. Currently, Co-based SACs attract much attention as next-generation hydroformylation catalysts, since the reactivity of Co SAC is very similar to that of Rh catalysts, as evidenced by the case of Co
1/
β-Mo
2C in the hydroformylation of propene
[19].
3. Constructing C-Y (Y = B, P, S, N) Bonds
Besides the above C-Si and C-C bonds forming, SACs have also been used in constructing some interesting C-Y (Y = B, P, S, N) bonds through hydroboration, hydrophosphinylation, sulfonation, and aziridination of olefins, respectively. The regioselective hydroboration of alkenes is a direct method to synthesize linear alkylboronic esters. In this reaction, the isolated Pt species are frequently used as active sites, and the catalytic activities are dramatically influenced by the coordination environment of Pt centers. For example, three different coordination structures of single-atom Pt species, denoted as SA Pt-O
3, SA Pt-O
2, and SA Pt-N
4, respectively, have been thoroughly investigated in the anti-Markovnikov hydroboration of 1-octene, and SA Pt-O
3 exhibited the highest activity (TON = 3288) along with good selectivity (97%)
[20]. DFT calculations suggested that the unique coordination structure of three O species with Pt could decrease the active energy.
Carbon nitride (CN) and TiO
2 are frequently used in photocatalysis as support materials for the capability to separate photogenerated electron pairs and holes. Recently, single-atom photocatalysts have been attracting increasing attention because of their improved photocatalytic performance. Wen and co-workers reported a biomimetic single Fe-atom photocatalyst CNH through coupling CN with hemin
[22]. Under visible light irradiation, up to 94% yield of
β-ketosulfone was obtained with the CNH-catalyzed sulfonation reaction. Later, the same group prepared another SAC, Ni/TiO
2, which can be easily scaled up for photocatalytic site-selective sulfonation, transforming enamide to
α-amidosulfones and
β-propionamidosulfones with TON levels as high as 18,963 under visible light
[23].
Recently, an unprecedented C-N bond formation of alkenes was revealed by a Co SAC, namely Co
SA-N/C, which was derived from a bimetal-organic framework. Co
SA-N/C showed good aziridination activity, transforming alkene to aziridine at 0 °C and further achieved direct oxyamination by adding methanol into the reaction system. The substrate scope can expand to a series of styrene derivatives and dienes, and some drug-derived olefins could also smoothly undergo the aziridination process
[24].
4. Hydrogenation
Hydrogenation represents a powerful method to convert alkenes to alkanes, generating a wide range of synthetic intermediates
[25]. The single-metal-atom-catalyzed hydrogenation reactions have been previously summarized by many groups
[26][27][28][29][30], so this text only focus on the recent development regarding selectively transforming butadiene to butene as an example, since single Pd/Pt atom-catalyzed selective hydrogenation of 1,3-butadiene plays a crucial role in the purification of dienes in the petrochemical industry, while NP analogues favor the complete hydrogenation process instead
[31]. Carbon-based materials are frequently used for their good thermal stability in the hydrogenation reaction
[32]. Graphene-supported Pd SAC showed excellent durability against sintering and coking for the 100 h duration and maintained ~70% 1-butene selectivity at 95% conversion, surpassing the Pd-NP catalysts
[33]. Moreover, C
3N
4 can also be used as a photocatalyst, and thus protons can be in situ generated through photo splitting of water. For example, carbon-nitride-supported Pd SAC (Pd
1-mpg-C
3N
4) performed the hydrogenation reaction under visible-light irradiation using water as a sustainable source of hydrogen
[34]. Besides Pt and Pd, an Ir SAC supported on the nitrogen-rich carbon substrate exhibited high activity to the hydrogenation of butadiene with perfect selectivity (~100%) to butenes even at 200 °C while the selectivity on Ir-NC gradually dropped to ~0% with increasing temperatures. The operando XAS demonstrated that Ir-X
3 (X = C/N/O) was an active site with great stability under working conditions
[35].
The oxide-supported single-atom alloy (SAA) is also capable of performing conversions with good selectivity. Lucci et al. fabricated the
γ-alumina-supported Pt-Cu SAA with isolated Pt atoms located in the Cu(111) surface. The as-prepared Pt/Cu(111) SAA could display high selectivity and excellent stability under working conditions
[36]. DFT calculations revealed that H
2 was readily activated on the single Pt atoms, and the dissociated H species could migrate to the Cu sites through the spillover effect. The selective hydrogenation to butene was attributed to the weak binding of the butadiene substrate on the Cu site.
5. Epoxidation
SACs can provide simplified models of catalytic centers, and DFT calculations have thus been frequently used to predict catalytic feasibility. Especially for some reactions utilizing explosive substrates, taking the epoxidation reaction as an example, the computational screening of catalysts is of great help in facilitating the traditional optimizing process, since the most-efficient catalyst can be conveniently selected from the catalyst library without performing many reactions.
The electronic structures of single-metal atoms supported on a phosphotungstic acid (PTA) cluster were systematically investigated with DFT calculations to predict the catalytic performance in ethylene epoxidation. Among the non-noble transition metals (Fe, Co, Ni, etc.), Fe preferred to anchor at the four hollow (4H) site of the PTA cluster
[37]. Moreover, the strong covalent metal-support interaction between Fe and PTA is the foundation for high stability. DFT calculations predicted that the activity of Co-N
3 SAC in styrene epoxidation could be further improved by constructing unsaturated defect sites, which underwent lower free energy compared to the Fe-N
4, Cu-N
4 and Co-N
4 structures
[38]. Using
tert-butyl hydroperoxide (TBHP) as the oxidant, 99.9% of the styrene conversion was achieved with 71% selectivity to styrene oxide.
In oxidation reactions, O
2 is an environmental benign replacement to organic peroxides such as TBHP. Therefore, Chen et al. constructed a vacancy-rich Co
1/NC-h SAC from a CoZn-ZIF precursor, aiming to use O
2 as the oxygen source. The Co-N
x active site showed high intrinsic activity in the epoxidation of cyclooctene at 140 °C, and the yield of the target product (1,2-epoxycycloheptane) reached 95%
[39]. Moreover, the oxide materials are also capable of introducing a large number of vacancies in SACs
[40]. For example, Bi vacancies in Ru
1/Bi
2−xWO
6 SACs can confine Ru species at an atomic scale and provide exceptional efficiency in the epoxidation of trans-stilbene to trans-stilbene oxide
[41]. In addition, Ir
1/
α-MnO
2 displayed a remarkable ~99% selectivity in ethylene epoxidation
[42]. In situ experiments and quantum-chemistry calculations indicated that the π-coordination structure between isolated Ir sites and substrates, such as ethylene, and molecular oxygen, can promote the formation of five-membered oxametallacycle intermediates and then accelerate the formation of ethylene oxide. Electrocatalysis is also applied in epoxidation, since Ir-MnO
x SAC exhibited a Faradaic efficiency of 46 ± 4% in cyclooctene epoxidation. Operando XAS characterizations suggested that the electron-deficient Ir sites induced the formation of highly electrophilic oxygen atoms, leading to the enhancement of electrocatalytic performance
[43].