Hypothesis
Cobalt phthalocyanine on carbon nanotubes should reduce CO2 to methanol regardless of whether the catalyst is monomeric or aggregated, which would explain why decades of earlier work missed the reaction.
What failed
Density-functional calculations of monomeric and dimeric CoPc found that for the dimer, CO dissociation is always favoured over the extra reduction needed to form the Co–CHO⁻ intermediate, so the methanol path is effectively blocked; only the monomer makes it competitive at more negative potentials.
Broken assumption
The catalyst's aggregation state does not change which CO2-reduction product it makes.
Why
The authors attribute the dimer's harder reduction to repulsive interactions between the CoPc units and weaker solvent stabilisation of charge in the aggregate.
From the published paper: Chen et al. Organometallics, 2021. ↗CC BY