Journal of Semiconductors,
Journal Year:
2024,
Volume and Issue:
45(12), P. 122302 - 122302
Published: Dec. 1, 2024
Abstract
Simulating
the
human
olfactory
nervous
system
is
one
of
key
issues
in
field
neuromorphic
computing.
Olfactory
neurons
interact
with
gas
molecules,
transmitting
and
storing
odor
information
to
center
brain.
In
order
emulate
complex
functionalities
neurons,
this
study
presents
a
flexible
synapse
transistor
(OST)
based
on
pentacene/C8-BTBT
organic
heterojunction.
By
modulating
interface
between
energy
bands
semiconductor
layers,
device
demonstrates
high
sensitivity
(ppb
level)
memory
function
for
NH
3
sensing.
Typical
synaptic
behaviors
triggered
by
pulses
have
been
successfully
demonstrated,
such
as
inhibitory
postsynaptic
currents
(IPSC),
paired-pulse
depression
(PPD),
long-term
potentiation/depression
(LTP/LTD),
transition
from
short-term
(STD)
(LTD).
Furthermore,
maintains
stable
functions
even
under
different
bending
conditions,
which
can
present
new
insights
possibilities
systems
bio-inspired
electronic
products.
In
modern
computing,
the
Von
Neumann
architecture
faces
challenges
such
as
memory
bottleneck,
hindering
efficient
processing
of
large
datasets
and
concurrent
programs.
Neuromorphic
inspired
by
brain's
architecture,
emerges
a
promising
alternative,
offering
unparalleled
computational
power
while
consuming
less
energy.
Artificial
synaptic
devices
play
crucial
role
in
this
paradigm
shift.
Various
material
systems,
from
organic
to
inorganic,
have
been
explored
for
neuromorphic
devices,
with
materials
attracting
attention
their
excellent
photoelectric
properties,
diverse
choices,
versatile
preparation
methods.
Organic
semiconductors,
particular,
offer
advantages
over
transition-metal
dichalcogenides,
including
ease
flexibility,
making
them
suitable
large-area
films.
This
review
focuses
on
emerging
artificial
based
discussing
different
branches
within
semiconductor
system,
various
fabrication
methods,
device
structure
designs,
applications
synapse.
Critical
considerations
achieving
truly
human-like
dynamic
perception
systems
semiconductors
are
also
outlined,
reflecting
ongoing
evolution
computing.
Ruddlesden–Popper
(RP)
tin
halide
perovskites
(THPs),
exemplified
by
PEA2SnI4,
are
promising
two-dimensional
semiconductors
for
optoelectronic
applications,
yet
their
field-effect
transistors
(FETs)
often
suffer
from
high
operating
voltages
and
stability
issues.
Addressing
these
challenges,
we
developed
a
novel
approach
integrating
ion
gel
dielectrics
composed
of
PVDF-HFP
[EMIM+][TFSI–]
with
achieving
FETs
record-low
as
low
2
V.
Additionally,
substituting
PEA+
BA+
in
BA2SnI4
FETs,
achieve
enhanced
device
stability,
devices
exhibiting
prolonged
functionality
exceeding
100
days.
Uniform
performance
was
also
observed
across
30
randomly
tested
fabricated
on
inch
silicon
wafer.
These
demonstrated
synaptic
behavior
ultralow
energy
consumption
(5
×
10–11
J
per
operation).
Leveraging
advancements,
constructed
artificial
neural
networks
item
classification,
accuracy
(97%).
Moreover,
the
single
training
process
based
ion-gel
gated
is
approximately
orders
magnitude
lower
than
that
similar
NVIDIA
GeForce
RTX
4060
GPU.
Our
results
highlight
potential
low-power,
high-stability
paving
way
next
generation
perovskite-based
electronic
neuromorphic
devices.
Nano Letters,
Journal Year:
2025,
Volume and Issue:
unknown
Published: Feb. 20, 2025
Developing
optoelectronic
synaptic
devices
with
low
power
consumption,
broadband
response,
and
biological
compatibility
is
crucial
to
simulate
the
functions
of
optic
nerve.
Here,
an
organic
synapse
transistor
based
on
C8-BTBT/PMMA/PbS
quantum
dots
(PbS
QDs)
fabricated,
which
has
good
stability,
consumption
(as
as
0.49
fJ
per
event
under
800
nm
near-infrared
optical
pulse),
response
from
ultraviolet
wavelengths.
Based
trap
release
photogenerated
carriers
by
PbS
QDs,
a
series
behaviors
are
simulated
device.
Furthermore,
we
use
artificial
neural
network
model
realize
recognition
facial
feature
image
in
broad
spectral
range;
rate
reached
96.25%
(350
ultraviolet),
92.14%
(580
visible),
90.03%
(800
near-infrared).
This
work
beneficial
for
advancing
development
future
intelligence
vision
sensing.
Advanced Functional Materials,
Journal Year:
2025,
Volume and Issue:
unknown
Published: Feb. 19, 2025
Abstract
Low‐power
and
efficiently
parallel
neuromorphic
computing
is
expected
to
break
the
bottleneck
of
von
Neumann
architecture.
Due
direct
responses
optical
signals,
photonic
synaptic
devices
can
work
as
core
components
artificial
visual
systems,
accelerating
development
neural
computing.
Furthermore,
community
looking
for
effective
coupling
electronic
behaviors
within
an
individual
organic
device
achieve
further
functional
integration.
Photoisomeric
molecules
with
photo‐regulatable
properties
are
facilitate
this
process.
Herein,
photoelectric
transistors
(OPSTs)
constructed
by
introducing
poly(2‐(3′,3′‐dimethyl‐6‐nitrospiro[chromene‐2,2′‐indolin]‐1′‐yl)
ethyl
methacrylate)
(PSPMA)
photoisomeric
groups,
which
effectively
improves
photo‐synaptic
response.
polarization
induction
light‐assisted
charge
trapping
PSPMA,
OPSTs
simulate
typical
significant
conductance
modulation
at
low
voltage
assistance
UV
light.
The
power
consumption
84
aJ
per
event.
Moreover,
mimic
nociceptors,
recognize
handwritten
digits
93.33%
accuracy,
decode
encrypted
information,
demonstrating
potential
applications
in
damage
perception
recognition.
These
findings
will
expand
application
devices,
open
up
new
possibilities
hardware
architectures
synapses.
Advanced Materials,
Journal Year:
2025,
Volume and Issue:
unknown
Published: March 26, 2025
In
the
brain,
both
recording
and
decaying
of
memory
information
following
external
stimulus
spikes
are
fundamental
learning
rules
that
determine
human
behaviors.
The
former
is
essential
to
acquire
new
knowledge
update
database,
while
latter
filters
noise
autorefresh
cache
data
reduce
energy
consumption.
To
execute
these
functions,
brain
relies
on
different
neuromorphic
transmitters
possessing
various
kinetics,
which
can
be
classified
as
nonvolatile
volatile
memory.
Inspired
by
electronic
devices
have
been
employed
realize
artificial
neural
networks
spiking
networks,
respectively,
emerged
tools
in
machine
learning.
Molecular
switches,
capable
responding
electrical,
optical,
electrochemical,
magnetic
stimuli,
display
a
disruptive
potential
for
emulating
storage
devices.
This
Review
highlights
recent
developments
responsive
molecules,
their
interfacing
with
low-dimensional
nanostructures
nanomaterials,
integration
into
By
capitalizing
concepts,
unique
account
neurotransmitter-transfer
based
molecules
ad
hoc
kinetics
provided.
Finally,
future
directions,
challenges,
opportunities
discussed
use
engineer
more
complex
logic
operations
computing
functions
at
hardware
level.
Advanced Functional Materials,
Journal Year:
2025,
Volume and Issue:
unknown
Published: April 17, 2025
Abstract
Neuromorphic
visual
systems
that
mimic
human
vision
can
efficiently
capture
and
process
information
with
an
in‐sensor
computing
paradigm.
However,
photoelectronic
devices
high
light
sensitivity,
stability
tunable
charge‐carrier
interactions
are
still
required
for
constructing
such
systems.
Herein,
a
heterojunction
of
two‐dimensional
(2D)
hydrogen‐substituted
graphdiyne
(HsGDY)
2D
(BA)
2
PbI
4
perovskite
neuromorphic
photoelectric
sensing
processing
is
demonstrated.
The
functions
as
light‐to‐electricity
conversion
layer,
the
HsGDY
forms
layer
well‐matched
energy
bands
to
allow
effective
charge
injection
in
at
zero‐voltage
bias.
unique
sp‐sp
hybridization
alkyne
bonds
thin
film
provide
abundant
interactive
sites
enable
efficient
processing.
device
utilized
environmentally
adaptive
experience
learning.
Excitatory
post‐synaptic
current
(EPSC)
gain
paired‐pulse
facilitation
(PPF)
index
stored
ambient
air
7
weeks
retained
95%
98%
initial
values.
A
array
fabricated,
applicable
perceive
dynamic
information.
These
results
basis
future
intelligent
hardware.
Advanced Functional Materials,
Journal Year:
2024,
Volume and Issue:
unknown
Published: Dec. 8, 2024
Abstract
Single‐mode
sensors
suffer
from
poor
robustness
and
insufficient
data
features
in
facial
expression
recognition,
so
fusing
multi‐sensor
signals
is
the
key
to
improving
accuracy
of
recognition
systems.
Here,
a
biocompatible
capacitive‐electromyographic
dual‐mode
sensor
(CEDS)
presented,
consisting
capacitive
pressure
sensing
unit
dry
electrodes
for
electrophysiological
signal
monitoring,
assembled
3D
stacking
fashion.
A
double‐coupled
microstructure
prepared
electrical
double‐layer
effect
realized
by
doping
ionic
liquid,
which
significantly
improves
performance
sensor.
The
application
effectively
solves
problems
hydrogel
that
are
prone
water
loss
skin
irritation.
Besides,
good
biocompatibility
antimicrobial
properties
CEDS
verified
through
cytotoxicity
bacteriostatic
tests.
Based
on
single
signal,
fatigue
driving
monitoring
system
manipulator
control
constructed
respectively.
By
further
integrating
functions
CEDS,
1D
convolutional
neural
network‐assisted
constructed,
demonstrates
great
potential
systems
based
flexible
technology
practical
applications.
Journal of Physics D Applied Physics,
Journal Year:
2025,
Volume and Issue:
58(13), P. 135110 - 135110
Published: Jan. 24, 2025
Abstract
Artificial
synaptic
devices
are
the
hardware
foundation
of
modern
computing
systems
which
have
shown
great
potential
in
overcoming
bottleneck
traditional
von-Neumann
architectures.
Organic
transistors
garnered
considerable
attention
due
to
their
merits,
such
as
low
cost,
weight,
and
mechanical
flexibility.
Various
materials
utilized
for
charge-capture
layer
organic
transistors.
Indium
gallium
zinc
oxide
(IGZO)
is
a
typical
metal
semiconductor
with
wide
bandgap,
high
carrier
mobility,
stable
characteristics.
Moreover,
IGZO
an
n-type
lower
highest
occupied
molecular
orbital
(HOMO)
lowest
unoccupied
(LUMO)
energy
level
compared
p-type
semiconductor,
has
capture
material
fabricate
high-performance
devices.
However,
application
trapping
received
limited
attention.
Consequently,
transistor
based
on
organic/inorganic
heterojunction
was
developed.
The
impact
program/erase
time
memory
performance
investigated,
revealing
that
window
ratio
increased
write/erase
extended.
Additionally,
behavior
were
successfully
emulated,
including
excitatory/inhibitory
postsynaptic
current,
paired-pulse
facilitation,
depression,
high-pass
filtering
characteristics,
transformation
short-term
plasticity
long-term
plasticity.
Notably,
inorganic–organic
bilayer
achieved
recognition
accuracy
89.2%
using
Modified
National
Institute
Standards
Technology
dataset
handwritten
digit
training.
This
study
provides
facile
route
fabricating
transistors,
paving
way
development
advanced
brain-like
computers.