Nature Communications,
Journal Year:
2020,
Volume and Issue:
11(1)
Published: July 17, 2020
Abstract
Experimental
procedures
for
chemical
synthesis
are
commonly
reported
in
prose
patents
or
the
scientific
literature.
The
extraction
of
details
necessary
to
reproduce
and
validate
a
laboratory
is
often
tedious
task
requiring
extensive
human
intervention.
We
present
method
convert
unstructured
experimental
written
English
structured
synthetic
steps
(action
sequences)
reflecting
all
operations
needed
successfully
conduct
corresponding
reactions.
To
achieve
this,
we
design
set
actions
with
predefined
properties
deep-learning
sequence
model
based
on
transformer
architecture
action
sequences.
pretrained
vast
amounts
data
generated
automatically
custom
rule-based
natural
language
processing
approach
refined
manually
annotated
samples.
Predictions
our
test
result
perfect
(100%)
match
60.8%
sentences,
90%
71.3%
75%
82.4%
sentences.
Advanced Materials,
Journal Year:
2020,
Volume and Issue:
32(30)
Published: June 4, 2020
Abstract
The
optimal
synthesis
of
advanced
nanomaterials
with
numerous
reaction
parameters,
stages,
and
routes,
poses
one
the
most
complex
challenges
modern
colloidal
science,
current
strategies
often
fail
to
meet
demands
these
combinatorially
large
systems.
In
response,
an
Artificial
Chemist
is
presented:
integration
machine‐learning‐based
experiment
selection
high‐efficiency
autonomous
flow
chemistry.
With
self‐driving
Chemist,
made‐to‐measure
inorganic
perovskite
quantum
dots
(QDs)
in
are
autonomously
synthesized,
their
yield
composition
polydispersity
at
target
bandgaps,
spanning
1.9
2.9
eV,
simultaneously
tuned.
Utilizing
eleven
precision‐tailored
QD
compositions
obtained
without
any
prior
knowledge,
within
30
h,
using
less
than
210
mL
total
starting
solutions,
user
experiments.
Using
knowledge
generated
from
studies,
pre‐trained
use
a
new
batch
precursors
further
accelerate
synthetic
path
discovery
compositions,
by
least
twofold.
knowledge‐transfer
strategy
enhances
optoelectronic
properties
in‐flow
synthesized
QDs
(within
same
resources
as
no‐prior‐knowledge
experiments)
mitigates
issues
batch‐to‐batch
precursor
variability,
resulting
averaging
1
meV
peak
emission
energy.
Angewandte Chemie International Edition,
Journal Year:
2019,
Volume and Issue:
59(52), P. 23414 - 23436
Published: Sept. 25, 2019
This
two-part
review
examines
how
automation
has
contributed
to
different
aspects
of
discovery
in
the
chemical
sciences.
In
this
second
part,
we
reflect
on
a
selection
exemplary
studies.
It
is
increasingly
important
articulate
what
role
and
computation
been
scientific
process
that
or
not
accelerated
discovery.
One
can
argue
even
best
automated
systems
have
yet
``discover''
despite
being
incredibly
useful
as
laboratory
assistants.
We
must
carefully
consider
they
be
applied
future
problems
order
effectively
design
interact
with
autonomous
platforms.
The
majority
article
defines
large
set
open
research
directions,
including
improving
our
ability
work
complex
data,
build
empirical
models,
automate
both
physical
computational
experiments
for
validation,
select
experiments,
evaluate
whether
are
making
progress
toward
ultimate
goal
Addressing
these
practical
methodological
challenges
will
greatly
advance
extent
which
make
meaningful
discoveries.
Chemical Reviews,
Journal Year:
2021,
Volume and Issue:
121(16), P. 9927 - 10000
Published: July 14, 2021
Transition-metal
complexes
are
attractive
targets
for
the
design
of
catalysts
and
functional
materials.
The
behavior
metal-organic
bond,
while
very
tunable
achieving
target
properties,
is
challenging
to
predict
necessitates
searching
a
wide
complex
space
identify
needles
in
haystacks
applications.
This
review
will
focus
on
techniques
that
make
high-throughput
search
transition-metal
chemical
feasible
discovery
with
desirable
properties.
cover
development,
promise,
limitations
"traditional"
computational
chemistry
(i.e.,
force
field,
semiempirical,
density
theory
methods)
as
it
pertains
data
generation
inorganic
molecular
discovery.
also
discuss
opportunities
leveraging
experimental
sources.
We
how
advances
statistical
modeling,
artificial
intelligence,
multiobjective
optimization,
automation
accelerate
lead
compounds
rules.
overall
objective
this
showcase
bringing
together
from
diverse
areas
computer
science
have
enabled
rapid
uncovering
structure-property
relationships
chemistry.
aim
highlight
unique
considerations
motifs
bonding
(e.g.,
variable
spin
oxidation
state,
strength/nature)
set
them
their
apart
more
commonly
considered
organic
molecules.
uncertainty
relative
scarcity
motivate
specific
developments
machine
learning
representations,
model
training,
Finally,
we
conclude
an
outlook
opportunity
accelerated
complexes.
Science,
Journal Year:
2020,
Volume and Issue:
370(6512), P. 101 - 108
Published: Oct. 2, 2020
Paper
in,
product
out
A
typical
chemist
running
a
known
reaction
will
start
by
finding
the
method
described
in
published
paper.
Mehr
et
al.
report
software
platform
that
uses
natural
language
processing
to
translate
organic
chemistry
literature
directly
into
editable
code,
which
turn
can
be
compiled
drive
automated
synthesis
of
compound
laboratory.
The
procedure
is
intended
universally
applicable
robotic
systems
operating
batch
architecture.
full
process
demonstrated
for
an
analgesic
as
well
common
oxidizing
and
fluorinating
agents.
Science
,
this
issue
p.
101
APL Materials,
Journal Year:
2020,
Volume and Issue:
8(8)
Published: Aug. 1, 2020
We
give
here
a
brief
overview
of
the
use
machine
learning
(ML)
in
our
field,
for
chemists
and
materials
scientists
with
no
experience
these
techniques.
illustrate
workflow
ML
computational
studies
materials,
specific
interest
prediction
properties.
present
concisely
fundamental
ideas
ML,
each
stage
workflow,
we
examples
possibilities
questions
to
be
considered
implementing
ML-based
modeling.
