Perfectly monodisperse clusters of oxides are critically important
model systems for catalysis studies because they allow the rigorous
analysis of reaction mechanisms, and variations at the single-atom
level can already be reflected in their reactivity. The generation and
intact immobilisation on a suitable substrate of such clusters is quite
challenging, and usually requires mass spectrometric size selection and sophisticated soft landing protocols to make such studies successful.
Tungsten (VI) oxide in particular holds promise as a visible-light photocatalyst, but is quite reactive and can be challenging to immobilise in
a well-defined manner in vacuum [1].
Here, we present a solution-based protocol for the preparation of
monodisperse cyclic tris (tungsten (VI) trioxide) clusters, (WO3) 3. The
clusters can be harvested efficiently on the boron nitride nanomesh
[2], an atomically thin layer of hexagonal boron nitride on Rh (111) with
strong corrugation, and a promising platform for self-assembly [3] and
electrochemical functionality [4]. The triangular (WO3) 3 clusters adsorb
in the ‚pores‘ of the nanomesh, where they were imaged with submolecular resolution using electrochemical scanning tunnelling microscopy.
The decorated surface was transferred to vacuum where the chemical
identity of the clusters was confirmed with XPS. To our knowledge, this
is the first successful example of self-assembly on the nanomesh from
solution. We expect that proper control over deposition conditions will
allow tuning of the number of clusters per pore, making this a promising model system for on-surface catalysis studies.
We contrast this finding with deposition of the same source material
on rutile TiO2 (110) in liquid, on which the clusters appear to react and
form chains, akin to some observations of sublimated WO3 in vacuum
[5]. Even though the clusters are likely hydroxylated in aqueous solution, this behaviour indicates surprising parallels with UHV and suggests
that, in many cases, solution-based procedures complement vacuum
methods.