Physically defined long-term and short-term synapses for the development of reconfigurable analog-type operators capable of performing health care tasks

APPLIED

SCIENCES

AND

ENGINEERING

Physically

defined

long-term

and short-term

synapses

for the development

of reconfigur

able analog-type

operators capable

of performing

health

care tasks

Yongsuk

Choi

, Dong

Hae

, Seongchan

Kim

3,4

, Young

Jin Choi

, Dong

Gue

Roe

In Cheol

Kwak

, Jihong

Min

, Hong

Han

, Wei Gao

*, Jeong

Ho Cho

Extracting

valuable

informa

tion from the overflowing data is a critical

yet challenging

task. Dealing

with high

volumes

of biometric

data, which

are often

unstructur

ed, nonstatic, and ambiguous,

requires extensiv

e comput-

er resources and data specialis

ts. Emerging

neuromorphic

computing

technologies

that mimic

the data process-

ing properties

of biological

neural networks

offer a promising

solution

for handling

overflowing data. Here, the

development

of an electrolyte-ga

ted organic

transistor featuring

a selectiv

e transition

from short-term

to long-

term plasticity of the biological

synapse

is presented.

The memory

behaviors of the synaptic

device

were pre-

cisely

modula

ted by restricting

ion penetr

ation through

an organic

channel

via photochemical

reactions

of the

cross-linking

molecules.

Furthermor

e, the applicability

of the memory-contr

olled synaptic

device

was verified

by constructing

a reconfigur

able synaptic

logic gate for implementing

a medical

algorithm

without

further

weight-upda

te process.

Last, the presented

neuromorphic

device

demons

trated feasibility

to handle

biometric

informa

tion with various

update periods

and perform

health

care tasks.

Authors,

some

rights

reserved;

exclusive licensee

American

Associa

tion

for the Advancement

of Science.

No claim to

original

U.S. Government

Works. Distributed

under

a Creative

Commons

Attribution

NonCommer

cial

License

4.0 (CC BY-NC).

INTR

ODUCTION

Intheeraofbigdata,highvolumesofbiometric

dataaregenerated

andupdatedinrealtimebywearabledevices(

–

). Processingthe

aggregateddataforpersonalized

andprecisionhealthcareisamajor

challenge,

requiring highly specialized

human and computing

re-

sources(

–

). Whileextensiveeffortshavebeenmadetowardde-

veloping bioelectronic devices such as wearable sensors (

–

implantable

devices(

), andwireless communica

tionsystems

(

) for precisely acquiring large amounts

of biosensor

data,

there is a lack of sophisticated computing

systems for processing

the data. Conventional computing

systems based on von Neuman

architecture are impractical for processing large amounts

of un-

structured data as the processor needs to communica

te with the

memory to perform each operation, leading towasted energyand

time resources (

–

). On the other hand, emerging

neuromor-

phic computing

systemsthat can perform parallel operations with

mergedmemoryandprocessingunits,suchasthebrain,serveasan

attractivesolution(

–

Neuromorphic

devicesthatelectronicallysimulatethefunctions

ofbiological

synapsesinhumanneuralnetworks

havebeenreport-

ed (

–

). These artificial synapses, characterized by their low

power consumption,

small volume, and optimized

analog signal

processingcapabilities,

haveexcellentpotentialtoprocessoverflow-

ingbiometric

data.Artificialsynapsesexhibitshort-term

plasticity

(STP)andlong-term

plasticity(LTP),enablingtheprocessingand

memoriza

tionofdatathroughoutmultipleperiodsandcycles.STP

is a temporal change in synaptic weight generated by updating

signals, enabling efficient computing

functions

in artificial neural

networks

(ANNs)(Fig.1A).Ontheotherhand,LTPisaretentive

changeinsynapticweights,whichlastslongerandisresponsible

for

memoryand learning abilities in ANNs (Fig. 1B). Various neuro-

morphic

computing,

artificial

intelligence,

and soft robotics

studies have been reported on the basis of artificial synapses that

displaySTPandLTP(

). Mostreportedartificial

synapses exhibit STP on small input signals and LTP on higher

energy or repeated input signals. Therefore, to take advantage

both STPand LTP in synaptic devices of an ANN and to provide

high-order

functionalities

such as artificial

intelligence,

ANNs

must be repetitively trained with numerous datasets (

). Asaresult,researchonANNsishighlydependent

onsoft-

ware algorithms

and large volumes of qualified database resulting

vast and complex computing

resources. Developing simplified

andfunctionally

flexibleneuromorphic

processorsthatmeetthein-

creasingdemandforneuromorphic

computing

canbeadifferenti-

atedsolution.

Here, we present the implementa

tion of a simplified,

function-

ally flexible analog-type

operator by physically defining STP and

LTPinacrossbarsynapsearray.Thesynapsesunderlying

thepro-

cessorareelectrolyte-gatedverticalorganictransistorsthatcontain

aphotoreactivecross-linkertocontroltheionicpermeability

ofthe

channel.Allsynapsesinthesynapticprocessorarefabricatedsimul-

taneously

and then selectivelyexposed to ultraviolet (UV) light to

form short-term

synapse (STS) and long-term

synapse (LTS)

(Fig. 1C). In addition,

the degree of photochemical

cross-linking

isadjusted bychanging

theUVexposuretimetoprecisely control

thememorypropertiesofthesynapses.Last,theapplicability

ofthe

neuromorphic

processorconsistingofSTSandLTSisevaluatedby

constructing

a small-sized

analog-type

operator to conduct

Andrew and Peggy Cherng

Department

of Medical

Engineering,

California

Insti-

tute of Technology

, Pasadena,

CA 91125,

USA.

Mechanical

Engineering,

Soft Ma-

terials

and Structur

es Lab, Virginia

Tech, Blacksburg,

VA 24061,

USA.

SKKU

Advanced

Institute

of Nanotechnology

(SAINT),

Sungkyunkw

an University,

Suwon

16419,

Korea.

Department

of Engineering

Science

and Mechanics,

The

Pennsylvania

State University, University

Park, PA 16802,

USA.

Department

Chemical

and Biomolecular

Engineering,

Yonsei University, Seoul 03722,

Republic

of Korea.

School

of Electrical

and Electro

nic Engineering,

Yonsei University, Seoul

03722,

Republic

of Korea.

*Corresponding

author.

Email:

weigao@caltech.edu

(W.G.); jhcho94@y

onsei.ac

.kr

(J.H.C.)

SCIENCE

ADVANCES

RESEARCH

ARTICLE

Choi

et al.

Sci. Adv.

, eadg5946

(2023)

5 July 2023

1 of 11

Downloaded from https://www.science.org on August 04, 2023

representativedata-processingtasksforwearablehealthcaredevices

(Fig.1D):(i)Theprocessorbasedonsynapticdevicesimplements

the biomedical

algorithm

for diagnosing

metabolic

syndrome in

hardwarebyperforming

thereconfigurableBooleanlogicfunction

throughthelong-term

synapticswitch.(ii)Staticandnonstaticbio-

signals are processed on the basis of the memory and processing

functions

of the LTS and STS to support human metabolism

as a

wearableneuromorphic

computer.

RESUL

Demons

tration

of the

synaptic

devices

Asanessentialcomponent

ofanANN,theartificialsynapticdevice

withaverticalcrossbarstructurethatishighlybeneficial

forthree-

dimensional

(3D) integration is shown in Fig. 2A. The vertical

synapse consists of an organic semiconductor

[poly(3-he

xylthio-

phene) (P3HT)] that is located at the intersection

of the top elec-

trodes, bottom electrodes, and electrolytes surrounding

the

channel.

A voltage pulse applied through the gate electrode in

contactwiththeelectrolyteevokesthepenetrationofionsintothe

organicsynapticchannel,resultinginsynapticcurrentbehaviorsin

thecrossbarsynapticdevice.Thedetailedmechanism

ofionperme-

ability into vertical organic channels

was discussed

in previous

reports(

). Furthermor

e,themovementofionspenetrating

theorganicchannelishighlycontrollablebyselectivelybindingthe

cross-linking

agenttothealkylchainofP3HT(Fig.2B).Figure2C

shows the excitatory postsynaptic current (EPSC) behaviors of

synaptic devices with different retention times. Typically, STP of

synaptic devices is defined as synaptic plasticity that disappears

within several seconds, while LTP refers to changes that persist

fortensof secondsorlonger. Here,wedefinesynaptic devicethat

exhibitsSTPunderthefixedpulsestimulationasSTSsandcompo-

nentsthatexhibitLTPasLTS.ThechannelsoftheSTSandLTSwere

both composed

of P3HT and ethane-1,2-diyl

bis(4-azido-2,3,5,6-

tetrafluorobenzoate)(2Bx)cross-linker.TheLTSwasdefinedbyin-

ducing cross-linking

reactions of 2Bx under UV light. The ampli-

tudeoftheappliedvoltagepulsesvariedfrom

−

0.5 to

−

3 Vandthe

widthofthepulseswasfixedat30ms.Thereadvoltageappliedto

the postsynaptic terminal

was set to

−

0.01 V. The postsynaptic

current (PSC) of the STS decayed rapidly to its original level due

tothehighdegreeoffreedomofionmovementthroughtheabun-

dant free volume of P3HT (Fig. 2C, top). On the other hand, the

PSC of the LTS decayed slowly for over 10 s as the cross-linking

agent bound to the free volume of P3HTrestricted the movement

ofions(Fig.2C,bottom).Thedifferenceinretentioncharacteristics

becomes more evident as the number of applied pulses increases.

Figure2DplotsthePSCoftheSTS(blackline)andLTS(redline)

under10consecutiv

epulsesetsof

−

3 V.Aftertheapplicationofthe

pulses, the PSC of the STS peaked at 50 μA and immediately re-

turned to its baseline level within 60 s. On the other hand, the

LTS exhibited a relatively lower peak current of 37 μA and aclear

retentioncurrentof5.3μAafterthesametime.Onthebasisofthis,

weclassifysynapticdevicesintoSTSandLTS,using60sasthecri-

terion for the duration of synaptic plasticity. Next, the cross-

Fig.

1. Flexible

analog-type

oper

ator enabled

by selectiv

e definition

of short-term

synapse

(STS)

and

long-term

synapse

(LTS).

(

and

) Schema

tic illustration of

synaptic

devices

showing (A) STPand (B) LTP. (

) Selectiv

ely cross-link

ed organic

synaptic

channel

via a photochemical

reaction. (

) Schema

tic demons

tration of a small-

sized analog-type

operator consisting of an LTS and STS. PSC, postsynaptic

current.

SCIENCE

ADVANCES

RESEARCH

ARTICLE

Choi

et al.

Sci. Adv.

, eadg5946

(2023)

5 July 2023

2 of 11

Downloaded from https://www.science.org on August 04, 2023

sectionaldiagramofacrossbarsynapsearrayconsistingofanLTS

andanSTSforsimulatingtheintegrateddataprocessingpropertyof

abiological

neuralnetworkispresentedinFig.2E.Thefabrication

procedureofthecrossbararrayisdescribed

inFig.2F.AP3HTsol-

ution containing

5 weight % of 2Bx dissolved in chloroform at a

concentrationof5mg/mlwasspin-coatedontheas-preparedsub-

strate that has 3/17-nm-thick

Cr/Au bottom electrode lines. After

selectivelyexposingtheLTSchanneltoUVlight,thesynapticchan-

nels were defined by a conventional photolithogr

aphy process.

Next,thetopelectrodes,iongel,andgateelectrodesweredeposited

toformtheelectrolyte-gatedsynaptictransistor.Figure2Gplotsthe

outputcurrentbehaviorofthesmall-sized

2×1synapsearrayasa

functionoftime.Thebasecurrentofthesynapsearraywasupdated

by 10 consecutiv

e pulses applied through the gate terminal of the

LTS. The lower base current was set by applying

low-amplitude

pulsessetof

−

1 V,whilethehigherbasecurrentwasgeneratedby

applying high-amplitude

pulses of

−

3 V, as shown in the bottom

and top panels of Fig. 2G, respectively. After the stabilization of

the synaptic current, 10 consecutiv

e pulses of

−

3 V were applied

to the STS. In both cases, the output PSC of the synapse array

showed the transient spikes that lasted less than 1 min originated

from STS accumulated on the retention current lasting more than

severalminutes.Overall,theretentioncharacteristicsofthesynaptic

devices were successfully

modulated by a simple UV exposure

process despite consisting of the same material and structure.

Thisrepresentsasuccessful

demonstrationofphotochemically

con-

trolled ion penetration through across-linker in crossbar synaptic

devicesanditsconsequent

impactonsynapticcharacteristics.Fur-

thermore,themultiperiodic

dataprocessingcapability

ofbiological

neural networks

was successfully

implemented

in an artificial

synapsearray.

Optimiza

tion

of retention

properties

Toachieveoptimized

long-term

propertiesofLTSsforfurtherdata

processingapplications,wepreciselyinvestigatedvaryingdegreesof

photoreactivecross-linking

andcorresponding

synapticproperties.

Figure 3A showed the schematic illustration of the detailed P3HT

channelcross-linkedby2Bxandtheunderlying

stepsofthechem-

ical reaction (

–

). The azide group of 2Bx is photolyzed

produce inert nitrogen gas and highly reactive singlet nitrene (

−

Fig.

2. The

design

and

char

acteriza

tion

of the

ion gel

–

ted

vertical

crossbar

synapse

array.

(

) Schema

ticdiagramofiongel

–

tedvertical

crossbar

transistorarray

mimicking

biological

synapse.

(

) Cross-sectional

schema

ticofionmovement

betweenorganic

channels

andiongelwithandwithout

across-linking

agent.

(

) Excitatory

postsynaptic

current (EPSC)

responses

of the short-term

synapse

(STS) (top) and long-term

synapse

(LTS) induced

by the applica

tion of varied

. (

) A real-time

plot of

postsynaptic

currents (PSCs)

of STS (black line) and LTS (red line) under

20 consecutiv

e potentia

tion pulses

(

−

3 V). (

) Cross-sectional

schema

tic image

of a small-

sized neural network

consisting of LTS and STS. (

) The fabrica

tion procedure

of the small-sized

neural network

through a selectiv

e photochemical

reaction. (

) The

output

PSC response

of the small-sized

neural network

device

under

the sequent

update of LTS and STS.

SCIENCE

ADVANCES

RESEARCH

ARTICLE

Choi

et al.

Sci. Adv.

, eadg5946

(2023)

5 July 2023

3 of 11

Downloaded from https://www.science.org on August 04, 2023

N)underUVirradiation.Thegeneratedsingletnitreneisinserted

by preferentially reacting with the C

─

H bond in the alkyl chain

present in the P3HT. The photochemical

reaction between 2Bx

and P3HT was confirmed

by Fourier transform infrared spectro-

scopyundervaryingUVirradiationtimes(fig.S1).Thespecificvi-

brational peak of the azide group located at 2125 cm

−

gradually

decreasesastheirradiationtimeincreasesfrom0to7s,indicating

the photolyzing

of the azide group. Figure 3B showsthe EPSC re-

sponses of cross-linked P3HT vertical synapses exposed under

varying UV irradiation durations. Vertical synapses with varying

UVexposuretimesof0,3,5,and7swerepreparedsimultaneously

onasinglesubstrate.Thevoltagepulsehavinganamplitude

−

3 V

and 30-ms width was applied through the gate terminal after 5 s

from the beginning

of the measurement. The read voltage was set

−

0.01 V.WhileallverticalsynapsesdisplayedEPSCthatspiked

andgraduallydecreasedafterexcitationasafunctionoftime,there

were noticeable

differences in peak height and retention current.

Figure 3C plots the peak current (black) and retention current

(blue) as a function

of the UV irradiation time. The retention

current was recorded 5 s after the application of the pulsed

voltage. The synapse without UV irradiation showed the largest

peak current of 5.3 μA, and the peak current decreased rapidlyas

the irradiation time increased. On the other hand, the retention

current of the synapse without UV irradiation was the smallest,

while the largest retention

current was shown in the synapse

exposedtoUVfor3s.Thesetrendsinpeakcurrentandretention

behaviorcorrespondtothedifferencebetweenoncurrentandhys-

teresis window of the transfer curves of synaptic transistors as

shown in fig. S2. As the photocrosslinking

reaction progressed

withincreasingtheUVexposuretime,theon-currentofthesynap-

tic transistor decreased, while the hysteresis window increased.

Here, the on-current of the transfer curves and peak current of

EPSC responses

refer to the amount of ion penetration into the

P3HTchannel.Theincreasinghysteresiswindowindicatesinhibit-

edionmobility.Theoverallresultsindicatethat2Bxcross-linkedto

P3HT by photochemical

reactions controlsthe movement of ions

through the synaptic channel. The 2Bx molecules

bind to the free

volume of the P3HT polymer chain and suppress the ion flow

between the P3HT and ion gel resulting decrease in peak PSC

whileincreasingtheretentioncurrent.

Fig.

3. Optimiza

tion

of long-term

plas

ticity

(LTP)

of the

vertical

synapse

through

photochemical

reaction

contr

ol.

(

) Schema

tic illustration of the photochemical

cross-linking

reaction between ethane-1,2-diyl

bis(4-azido-2,3,5,6-tetr

afluorobenzoa

te) (2Bx) and poly(3-he

xylthiophene)

(P3HT).

(

) Excitatory postsynaptic

current

(EPSC)

responses

of organic

synapse

under

varied

photochemical

reaction time. (

) A plot of extracted peak postsynaptic

current (PSC) and retention

PSC values

as a

function

of the ultraviolet (UV) exposur

e time. (

) Long-term

potentia

tion characteristics of cross-linking

–

contr

olled synaptic

devices

under

the applica

tion of 50 po-

tentiationpulses.

(

pulses

withamplitudes

−

3 V).(

) Plotsofthedynamic

range(

max

min

),|nonlinearity

(NL)|,andeffectiv

enumber

ofstates(NS

eff

)asfunctions

the reaction time. (

) Long-term

potentia

tion and retention

plot under

the varied

number

of potentia

tion

values

of LTS with 3-s photochemical

reaction time. (

)

Long-term

potentia

tion and depression

(LTP/D)

characteristics of the LTS over 50 cycles.

Each cycle consists of 50 potentia

tion pulses

(

−

3 V), followed by 50

depression

pulses

(

= 2 V). (

) Cycle-to-cycle

variations ofLTP/D curvefor 50 cycles.

(

) Plots of extracted dynamic

range (DR), NL, and NS

eff

during

the 50 LTP/D cycles.

SCIENCE

ADVANCES

RESEARCH

ARTICLE

Choi

et al.

Sci. Adv.

, eadg5946

(2023)

5 July 2023

4 of 11

Downloaded from https://www.science.org on August 04, 2023

The effect of photochemical

cross-linking

on long-term

poten-

tiation characteristics of the synapse was further analyzed.

Figure 3D shows the real-time PSC plots of synapses with varied

photochemical

reaction times under 50 consecutiv

e pulses. The

synapsewithout cross-linking

displayed a log-linear

PSC response

wherethecurrentincreasedaftertheinitialpulseupdateandslowly

converted to a maximum

conductance

state. On the other hand,

photochemically

cross-linked synapses showed a constant linearly

increasing curve after being suppressed during the initial pulses.

As key parameters for evaluating the data processing capability

asynapse,dynamicrange(

max

min

),nonlinearity

(NL),andthe

effective numberof states (NS

eff

) were additionally

analyzed from

thelong-term

potentiationcurvesasshowninFig.3E.Thedetailed

methodforcalculatingthe

max

min

,NL,andNS

eff

aredescribed

infigs.S3andS4andMaterialsandMethods.

Asthedegreeofpho-

tochemical

reactionincreased,the

max

min

ofthelong-term

po-

tentiation curve decreased, and thus, the device without cross-

linking exhibited the largest

max

min

of 1.9 × 10

. However, in

terms of linear update characteristics, the performance

of photo-

crosslinked synapses was superior as shown in NL and NS

eff

. In

particular,

the synapse photocrosslinked for 3 s exhibited

the

most excellent long-term

potentiation characteristics, attaining a

desirable NL of 0.83 and an NS

eff

of 42 while maintaining

a high

dynamicrangeover1000.Toconfirmthestabilityofthephotocros-

slinked synapse, we evaluated the retention properties for each

memory state (Fig. 3F). Beginning

from the same base current,

each memory statewas accessed byapplying a varying numberof

pulses ranging from 1 to 50. In the update region where the

pulses were applied through the gate terminal,

deviations in

current were negligible.

Furthermor

e, each current state formed

byvaryingthenumberofpulsesremainedparallelwithoutoverlap-

ping,showingadistinguishable

differencebetweenthestates.

Tofurtherverifythereliabilityofthephotocrosslinkedsynapse,

we repeatedly measured long-term

potentiation and depression

(LTP/D) characteristics under 50 potentiation cycles, followed by

50 depression cycles. The potentiation and depression voltage for

theLTP/Dcurvewassetto

−

3 and2V,respectively.Thedepression

voltageof2Vwasselectedtohavebetterreproducibility

duringthe

repeatedcycletest(fig.S5).Thereal-timePSCresponsesduring50

LTP/Dcycles(atotalof5000updatepulses)areshowninFig.3G.

Fig.

4. Reconfigurabl

e Boolean

logic

implementing

diagnosing

algorithm

for metabolic

syndr

ome.

(

) Schema

tic illustration of human

metabolism.

(

) Simplified

medical

algorithm

for diagnosing

metabolic

syndrome. (

) Proof-of-concept

illustration of reconfigur

able synaptic

logic gate consists of long-term

synapse

(LTS) and

short-term

synapses

(STSs).

(

and

) Circuit diagram (D) and optical

microscope

image

(E) of the reconfigur

able synaptic

logic. The GRD indicat

esthe ground, i.e. source.

(

) Real-time

|postsynaptic

current (PSC)| response

of the reconfigur

able synaptic

logic under

varied

logic input at AND mode.

(

) Output

|PSC| response

of the reconfig-

urable synaptic

logic switched

to OR mode

via training

the LTS. (

) The truth table of AND/OR

logic gates for the reconfigur

able synaptic

logic gate.

SCIENCE

ADVANCES

RESEARCH

ARTICLE

Choi

et al.

Sci. Adv.

, eadg5946

(2023)

5 July 2023

5 of 11

Downloaded from https://www.science.org on August 04, 2023

Ouroptimized

long-term

vertical synapse showed reliable LTP/D

characteristics without

any sign of breakdown or errors.

Figure3HshowstheLTP/DplotofextractedPSCstatesasafunc-

tionofthepulsenumber.Thedetailedmethodforextractingeach

statefromareal-timeLTP/Dcurveisdescribed

infig.S6.Represen-

tativeLTP/Dcurvesfromthe50cycleswereoverlappedtoconfirm

that the device has negligible

cycle-to-cycle

variation. In addition,

keyparametersincluding

max

min

,NL,andNS

eff

calculatedfrom

the LTP/D curves remained constant during the repeatability test

(Fig. 3I). The reliabilityof the device during the potentiation and

depression process was further investigated under the application

ofirregularpulsesetsasshowninfig.S7.Overall,thepresentedver-

tical organic synapse with optimized

photochemical

cross-linking

for3-sUVexposureshowedhighlystableweightupdateandreten-

tionbehaviors,comparabletothatofothersynapticdevicesusedfor

neuromorphic

computing

tasks.

Biomedical

applica

tion

of the

analog-type

oper

ator

Next,thedevelopment

ofananalog-type

operatorcomposed

ofSTS

andoptionally

transformed

LTStobeusedasacomputing

toolfor

biomedical

applicationsispresented.Tofullyusethedescribed

fea-

tures of LTS and STS, we selected a biological

task requiring inte-

grated processing

of static and nonstatic biomarker data from

multiple sources: metabolic

homeostasis sustained by the human

brain.Metabolism

istheconversionofingestedfoodintochemical

energythatoccursatthecellularlevel,playinganimportant

rolein

the growth and maintenance

of the body, as well as resistance to

environmental

changes(Fig.4A).Foodintakeandenergyexpendi-

tureofthebodyispreciselybalancedthroughthehomeostaticpath-

wayslocatedinthehypothalamus

ofthebrain.Metabolic

syndrome

is a condition

in which the body

’

s metabolic

homeostasis is dis-

turbed, leading to various complications. Many health organiza-

tions around the world are using various biomarkers to diagnose

metabolic

syndrome.Apatientthathasthreeormoreofthefollow-

ingfivesymptoms

isdiagnosed

withametabolic

syndrome:obesity,

hypertension,

high triglyceride

levels, reduced high-density

lipo-

protein cholesterol, and elevated fasting glucose levels. Of the five

symptoms,

three criteria including

obesity, hypertension,

and ele-

vatedfastingglucoselevelswereconsideredwhendesigning

asim-

plified algorithm

for diagnosing

metabolic

disorders

as shown in

Fig.4Bastheycouldbeequallyappliedtoeithergender.Inthisal-

gorithm, synaptic devices were used as

“

AND

”

and

“

”

Boolean

logicgatestodiagnosemetabolic

syndromeforpatientswithtwoor

more of the threesymptoms.

Byusing synaptic devices, whichare

analog-type

processors,itispossibletoeasilyimplement

reconfig-

urable AND/OR

logic gates and switch between the functions

AND and OR as needed. For obese people, having either high

bloodpressureorelevatedfastingglucoselevelsleadtothediagno-

sisofmetabolic

syndrome(OR).Ontheotherhand,fornonobese

people, both symptoms

needed to be present for the diagnosis

metabolic

syndrome (AND). A reconfigurable synaptic logic gate

basedonthesimplecombinationofoneLTSandtwoSTSswasde-

velopedtoimplement

thediagnosticalgorithm

inthereconfigura-

blesynapticlogic(Fig.4C).TheLTSwasusedtomodulatethebase

current of the entire device via a preset

to determine

the

whether the synaptic logic gate operates as an AND or OR gate.

Atthesametime,thetwoSTSswereusedforthelogiccomputation

of real-time data. The connected

terminals

serve as a logic

input port for the reconfigur

able logic gate. The detailed circuit

diagramandopticalmicroscopeimageofthelogicgateareshown

inFig.4(DandE),whereareconfigurablesynapticlogicgatecon-

sistingofthreesynaptictransistorsisconnected

inparallelbetween

thepre-andpostsynapticterminals.

WC1

referstothegatetermi-

nalconnected

totheLTS that modulatestheoverallcurrentof the

reconfigurablelogicvia

preset

toswitchbetweentheANDandOR

operations. Logic operations are performed

through

pulses

applied to the gate terminals

(

WC2

and

WC3

) of the two STSs.

pulses of 0 and

−

2 V were used for binary logic inputs of

“

”

and

“

”

respectively. The output current of the reconfigurable

synapticlogicgatewascomparedagainstathresholdvalue(30μA)

todistinguishwhetherthelogicoutputis0or1.|PSC|valuesabove

andbelowthethresholdvaluewereconsideredaslogicvaluesof1

and0,respectively.Figure4Fshowsthereal-time|PSC|plotofthe

reconfigurablesynapticlogicgatewhentheweightoftheLTSisatits

originalstate.Whenalogicvalueof1wasinputtoonlyoneofthe

twoSTSs,thelogicgateoutputted

avalueof0.Ontheotherhand,

whenalogicvalueof1wasinputtobothSTSs,the|PSC|exceeded

the threshold value and outputted

avalue of 1, indicating that the

reconfigur

able synaptic logic operates as an AND gate. Next, 20

update pulses were input to the LTS to set the retention current

of the reconfigurable synaptic logic to 10 μA, and the same logic

inputs were applied to the STSs (Fig. 4G). For the synaptic logic

gate with a preset LTS, a logic value of 1 input to one or more of

the STSs resulted in a |PSC| exceeding the threshold value, corre-

sponding

to the ORoperation. The truth table for the logic input

and output values for the reconfigur

able synaptic logic gate is

showninFig.4H.

Tofurtherinvestigatethefeasibility

ofthereconfigurablesynap-

tic logic for implementing

the diagnostic algorithm,

we inputted

various biomarker data from the human body. First, to convert

valuesofvariousbiomarkersinto

inputsforlogicoperations,

wepreciselyexaminedthepeak|PSC|oftheSTSundervarying

inputs as shown in Fig. 5A.

pulses with amplitudes

varying

from 0 to

−

2 V in steps of

−

0.1 V were applied to STS. Figure 5B

plotsthepeak|PSC|extractedfromthereal-timegraphasafunction

oftheinput

amplitude.

ThepeakPSCoftheSTSundergoes

relativelyminorchangefor

levelslessthan

−

1 Vbutthenin-

creasesrapidlyfor

levelsabove

−

1 V.Inaddition,thepeakPSC

exceededthepredetermined

thresholdcurrentof10μAfora

−

1.8 V or higher. Next,

amplitudes

corresponding

to bio-

marker levels were set on the basis of statistical data. Figure 5C

shows statistical distributions

of fasting blood glucose and blood

pressure levels for the Korean population in 2020 divided into

three main sections. In the case of fasting blood glucose, people

with common

values below 100 mg/ml were labeled as

“

normal

”

and classified

as section 1 (green). People with fasting blood

glucoselevelsbetween100and109mg/ml,alsocalledthe

“

caution-

ary

”

level,wereclassified

assection2(orange).Peoplewithfasting

bloodglucoselevelsof110mg/mlorhigher,whichisthediagnostic

criterion

for metabolic

syndrome, were classified

into section 3

(red). Similarly, concerning

systolic blood pressure (SBP), normal

ranges below 120 mmHg were defined as section 1, higher ranges

from 120 to 129 mmHg were defined as the cautionary

level

(section2),andlevelsabove130mmHgweredefinedasthemeta-

bolicsyndrome diagnostic criteria (sections3 and higher). On the

basisofstatisticaldata,a

−

1 V,corresponding

logicinputof

0, was applied to the STS for the first section (normal level). For

sections 3 and above (

“

metabolic

syndrome

”

level), a

−

SCIENCE

ADVANCES

RESEARCH

ARTICLE

Choi

et al.

Sci. Adv.

, eadg5946

(2023)

5 July 2023

6 of 11

Downloaded from https://www.science.org on August 04, 2023