Academic Edi o s: F ancesco Della
Co e and An onio Di Ba olomeo
Recei ed: 14 Ma ch 2025
Re ised: 4 May 2025
Accep ed: 8 May 2025
Published: 10 May 2025
Ci a ion: Las o icka-Medin, G.;
K ambe ge , G.; K oll, J.; Reba z, M.
Impac o he Gua d Rings on
Sel -Induced Signal and Leakage
Cu en in T ench-Isola ed Low Gain
A alanche Diodes. Senso s 2025,25,
3006. h ps://doi.o g/10.3390/
s25103006
Copy igh : © 2025 by he au ho s.
Licensee MDPI, Basel, Swi ze land.
This a icle is an open access a icle
dis ibu ed unde he e ms and
condi ions o he C ea i e Commons
A ibu ion (CC BY) license
(h ps://c ea i ecommons.o g/
licenses/by/4.0/).
A icle
Impac o he Gua d Rings on Sel -Induced Signal and Leakage
Cu en in T ench-Isola ed Low Gain A alanche Diodes
Go dana Las o icka-Medin 1,* , G ego K ambe ge 2, Ji i K oll 3and Ma eusz Reba z 4
1Facul y o Na u al Sciences and Ma hema ics, Uni e si y o Mon eneg o, Dzo dza Vashing ona,
81000 Podgo ica, Mon eneg o
2Joze S e an Ins i u e, Jamo a Ces a 39, 1000 Ljubljana, Slo enia; g ego .k ambe [email p o ec ed]
3Ins i u e o Physics, Academy o Sciences o he Czech Republic, Na Slo ance 2,
18221 P ague, Czech Republic; [email p o ec ed]
4ELI Beamlines Facili y, The Ex eme Ligh In as uc u e ERIC, Za Radnicí 835,
25241 Dolní Bˇ ežany, Czech Republic; [email p o ec ed]
*Co espondence: go [email p o ec ed]
Abs ac : In his con ibu ion, we explo ed he in e play o gua d ing (GR) con igu a ion
and isola ion s uc u es, as well as i adia ion e ec s, which all oge he c ea e a ich
landscape o phenomena such as sel -induced signals (“ghos s”) in ench-isola ed Low-
Gain A alanche Diodes (TI-LGADs). The ghos e ec is ela ed o he inc eased su ace
cu en due o p esence o SiO
2
enches (and de ec s) in s udied diodes, bu i is also
a ec ed by in e play be ween he gua d ing(s) and he n
+
bias ing, implan ed in in e -
pixel egion o hese de ices. In double- enched senso s, he n
+
bias ing is inse ed in
be ween he wo enches. We p esen he in es iga ion on he ole o hese s uc u es on he
sel -induced signals in ench-isola ed senso s om wo di e en p oduc ions (RD50 and
AIDAinno a). The senso s om he i s p oduc ion ha e mul iple gua d ings, whe eas
he second ype o de ices ea u e only one. De ailed examina ion o he ghos e ec and
leak cu en was pe o med when gua d ings we e le loa ing o connec ed o he pixels
(b ough o he same po en ial). The esul s show ha gua d ing con igu a ion in enched
senso s can be c i ical o he leak cu en and he p esence o a ghos signal. To ou bes
knowledge, he la e p oblem has no been in es iga ed ye .
Keywo ds: LGAD; in e -pixel egion; ench isola ion; sel -induced signals; gua d ing;
bias ing; IV cha ac e is ics; TCT echnique
1. In oduc ion
A Low-Gain A alanche De ec o (LGAD) is an n-on-p diode (n
++
/p
+
/p/p
++
s uc u e)
wi h mode a e in e nal gain enabled by a hin (~2
µ
m) highly doped p
+
laye inse ed
benea h he n
++
laye . These de ices we e designed as minimum ionizing pa icle ime de-
ec o s (MIP TD) and high-g anula i y ime de ec o s (HGDT) o Compac Muon Solenoid
(CMS) and ATLAS expe imen s [
1
]. A he physical edge o he s anda d segmen ed LGAD,
he e a e gua d ings (GRs). These s uc u es ha e he ask o g ading he ol age om he
senso edge o he i s ead-ou pad, held a i ual g ound by he ead-ou elec onics.
Each GR consis s o an n
++
doped implan , equipped wi h me al ield pla es. A p
+
implan
(ei he a p-s op o p-sp ay) is (in mos designs) in e posed be ween each pai o GRs, wi h
he ou e one le loa ing, and he inne one gene ally g ounded in o de o collec he
leakage cu en gene a ed ou side he co e egion o he de ice. The e a e a ew challenges
in designing he pe iphe y and in e pad egion in LGADs.
Senso s 2025,25, 3006 h ps://doi.o g/10.3390/s25103006
Senso s 2025,25, 3006 2 o 14
1.
Junc ion Te mina ion Ex ension (JTE): In s anda d LGADs, he gain laye is su -
ounded by a deep n
++
implan (JTE), which includes a me al ield pla e. The JTE is
placed a ound each pad o p e en cha ge ca ie s gene a ed in he in e pad egion
om eaching he gain laye . When a pa icle a e ses he gain laye , elec ons imme-
dia ely igge he a alanche mul iplica ion p ocess. Howe e , i cha ge ca ie s a e
gene a ed in he in e pad egion, hey mus i s d i o he gain implan , in oducing
a signi ican delay. The JTE mi iga es his issue by con ining he senso ’s ac i e a ea
o he egions whe e he gain implan is p esen .
2.
P-s op isola ion: In s anda d segmen ed LGADs, n
++
implan s mus be elec ically
isola ed om one ano he . This can be achie ed using an addi ional p
++
implan
known as a p-s op. In some designs, isola ion is enhanced by inco po a ing wo p-
s ops along wi h a cen al n
+
bias ing posi ioned be ween hem. In o de o imp o e
he ill ac o in segmen ed senso s, he isola ion s uc u es o small dimensions a e
desi ed. Howe e , hey lowe senso capabili y o hold high bias ol ages. The p-
s op s uc u e is loa ing, so i loa s o a po en ial be ween ha o he n
++
implan
and he bias. The e is, he e o e, a s ong elec ic ield be ween he p-s op and he
n
++
pad: he sho e his dis ance, he highe he ield. A ew in e es ing ea u es
ha e been obse ed: (i) la ge in e pad dis ances lead o highe b eakdown ol ages,
and (ii) he design wi h he wides in e pad dis ance exhibi s a b eakdown due o
gain. The high alue o he p-s op doping is de imen al o senso s abili y due o
he p esence o posi i e cha ges in SiO
2
(a he Si-SiO
2
in e ace). In con as , low
doping alues inc ease he capabili y o holding high bias ol ages e en wi h sho
in e pad dis ances. Howe e , oo low alues o he p-s op doping migh no assu e
pad isola ion [2].
3.
Gua d ings: The ole o GRs becomes inc easingly complex as senso hickness
dec eases. In ul a as senso s, he la e al expansion o he deple ion egion is app ox-
ima ely equal o he senso hickness. Consequen ly, loa ing GRs mus be posi ioned
wi hin a la e al dis ance o abou 2–3 imes he ac i e hickness om he inne mos GR.
Reducing he la e al sp ead o GRs inc eases he elec ic ield be ween hem, which
can lead o p ema u e b eakdown. This poses a signi ican challenge in he design o
ul a hin (~20–30
µ
m) ul a as senso s. E en he inne mos GR can be le loa ing,
bu in such a con igu a ion, he adjacen pads will exhibi highe leakage cu en . The
p-s ops be ween GRs loa o a po en ial be ween he bias ol age and g ound; gi en
he much- educed hickness o he bulk, he p-s op in his case migh loa o po en ial
qui e close o he bias alue. Unde his condi ion, he sequence o GRs migh no be
able o sus ain a la ge ol age d op. In RD50 TI-LGADs mul iple GRs designs we e
applied (Figu e 1a). In he s anda d e sion, he inne mos ing is hicke han he
o he s and can be con ac ed h ough an opening in he passi a ion. The ings should
be isola ed om each o he ; howe e , his depends on he deple ion o he senso .
4.
GR op imiza ion: The op imiza ion o he senso ’s pe iphe y is c ucial o enable he
senso (i e y hin (
≤
4
µ
m) o wi h a e y na ow ( ew
µ
m) in e pad egion) o ope -
a e e icien ly up o e y high luence (~1
×
10
17
1 MeV n
eq
/cm
2
). I he pe iphe al
egion is oo sho , he deple ion egion can e en ually each he chip edge, leading o
a high cu en injec ion in o he co e egion. De ices wi h sho e enches a e e en
mo e ulne able o his p oblem. When designing GR layou , while a he same ime
we a e eaching senso ’s limi s (geome ic and ope a ional) some senso ’s pa ame e s
may coun e ac and cancel each o he . Fo ins ance, he op imiza ion o he gua d ing
(GR) p o ec ion s uc u e, especially when small subs a e hicknesses a e used, was
ex ensi ely and sys ema ically s udied in a ecen R&D ba ch p oduced a Fondazione
B uno Kessle (FBK) in he amewo k o he “eXFlu” p ojec —INFN CSN5 g an
Senso s 2025,25, 3006 3 o 14
o Young Resea che s, whe e di e en op imiza ion s udies o GR s uc u es o
hin subs a es (45, 30, 20 and 15
µ
m) up o high luence (
2.5 ×1015 1 MeV neq/cm2
)
we e conduc ed [
3
,
4
]. Those s udies ha e been enabled hanks o ad anced Technol-
ogy CAD (TCAD) modelling o di e en GR design s a egies, accoun ing o he
comp ehensi e bulk and su ace adia ion-induced damage e ec s. Typically, he
senso ’s pe iphe y egion is (as al eady men ion) composed by a collec o ing, i.e.,
he bias ing, and a loa ing gua d ing whe e GR ing is de ised wi h bo h n-deep
and p-s op implan . P-s op cu s o he possible buildup o he “in e sion laye ”, i.e., a
su ace leakage cu en pa h be ween he collec o ing and he n-deep implan i sel ,
hus a oiding connec ion be ween hem. One o he mos impo an massages and
ou comes om he ex ensi e s udy on GR s uc u e layou s (on he samples om he
eXFlu1 ba ch) was ha a e a luence o 1
×
10
16
n
eq
/cm
2
, a senso pe iphe y wi h
loa ing GR seems o be mo e e ec i e wi hou any p-s op implan , bene i ing om
he concu en ac ion o bulk and su ace damage o mi iga e he build-up o he in-
e sion laye . This esul was e y use ul o he FBK decision on GR s uc u e design
implemen ed in he second TI-LGAD p oduc ion (AIDAinno a ba ch). In he second
TI-LGAD p oduc ion (AIDAinno a ba ch), i was decided no o use he mul i-GR-
layou wi h mul i- loa ing GRs anymo e, bu ins ead o use a single GR layou (shown
in Figu e 1b). The p e en ion o injec ion o su ace leakage cu en in he co e egion
o he senso i he pe iphe y egion is oo sho [5] is s ill challenging ask.
5.
T ench Isola ion: In 2019 he easibili y o implemen ing hin LGADs segmen ed
using he SiO
2
ench isola ion echnology was demons a ed and he i s TI-LGAD
ba ch was ab ica ed a Fondazione B uno Kessle (FBK) wi hin he RD50 Common
p ojec [
6
,
7
]. The wa e layou consis s o pai s o pads (Figu e 1a) wi h pad size
375
×
250
µ
m
2
, ab ica ed on 55 and 45
µ
m- hick p-epi axial bulk. The layou spli s
implemen ed abou 30 pai s, di e ing in he numbe o enches (1 o 2), he dimension
o he bo de s, ench p ocess pa ame e s, dep h o ench (D1 < D2 < D3), and he
dis ance be ween pixel bo de s (V1 < V2 < V3). Two di e en ench designs ha e been
implemen ed in he i s TI-LGADs p oduc ion. The i s one has a ench g id be ween
pads (1TR), while in he second one, each pad is su ounded by an independen pai
o ench ings (2TR). The nominal dis ance be ween he gain implan s in he 1TR and
2TR designs is ~4
µ
m and ~6
µ
m [
8
], espec i ely. The second TI-LGAD p oduc ion
was eleased 2 yea s la e , wi hin he AIDAinno a p ojec (Figu e 1b) [
9
], also a FBK.
In his p oduc ion a medium dep h o ench (equi alen o D2 in RD50 p oduc ion)
was chosen as s anda d while he wid h o enches was a ied. This p oduc ion was
also he i s whe e ca bon was implemen ed o inc ease he adia ion ha dness.
In ou p e ious s udies on TI-LGADs om bo h RD50 and AIDAinno a p oduc-
ions [
10
–
12
], we obse ed he ex ao dina y sel -induced (wi hou ex e nal s imula ion)
pulse signal, ha we call “ghos ”, appea ing in all TI-LGADs a e exceeding a ce ain
alue o applied bias. The example o a ghos signal wi h compa ison o a s anda d signal
gene a ed in he same LGAD senso is shown in Figu e 1d. The p ope ies o hese signals,
such as ampli ude, du a ion, and occu ence a e, depend on many ac o s.
T ench con igu a ion (1TR o 2TR), empe a u e, and i adia ion a ec he ghos s’
cha ac e is ics. The obse ed phenomenon is a complex in e play o many pa ame e s
whe e he manu ac u ing and p ocess pa ame e s o he s udied senso s a e no publicly
accessible (due o non-disclosu e ag eemen s wi h endo FBK). Fo his eason, pe o ming
simula ions ha would eliably suppo ghos e ec elucida ions is no ye possible. The
cu en unde s anding o he discussed phenomena is ully based on he expe imen al
obse a ions, and a summa y o he key esul s is p esen ed in Sec ion 3.1.1. Ex ending he
p e ious esea ch, he e we in es iga e and discuss he in luence o he GR con igu a ion
Senso s 2025,25, 3006 4 o 14
on he ghos signal in TI-LGADs. Senso s wi h mul iple GRs (RD50) and a single GR
(AIDAinno a) we e in es iga ed o his pu pose. In addi ion, he impac o loa ing o
biasing (connec ing o pad) o he GRs on ghos e ec is examined, and ela ed changes in
leakage cu en a e discussed.
Senso s 2025, 25, x FOR PEER REVIEW 4 o 15
Figu e 1. Layou o 1 × 2 pixels a ay o TI-LGADs: (a) RD50 senso wi h mul iple gua d ings. (b)
AIDAinno a senso wi h single gua d ing. (c) C oss-sec ion o LGAD senso wi h single ench.
(d) Compa ison o sel -induced ghos signal in TI-LGAD de ice a 100 V wi h no mal signal gen-
e a ed in pad o his senso by 5 pJ lase pulse a he same bias.
T ench configu a ion (1TR o 2TR), empe a u e, and i adia ion affec he ghos s’
cha ac e is ics. The obse ed phenomenon is a complex in e play o many pa ame e s
whe e he manu ac u ing and p ocess pa ame e s o he s udied senso s a e no publicly
accessible (due o non-disclosu e ag eemen s wi h endo FBK). Fo his eason, pe -
o ming simula ions ha would eliably suppo ghos effec elucida ions is no ye pos-
sible. The cu en unde s anding o he discussed phenomena is ully based on he ex-
pe imen al obse a ions, and a summa y o he key esul s is p esen ed in Sec ion 3.1.1.
Ex ending he p e ious esea ch, he e we in es iga e and discuss he influence o he GR
configu a ion on he ghos signal in TI-LGADs. Senso s wi h mul iple GRs (RD50) and a
single GR (AIDAinno a) we e in es iga ed o his pu pose. In addi ion, he impac o
floa ing o biasing (connec ing o pad) o he GRs on ghos effec is examined, and ela ed
changes in leakage cu en a e discussed.
2. Ma e ials and Me hods
The senso s used in his s udy we e designed and manu ac u ed by Fondazione
B uno Kessle (FBK, T en o, I aly) wi hin wo diffe en p oduc ions: RD50 and AIDAin-
no a. E e y senso had a o m o 2 × 1 a ay ( wo pixels isola ed by SiO2 enches), as
shown in Figu e 1. Selec ed senso s we e i adia ed wi h neu ons a he T iga II Reac o
o he Joze S e an Ins i u e (JSI) in Ljubljana, Slo enia. A ull lis o he senso s examined
in his s udy is p esen ed in Table 1.
The sel -induced (ghos ) signal gene a ed upon he applied ol age was in es iga ed
by ansien cu en echnique (TCT) using he expe imen al se up a he ELI Beamlines
acili y (Dolni B ezany, Czech Republic) [13]. The aluminum housings, ab ica ed and
designed a he Joze S e an Ins i u e, we e used o he samples moun ing. The image o
example housing is shown in Figu e 2a. The samples we e conduc i ely connec ed o he
housing i sel ( his all is hen connec ed o he g ound) and biasing o he senso s was
achie ed om he op side. Bo h signal pads o he sample we e wi e bonded o he pin o
Figu e 1. Layou o 1
×
2 pixels a ay o TI-LGADs: (a) RD50 senso wi h mul iple gua d ings.
(b) AIDAinno a senso wi h single gua d ing. (c) C oss-sec ion o LGAD senso wi h single ench.
(d) Compa ison o sel -induced ghos signal in TI-LGAD de ice a 100 V wi h no mal signal gene a ed
in pad o his senso by 5 pJ lase pulse a he same bias.
2. Ma e ials and Me hods
The senso s used in his s udy we e designed and manu ac u ed by Fondazione B uno
Kessle (FBK, T en o, I aly) wi hin wo di e en p oduc ions: RD50 and AIDAinno a.
E e y senso had a o m o 2
×
1 a ay ( wo pixels isola ed by SiO
2
enches), as shown in
Figu e 1. Selec ed senso s we e i adia ed wi h neu ons a he T iga II Reac o o he Joze
S e an Ins i u e (JSI) in Ljubljana, Slo enia. A ull lis o he senso s examined in his s udy
is p esen ed in Table 1.
The sel -induced (ghos ) signal gene a ed upon he applied ol age was in es iga ed
by ansien cu en echnique (TCT) using he expe imen al se up a he ELI Beamlines
acili y (Dolni B ezany, Czech Republic) [
13
]. The aluminum housings, ab ica ed and
designed a he Joze S e an Ins i u e, we e used o he samples moun ing. The image
o example housing is shown in Figu e 2a. The samples we e conduc i ely connec ed o
he housing i sel ( his all is hen connec ed o he g ound) and biasing o he senso s was
achie ed om he op side. Bo h signal pads o he sample we e wi e bonded o he pin o
he SMA connec o (which is elec ically isola ed om aluminum housing) as depic ed in
Figu e 2b.
The pin was connec ed o HV powe supply (EBS C0_30SHV, ISEG Spezialelek onik,
Radbe g, Ge many) ia a bias ee and he signal ou pu was coupled di ec ly o he oscillo-
scope (In iniiVision DSOX6004A, 6 GHz, 20 GS, Keysigh , San a Rosa, CA, USA) wi hou
addi ional ampli ie s. The gene a ed wa e o ms we e eco ded a di e en bias up o
maximal non-des uc i e alues o a gi en senso . The occu ence a e o he ghos signal
was also measu ed by in eg a ed coun e wi h o alize unc ion (maximal measu able a e
Senso s 2025,25, 3006 5 o 14
25 MHz). The housings wi h senso s we e placed on he cooling pla e and all measu emen s
we e conduc ed a oom (+20 ◦C) and low (−20 ◦C) empe a u e.
Table 1. T ench-isola ed senso s used in his s udy.
P oduc ion Senso Wa e Type I adia ion
RD50
LGAD W11 C1-V2-2TR No
PIN W11 C1-V4-2TR No
LGAD W3 C1-V2-2TR 0.8 ×1015 neq/cm2
PIN W3 C1-V2-2TR 0.8 ×1015 neq/cm2
AIDAinno a
LGAD W1 V2-2TR TW5 Cell-D TS1 No
PIN W1 1TR TW5 Cell-D TS1 No
LGAD W1 V2-1TR TW5 Cell-C TS3 0.8 ×1015 neq/cm2
PIN W1 1TR TW2 Cell-C TS3 0.8 ×1015 neq/cm2
LGAD W1 V3-1TR TW5 Cell-C TS8 1.5 ×1015 neq/cm2
PIN W1 1TR TW6 Cell-C TS4 1.5 ×1015 neq/cm2
Senso s 2025, 25, x FOR PEER REVIEW 5 o 15
he SMA connec o (which is elec ically isola ed om aluminum housing) as depic ed in
Figu e 2b.
Table 1. T ench-isola ed senso s used in his s udy.
P oduc ion Senso Wa e Type I adia ion
RD50
LGAD W11 C1-V2-2TR No
PIN W11 C1-V4-2TR No
LGAD W3 C1-V2-2TR 0.8 × 1015 neq/cm2
PIN W3 C1-V2-2TR 0.8 × 1015 neq/cm2
AIDAinno a
LGAD W1 V2-2TR TW5 Cell-D TS1 No
PIN W1 1TR TW5 Cell-D TS1 No
LGAD W1 V2-1TR TW5 Cell-C TS3 0.8 × 1015 neq/cm2
PIN W1 1TR TW2 Cell-C TS3 0.8 × 1015 neq/cm2
LGAD W1 V3-1TR TW5 Cell-C TS8 1.5 × 1015 neq/cm2
PIN W1 1TR TW6 Cell-C TS4 1.5 × 1015 neq/cm2
The pin was connec ed o HV powe supply (EBS C0_30SHV, ISEG Spezi-
alelek onik, Radbe g, Ge many) ia a bias ee and he signal ou pu was coupled di-
ec ly o he oscilloscope (InfiniiVision DSOX6004A, 6 GHz, 20 GS, Keysigh , San a Rosa,
CA, USA) wi hou addi ional amplifie s. The gene a ed wa e o ms we e eco ded a
diffe en bias up o maximal non-des uc i e alues o a gi en senso . The occu ence
a e o he ghos signal was also measu ed by in eg a ed coun e wi h o alize unc ion
(maximal measu able a e 25 MHz). The housings wi h senso s we e placed on he cool-
ing pla e and all measu emen s we e conduc ed a oom (+20 °C) and low (−20 °C) em-
pe a u e.
Figu e 2. (a) Aluminum housing used o senso s moun ing; (b) example 1 × 2 pixel senso wi h
indica ed wi e bonds; (c) AIDAinno a and (d) RD50 senso s du ing IV measu emen s.
Figu e 2. (a) Aluminum housing used o senso s moun ing; (b) example 1
×
2 pixel senso wi h
indica ed wi e bonds; (c) AIDAinno a and (d) RD50 senso s du ing IV measu emen s.
I–V measu emen s we e pe o med in he Ins i u e o Physics a Czech Academy o
Science (P ague, Czech Republic) using an au oma ic p obe s a ion (TESLA200, Fo mFac o )
wi h he iaxial chuck enclosed in he en i onmen al chambe . The chuck wi h a golden
su ace (diame e o 200 mm) and implemen ed acuum ci cui s was connec ed by he
HV compa ible iaxial cable wi h he HI o he sou ce measu e uni (SMU) ia he 1 M
Ω
p o ec ion esis o . The es ing pad and GR we e con ac ed by he p obes (see Figu e 2c,d),
which we e en e ing he p obe s a ion en i onmen al chambe h ough he so-called op
ha . The p obe con ac ing he es ing pad was b ough o he SMU LO by he iaxial cable.
Senso s 2025,25, 3006 6 o 14
The GR o he sample was b ough o he labo a o y g ound by he second p obe, again
by he iaxial cable. Du ing he measu emen , he ela i e humidi y in he en i onmen al
chambe o he p obe s a ion was dec eased o close o ze o alues.
All TCT and I–V measu emen s we e pe o med o wo di e en con igu a ions o
gua d ings connec ions. In he i s a ian , all GRs in he senso s we e le loa ing. In he
second case, he inne mos ing was b ough o he pad po en ial by wi e bonding o he
co esponding a eas (see Figu e 2b).
3. Resul s
3.1. Ghos s in T ench-Isola ed Senso s wi h Floa ing Gua d Rings
3.1.1. P e ious Resul s
The occu ence o sel -induced signals is a complex phenomenon in ol ing a ious
ac o s so we i s p o ide a sho summa y o ou p e ious esul s. We obse ed he i s
ghos s in he non-i adia ed LGAD o igina ed om RD50 p oduc ion wi h double- ench
IP isola ion. In he nex s eps, we in es iga ed he ole o he ollowing ac o s [10–12]:
•p esence o he gain laye (LGAD s. PIN)
•numbe o enches (2TR s. 1TR)
•i adia ion e ec s (gain loss)
• empe a u e e ec s ( oom empe a u e s. low empe a u e)
•gain laye ca boniza ion (RD50 s. AIDAinno a)
All hese s udies e ealed ha di e en ypes o ghos s exis in ench-isola ed senso s
depending on he senso ype and expe imen al condi ions. Th ee main ypes o ghos s
we e iden i ied.
•
Type A: Signal appea ing a low bias ( ypically 30–100 V in non-i adia ed senso s) wi h
occu ence a e dec easing wi h inc easing bias (a ypical ea u es o au o- igge ed
signals in de ices wi h in e nal gain and cha ge mul iplica ion). This signal is s ic ly
ela ed o he p esence o he gain laye . I is no obse ed in PINs (no gain) o
i adia ed LGADs ha los hei gain (RD50 LGADs). We assumed ha onse o his
ype o ghos s is a ound he GR s uc u es. These discha ges, expe imen ally e i ied
and p esen ed in [
13
], a e enhanced by i adia ion bu signi ican ly supp essed a low
empe a u es ( ied o he mally ac i a ed p ocesses, such as ca ie gene a ion ia
shallow aps o su ace s a es).
•
Type B: Signal appea ing in medium bias ange ( ypically >80 V in non-i adia ed
senso s) exclusi ely in 2TR LGADs. The signal o he same cha ac e is ics (ampli ude
and shape) can be also gene a ed by lase s imula ion when in e -pixel egion is illu-
mina ed. This ype o signal comes om he de ec s c ea ed be ween he enches ha
can be he mally popula ed a oom empe a u e ( his signal is s ongly supp essed a
low empe a u e). Mode a e dependence o i s occu ence a e on bias is a ibu ed o
he p esence o n+implan placed be ween he enches.
•
Type C: Signal appea ing a high bias ( ypically >200 V) exhibi ing s ong ampli ude
(o de o magni ude highe han o he s). Occu ence a e o hese e en s is s ongly
bias dependen and inc eases wi h ising ol age. This ype o signal is associa ed wi h
high- ield phenomena such as impac ioniza ion, a alanche p ocesses, o enhanced
ap-assis ed ca ie gene a ion unde s ong elec ic ields. I is obse ed in all ypes
o TI senso s when high enough bias alue is eached.
The exis ence o indi idual ypes o ghos signals in di e en ench-isola ed senso s
is summa ized in Table 2. In he ollowing subsec ions, he new esul s o he senso s wi h
di e en GRs con igu a ions a e p esen ed.
Senso s 2025,25, 3006 7 o 14
Table 2. Occu ence o di e en ypes o ghos signals in di e en ench-isola ed senso s.
Senso
RD50 P oduc ion
(Mul iple GRs, Non-Ca bonized)
AIDAinno a P oduc ion
(Single GR, Ca bonized)
Non-I adia ed I adia ed Non-I adia ed I adia ed
LGAD 1TR Type C Type A, C no exp. da a Type A, C
2TR Type A, B, C Type C Type A, B, C Type A, B, C
PIN 1TR no ghos s Type C no ghos s Type C
2TR no ghos s Type C no ghos s no exp. da a
3.1.2. RD50 Senso s wi h Mul iple GRs
In case o LGADs om RD50 p oduc ion, he ghos s appea ed when he bias eached
di e en le el (depending on he senso ) in he ange 30–90 V. These sel -induced discha ges
a e ep esen ed by he wa e o ms wi h wid h o se e al nanoseconds and ampli udes in
he ange o hund ed hs o mVs (see Figu e 3a). Abo e he gene a ion h eshold, he ghos s
a e p esen in en i e bias ange, up o maximal non-des uc i e alues ( ypically be ween
160 and 220 V, depending on he senso and empe a u e). The ghos signal in his ange
anishes comple ely in i adia ed senso s. Howe e , e y s ong sel -induced discha ges
occu when he bias eaches he alues close o he b eakdown limi s. In case o he senso s
i adia ed o luence o 0.8
×
10
15
n
eq
/cm
2
, whe e he b eakdown bias is ypically in he
ange 500–550 V, he ghos s appea a he ol age abo e 450 V. In compa ison o hei
coun e pa s in non-i adia ed samples, he ep esen a i e wa e o ms ha e much highe
ampli ude (up o 4 V), mul ipeak shape and hey a e much b oade ex ending up o 40–45 ns
(see Figu e 3a). The e y simila signal was also obse ed in co esponding PIN i adia ed
a he same luence (compa e wi h Figu e 3b). I is no ewo hy ha in non-i adia ed PINs
he ghos s we e no obse ed e en a he highes applicable bias and hei p esence in
LGADs was a ibu ed o he gain laye . Howe e , in i adia ed PIN, he sel -induced
signal is obse ed a he bias abo e 450 V in spi e o lack o gain. Simila ly o he i adia ed
LGADs, he ghos s appea a bias ol ages e y close o he onse o adia ion-induced
b eakdown in he bulk. The mul ipeak cha ac e o his signal ei he in PIN o LGAD can
be a ibu ed o he p esence o mul iple GRs in all RD50 senso s (see Sec ion 4below).
Senso s 2025, 25, x FOR PEER REVIEW 7 o 15
wi h high-field phenomena such as impac ioniza ion, a alanche p ocesses, o en-
hanced ap-assis ed ca ie gene a ion unde s ong elec ic fields. I is obse ed in
all ypes o TI senso s when high enough bias alue is eached.
The exis ence o indi idual ypes o ghos signals in diffe en ench-isola ed senso s
is summa ized in Table 2. In he ollowing subsec ions, he new esul s o he senso s
wi h diffe en GRs configu a ions a e p esen ed.
Table 2. Occu ence o diffe en ypes o ghos signals in diffe en ench-isola ed senso s.
Senso
RD50 P oduc ion
(Mul iple GRs, Non-Ca bonized)
AIDAinno a P oduc ion
(Single GR, Ca bonized)
Non-I adia ed I adia ed Non-I adia ed I adia ed
LGAD 1TR Type C Type A,C no exp. da a Type A, C
2TR Type A, B, C Type C Type A, B,C Type A, B, C
PIN 1TR no ghos s Type C no ghos s Type C
2TR no ghos s Type C no ghos s no exp. da a
3.1.2. RD50 Senso s wi h Mul iple GRs
In case o LGADs om RD50 p oduc ion, he ghos s appea ed when he bias
eached diffe en le el (depending on he senso ) in he ange 30–90 V. These
sel -induced discha ges a e ep esen ed by he wa e o ms wi h wid h o se e al nano-
seconds and ampli udes in he ange o hund ed hs o mVs (see Figu e 3a). Abo e he
gene a ion h eshold, he ghos s a e p esen in en i e bias ange, up o maximal
non-des uc i e alues ( ypically be ween 160 and 220 V, depending on he senso and
empe a u e). The ghos signal in his ange anishes comple ely in i adia ed senso s.
Howe e , e y s ong sel -induced discha ges occu when he bias eaches he alues
close o he b eakdown limi s. In case o he senso s i adia ed o fluence o 0.8 × 1015
neq/cm2, whe e he b eakdown bias is ypically in he ange 500–550 V, he ghos s appea
a he ol age abo e 450 V. In compa ison o hei coun e pa s in non-i adia ed sam-
ples, he ep esen a i e wa e o ms ha e much highe ampli ude (up o 4 V), mul ipeak
shape and hey a e much b oade ex ending up o 40–45 ns (see Figu e 3a). The e y
simila signal was also obse ed in co esponding PIN i adia ed a he same fluence
(compa e wi h Figu e 3b). I is no ewo hy ha in non-i adia ed PINs he ghos s we e
no obse ed e en a he highes applicable bias and hei p esence in LGADs was
a ibu ed o he gain laye . Howe e , in i adia ed PIN, he sel -induced signal is ob-
se ed a he bias abo e 450 V in spi e o lack o gain. Simila ly o he i adia ed LGADs,
he ghos s appea a bias ol ages e y close o he onse o adia ion-induced b eakdown
in he bulk. The mul ipeak cha ac e o his signal ei he in PIN o LGAD can be
a ibu ed o he p esence o mul iple GRs in all RD50 senso s (see Sec ion 4 below).
Figu e 3. Ghos signals in i adia ed and non-i adia ed (a) LGADs (W11 C1-V2-2TR) and (b) PIN
(W11 C1-V4-2TR) om RD50 p oduc ion.
3.1.3. AIDAinno a Senso s wi h Single GR
In he ench-isola ed senso s om he AIDAinno a p oduc ion, wo impo an aspec s
make hem di e en om RD50 samples. LGADs om his p oduc ion ha e a gain laye
Senso s 2025,25, 3006 8 o 14
en iched wi h ca bon. In addi ion, all AIDAinno a senso s ea u e only a single gua d
ing, con a y o he mul iple GR s uc u e in RD50 samples. These al e a ions can lead
o di e en ghos s’ beha io . In non-i adia ed TI-LGADs om AIDAinno a p oduc ion,
sel -induced signal is qui e simila o ha obse ed in RD50. I appea s abo e 30 V and
inc eases wi h bias up o 90 V. Abo e ha alue, he signal changes i s cha ac e , becoming
b oade and ea u ing lowe ampli ude. These signals, assigned be o e as Type A and
Type B (see Sec ion 3.1.1), a e shown in Figu e 4a. In he co esponding non-i adia ed
PIN senso , he ghos s we e no obse ed. A e i adia ion, he addi ional ype o ghos ,
p esen a high bias alues, appea s in TI-LGAD bu also in TI-PIN [
11
]. Al hough his
signal exhibi s e y high ampli ude (up o 2 V), i is e y di e en om i s coun e pa
eco ded in RD50 i adia ed senso s. The wa e o m is e y na ow (see Figu e 4b) and
does no display he mul ipeak cha ac e . This ea u e is assigned o he ac ha only a
single gua d ing is p esen in he senso s om AIDAinno a p oduc ion. Al hough he
di e ence be ween he wo discussed p oduc ions lays no only in GRs con igu a ion bu
also in gain ca boniza ion ( o educe losses due o i adia ion), he obse ed di e ence in
ghos wa e o ms is mani es ed no only in LGADs bu also in PINs wi hou gain laye .
Hence, he obse ed e ec is only a ibu ed o he di e ence in GRs.
Senso s 2025, 25, x FOR PEER REVIEW 8 o 15
Figu e 3. Ghos signals in i adia ed and non-i adia ed (a) LGADs (W11 C1-V2-2TR) and (b) PIN
(W11 C1-V4-2TR) om RD50 p oduc ion.
3.1.3. AIDAinno a Senso s wi h Single GR
In he ench-isola ed senso s om he AIDAinno a p oduc ion, wo impo an as-
pec s make hem diffe en om RD50 samples. LGADs om his p oduc ion ha e a gain
laye en iched wi h ca bon. In addi ion, all AIDAinno a senso s ea u e only a single
gua d ing, con a y o he mul iple GR s uc u e in RD50 samples. These al e a ions can
lead o diffe en ghos s’ beha io . In non-i adia ed TI-LGADs om AIDAinno a p o-
duc ion, sel -induced signal is qui e simila o ha obse ed in RD50. I appea s abo e 30
V and inc eases wi h bias up o 90 V. Abo e ha alue, he signal changes i s cha ac e ,
becoming b oade and ea u ing lowe ampli ude. These signals, assigned be o e as Type
A and Type B (see Sec ion 3.1.1), a e shown in Figu e 4a. In he co esponding
non-i adia ed PIN senso , he ghos s we e no obse ed. A e i adia ion, he addi ional
ype o ghos , p esen a high bias alues, appea s in TI-LGAD bu also in TI-PIN [11].
Al hough his signal exhibi s e y high ampli ude (up o 2 V), i is e y diffe en om i s
coun e pa eco ded in RD50 i adia ed senso s. The wa e o m is e y na ow (see
Figu e 4b) and does no display he mul ipeak cha ac e . This ea u e is assigned o he
ac ha only a single gua d ing is p esen in he senso s om AIDAinno a p oduc ion.
Al hough he diffe ence be ween he wo discussed p oduc ions lays no only in GRs
configu a ion bu also in gain ca boniza ion ( o educe losses due o i adia ion), he ob-
se ed diffe ence in ghos wa e o ms is mani es ed no only in LGADs bu also in PINs
wi hou gain laye . Hence, he obse ed effec is only a ibu ed o he diffe ence in GRs.
Figu e 4. Ghos signals in i adia ed and non-i adia ed (a) LGADs (V2-2TR TW5 and V3-1TR
TW5) and (b) PIN (1TR TW6) om AIDAinno a p oduc ion.
3.2. Ghos s in T ench-Isola ed Senso s wi h GR Connec ed o Pad
I was clea ly obse ed expe imen ally ha he sel -induced signals in RD50 and
AIDAinno a ench-isola ed senso s beha e diffe en ly. The obse ed diffe ences indi-
ca e ce ain co ela ion wi h gua d ings s uc u e in bo h amilies o samples (mul iple
GRs in RD50 and single GR in AIDAinno a). Since, all obse a ions we e eco ded o
he senso s wi h floa ing ings, we decided o es wha is he impac o GR po en ial o
sel -induced signals. Fo his pu pose, he inne mos GR in RD50 senso s and single GR
in AIDAinno a samples we e connec ed by addi ional wi e wi h he neighbo pad (as
p esen ed in Figu e 2b). A e ha , all he senso s we e examined in exac ly he same
condi ions (bias, empe a u e) as p e iously. I u ned ou ha b inging he inne mos
GR o he same po en ial as he pad esul ed in comple e anishing o ghos s. They we e
no obse ed in any ci cums ances in ei he LGADs o PINs.
Figu e 4. Ghos signals in i adia ed and non-i adia ed (a) LGADs (V2-2TR TW5 and V3-1TR TW5)
and (b) PIN (1TR TW6) om AIDAinno a p oduc ion.
3.2. Ghos s in T ench-Isola ed Senso s wi h GR Connec ed o Pad
I was clea ly obse ed expe imen ally ha he sel -induced signals in RD50 and
AIDAinno a ench-isola ed senso s beha e di e en ly. The obse ed di e ences indica e
ce ain co ela ion wi h gua d ings s uc u e in bo h amilies o samples (mul iple GRs
in RD50 and single GR in AIDAinno a). Since, all obse a ions we e eco ded o he
senso s wi h loa ing ings, we decided o es wha is he impac o GR po en ial o
sel -induced signals. Fo his pu pose, he inne mos GR in RD50 senso s and single GR
in AIDAinno a samples we e connec ed by addi ional wi e wi h he neighbo pad (as
p esen ed in Figu e 2b). A e ha , all he senso s we e examined in exac ly he same
condi ions (bias, empe a u e) as p e iously. I u ned ou ha b inging he inne mos GR
o he same po en ial as he pad esul ed in comple e anishing o ghos s. They we e no
obse ed in any ci cums ances in ei he LGADs o PINs.
3.3. I–V Cha ac e is ics o T ench-Isola ed LGADs and PINs
Since he ghos signal anishes in all enched de ices a e GR connec ion, he only
way o ob ain some quan i a i e di e ences be ween in es iga ed GR con igu a ions is o
moni o he ela ed changes in leakage cu en .
Senso s 2025,25, 3006 9 o 14
3.3.1. Non-I adia ed T ench-Isola ed Senso s
IV cha ac e is ics o ench-isola ed LGADs and PINs we e i s measu ed o non-
i adia ed de ec o s. In each case, IV cu es we e eco ded o he senso wi h loa ing GR
and he same sample when GR is connec ed o i s pad. Fo ha pu pose, wo di e en
senso s om AIDAinno a p oduc ion we e chosen, one LGAD (V2-1TR TW5 Cell-D TS1)
and a co esponding PIN (1TR TW5 Cell-D TS1), bo h loca ed on he same a ay. The esul s
o hese senso s a e p esen ed in Figu e 5a,b. In he case o he LGAD, he b eakdown
ol age shi s abou 20 V o highe alues when he GR is a he same po en ial as he pad
(Figu e 5a). This e ec is s onge in he case o he PIN, whe e he b eakdown limi is
shi ed abou 100 V (Figu e 5b). All LGAD samples (including hose no shown he e) wi h
loa ing GR exhibi a small local maximum in IV cu e a he bias a ound 35 V. This e ec
disappea s when he GR is connec ed o he pad and i is no obse ed in he PIN a all.
I is p obably ela ed o he exis ence o di e en deple ion egions in LGADs and hei
me ging a ce ain bias (see discussion in Sec ion 4.1.3).
Senso s 2025, 25, x FOR PEER REVIEW 9 o 15
3.3. I–V Cha ac e is ics o T ench-Isola ed LGADs and PINs
Since he ghos signal anishes in all enched de ices a e GR connec ion, he only
way o ob ain some quan i a i e diffe ences be ween in es iga ed GR configu a ions is o
moni o he ela ed changes in leakage cu en .
3.3.1. Non-I adia ed T ench-Isola ed Senso s
IV cha ac e is ics o ench-isola ed LGADs and PINs we e fi s measu ed o
non-i adia ed de ec o s. In each case, IV cu es we e eco ded o he senso wi h
floa ing GR and he same sample when GR is connec ed o i s pad. Fo ha pu pose, wo
diffe en senso s om AIDAinno a p oduc ion we e chosen, one LGAD (V2-1TR TW5
Cell-D TS1) and a co esponding PIN (1TR TW5 Cell-D TS1), bo h loca ed on he same
a ay. The esul s o hese senso s a e p esen ed in Figu e 5a,b. In he case o he LGAD,
he b eakdown ol age shi s abou 20 V o highe alues when he GR is a he same
po en ial as he pad (Figu e 5a). This effec is s onge in he case o he PIN, whe e he
b eakdown limi is shi ed abou 100 V (Figu e 5b). All LGAD samples (including hose
no shown he e) wi h floa ing GR exhibi a small local maximum in IV cu e a he bias
a ound 35 V. This effec disappea s when he GR is connec ed o he pad and i is no
obse ed in he PIN a all. I is p obably ela ed o he exis ence o diffe en deple ion
egions in LGADs and hei me ging a ce ain bias (see discussion in Sec ion 4.1.3).
Figu e 5. IV cha ac e is ics o non-i adia ed (a,b), i adia ed a 0.8 × 1015 neq/cm2 (c,d), and i adi-
a ed a 1.5 × 1015 neq/cm2 (e, ) ench-isola ed senso s (LGADs and PINs) om AIDAinno a p o-
duc ion.
Figu e 5. IV cha ac e is ics o non-i adia ed (a,b), i adia ed a 0.8
×
10
15
n
eq
/cm
2
(c,d), and
i adia ed a 1.5
×
10
15
n
eq
/cm
2
(e, ) ench-isola ed senso s (LGADs and PINs) om AIDAin-
no a p oduc ion.
3.3.2. I adia ed T ench-Isola ed Senso s
To examine i adia ion impac on IV cha ac e is ics o he senso s wi h loa ing and
pad connec ed GR, ou senso s o di e en ea u es we e selec ed om AIDAinno a
p oduc ion. Fi s , LGAD (V2-1TR TW5 Cell-C TS3) and co esponding PIN (1TR TW2
Cell-C TS3) i adia ed o luence o 0.8
×
10
15
n
eq
/cm
2
we e es ed. In bo h cases, he
b eakdown ol age was upshi ed a e connec ion o GR o pad (see Figu e 5c,d). S onge
i adia ion a luency o 1.5
×
10
15
n
eq
/cm
2
does no change his gene al end al hough i
