Environmental Science & Technology,
Journal Year:
2024,
Volume and Issue:
58(15), P. 6628 - 6636
Published: March 18, 2024
Biomass
waste-derived
engineered
biochar
for
CO2
capture
presents
a
viable
route
climate
change
mitigation
and
sustainable
waste
management.
However,
optimally
synthesizing
them
enhanced
performance
is
time-
labor-intensive.
To
address
these
issues,
we
devise
an
active
learning
strategy
to
guide
expedite
their
synthesis
with
improved
adsorption
capacities.
Our
framework
learns
from
experimental
data
recommends
optimal
parameters,
aiming
maximize
the
narrow
micropore
volume
of
biochar,
which
exhibits
linear
correlation
its
capacity.
We
experimentally
validate
predictions,
are
iteratively
leveraged
subsequent
model
training
revalidation,
thereby
establishing
closed
loop.
Over
three
cycles,
synthesized
16
property-specific
samples
such
that
uptake
nearly
doubled
by
final
round.
demonstrate
data-driven
workflow
accelerate
development
high-performance
broader
applications
as
functional
material.
Biochar,
Journal Year:
2022,
Volume and Issue:
4(1)
Published: Nov. 29, 2022
Abstract
Biochar
produced
from
pyrolysis
of
biomass
has
been
developed
as
a
platform
carbonaceous
material
that
can
be
used
in
various
applications.
The
specific
surface
area
(SSA)
and
functionalities
such
N-containing
functional
groups
biochar
are
the
most
significant
properties
determining
application
performance
carbon
areas,
removal
pollutants,
adsorption
CO
2
H
,
catalysis,
energy
storage.
Producing
with
preferable
SSA
N
is
among
frontiers
to
engineer
materials.
This
study
attempted
build
machine
learning
models
predict
optimize
(SSA-char),
content
(N-char),
yield
(Yield-char)
individually
or
simultaneously,
by
using
elemental,
proximate,
biochemical
compositions
conditions
input
variables.
predictions
Yield-char,
N-char,
SSA-char
were
compared
random
forest
(RF)
gradient
boosting
regression
(GBR)
models.
GBR
outperformed
RF
for
predictions.
When
parameters
included
elemental
proximate
well
conditions,
test
R
values
single-target
multi-target
0.90–0.95
except
two-target
prediction
Yield-char
which
had
0.84
three-target
model
0.81.
As
indicated
Pearson
correlation
coefficient
between
variables
feature
importance
these
models,
top
influencing
factors
toward
predicting
three
targets
specified
follows:
temperature,
residence
time,
fixed
Yield-char;
ash
N-char;
temperature
SSA-char.
effects
on
different,
but
trade-offs
balanced
during
ML
optimization.
optimum
solutions
then
experimentally
verified,
opens
new
way
designing
smart
target
oriented
potential.
Graphical
Green Chemical Engineering,
Journal Year:
2022,
Volume and Issue:
4(1), P. 123 - 133
Published: May 27, 2022
Gasification
is
a
sustainable
approach
for
biomass
waste
treatment
with
simultaneous
combustible
H2-syngas
production.
However,
this
thermochemical
process
was
quite
complicated
multi-phase
products
generated.
The
product
distribution
and
composition
also
highly
depend
on
the
feedstock
information
gasification
condition.
At
present,
it
still
challenging
to
fully
understand
optimize
process.
In
context,
four
data-driven
machine
learning
(ML)
methods
were
applied
model
prediction
interpretation
optimization.
results
indicated
that
Gradient
Boosting
Regression
(GBR)
showed
good
performance
predicting
three-phase
syngas
compositions
test
R2
of
0.82–0.96.
GBR
model-based
suggested
both
feed
condition
(including
contents
ash,
carbon,
nitrogen,
oxygen,
temperature)
important
factors
influencing
char,
tar,
syngas.
Furthermore,
found
higher
carbon
(>
48%),
lower
nitrogen
(<
0.5%),
ash
(1%–5%)
under
temperature
over
800
°C
could
achieve
yield
H2-rich
It
shown
optimal
conditions
by
an
output
containing
60%–62%
H2
44.34
mol/kg.
These
valuable
insights
provided
from
aid
understanding
optimization
guide
production
ACS Nano,
Journal Year:
2023,
Volume and Issue:
17(11), P. 9763 - 9792
Published: June 2, 2023
Zero-carbon
energy
and
negative
emission
technologies
are
crucial
for
achieving
a
carbon
neutral
future,
nanomaterials
have
played
critical
roles
in
advancing
such
technologies.
More
recently,
due
to
the
explosive
growth
data,
adoption
exploitation
of
artificial
intelligence
(AI)
as
part
materials
research
framework
had
tremendous
impact
on
development
nanomaterials.
AI
has
enabled
revolutionary
next-generation
paradigms
significantly
accelerate
all
stages
material
discovery
facilitate
exploration
enormous
design
space.
In
this
review,
we
summarize
recent
advancements
applications
discovery,
with
special
emphasis
selected
nanotechnology
net-zero
future
including
solar
cells,
hydrogen
energy,
battery
renewable
CO2
capture
conversion
capture,
utilization
storage
(CCUS)
addition,
discuss
limitations
challenges
current
area
by
identifying
gaps
that
exist
development.
Finally,
present
prospect
directions
order
large-scale
Environmental Science & Technology,
Journal Year:
2023,
Volume and Issue:
57(46), P. 17971 - 17980
Published: April 8, 2023
Apparent
quantum
yields
(Φ)
of
photochemically
produced
reactive
intermediates
(PPRIs)
formed
by
dissolved
organic
matter
(DOM)
are
vital
to
element
cycles
and
contaminant
fates
in
surface
water.
Simultaneous
determination
ΦPPRI
values
from
numerous
water
samples
through
existing
experimental
methods
is
time
consuming
ineffective.
Herein,
machine
learning
models
were
developed
with
a
systematic
data
set
including
1329
points
predict
the
three
ΦPPRIs
(Φ3DOM*,
Φ1O2,
Φ·OH)
based
on
DOM
spectral
parameters,
conditions,
calculation
parameters.
The
best
predictive
performances
for
Φ3DOM*,
Φ·OH
achieved
using
CatBoost
model,
which
outperformed
traditional
linear
regression
models.
significances
wavelength
range
parameters
predictions
revealed,
suggesting
that
lower
molecular
weight,
aromatic
content,
more
autochthonous
portion
possessed
higher
ΦPPRIs.
Chain
constructed
adding
predicted
Φ3DOM*
as
new
feature
into
Φ1O2
models,
consequently
improved
performance
but
worsened
prediction
likely
due
complex
formation
pathways
·OH.
Overall,
this
study
offered
robust
across
interlaboratory
differences
provided
insights
relationship
between
PPRIs
properties.
