CMOS SPADs selection, modeling and characterization towards image sensors implementation
Abstract
The selection, modeling and characterization of Single Photon Avalanche Diodes (SPADs) are presented. Working with the standard 180nm UMC CMOS process, different SPAD structures are proposed in combination with several quenching circuits in order to compare their relative performances. Various configurations for the active region and the prevention of the premature edge breakdown are tested, looking for a miniaturization of the devices to implement image sensor arrays without loses in their performance.
Full text
CMOS SPADs Selec ion, Modeling and
Cha ac e iza ion Towa ds Image Senso s
Implemen a ion
Manuel Mo eno Ga c´
ıa, ´
Osca Gue a Vinuesa, Roc´
ıo del R´
ıo Fe n´
andez, Bel´
en P´
e ez Ve d´
u
and ´
Angel Rod ´
ıguez V´
azquez
Ins i u e o Mic oelec onics o Se illa, IMSE-CNM (CSIC/Uni e sidad de Se illa)
C/Ame ico Vespucio, s/n, 41092 Se ille (Spain)
Email: [email p o ec ed]
Abs ac —The selec ion, modeling and cha ac e iza ion o
Single Pho on A alanche Diodes (SPADs) a e p esen ed. Wo king
wi h he s anda d 180nm UMC CMOS p ocess, di e en SPAD
s uc u es a e p oposed in combina ion wi h se e al quenching
ci cui s in o de o compa e hei ela i e pe o mances. Va ious
con igu a ions o he ac i e egion and he p e en ion o he
p ema u e edge b eakdown a e es ed, looking o a minia u i-
za ion o he de ices o implemen image senso a ays wi hou
loses in hei pe o mance.
I. INTRODUCTION
Single Pho on A alanche Diodes (SPADs) a e p/n junc ions
biased abo e hei b eakdown ol age [1], in he so-called
Geige mode [2]. In his si ua ion he elec ic ield is so high
ha a single ca ie in he deple ion egion can o igina e a
sel -sus ained a alanche cu en in he milliampe es ange [3].
An addi ional ci cui is equi ed o use he SPAD as a single-
pho on de ec o . I is he quenching ci cui , and i s mission is o
ex inguish he a alanche be o e damage o he s uc u e occu s.
The simples one consis s o a high- alue ballas esis o
connec ed in se ies o he diode (Rqin Fig. 1(a)), which is
aimed o lowe he bias ol age below he b eakdown ol age
[4], [5]. When an a alanche is igge ed, he cu en lowing
h ough he diode inc eases, inducing a ol age d op o e he
ballas esis o . The cu en s a s dec easing un il i eaches a
h eshold alue called la ching cu en , a which he a alanche
is no mo e sel -sus ained [4]. Fig. 1(b) shows he ypical I-V
piecewise linea cha ac e is ic o he SPAD.
SPADs a e a dedica ed ype o pho ode ec o s ha a e
a ac ing inc easing in e es . Thei ab ica ion in s anda d
CMOS echnologies allows o in eg a e all pe iphe al blocks
equi ed o ision sys ems (a ays o SPADs and digi al
p ocessing elec onics) on he same monoli hic chip. Thus,
he combina ion o high gain and quan um e iciency, along
wi h hei small size, low ope a ion ol age and insensi i i y
o magne ic ields make SPADs pa icula ly a ac i e o
di e en imaging and medical applica ions [6].
In his pape we p esen a es chip wi h se e al SPAD
s uc u es o be es ed, oge he wi h di e en quenching
ci cui s ha ha e been chosen using he in o ma ion p o ided
by a p op ie a y model o SPADs based on [7]. The main
(a)
Vb
Is
a alanche
no a alanche
Rb
Ila
VD
ID
GEIGER MODE
OF OPERATION
(b)
Fig. 1. (a) Scheme o illus a ing he ope a ion o he SPAD wi h passi e
quenching. (b) I-V cha ac e is ic o he SPAD.
objec i e o his wo k is o s udy he po en iali y o di e en
s uc u es and quenching ci cui s o gene a ing image senso
a ays, using he s anda d 180nm UMC CMOS p ocess.
II. SELECTION OF CMOS SPADSSTRUCTURES
In his sec ion a a ie y o SPAD s uc u es ha ha e been
used so a in di e en echnological nodes is p esen ed. The
di e en modi ica ions in he s uc u e a e in ended o a oid
he p ema u e edge b eakdown (PEB). I consis s o a spu ious
a alanche ha a o able akes places a he edges o he de ice
ac i e a ea, educing i s sensi i i y [1]. The s uc u es a e he
ollowing:
1) Simple p+/nwell diode: This is he basis s uc u e and
does no include any mechanism o a oid PEB.
2) P+/nwell diode wi h well gua d ing: A deepe and
lowe doped implan is in oduced in he pe iphe y o
he junc ion o aise he b eakdown ol age in his egion
wi h ega d o he cen e o he ac i e a ea [8].
3) Twell/nwell diode: The junc ion is o med be ween he
lowe doped and deepe well, and he nwell o dec ease
he da k coun a e (DCR) and change he maximum o
he pho on de ec ion e iciency (PDE) [9].
978-1-4673-1260-8/12/$31.00 ©2012 IEEE 332
4) Con ol ga e gua d ing: A biased poly laye ha
changes he shape o he deple ion egion oge he wi h
a well implan a a ce ain dis ance o he ac i e a ea
a e in oduced o inc ease he b eakdown ol age a he
edge o he junc ion [8].
5) Vi ual gua d ing: A highe doped implan is in o-
duced in he cen al egion o he ac i e a ea o enhance
he elec ic ield in his zone [9].
6) Di used gua d ing: A lowe doped and deepe gua d
ing in he pe iphe y o he de ice is o med by di u-
sion, which imposes minimum size cons ain s [10].
7) Shallow T ench Isola ion gua d ing: An STI is used
o implemen ing an a ea-e icien gua d ing, inc easing
he ill ac o and educing he pixel a ea [11].
Al hough cases 5 and 7 seem o be p omising s uc u es
in e ms o scalabili y, hey will no longe be aken in o
accoun since he use o s anda d CMOS echnologies, o
a oid excessi e ab ica ion expenses, imposes es ic ions on
he a ailabili y o laye s. Mo eo e , case 6 will be neglec ed
because o i s minimum size cons ain s and high DCR [5],
which a e e y impo an d awbacks o image applica ions
using senso a ays.
Besides o he a ia ions in he s uc u e, changes in he
SPAD sizes and geome ies a e o in e es in o de o de-
e mine he op imal image sensing esponse. This is done
by a ying he diame e o he ac i e a ea and implemen ing
SPADs wi h ci cula and oc agonal shapes.
Wi h ega d o he inal applica ion, i is in e es ing o
compa e he ela i e pe o mances o he di e en SPADs in
e ms o se e al pa ame e s be o e designing he senso a ay.
These main pa ame e s ha cha ac e ize SPADs ope a ion a e:
•S a ic I-V cha ac e is ic: This measu emen allows o
de e mine he b eakdown ol age o he SPAD and i s
in e nal esis ance o check he alidi y o he assumed
piecewise linea cha ac e is ic.
•Noise: The main sou ce o in e nal noise is usually asso-
cia ed o da k coun s, which a e p oduced by pho ons and
he mally gene a ed ca ie s ha igge a new a alanche.
The a e pulses also con ibu e o he noise. They happen
when he gene a ed ca ie s a e apped by c ys al de ec s
and a e a ce ain ime delay hey a e eleased p oducing
an a alanche [8]. In he case o SPAD a ays, he c oss alk
will also in luence he o al in e nal noise o he de ice.
I occu s i ca ie s gene a ed in one pixel igge an
a alanche in a nea one.
•Dead ime: Lapse o ime du ing which he SPAD
canno de ec he a i al o new pho ons. I comp ises
he quenching, du ing which i akes place he ex inc ion
o he a alanche, and he ese phase, whe e he ini ial
bias condi ions a e es o ed. Some imes, i is possible o
con ol he pe iod o ime be ween hese wo s ages, he
hold o ime, and see how i a ec s he a e pulsing [4].
•PDE: I ep esen s he p obabili y ha a single abso bed
pho on gene a es an a alanche cu en . I s alue depends
on he wa eleng h o he inciden pho ons [2].
Fig. 2. Cha ge con ol model used o ep esen ing he SPAD beha io .
•Fill ac o : When dealing wi h SPAD a ays, i is impo -
an o inc ease he ill ac o , which is he a io be ween
he ac i e a ea and he o al a ea o he SPAD [12].
•Time esolu ion: Va ia ion o he posi ion in ime o
he di e en ca ie a alanche pulses when he SPAD is
illumina ed wi h a pulsed lase [12].
III. MODEL PARAMETRIZATION
T adi ionally, he model used o he SPAD is ha shown
in [13], bu he e a e new e sions like ha in [7] wi h a
mo e accu a e desc ip ion o he de ice. In his wo k he
model in oduced in [7] has been upda ed o deal wi h in e-
g a ed de ices. This has been done by including he pa asi ic
capaci ances p esen in each s uc u e and wi h a p ope
pa ame iza ion o ou speci ic echnology.
Using he in o ma ion p o ided by he simula ion o a
p+/nwell diode om he echnology, a wo-s ep linea I-V
cu e o each one o he SPAD s uc u es has been ob ained.1
Ispad =(Is, VD< Vb
Is+Vn
Rb
ln 1 + e
VD−Vb
Vn, VD> Vb
(1)
Eq. (1) models he ol age dependence o he s a ic cu en
h ough he SPAD, whe e Isis he sa u a ion cu en , Rb is
he in e nal esis ance, Vnis a no maliza ion ol age, Vb is he
b eakdown ol age, and VDis he ol age ac oss he SPAD.
Looking a he di e en diode s uc u es (see Fig. 3), i can
be seen ha he e exis s a pa asi ic nwell/psubs a e diode. As
illus a ed in Fig. 2, his diode has also been inco po a ed in
he model h ough he inclusion o he ca hode- o-subs a e
capaci o o allow a comple e desc ip ion o he dynamic
beha io o he de ice.
The in o ma ion ob ained by simula ing he SPAD h ough
i s model, oge he wi h he di e en quenching ci cui s in-
cluded in he es chip has been used o design he la e
p ope ly. As i will be shown in he nex sec ion, h ee di e en
opologies o he quenching ci cui will be s udied in o de
o compa e hei pe o mance in e ms o noise (da k coun s
and a e pulsing) and ime.
1The diode wid h and leng h ha e been chosen ying o minimize he
ela i e e o o a ea and pe ime e compa ed o ci cula and oc agonal shapes.
333
(a) (b)
(c) (d)
Fig. 3. SPAD s uc u es implemen ed in he es chip: (a) p+/nwell diode
wi h well gua d ing, (b) well/nwell diode, (c) diode wi h con ol ga e gua d
ing, (d) simple p+/nwell diode.
The compa ison be ween he di e en SPAD s uc u es
oge he wi h hei co esponding quenching ci cui s will be
done in e ms o some o he pa ame e s shown p e iously.
To deal wi h he desi ed applica ion, special a en ion will
be paid o he s a ic I-V cha ac e is ic o he di e en es
s uc u es, he equency o he noise sou ces, he dead ime
and he change in he PDE depending on he wa eleng h o
he inciden pho ons and he s uc u e o he SPAD.
IV. CHARACTERIZATION CIRCUITRY AND TESTCHIP
As s a ed be o e, o he ab ica ion o he es chip he s an-
da d 180nm UMC CMOS p ocess has been chosen. Due o he
cons ain s imposed by he laye a ailabili y o he echnology
i is no possible o implemen he i ual gua d ing and
he STI gua d ing s uc u e. Mo eo e , he minimum size
cons ain s o he di used gua d ing s uc u e is a p oblem
when implemen ing SPAD a ays, so i was disca ded. Only
he i s ou SPAD s uc u es al eady desc ibed in sec ion II
will be implemen ed. They a e shown in Fig. 3.
Rega ding he quenching ci cui y, h ee di e en al e na-
i es ha e been implemen ed:
•Simple passi e quenching ci cui , implemen ing he ba-
llas esis o wi h ei he an NMOS ansis o o a passi e
esis ance.
•Addi ion o he p e ious s uc u e o an ac i e ese
b anch o speed up his s age.
•Passi e quenching and ac i e ese ci cui , which includes
a mechanism o con ol he hold-o ime.
This i s wo quenching ci cui s a e pu oge he (see Fig.
4), wi h he possibili y o ac i a ing o no he ac i e b anch.
This allows o s a om he simples case and hen see
he changes when wo king wi h a mo e complex quenching
ci cui . The compa ison will be made in e ms o :
•Time: Va ia ion o he dead ime and ime esolu ion.
(a)
(b)
Fig. 4. (a) Fi s quenching ci cui . (b) Second quenching ci cui .
TABLE I
MAIN CHARACTERISTICS OF THE DIFFERENT SPAD STRUCTURES.
Tes Type Gua d Type o Ac i e Quenching
s uc u e o diode ing shape a ea (φ) ci cui
#1 p+/nwell well ci cula 7µm Fig. 4(a)
#2 p+/nwell well ci cula 14µm Fig. 4(a)
#3 p+/nwell well ci cula 21µm Fig. 4(b)
#4 well/nwell – ci cula 14µm Fig. 4(a)
#5 p+/nwell well oc agonal 14µm Fig. 4(a)
#6 p+/nwell – ci cula 14µm –
p+/nwell well ci cula 14µm Fig. 4(b)
#7 p+/nwell con ol ci cula 14µm Fig. 4(b)
ga e
•Noise: F equency o da k coun and a e pulsing e en s
depending on he quenching ci cui y.
•Abili y o de ec ing he incoming pho ons a di e en
wa eleng hs when adding complexi y o he quenching
ci cui s, which will depend on he dead ime and PDE.
The quenching ci cui al e na i es and he combina ion o
di e en SPAD s uc u es, sizes and shapes esul in o se en
di e en es s uc u es ha a e de ailed in Table I. Tables II
and III show he alues o he geome ic pa ame e s2and he
s a ic I-V pa ame e s o he di e en s uc u es. As said be o e,
his in o ma ion was used o he design o he wo quenching
ci cui s.
Fig. 5 shows he layou o he di e en s uc u es oge he
wi h he es ci cui y. The measu emen s expec ed o be done
wi h he es chip a e he ollowing:
•The I-V cha ac e iza ion will be ob ained o each s uc-
2ADand PD e e s o he dimensions o he junc ion diode, while Awell
and Pwell a e associa ed o he nwell/psubs a e pa asi ic diode.
334
TABLE II
GEOMETRIC PARAMETERS OF THE DIFFERENT SPAD STRUCTURES.
Tes s uc u e PD(µm) AD(µm2)Pwell (µm) Awell (µm2)
#1 22.0 38.5 36.4 105.7
#2 44.0 154.0 58.4 271.7
#3 66.0 346.4 80.4 514.7
#4 46.4 162.4 61.6 286.3
#5 44.0 154.0 55.9 248.8
#6 44.0 154.0 61.6 301.7
#7 44.0 154.0 51.9 214.3
TABLE III
FITTING PARAMETERS OF THE I-V CHARACTERISTIC TO EQ. (1).
Tes S uc u e Vn(mV) Vb (V) Rb (Ω)Is(A)
#1 10 9.68 823.4 9.7×10−18
#2 10 9.70 207.2 2.8×10−17
#3 10 9.73 91.9 5.6×10−17
#4 10 9.70 207.2 2.8×10−17
#5 10 9.70 207.2 2.8×10−17
#6 10 9.70 207.2 2.8×10−17
#7 10 9.70 207.2 2.8×10−17
u e o see how good he i wi h he model is.
•A measu emen o he dead ime in he di e en con-
igu a ions o he quenching ci cui s will be also done,
oge he wi h he ime esolu ion using a pulsed lase .
•The noise will be cha ac e ized a e lea ing he SPADs
in he da k o se e al hou s. The a e pulsing can be
ob ained by compa ing he esul s wi h he ac i e and
passi e echa ge i i s p obabili y is high. I no , i has o
be s udied in a ime scale a ound one o de o magni ude
smalle han o he DCR (besides educing he dead ime,
as s a ed be o e).
•The PDE will be measu ed a e emo ing he in luence
o he da k coun and a e pulsing e en s.
Once he ela i e pe o mance o he di e en es s uc u es
is compa ed, he e will be mo e in o ma ion o be used ha
will allow o ad ance in he design o image senso a ays.
V. CONCLUSIONS
A e iew o di e en SPAD s uc u es and hei p incipal
cha ac e is ics is p esen ed. Taking in o accoun he es ic ions
in e ms o laye a ailabili y imposed by he s anda d 180nm
UMC CMOS echnology, and he inal a ge o implemen ing
image senso CMOS SPAD a ays, h ee o he s uc u es a e
ejec ed. A p op ie a y model based on [7] and adap ed o
in eg a ed de ices is de eloped o he design o he quenching
ci cui y ha is connec ed o he SPADs. The combina ion
o se e al SPADs wi h dis inc size, shape o s uc u e, along
wi h di e en quenching ci cui s esul s in se en es s uc u es
whose ela i e pe o mance in e ms o he main pa ame e s
o he SPADs will be s udied. This wo k p o ides use ul da a
owa ds he implemen a ion o image senso s in a s anda d
CMOS echnology.
ACKNOWLEDGMENT
The au ho s would like o hank he Spanish Na ional
Resea ch Council (Consejo Supe io de In es igaciones
Fig. 5. Layou iew o he di e en s uc u es, oge he wi h he quenching
and es ci cui y.
Cien ´
ı icas, CSIC) o he JAE-PREDOC esea ch g an ha
suppo s he wo k he e p esen ed.
REFERENCES
[1] P. Sei z and A. J. P. Theuwissen, Single-Pho on Imaging. Sp inge , 2011.
[2] F. Zappa e al., “P inciples and Fea u es o Single Pho on A alanche
Diode A ays,” Senso s and Ac ua o s A: Physical, ol. 140, pp. 103–
112, Oc obe 2007.
[3] I. Rech e al., “Monoli hic F on -End Sys em o Pho on Timing
Applica ions,” P oc. o he IEEE LEOS Annual Mee ing Con ., pp. 299–
300, 2009.
[4] A. Galli anoni, I. Rech, and M. Ghioni, “P og ess in Quenching Ci cui s
o Single Pho on A alanche Diodes,” IEEE T ans. on Nuclea Science,
ol. 57, pp. 3815–3826, Decembe 2010.
[5] C. Niclass e al., “A Single Pho on A alanche Diode Implemen ed in
130-nm CMOS Technology,” IEEE J. o Selec ed Topics in Quan um
Elec onics, ol. 13, pp. 863–869, July/Augus 2007.
[6] D. Palubiak e al., “High-Speed, Single-Pho on A alanche-Pho odiode
Image o Biomedical Applica ions,” IEEE Senso s Jou nal, ol. 11,
pp. 2401–2412, Oc obe 2011.
[7] G. Gius olisi, R. Mi a, and G. Palumbo, “Beha io al Modeling o S a-
is ical Phenomena o Single-Pho on A alanche Diodes,” In e na ional
J. o Ci cui Theo y and Applica ions, ol. 40, pp. 661–679, July 2012.
[8] A. Rochas, Single Pho on A alanche Diodes in CMOS Technology. PhD
Thesis, EPFL, 2003.
[9] J. A. Richa dson e al., “Scaleable Single-Pho on A alanche Diode
S uc u es in Nanome e CMOS Technology,” IEEE T ans. on Elec on
De ices, ol. 58, pp. 2028–2035, July 2011.
[10] N. Fa ama zpou e al., “Fully In eg a ed Single Pho on A alanche
Diode De ec o in S anda d CMOS 0.18-µm Technology,” IEEE T ans.
on Elec on De ices, ol. 55, pp. 760–767, Ma ch 2008.
[11] H. Finkels ein, M. Hsu, and S. Esene , “An Ul a as Geige -Mode
Single-Pho on A alanche Diode in 0.18-µm CMOS Technology,”
Ad anced Pho on Coun ing Techniques, P oc. o SPIE, ol. 6372,
p. 63720W, 2006.
[12] G. F. Dalla-Be a e al., “A alanche Pho odiodes in Submic on CMOS
Technologies o High-Sensi i i y Imaging,” Chap e 11 in Ad ances in
Pho odiodes (G. F. Dalla Be a, Edi o ), InTech, 2011.
[13] S. Co a e al., “High-accu acy picosecond cha ac e iza ion o gain-
swi ched lase diodes,” Op ics Le e s, ol. 14, pp. 1341–1343, Decem-
be 1989.
335