High-performance light-driven heterogeneous CO2 catalysis with near-unity selectivity on metal phosphides

Akin to single-site homogeneous catalysis, a long sought-after goal is to achieve reaction site precision in heterogeneous catalysis for chemical control over patterns of activity, selectivity and stability. Herein, we report on metal phosphides as a class of material capable of realizing these attributes and unlock their potential in solar-driven CO2 hydrogenation. Selected as an archetype, Ni12P5 affords a structure based upon highly dispersed nickel nanoclusters integrated into a phosphorus lattice that harvest light intensely across the entire solar spectral range. Motivated by its panchromatic absorption and unique linearly bonded nickel-carbonyl-dominated reaction route, Ni12P5 is found to be a photothermal catalyst for the reverse water gas shift reaction, offering a CO production rate of 960 ± 12 mmol gcat−1 h−1, near 100% selectivity and long-term stability. Successful extension of this idea to Co2P analogs implies that metal phosphide materials are poised as a universal platform for high-rate and highly selective photothermal CO2 catalysis.

| High-resolution XPS plots of a, Ni 2p and b, P 2p regions before and after photocatalytic testing. Note a peak observed at 856.1 eV was attributed to the formation of oxidized Ni, while another at 133.6 eV was attributed to P in phosphate, suggesting that some partial surface oxidation occurred when exposing the sample to air.  The accuracy of the ASPEN simulation depends on 1) the validity of the ideal gas approximation 2) if water stays in the gas phase for the duration of the experiment. Typically, we assume that 1) & 2) are met. 1) is valid for high temperature (>100 o C) and low-pressure systems (<2 atm).
For the estimation we used the ASPEN NRTL property package (assumes ideal gas phase) and the Gibbs reactor block that assumes all reaction between components (H2, CO2, CO, H2O). A sweep was performed in 50 o C increments and the local temperature was determined by matching our GC results with the temperature dependent ASPEN output.
Wet basis mode (assuming product water reaches GC for detection):

Supplementary Note 1. Details of TOF calculation
The TOF is calculated in terms of per nickel metal site, according to the following equation:

Total CO turnovers per second Total Ni site numbers
Where the total CO turnovers per second is calculated as Total CO turnovers per second r mass 3600 r mass 3600 Where r is the CO production rate (mmol gcat -1 h -1 ) presented in Table 1, masstotal is the mass of the used catalyst (including Ni12P5 and SiO2 supports), x is the weight percent of Ni12P5 in the Ni12P5/SiO2, NA is the Avogadro number.
To determine Ni site numbers, there are two methods:

Total Ni site numbers mass M
where y is the weight percent of Ni in the sample, as determined from ICP-OES, D is the metal dispersion which was obtained from a pulse H2 chemisorption experiment. Note the metal dispersion herein is determined with respect to the Ni metal, and assume atomic hydrogen only binds to surface nickel atoms with a H:Ni stoichiometry of 1. [10][11] M is the atomic weight of Ni, which is 58.69 g mol -1 .
Therefore, the TOF can be calculated according to the following equation: The results are listed in the following Table   Sample Metal dispersion (%) TOF (s  12 The theoretical metal site concentration assumes that the samples are composed of uniform spherical particles. This L is calculated by where n is the average surface metal atom density, and S is the effective surface area.
The effective surface area was calculated as where ρ is the material density (7.53 g cm -3 for Ni12P5), D is the average particle size extract from the size distribution statistics based on TEM images (as shown in Table 1 and Supplementary Fig. 3).
Average surface metal atom density is estimated via a published method. 3 For the Ni12P5 there are six, six, and six Ni atoms on the ac, ab, and bc unit cell faces, respectively ( Supplementary Fig. 20

Supplementary Note 2. The calculation of the internal quantum yield of CO (IQY CO )
In this study the internal quantum yield was calculated as it can exclude the light absorption variations to make a fair comparison. The internal quantum yield of CO was defined as 13 Internal quantum yield produced CO molecules per unit time absorbed photon numbers per unit time Where the absorbed photon numbers per unit time, Nphoton, is estimated from the light intensity dispersion of the Xe lamp ( Supplementary Fig. 21) and the UV-vis-NIR absorption spectra.
Light intensity * I% * A% * illmination area * time Average single photon energy * Where the light intensity is 2.3 W, illumination area is 1 cm 2 , I% is the percentage of the Xe light intensity at certain wavelength, A% is the light harvesting efficiency at certain wavelength according to the absorption spectra ( Fig. 1b in