J. Neu al Eng. 21 (2024) 046003 h ps://doi.o g/10.1088/1741-2552/ad593d
Jou nal o Neu al Enginee ing
OPEN ACCESS
RECEIVED
6 Feb ua y 2024
REVISED
9 June 2024
ACCEPTED FOR PUBLICATION
17 June 2024
PUBLISHED
2 July 2024
O iginal con en om
his wo k may be used
unde he e ms o he
C ea i e Commons
A ibu ion 4.0 licence.
Any u he dis ibu ion
o his wo k mus
main ain a ibu ion o
he au ho (s) and he i le
o he wo k, jou nal
ci a ion and DOI.
PAPER
Coupling o pho o ol aics wi h neu os imula ion
elec odes—op ical o elec oly ic ansduc ion
Ma ie Jakeˇ
so á1, Ondˇ
ej Kuno sk´
y1, Im ich Gablech1, Dion Khodagholy2, Jenni e Gelinas3,4
and E ic Daniel Głowacki1,∗
1Bioelec onics Ma e ials and De ices Labo a o y, Cen al Eu opean Ins i u e o Technology CEITEC, B no Uni e si y o Technology,
Pu kyˇ
no a 123, 61200 B no, Czech Republic
2Depa men o Elec ical Enginee ing, Columbia Uni e si y, New Yo k, NY 10027, Uni ed S a es o Ame ica
3Depa men o Biomedical Enginee ing, Columbia Uni e si y, New Yo k, NY 10027, Uni ed S a es o Ame ica
4Depa men o Neu ology, Columbia Uni e si y, New Yo k, NY 10032, Uni ed S a es o Ame ica
∗Au ho o whom any co espondence should be add essed.
E-mail: glowacki@ u b .cz
Keywo ds: bioelec onics, neu os imula ion, pho o ol aics, wi eless powe ans e , mic oelec odes
Abs ac
Objec i e. The wi eless ans e o powe o d i ing implan able neu al s imula ion de ices has
ga ne ed signi ican a en ion in he bioelec onics ield. This s udy explo es he po en ial o
pho o ol aic (PV) powe ans e , u ilizing issue-pene a ing deep- ed ligh —a no el and
p omising app oach ha has ecei ed less a en ion compa ed o adi ional induc ion o
ul asound echniques. Ou objec i e is o c i ically assess key pa ame e s o di ec ly powe ing
neu os imula ion elec odes wi h PVs, con e ing ligh impulses in o neu os imula ion cu en s.
App oach. We sys ema ically in es iga e a ying PV cell size, op ional se ies con igu a ions, and
coupling wi h mic oelec odes ab ica ed om a ange o ma e ials such as P , TiN, I Ox, Ti, W,
P Ox, Au, o poly(3,4 e hylenedioxy hiophene):poly(s y ene sul ona e). Addi ionally, wo ypes o
PVs, ul a hin o ganic PVs and monoc ys alline silicon PVs, a e compa ed. These combina ions
a e employed o d i e pai s o elec odes wi h di e en sizes and impedances. The eadou me hod
in ol es measu ing elec oly ic cu en using a s aigh o wa d ampli ie ci cui . Main esul s.
Op imal PV selec ion is c ucial, necessi a ing su icien ly la ge PV cells o gene a e he desi ed
pho ocu en . A anging PVs in se ies is essen ial o p oduce he app op ia e ol age o d i ing
cu en ac oss elec ode/elec oly e impedances. By ca e ully choosing he PV a angemen and
elec ode ype, i becomes possible o emula e elec ical s imula ion p o ocols in e ms o cha ge
and equency. An impo an conside a ion is whe he he ci cui is pho o ol age-limi ed o
pho ocu en -limi ed. High cha ge-injec ion capaci y elec odes made om pseudo- a adaic
ma e ials impose a pho ocu en limi , while mo e capaci i e ma e ials like P a e
pho o ol age-limi ed. Al hough o ganic PVs exhibi lowe e iciency han silicon PVs, in many
p ac ical scena ios, s imula ion cu en is p ima ily limi ed by he elec odes a he han he PV
d i e , leading o po en ial pa i y be ween he wo ypes. Signi icance. This s udy p o ides a
ounda ional guide o designing a PV-powe ed neu os imula ion ci cui . The insigh s gained a e
applicable o bo h in i o and in i o applica ions, o e ing a esou ce o he neu al enginee ing
communi y.
© 2024 The Au ho (s). Published by IOP Publishing L d
J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
1. In oduc ion
1.1. Backg ound on implan able pho o ol aics
(PVs) in he con ex o wi eless powe
ansmission me hods
Implan able neu os imula ion de ices a e a ubiqui -
ous ool o in i o neu oscience esea ch and a c i -
ical componen in bioelec onic medicine [1–3]. In
all hese applica ions, minia u iza ion and wi eless
powe ing a e in high demand [4–6]. This is impo -
an o make nex -gene a ion bioelec onics medi-
cine de ices su gically less in asi e. Mo eo e , o
basic esea ch using small animal models, de ice oo -
p in is o en a se ious limi a ion, he e o e ind-
ing e icien ways o accomplish wi eless elec ical
neu os imula ion can also ansla e o no el esea ch
expe imen s in neu oscience [7]. Many elec omag-
ne ic induc ion p o ocols exis o wi elessly ans-
mi powe o implan able de ices [5,8,9]. These
solu ions, howe e , su e om low powe ans e
e iciency and geome ic size/posi ion cons ain s o
bo h sending and ecei ing an ennas. A leas some
le el o analog ci cui y is always necessa y on he
ecei e side o con e he incoming elec omag-
ne ic signal in o a cha ge-balanced, biphasic elec ical
s imula ion pulse. The e o e, he de ice mus con-
sis , a minimum, o ecei ing coil, elec onic ci cui
componen s like ec i ie s, and inally he s imula-
ion elec odes in e aced wi h he a ge issue. A
p omising al e na i e echnique is ul asound powe
ans e , which, pe uni a ea, o en ep esen s a
mo e e icien al e na i e o induc i e powe ans-
e , bu wi h he necessi y o close con ac o he
ansmi e on he su ace o he skin [10,11]. These
de ices equi e a piezoelec ic ansduce o con e
acous ic ene gy in o elec ical signals. Un o una ely,
mos o he e icien piezoelec ic ma e ials a e oxic
compounds, c ea ing a ba ie o p ac ical adop ion.
A combina ion o esonan magne ic powe ans-
e and piezoelec ics is he magne oelec ic powe
ans e concep elabo a ed by Singe e al [12].
These app oaches show an imp essi e sh inking o
de ice oo p in , ne e heless he piezoelec ic ma e -
ial emains an issue, and ela i ely la ge and ine i-
cien ansmi ing coils mus be used. Op ical powe
ans e using ligh wa eleng hs in he issue anspa -
ency window ep esen s an eme ging concep which
has been explo ed ela i ely less han induc i e o
ul asound powe ans e . This is likely because in u-
i i ely anyone can obse e ha mos biological issues
a e highly opaque due o abso p ion and sca e ing.
Howe e , he e a e egions o he isible and nea -
in a ed spec um whe e local maxima o ligh ans-
mi ance exis . These a e some imes e e ed o as is-
sue anspa ency windows, whe e abso p ion om
biomolecules is la gely absen , and sca e ing is also
minimized. The i s ‘window’ is be ween oughly
630–900 nm, and he second is be ween 1000 and
1350 nm [13,14]. Wa eleng hs in hese egions can
pene a e in o issue o some deg ee. Se e al examples
o implan able PVs ha e been shown o e ec i ely and
sa ely ope a e a dep hs o a ew mm below he skin,
e en down o a ound 10 mm, wi h ed o in a ed
ligh being deli e ed om ou side o he body [15,
16]. Since PVs a e diodes, he simple connec ion
o a PV o wo elec odes sho ed by an elec oly e
al eady ac s as a ec i ie ci cui ha will ansduce
a squa e pulse o ligh (basically ligh on/o ), in o a
biphasic cu en pulse wi h a high deg ee o cha ge
balance jus om he na u e o his simple ci cui
[17,18]. I he s imula ion elec odes a e in eg a ed
di ec ly on o he PV i sel as hin- ilms, his ep es-
en s a e y simple and compac sys em wi h compe -
i i e cha ge pe uni a ea (o pe uni olume) ha
exceeds he a o emen ioned echnologies [16]. The
concep o PV neu os imula ion has been deployed
in a ious ways o e he pas wo decades. The mos
p ominen a e examples o PV de ices o op oelec-
onic s imula ion o he e ina, which mus be seen
as a special use case. He e, ligh ing ess is i ial due
o he pupil, and ed/in a ed ligh is no absolu ely
equi ed [19–21]. In hese applica ions, a ocused
ligh beam is used o ac ua e a PV pixel which in
u n locally s imula es he e ina. The low sca e -
ing in he eye i eous allows lase exci a ion o be
highly collima ed and hus i is possible o as e
scan ac oss an implan o p oduce spa ially-p ecise
s imula ion. These de ices comp ise a PV cell con-
nec ed o a p ima y and e u n s imula ion elec-
ode. In he sys em epo ed by Palanke e al, sil-
icon PV cells (a se ies o 2 o 3 o inc ease he ol age)
a e monoli hically in eg a ed wi h s imula ion/ e u n
mic oelec odes [19,22,23]. These PV s imula o s
ha e been es ed in animal expe imen s and he e a e
p omising esul s om clinical ials [24]. Ghezzi e al
[20,25] and Rand e al [17] ha e epo ed o ganic
PV analogs, success ully es ing hese de ices in i o
and in i o.
While he eye p esen s a kind o ‘ideal’ scena io
o p ecise ligh -based s imula ion, e e y o he kind
o implan able a ge (i.e. b ain, pe iphe al ne e)
in ol es anscu aneous ope a ion and will be aced
wi h highe loss o ansmi ed ligh powe , and also
ligh sca e ing in he in e ening issues. This ac
necessi a es he use o wa eleng hs in he ed/nea
in a ed egion, o mo e e ec i ely pene a e ha d
and so issues. Thin-con o mable PVs ha e been
shown o ope a e as co ical s imula o s, ac ua ed
ansc anially in mice [26], and se e al examples
o ansde mal PV s imula o s ha e been published
[27], ope a ing down o a dep h o oughly 1 cm.
The e ha e been ecen epo s o using PV a ays
o echa ge implan able pacemake de ices, e en ia
ha es ing ambien ligh [28,29]. The e a e wo
app oaches o each deepe s imula ion a ge s. The
i s is o c ea e leads be ween he s imula ion elec-
odes and he PV, which can be implan ed in a shal-
lowe egion close o he skin in e ace [30]. The
2
J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
second is o use implan able ibe op ics o channel
he ligh o he egion o in e es [15].
1.2. Semiconduc o ma e ials o implan able PVs
Th ee ca ego ies o ligh -abso bing semiconduc o
ma e ials ha e been explo ed o implan able PVs:
Silicon, ino ganic compound semiconduc o s, and
o ganic semiconduc o s. The dis inc ions be ween
hese di e en PV ypes a e impo an , as each
app oach has ad an ages and disad an ages ha will
a ec which ype o PV is he bes o a gi en applica-
ion. The basic igu es o me i o conside a e powe
con e sion e iciency and quan um e iciency. Silicon
PVs ha e 2–10 imes highe powe con e sion e i-
ciency and quan um e iciency han OPVs, depend-
ing on he pa icula o ganic o silicon de ice con-
side ed. Meanwhile, op imized compound ino gan-
ics like GaAs can ha e e en highe powe con e -
sion and quan um e iciency numbe s, highe han
silicon by se e al pe cen . In e ms o e iciency,
GaAs >Si ≫O ganics. The nex conside a ion is
abso p ion coe icien , which will dic a e he p ac-
ical hickness o he semiconduc o laye o e i-
cien ly abso b ligh . O ganic semiconduc o s ypic-
ally excel in his ca ego y, ha ing abso p ion coe i-
cien s o e an o de o magni ude highe han Silicon,
and on pa o sligh ly highe han he bes ino -
ganic compound ma e ials. The e o e, i minimiza-
ion o hickness and olume is impo an o a gi en
applica ion, compound semiconduc o s o o ganics
will ou pe o m silicon. Silicon app oaches bene i
om a ma u e and highly-accessible mic o ab ica-
ion pa hway, and he oppo uni y o use di e en
comme cial silicon wa e s o hin- ilm silicon om
chemical apo deposi ion me hods. The main d aw-
backs o silicon a e mechanical igidi y, low op ical
abso p ion coe icien , and necessi y o high-quali y
passi a ion o p e en su ace ecombina ion losses
and hus pho o ol age d ops. Finally, he e is he
issue o s abili y. Silicon needs ca e ul passi a ion
and encapsula ion o emain s able o co osion in
physiological en i onmen [31]. On he o he hand,
his can be an ad an age. Silicon can dissol e in he
body in a ela i ely ha mless way, p omp ing demon-
s a ions o ansien , dissol able, implan able powe
sou ces [31]. Recen ly, he g oup o Bozhi Tian has
published a high-pe o mance and e sa ile silicon
pla o m o his ype [32]. OPVs su e om lowe
e iciency, ye can be made ex emely hin (sub-
100 nm), on accoun o he high abso bance coe -
icien , and a e in insically mechanically so and
lexible. Compa ibili y wi h hin- ilm polyme ic sub-
s a es like pa ylene-c o polyimide is ensu ed by
low- empe a u e p ocessibili y o OPVs. Mo eo e ,
some o ganic semiconduc o s appea o be e y
s able in di ec con ac wi h physiological medium,
wi hou any passi a ion. Thus, o applica ions whe e
con o mal applica ion is desi ed, such as a ne e
cu o co ical su ace elec ode, hin and lexible
de ices based on o ganic semiconduc o s can be
a o able. While compound ino ganic semiconduc -
o s can be e y hin and o highes pe o mance, hey
p esen p ocessing di icul ies. GaAs and i s de i a -
i es a e p oduced ia specialized and expensi e epi-
axial deposi ion p ocesses equi ing oxic gases and
high empe a u es. The p esence o po en ially oxic
a senic is a ba ie o adop ion, hough i has been
epo ed ha leeching o a senic om he implan is
wi hin sa e limi s [33]. To-da e, he smalles - oo p in
de ices in e ms o olume and hickness a e made
o GaAs [15] o o ganic semiconduc o s [16,34].
In e ms o op imized high-pe o mance compound
ino ganic PV s acks, Sahin and colleagues ha e pub-
lished a se ies o pape s on loa ing ligh ac i a ed
mic o-elec ical s imula o s (FLAMESs), comp ising
op imized he e ojunc ions o GaAs, gi ing pho o-
ol age up o 0.7 V [15,35,36]. The FLAMES de ices
we e on he size o de o se e al hund ed mic ome -
e s, and could achie e wi eless in aspinal s imula-
ion in he a spinal co d. La ge a ays o GaAs cells
we e ecen ly demons a ed in a ansde mal powe
ans e sys em o wi eless powe ing o implan -
able elec onics. He e, se ies and pa allel connec ions
we e used o gi e a max o 2.4 V and o e all se -
e al mic owa s o powe [8]. On he OPV side, ou
esea ch g oup has epo ed e icacious o ganic s im-
ula ion de ices implan ed up o 15 mm below skin
and so issue/bone o he s imula ion o pe iphe al
ne es o he co ical su ace [16,26,30]. These la e
de ices a e based on he minimalis ic o ganic pho-
ocapaci o design, whe e he PV diode is i sel he
s imula ion elec ode, he whole de ice is ul a lex-
ible and has a hickness o unde 5 µm [34]. Fe lau o
e al ha e de eloped lexible and oldable OPV-based
s imula o s [20], which p esen a su gical ad an -
age o being implan able h ough a small incision
and hen un olded in he a ge a ea. O e all, o ganic
semiconduc o s ha e good indica ions om com-
bined biocompa ibili y and non oxici y, while being
s able and ine in he body wi hou need o ex ens-
i e passi a ion which would be equi ed o ino ganic
semiconduc o s. Howe e , he ques ion o long- e m
eliabili y and sa e y o implan able PVs emains an
open one ha mus be igo ously con on ed o allow
p og ess in his ield.
I is no ewo hy ha in pa allel o PV-d i en
neu os imula o de ices, he e is a g owing body o
esea ch desc ibing mic o o nanopa icle in e aces
which, when exci ed by ligh , can deli e s imula-
ion o cells and issues wi h a a ie y o mechan-
isms, including o en an in e play o pho o he mal
and pho oelec ic e ec s [37–45]. Such mic oscaled
ma e ials can o m single-cell in e aces. These ypes
o pa icle-based pho os imula ion concep s hold
g ea p omise in pushing he limi s o minia u iza ion
and ob aining he leas su gically-in asi e solu ion.
Ideas o un e he ed and injec able pa icles ha e been
p oposed. The a ious s imula ion mechanisms ha e
3
J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
been e iewed in de ail o some ex en o mic o/n-
anopa icles [42,46–50]. A ac ion o he epo ed
pa icles ope a e ia a PV mechanism. Fo hose ha
do, he indings discussed in his pape should in
p inciple apply as well. To pe o m op imal ex acel-
lula s imula ion, a mic opa icle should ha e a ca h-
ode and anode componen o op imize pho ocha ge
densi y and spa ially sepa a e he cha ges o maxim-
ize he esul an elec ic ield. In his way, he de ice
ope a es acco ding o he measu emen s laid ou in
his wo k.
This wo k conce ns di ec PV s imula ion, whe e
a PV diode is connec ed o a s imula ion neu al in e -
ace. To da e, he quan i y o success ul implan ed PV
s imula ion demons a ions is ela i ely small, and
a p ima y eason is ha specialis s om he op o-
elec onics/PV ield do no o en o e lap wi h he
neu al in e ace de ice ield. We seek o b idge his
gap and we ha e pe o med his s udy o map and
explain he pa ame e s needed when combining a PV
d i e wi h neu os imula ion elec odes ( igu e 1(a)).
In his wo k, we ha e endea o ed o gi e a oadmap
o designing PV neu os imula ion de ices, gi ing
a en ion o c i ical pa ame e s: s imula ion elec ode
ype, elec ode impedance, PV d i e ype, se ial/pa -
allel connec ions, and op ical pulse in ensi y and e-
quency. Wi h hese de ails de e mined, i is possible
o design a PV neu os imula ion de ice ailo ed o a
speci ic applica ion.
1.3. Measu ing he PV-d i en neu os imula ion
ci cui
An illumina ed PV is a powe sou ce, beha ing
nei he as an ideal cu en no an ideal ol age sou ce.
The cu en d awn om he PV will depend on he
amoun o pho ocu en gene a ed by he cell, as
well as he impedance o he load ha is a ached o
he cell. The maximum pho o ol age p oduced by a
single PV is limi ed by he modynamics, speci ically
by he pa ame e o bandgap o he semiconduc o
ma e ial comp ising he cell [51]. O he losses, like
cha ge ecombina ion and se ies esis ance will p ac-
ically limi cell ol age. Gene ally, a single-junc ion Si
PV can p o ide a ound 0.5 V a open-ci cui . OPVs,
depending on he ma e ials used, gene a e be ween
0.4 and 0.8 V [52]. The OPV es ed in his wo k is
made om a he e ojunc ion o ph halocyanine and
pe ylene e aca boxylicdiimide de i a i es [53], a
combina ion we ha e op imized, and which yields an
open-ci cui ol age o 0.5 V [54]. The e o e, he wo
PV ypes compa ed in his wo k a e essen ially equi-
alen in e ms o open-ci cui ol age. The OPV and
Si PV used in his wo k a e schema ized in igu e 1(b).
We used comme cial monoc ys alline silicon sola
cell wa e s, wi h con ac lines and passi a ion al eady
deposi ed. These we e diced o yield di e en chip
sizes. The OPVs we e ab ica ed a he desi ed size
di ec ly in-house. A ays o (mic o)elec odes we e
p epa ed on glass o silicon wa e s, using pho oli ho-
g aphy echniques ( igu e 1(c)). The cha ac e iza ion
o PV-d i en neu os imula ion elec odes was pe -
o med by connec ing he e minals o he PV wi h
a gi en combina ion o elec odes which a e b idged
by elec oly e, and measu ing he cu en ha lows
h ough his ci cui upon applica ion o a pulse o ed
ligh ( igu e 1(d)). This se up allows e alua ion o di -
e en pe mu a ions o PV d i e s wi h di e en elec-
odes. The o e all con igu a ion can be ega ded as a
PV diode sho -ci cui ed by a load consis ing o wo
elec oly ic con ac s and he in e ening elec oly e
( igu e 1(e)). The elec oly ic con ac s can be unde -
s ood in e ms o he Randles model, o a capaci -
i e componen in pa allel wi h a esis i e compon-
en . The esis i e componen s co espond o a adaic
eac ions, while he capaci ance will be gi en by he
cha ge injec ion capaci y o he elec ode ma e ial. An
illumina ed PV will ac as a powe sou ce, capable o
gene a ing a pho o ol age and a pho ocu en . The
magni ude o cu en ha lows o e he whole ci -
cui will depend on he impedance and cha ge injec-
ion capaci y o he elec ode/elec oly e in e aces.
The undamen al limi a ion o PV neu os imula o
app oachesis ha i ischallenging oob ainhigh d i -
ing ol ages—a single PN junc ion, in p ac ice, can
a ely p oduce mo e han >1 V.
2. Ma e ials and me hods
2.1. Fab ica ion o OPVs
Glass wa e s (500 ±20 µm) we e coa ed wi h a
base laye o 2 µm pa ylene-C g own by CVD (SCS
Labco e PDS 2010). Then, a s ack o 1 nm Pd, 9 nm
Au, 30 nm Ti, 50 nm Au, 10 nm Ti we e deposi ed
ia magne on spu e ing (Bes ec GmbH). The i s
pho oli hog aphy s ep de ined he shape o he PV
bo om elec ode, anode/ca hode leads, and con ac
pads. AZ 1518 pho o esis spin-coa ed a 4000 pm
was exposed h ough a soda lime mask using a S¨
USS
Mic oTec MA8 mask aligne equipped wi h an i-line
il e . The esis was de eloped in AZ 400 K de elope
dilu ed 1:4 in deionized wa e (DI). The me al laye s
we e e ched in a KI/I2(Au, Pd) and HF/H2O2/H2O
(Ti) e ch mix u es. The esis was s ipped using ace -
one. Using he same p ocessing, he second pho o-
li hog aphy s ep was used o emo e he op Ti/Au/Ti
laye s o yield a de ined, semi- anspa en PV bo om
elec ode a ea. Nex , he o ganic PN laye was depos-
i ed h ough a shadow mask. Me al ee ph halocyan-
ine (H2Pc, Al a Aesa ) and N,N′-dime hyl-3,4,9,10-
pe ylene e aca boxylic diimide (PTCDI, BASF) we e
pu i ied by h ee old empe a u e-g adien sublima-
ion. Laye s o 40 nm H2Pc and 50 nm PTCDI we e
he mally e apo a ed om esis i elyhea edc ucibles
(Edwa ds 306, <2×10−6To , a es o 1–6 Å s−1).
The OPV op elec ode was a 70 nm laye o Ti, which
was deposi ed h ough a shadow mask using an E-
beam e apo a o (Bes ec GmbH, <1×10−7To , a a
4
J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
Figu e 1. (a) Concep ual illus a ion o a possible use case o an implan able pho o ol aic-d i en neu os imula o , whe e
issue-pene a ing ed ligh eaches a subde mally placed PV, which con e s he ligh impulse in o an elec ical cu en ha is
ans e ed o a pai o s imula ion elec odes. These elec odes can be in e aced wi h a pe iphe al ne e ia a cu elec ode (1),
o a deep b ain a ge by a dep h p obe (2). (b) C oss-sec ional diag ams o he he e ojunc ion OPV used in his wo k, and a
monoc ys alline Si PV. OPV is less e icien , bu is ≈100 imes hinne and highly lexible. (c) Tes pa e n o di e en ly-sized
elec odes, deposi ed on glass wa e . These allow c ea ion o a ‘d y’ con ac o an elec ode pad using a mic op obe needle, and
using d ops o elec oly e o b idge a pai o es ed s imula ion elec odes. (d) Schema ic o he expe imen al se up. A gi en PV
d i e is connec ed ia mic op obe needles o a pai o s imula ion elec odes, which a e in phospha e-bu e ed saline solu ion. A
ansimpedance ampli ie (TIA) is in se ies wi h his ci cui , and con e s he measu ed cu en in o an ou pu ol age, which is
ed in o a high- esolu ion digi al oscilloscope (osc). The oscilloscope is igge ed by an LED pulse gene a o , which in u n
p o ides con olled cu en o an LED ha ou pu s he op ical impulse o he PV (660 nm ligh was used). (e) Simpli ied
equi alen ci cui diag am o he PV d i ing a pai o s imula ion elec odes. PVs can be connec ed in se ies, o sum up highe
pho o ol age, o in pa allel, o sum up o highe pho ocu en . The neu os imula ion ci cui is ep esen ed by he se ies esis ance
o de ice leads (RS) and elec oly e (RE), and he wo mic oelec odes, µE, which a e ega ded as Randles ci cui s (pa allel
capaci ance, CµEand a adaic cha ge- ans e esis ance, RµE).
a e o 5 Å s−1). The de ices we e encapsula ed wi h a
2µm pa ylene-C laye wi h 3-( ime hoxysilyl)p opyl
me hac yla e (A-174) p esen in he CVD chambe as
an adhesion p omo e . The hi d pho oli hog aphy
s ep was used o expose he con ac pads. A hicke
laye o AZ 1518 was spin coa ed a 1000 pm o se e
as an e ch mask du ing eac i e ion e ching (RIE,
Ox o d Ins umen s PlasmaP o 80, 200 W, 50 sccm
O2, 100 mTo ) o he pa ylene-C laye . The esis was
s ipped in ace one and he op 10 nm Ti laye was we
e ched o expose Au-coa ed con ac pads.
2.2. Fab ica ion o elec ode a ays
Glass o Si wa e s wi h a he mally g own SiO2laye
(525 ±25 µm and 2.6 µm, espec i ely) we e used as
subs a es. In all cases, he elec ode a ea, leads, and
con ac pads we e composed o E-beam e apo a ed
20 nm o Ti and 50 nm Au (and 30 nm Ti o i anium
elec ode a ays) deposi ed on o O2plasma ac i a ed
subs a es (Diene NANO Plasma Cleane ). The lay-
ou was pa e ned wi h AZ 1518 (4000 pm) and we
e ching as speci ied in OPV ab ica ion sec ion. Nex ,
he su ace was ac i a ed by O2plasma (Diene ) and
a 2 µm pa ylene-C encapsula ion laye was deposi ed
using he A-174 adhesion p omo e . The con ac pads
we e exposed h ough a hick AZ 1518 (1000 pm)
e ch mask using RIE (200 W, 50 sccm O2, 100 mTo ).
In case o he Au and Ti elec ode a ays, he elec-
ode a eas we e opened di ec ly a e using he hi d
li hog aphy mask. All o he elec ode ma e ials we e
5

J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
pa e ned by he pa ylene peel-o echnique. Fo
ha pu pose, an an i-adhesi e laye o 2% Mic o
90 soap (In e na ional P oduc s) was spin coa ed a
1000 pm and le o ai d y be o e deposi ion o
he sac i icial 2 µm pa ylene-C laye . The elec ode
a ea was opened h ough an AZ 12XT (3000 pm,
AZ 326 MIF de elope ) hick e ch mask using RIE
(200 W, 50 sccm O2, 100 mTo ). The esidual es-
is was s ipped in ace one. Be o e deposi ion o he
elec ode ma e ial, he subs a es we e O2plasma
ac i a ed. W was deposi ed by magne on spu e -
ing (100 nm, Bes ec GmbH), P and TiN by ion beam
spu e ing (100 nm wi h 5 nm Ti s icking laye , Bes ec
GmbH) [55], I Oxand P Oxby eac i e magne on
spu e ing (240 nm, [56]). The sac i icial pa ylene-
C was ca e ully peeled o unde DI. The poly(3,4-
e hylenedioxy hiophene):poly(s y ene sul ona e)
(PEDOT:PSS) laye was deposi ed by spin coa ing.
The solu ion was p epa ed eshly be o e depos-
i ion and con ained PEDOT:PSS (PH1000, He aeus
GmbH), 5 w % e hylene glycol (EG), 0.1 w % 4-
dodecylbenzenesul onic acid (DBSA) and 1 w %
(3-glycidyloxyp opyl) ime hoxysilane (GOPS). The
mix u e o PH1000, EG and DBSA was sonica ed o
45 min o ensu e homogenei y, GOPS was added
jus be o e use and sonica ed o 2 min. PEDOT:PSS
il e ed h ough a 0.45 µm poly inylidene luo ide
memb ane il e was spincoa ed in h ee laye s (3000,
750 and 750 pm) wi h a 40 s bake a 90 ◦C a e each
laye . A e he las coa , he sac i icial pa ylene-C
laye was peeled o yielding he mic oelec odes. The
wa e was inally annealed a 140 ◦C o 1 h and hen
washed in DI o emo e he esidual soap and o he
chemicals.
2.3. P epa a ion o Si PV chips
125 mm ×125 mm monoc ys alline silicon PVs,
a ed a 2.8 W, we e ob ained om CincoSola
(China). 16 a ia ions o silicon PV cells we e p e-
pa ed by moun ing hem on alumina ce amic sub-
s a es (Elce am, Czech Republic). To achie e in e -
connec ion o 1, 2, o 3 se ies o 2 pa allel combina-
ions o silicon PV cells wi h ac i e su ace a ea o (1.0,
4.5, 25, and 100) mm2, alumina subs a es wi h sil e
leads and pads elying on hick- ilm echnology we e
employed. The silicon PV cells we e o iginally manu-
ac u ed wi h Al me alliza ion ha is al eady pa ially
oxidized upon deli e y. The e o e, a ha d mask was
used o expose only he Al me alliza ion o he chip
and RIE was pe o med o emo e aluminum oxides
on bo h sides. Subsequen ly, 100 nm o Au we e e ap-
o a ed h ough a shadow mask (Bes ec GmbH) o
minimize he isk o undesi ed con ac esis ance. In
he nex s ep, he PV cells we e co e ed wi h pos-
i i e pho o esis and cu in o single chips using a
dicing saw (ESEC 8003). The las s ep o silicon PV
cell ab ica ion in ol ed emo al o he pho o esis
be o e moun ing on he alumina subs a e. Such p e-
pa ed PV cells we e glued o alumina wi h sil e -
illed epoxy glue (Epo- ek H21D, Epoxy echnology,
USA) om he bo om side p o iding s ong mech-
anical adhesion and low-ohmic elec ical con ac ,
while he op side o he PV cells was connec ed o
he Ag pad on alumina using 25 µm Au wi e wi h
wedge–wedge he mosonic wi e-bonding echnique
(TPT HB 16).
2.4. Elec ochemical and op oelec ochemical
cha ac e iza ion
The i s expe imen was ocused on elec ical imped-
ancespec oscopy measu emen so ab ica ed mic o-
elec odes om Ti, TiN, P , Au, W, I Ox, PEDOT:PSS
and P Oxin a domain o equencies anging om
100 kHz o 1 Hz a a ol age ampli ude o 10 mV. Ou
op oelec ical de ices we e hen u ilized in cus om-
made docking s a ion c ea ed by 3D p in ing me hod
om polylac ic acid. Alumina wi h silicon PV cells
was glued o he sample holde and inse ed in o a
s a ion equipped wi h pogo-pins o elec ical con-
ac and ed LED a ay o sample i adia ion wi h
wa eleng h o ≈656–660 nm peak. In ensi y was cal-
ib a ed using a Tho labs SM1PD1A pho odiode. The
LED was d i en using a Tho labs DC2200, using he
in e nal pulse gene a o o igge he oscilloscope.
The unc ionali y o p epa ed de ices was e i ied ia
J(V) cha ac e is ic measu emen s in a ange om 0 o
2 V a di e en con inuous ed ligh i adia ion powe
densi ies o (0, 0.38, 0.62, 0.86 and 1.20) mW mm−2.
The inal expe imen was ca ied ou by op oelec-
ochemical cha ac e isa ion o PV cells connec ed o
mic oelec odes using mic op obe s. Dynamic meas-
u emen s o pho ocu en we e ansduced o ol age
using a ansimpedance ampli ie (FEMTO GmbH,
DLPCA-200, gain se o 1000) and cap u ed by oscil-
loscope (Picoscope 4262).
3. Resul s
3.1. Key cha ac e is ics o he PV and s imula ion
elec odes measu ed independen ly
The mos common me hod o PV cha ac e iza ion
is a J(V) sweep, whe e cu en densi y (J) is meas-
u ed while applying di e en ol ages o he cell.
Rep esen a i e J(V) cu es o Si PVs a e plo ed in
igu e 2(a), showing he dependance on ligh in ens-
i y and he se ial connec ion o 2 and 3 PVs o boos
he ol age. J(V) plo s can be used as a guide o es im-
a e how much cha ge a PV can deli e o e a gi en
s imula ion ci cui , and o choose PV size and se ies
a angemen app op ia ely. High-impedance loads,
such as small mic oelec odes, will equi e se ies con-
nec ions o p o ide su icien pho o ol age o deli e
6
J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
Figu e 2. (a) Cu en densi y (Vol age), J(V), cha ac e is ics o silicon PVs: single PVs and 2-PV and 3-PV se ies connec ions, a
di e en ligh in ensi ies (660 nm) ep esen a i e o hose ansmi able h ough se e al mm o skin and so issue. The do ed
lines ep esen di e en esis i e loads, concep ually showing he example o low-impedance loads esul ing in a
pho ocu en -limi and high-impedance loads imposing a pho o ol age limi . (b) Bode plo s o absolu e impedance |Z| as a
unc ion o equency o he 8 elec ode ma e ials es ed in his s udy, wi h esul s o 105µm2a ea elec odes shown. N=12
samples, ±SD. (c) Impedance bode plo s o |Z| o PEDOT:PSS elec odes o di e en sizes N=12 samples, ±SD.
he cu en /cha ge necessa y o a gi en neu al a -
ge . To ep oduce p ac ically- ele an scena ios in
neu al enginee ing applica ions, we used eigh di -
e en neu al in e ace elec ode ma e ials, which
gi e a ange o impedances and cha ge injec ion
mechanisms (e.g. capaci i e e sus a adaic e sus
pseudo-capaci i e). A compa ison o absolu e imped-
ance alues, |Z|, as a unc ion o equency, is shown
in igu e 2(b). The lowes impedance alues a e gen-
e ally p o ided by he high cha ge-injec ion capaci y
ma e ials PEDOT:PSS, I Ox, and P Ox(nanopo ous
P ). The o me wo ma e ials suppo bulk olu-
me ic capaci ance due o highly e e sible a adaic
eac ions, while he la e P Oxis simply a e y high
su ace-a ea ma e ial. O he ma e ials, like P and
TiN which a e commonly used in clinical implan s,
ha e mode a e impedance alues. Ti has he highes
impedance, due o i s na i e TiOxlaye . Elec ode
size-dependence impedance plo s a e shown o he
o e all bes -pe o ming ma e ial in e ms o low
impedance: PEDOT:PSS ( igu e 2(c)).
3.2. Pho ocu en measu ed om
PV +s imula ion elec ode combina ions
Di e en PV d i e s we e wi ed oge he wi h s im-
ula ion elec ode pai s as shown in igu e 1(d) and
de ailed in sec ion 2.4. As a s a ing poin , we conside
a single 4.5 mm2Si PV, illumina ed wi h 0.1 o 1 ms
impulses o ligh wi h an in ensi y o 1.2 mW mm−2,
and we a y he size o he s imula ion elec odes. The
7
J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
Figu e 3. (a) Pho ocu en ansien s egis e ed o a 1 ms ligh pulse on 105µm2elec odes (co esponds o da a poin s in g ey
box o panel (b)). (b) In eg a ed cha ge deli e ed by a 1 ms ligh pulse inciden on a 4.5 mm2PV as a unc ion o elec ode size
and ma e ial (c) in eg a ed cha ge deli e ed o e 0.1 ms.
measu edcu en ansien sand in eg a edcha gea e
shown in igu e 3. S a k di e ences in cu en an-
sien s occu due o he ma e ial ype ( igu e 3(a)).
The ela i ely low ol age a o ded by PVs (com-
pa ed o a ypical neu os imula ion cu en isola o
ha can ha e compliance ol age o se e al ol s o
mo e) means ha o mos ypical me allic elec ode
ma e ials (Ti, TiN, P , W, Au) he cu en ansien s
a e essen ially pu ely double-laye capaci i e, wi h
minimal a adaic componen . The pseudo-capaci i e
ma e ials I Ox, PEDOT:PSS, and P Ox, all ha e high
capaci ance alues and hus can accommoda e much
mo e cha ge a a gi en ol age. Fo his eason, he
o aldeli e ed cha ges can be10 imeshighe o hese
ma e ials ( igu e 3(b)). Fo sho pulse imes, 0.1 ms,
he ad an age o he pseudo-capaci i e ma e ials is
no as g ea ( igu e 3(c)). I is clea , howe e , ha o
a single PV whe e he o al a ailable d i ing ol age
will no exceed 0.5 V, choosing as low-impedance
elec ode ma e ials as possible is c i ical o building
a p ac ical s imula o .
In he nex se o p esen ed expe imen s, we a y
Si PV size (1 mm2 e sus 4.5 mm2) and e alua e
he e ec o se ies connec ion o 2 o 3 cells, as
well as a pa allel connec ion o wo disc e e cells
( igu e 4). Ligh in ensi y and pulse leng h a e held
cons an . Se e al clea ends eme ge. Adding se ies
connec ion inc eases he amoun o deli e ed cha ge
only in he case o ol age-limi ed ci cui s, ha
is hose comp ising small elec odes and/o highe
impedance ma e ials. Fo ma e ials like PEDOT:PSS,
using a double o iple se ies only helps o elec-
odes <104µm2. La ge PV size boos s cha ge o
low-impedance elec odes, whe eas i he ci cui is
ol age-limi ed inc easing he PV size will no es-
ul in any mo e deli e ed cha ge. This is c i ical, as
inc easing PV de ice size can be was e ul and only
makes sense i impedance o he s imula ion ci cui is
8
J. Neu al Eng. 21 (2024) 046003 M Jakeˇ
so á e al
Figu e 4. To al measu ed cha ge in nC deli e ed o e a 1 ms pulse a 1.2 mW mm−2. Panels (a)–(d) co espond o a 1 mm2PV in
single PV, 2-se ies, 3-se ies, and 2-pa allel con igu a ion, espec i ely. The diode diag ams in he inse o he panels show he PV
con igu a ion. Panels (e)–(h) a e measu ed wi h 4.5 mm2PVs, in single PV, 2-se ies, 3-se ies, and 2-pa ellel con igu a ion,
espec i ely.
low enough. This p inciple also applies when consid-
e ing con e sion e iciency o PVs: highe e iciency
may no be o signi ican bene i i he ci cui is
pho o ol age-limi ed.
Wi h hese p inciples o pho o ol age and pho-
ocu en limi a ions clea ly demons a ed, we nex
mo ed o compa e Si PVs wi h OPVs (all PVs had
he same a ea =4.5 mm2), as well as he e ec o
ligh pulse equency, using PEDOT:PSS and P as
wo ep esen a i e elec ode ma e ials. Bo h ypes o
PVs gi e he same open-ci cui ol age. In he case
o high-impedance elec odes, (small PEDOT:PSS o
all bu he la ges P elec odes) OPV and Si PV
deli e he same cha ge ( igu es 5(a) and (b)). Adding
se ies connec ions o ei he ype o PV ollows he
same gene al end as es ablished p e iously, mo e
cha ge can be loaded on o highe -impedance elec-
odes ( igu es 5(c) and (d)). The ad an age p o ided
by he mo e e icien Si PVs becomes appa en only
in cases o la ge s imula ion elec odes, whe e he
o al cha ge will become limi ed by he pho ocu -
en gene a ion e iciency. We es ed pulse equen-
cies be ween 1 and 60 Hz ( igu es 5(e) and ( )). In
gene al, he e is always some decline in he cha ge
deli e ed in each 1 ms pulse as equency inc eases.
This is due o he issue o PV diodes discha ging in
he da k be ween ligh pulses. I he nex pulse a i es
be o e he ci cui is ully discha ged, he subsequen
o al cha ging ol age will be lowe . This e ec is
mild in Si PV-d i en ci cui s, while in OPV he e-
quency oll-o in cu en ampli ude is signi ican ly
la ge . This is caused by he ac ha OPVs ha e
e y low cha ge mobili y in he da k, he e o e he
OPVs do no discha ge quickly. Si has high mobil-
i y and cha ges can ecombine inside he de ice ela -
i ely apidly, ensu ing ha he whole ci cui can dis-
cha ge be o e he nex pulse a i es. In case o he
la ge PEDOT:PSS elec odes, we can also obse e a
equency oll-o also wi h Si PV, as his is imposed
by he slow discha ge kine ics o he PEDOT:PSS
elec ode. The issue o discha ge kine ics causing
equency oll-o can be esol ed by enginee ing a
shun esis o in pa allel wi h he PV diode, a solu-
ion sugges ed by Palanke e al o PV-d i en e inal
s imula o s [19]. Explo ing highe - equency ope a-
ion (>1 kHz) may be uniquely ele an in he case o
PV s imula o s which ha e an in insic ol age limi -
a ion. The use o high- equency pulse ains, whe e
nume ous sub h eshold-ampli ude pulses a i e a
he cell in sho succession, can esul in he sum-
ma ion e ec , causing e ec i e depola iza ion o cell
memb ane despi e he ac ha each indi idual pulse
has oo low ampli ude o depola ize cells [57,58].
This kind o high- equency app oach has no been
used in PV s imula ion o ou knowledge, bu would
be in e es ing o y.
9