# Hot Carrier Injection-Driven Nano-Interface Assembly for Hydrogen Generation

## TL;DR

This study explores using plasmonic nanoparticles with FeVO4 to improve solar-driven hydrogen generation by optimizing charge transfer and reducing electron-hole recombination.

## Contribution

A novel sequential optimization procedure for 1D FeVO4 integrated with plasmonic nanoparticles is introduced for enhanced photocatalytic performance.

## Key findings

- Plasmonic nanoparticles enhance visible light absorption and charge separation in FeVO4.
- Au, Au-urchin, Ag, and Au+Ag NPs show distinct effects on photocatalytic efficiency based on size, shape, and material.
- A machine learning model predicts optimal parameters for band gap tuning and photon-to-current efficiency.

## Abstract

Harnessing hot electron
transfer (HET) at plasmonic–semiconductor
interfaces is a promising route to modulate charge carrier dynamics
toward solar energy-driven water splitting for hydrogen generation.
Popular semiconductor photocatalysts driving solar-to-hydrogen conversion,
such as FeVO4, suffer from limited visible light absorption,
electron–hole recombination, and aqueous instability, often
seeking band-gap engineering or cation doping to improve their catalytic
prowess. In this first-of-its-kind comprehensive study, we demonstrate
a sequentially optimized procedure to obtain one-dimensional (1D)
FeVO4 that is integrated with plasmonic nanoparticles (PNPs)
to address these limitations. Anchored on the surface of the semiconductor,
PNPs generate hot electrons upon visible light irradiation, that are
then transferred to FeVO4. Finite-difference time-domain
simulations verify the electromagnetic field distribution around the
FeVO4–PNP. Additionally, Au, Au-urchin, Ag, and
Au+Ag NPs were used to understand the effect of varying sizes, shapes,
and plasmonic metals on the photocatalytic efficiency of FeVO4. Circularly polarized photon-triggered asymmetric hot carrier
injection (from Au, Au-urchin, Au+Ag) and plasmon-induced resonance
energy transfer (from Ag) reveal voltage-dependent interfacial dynamics
that govern charge separation and hydrogen evolution efficiency. The
experimental data was used to train a generative reinforcement learning
(GRL)-based machine learning model to predict the optimum parameters
for tunable band gaps and applied bias photon-to-current efficiency.
This study thus lays the foundation for determining appropriate combinations
of PNPs and other semiconductor materials for photoelectrochemical
(PEC) applications.