Abstract
In
this
study,
we
identified
features
with
the
largest
contributions
and
property
trends
in
predicting
adsorption
energies
of
carbon,
hydrogen,
oxygen
adsorbates
on
transition
metal
(TM)
surfaces
by
performing
Density
Functional
Theory
(DFT)‐based
calculations
Machine
Learning
(ML)
regression
models.
From
26
monometallic
400
bimetallic
fcc(111)
TM
obtained
from
Catalysis‐hub.org,
three
datasets
consisting
fourteen
elemental,
electronic,
structural
properties
were
generated
using
DFT
calculations,
site
online
databases.
The
number
was
reduced
feature
selection
then
finely‐tuned
random
forest
(RFR),
gaussian
process
(GPR),
artificial
neural
network
(ANN)
algorithms
implemented
for
energy
prediction.
Finally,
model‐agnostic
interpretation
methods
such
as
permutation
importance
(PFI)
shapely
additive
explanations
(SHAP)
provided
rankings
directional
trends.
For
all
datasets,
RFR
GPR
demonstrated
highest
prediction
accuracies.
addition,
that
contributing
models
consistent
structure‐property‐performance
relationships
TMs
like
d‐band
model,
Friedel
higher‐fold
sites.
Overall,
interpretable
ML–DFT
approach
can
be
applied
to
their
derivatives
atomic
model
explainability.
Energy & Environmental Science,
Год журнала:
2024,
Номер
17(14), С. 5200 - 5215
Опубликована: Янв. 1, 2024
We
demonstrate
the
practical
applicability
of
Ni–Co–Mn–P
as
an
efficient
electrocatalyst
active
in
all
HER,
OER,
and
ORR
processes
even
under
ultra-high
mass
loading
over
22
mg
cm
−2
.
Due
to
the
high
cost
of
available
Pt
electrocatalysts,
large-scale
water
electrolysis
production
hydrogen
has
been
hindered.
Hydrogen
generation
via
electrochemical
splitting
is
a
renewable
energy
essential
sustainable
society,
creating
distinct
material
interface
that
shows
Pt-like
properties
with
long-term
stability
crucial
evolution
reactions
(HERs).
Here,
we
synthesized
guanine-assisted
facile
synthesis
1
wt
%
Pt/Mo2C/C
having
layered
type
morphology
solid
state
calcined
process
followed
by
chemical
reduction.
The
well-developed
heterostructure
analyzed
inductively
coupled
plasma
optical
emission
spectroscopy
(ICP-OES)
understand
percentage
doped
on
Mo2C/C.
as-synthesized
exhibits
better
HER
activity
than
commercial
Pt/C
small
overpotential
19
mV
reach
current
density
at
10
mA
cm–2
Tafel
slope
28
mV/dec.
catalyst
42
h
in
0.5
M
H2SO4.
sheet
structure
N-doped
carbon
(C)
nanosheet,
encapsulating
well-dispersed
within
layers,
significantly
enhances
reaction
kinetics
Pt/Mo2C/C.
This
design
creates
synergistic
effect
among
Mo2C,
Pt,
and
matrix,
improving
catalytic
performance.
Theoretical
calculations
using
functional
theory
(DFT)
indicate
active
sites
for
Pt-integrated
possessed
reduced
ΔGH*
value
(−0.06
eV)
as
compared
pristine
Mo2C/C
(ΔGH*
=
0.34
eV),
suggesting
higher
activity.
simple
method
offers
fresh
means
make
clearly
defined
carbides
sheds
light
low-Pt
catalysts
scalable
HER.