Surface passivation of crystalline silicon by amorphous silicon carbide films for photovoltaic applications

“SURFACE PASSIVATION OF CRYSTALLINE SILICON BY
AMORPHOUS SILICON CARBIDE FILMS FOR
PHOTOVOLTAIC APPLICATIONS”
Tesi doc o al p esen ada pe a l’ob enció
del í ol de Doc o
Ra el Fe é i Tomàs
Co-di ec o s: Ramon Alcubilla González
Michael Ve e
Gene 2008
Acknowledgemen s
I would like o hank o my supe iso , Ramon Alcubilla, o in oducing me in he
sola cells ield. He has been always a ailable, a any ime, eaching, discussing and
helping. I canno exp ess all my g a i ude o him o e e y hing he did o me. Michael
Ve e , also my supe iso , spen many hou s wi h me inside he clean oom and b ough
me se e al imes o Ge many, whe e I could widen my ision o sola cells echnology.
Isid o, Pablo and Albe wo ked ha d, bu also spen , like S ephanie, C is óbal, Kim,
Sand a, and Se ane, nice momen s du ing hese las ou yea s. And o cou se, I need o
men ion Moisés, Jo di, Da id, Del ina, Ge a d, I aldo, Lo e o, Lukasz, Mónica and all
he o he s (many) who sha ed as PhD s uden s unce ain ies abou u u e, discussions,
expe iences and a nice clima e o ice. I canno o ge ou clean oom echnicians: Miguel,
Xa i, Ja i and Juan Ca los.
I am ex emely g a e ul o And es Cue as o hos ing me in his g oup a he ANU,
Canbe a, and also o his pa ience a my uncoun able amoun o ques ions. The e I me
many in e es ing people like De k, Dan, Jason, E an, Helmu , and specially Fede, who
was an excellen eache , colleague and iend. To him I also acknowledge he
collabo a ion o his g oup a he EHU, Bilbao, led by Juan Ca los Gimeno. Tho s en
T upke and Robe Ba dos, om UNSW, Sydney, p o ided pho oluminescence
measu emen s o silicon ca bide passi a ed samples.
I also would like o exp ess my acknowledgemen o Jan Schmid o hos ing me a
he ISFH, Hameln, and o his always e y in e es ing commen s. Also om ISFH hanks
o Robe Bock, who p o ided ECV measu emen s, Ba ba a Te heiden, who e iewed he
manusc ip o he p esen hesis, and Ka s en Bo he, o help in co ona cha ge
measu emen s. And o all he echnicians, who kindly o e ed his help a any ime wi h
ex eme p o essionalism. Quique became a good iend sha ing nice discussions, no
always abou science, which we e specially help ul o somebody who does no speak
Ge man, ye .
Thanks o S e an Glunz and his g oup in ISE, F eibu g, whose collabo a ion le us
achie e oge he he i s 20% e icien sola cell wi h silicon ca bide ea side
passi a ion. Thanks o Pe e Roca i Caba ocas and Jé ôme Damon-Lacos e, om LPICM,
École Poli echnique, Pa is, o ellipsome y measu emen s. Thanks o Jo di And eu and
his g oup a UB, Ba celona, o o e ing his collabo a ion and o Jo di Esca é o
simula ions o op ical measu emen s. Also a UB I le many iends, like Olga and
And eu.
Apa om he p o essional help, I would like o exp ess my acknowledgemen o all
my amily, specially my mo he , bu also my a he , sis e , b o he in law and Alex, who
was bo n jus when his hesis s a ed. To all my iends in Ampos a and now, om many
places in he wo ld, in all since i y: g àcies! And he las , bu o me he mos impo an
one: hanks Ri a o sha ing hese momen s wi h me.
This wo k was unded by he Spanish Minis y o Educa ion and Science unde
con ac s TIC2002-04184-C02-01 and TEC2005-02716/MIC.

Abs ac
Du ing he las wo decades he pho o ol aic ma ke has apidly inc eased, pa ly
because o a new ision o sola ene gy as a use ul and compe i i e sou ce o elec ici y
p oduc ion and pa ly because o he educ ion o manu ac u e cos s while keeping
e iciencies high. C ys alline silicon sola cells a e he mos widely s udied cells and up o
da e hey o e one o he bes pe o mances in e es ial applica ions. The key issue o
cos educ ion in hese cells is he educ ion o silicon cos s, by using hinne subs a es
and/o by wo king wi h sola -g ade low-cos silicon ma e ial. In bo h cases educ ion o
ecombina ion o cha ge ca ie s a he su ace, wha is called su ace passi a ion, is
compulso y o achie e he high e icencies (> 20%) expec ed.
The s a e o he a in su ace passi a ion is done by hin ilms o amo phous silicon
ni ide g ow by Plasma Enhanced Chemical Vapou Deposi ion (PECVD). This ma e ial
o e s excellen su ace passi a ion in mos o he sola cell schemes and an i e lec i e
p ope ies a he same ime. Despi e silicon ni ide is e y well es ablished in pho o ol aic
ield, in his hesis we o e an al e na i e ha is based on amo phous silicon ca bide (a-
SiC), also g own by PECVD. The passi a ion p ope ies o silicon ca bide ha e been
al eady s udied in ou g oup inding ha excellen esul s can be ob ained when he ilms
a e ich in silicon, especially o hose doped wi h phospho us o o m a n- ype ma e ial.
Because his ea u e leads o undesi able abso p ion o sola ligh wi hin he ilms ha
does no con ibu e o he pho ocu en , silicon ca bide would hen be elega ed o
passi a e only he ea side o he sola cell.
The aim o his wo k is o imp o e su ace passi a ion p ope ies de eloped
p e iously a UPC, and add compulso y equisi es o he applica ion o c ys alline sola
cells. These equisi es a e: uni o mi y, anspa ency and an i e lec i e p ope ies, s abili y
unde long e m ope a ion and s abili y unde high empe a u e s eps (allowing sc een
p in ing p ocesses). Also i is he willing o p o ide a be e unde s anding o he
undamen al p ope ies o hese ilms.
The main esul s achie ed a e enume a ed he ea e :
- Su ace passi a ion imp o es wi h he ilm hickness and hen sa u a es o ilms
hicke han 50 nm. The mechanism esponsible o his imp o emen is no an
inc ease o he elec ic cha ge in he ilm, as in p inciple could be hough , bu a
be e sa u a ion o de ec s by he p esence o hyd ogen. The amoun o cha ge
densi y seems o be independen o he ilm.
- Expe imen s o co ona cha ge e eal some ea s abou he na u e o he cha ge
densi y o p o ide he ield e ec passi a ion. The o igin o he cha ge seems o
be a con inuous densi y o s a es a he in e ace, a he a ixed cha ge alloca ed in
he ilm.
- None o he a emp s using ca bon ich ilms, which a e anspa en and wi h
an i e lec i e p ope ies, esul ed in excellen su ace passi a ion. Such a emp s
included a ia ion o he deposi ion pa ame e s, dilu ion o plasma wi h
hyd ogen, and in oduc ion o ni ogen o in he phospho us doped a-SiC ilms.
The e o e, up o now i becomes appa en ha i is a undamen al p ope y o
silicon ca bide ilms he necessi y o be ich in silicon o pe o m su ace
passi a ion.
- The way o combine su ace passi a ion and an i e lec i e p ope ies was
applying s acks o di e en a-SiC laye s: one silicon ich and one ca bon ich.
The hickness o he silicon ich laye was op imized o each a ade-o be ween
passi a ion and los o pho ocu en due o he abso p ion in he ilm. The s acks
we e used o passi a e p- ype bases, wi h easonably good esul s, and n
+
- ype
emi e s, wi h e y good esul s.
- A new ma e ial was es ed: a e na y alloy o silicon, ca bon and ni ogen doped
wi h phospho us. This ma e ial was applied o n- and p- ype bases and n
+
- ype
emi e s, p esen ing he bes esul s in su ace passi a ion achie ed by ou g oup,
and compa able o su ace passi a ion eco d achie ed by amo phous silicon
ca bide. Bes composi ion was ich in silicon, and again he use s acks o silicon
ich and ca bon ich ilms was combined success ully.
- S abili y agains he mal p ocesses was es ed on di e en passi a ion schemes.
A e he ea men , he passi a ion is s ongly educed o single silicon ich
ilms, which we e o e ing good ini ial esul s. On he o he hand, he s acks wi h
a second ca bon ich ilm main ain easonably well he su ace passi a ion
p ope ies.
- The good dielec ic p ope ies o he ca bon ich ilms, no p o ided by he
silicon ich ones, esul ed c ucial o pe o m he i s silicon sola cell wi h a-SiC
ea side passi a ion wi h e iciencies abo e 20%.
Table o con en s
In oduc ion..................................................................................................................... 13
CHAPTER 1 .................................................................................................................... 17
1.1 In oduc ion.............................................................................................................17
1.2 Recombina ion mechanisms in silicon ...................................................................18
1.2.1 Radia i e ecombina ion..................................................................................19
1.2.2 Auge ecombina ion.......................................................................................21
1.2.2.1 Basic model o Auge ecombina ion.....................................................21
1.2.2.2 Coulomb enhanced Auge ecombina ion................................................22
1.2.2.3 Gene al models o Auge li e ime ..........................................................23
1.2.3 Recombina ion h ough de ec s. Shockley Read Hall heo y..........................25
1.2.3.1 SRH ecombina ion in he bulk................................................................29
1.2.3.2 Su ace ecombina ion .............................................................................31
1.2.4 Recombina ion in hea ily doped (emi e s) egions........................................34
1.3 Su ace passi a ion echniques...............................................................................36
1.3.1 Field e ec passi a ion....................................................................................37
1.3.2 Sa u a ion o de ec s........................................................................................38
1.4 The e ec i e li e ime as a cha ac e iza ion ool o su ace passi a ion................39
1.4.1 Gene al ela ionship be ween e ec i e li e ime and su ace ecombina ion
eloci y.....................................................................................................................39
1.4.2 Gene al ela ionship be ween e ec i e li e ime and excess ca ie densi y....41
1.4.2.1 S eady s a e ..............................................................................................44
1.4.2.2 T ansien s a e..........................................................................................45
1.4.2.3 Gene al s a e o Quasi S eady S a e (QSS) ..............................................45
1.5 Li e ime measu emen s h ough pho oconduc ance based me hods.......................46
1.5.1 Pho o Conduc ance Decay (PCD) ...................................................................48
1.5.2 Quasi-S eady S a e Pho o Conduc ance (QSS-PC) decay...............................49
1.5.3 Pho oconduc ance measu emen echniques ...................................................49
1.5.3.1 Mic owa e de ec ed.................................................................................49
1.5.3.2 Induc i e coupling....................................................................................51
1.5.4 A i ac s in pho oconduc ance based me hods ................................................53
1.5.4.1 T apping o mino i y ca ie s...................................................................54
1.5.4.2 Deple ion Region Modula ion (DRM).....................................................55
1.6 Pho oluminescence based me hods.........................................................................56
1.7 Simula ion o li e ime cu es o passi a ed wa e s...............................................57
1.7.1 Model o a dielec ic/c-Si in e ace (Gi isch model) .....................................58
1.7.2 Modelling Deple ion Region Modula ion (DRM) e ec .................................61
1.7.2.1 DRM on p-n junc ions..............................................................................62
1.7.2.2 DRM on cha ged insula o s .....................................................................63
1.8 Chap e conclusions................................................................................................65
CHAPTER 2 .................................................................................................................... 67
2.1 In oduc ion.............................................................................................................67
2.2 The mally g own silicon dioxide (SiO
2
).................................................................68
2.3 Amo phous silicon-based compounds....................................................................72
2.3.1 Amo phous silicon ni ide...............................................................................74
2.3.2 Amo phous silicon ..........................................................................................76
2.3.3 Amo phous silicon ca bide..............................................................................78
2.3 Aluminium back su ace ield.................................................................................81

CHAPTER 1
Recombina ion and su ace
passi a ion in c ys alline silicon
1.1 In oduc ion
In c ys alline silicon gene a ion o cha ge ca ie s in excess, elec ons and holes, is
p o ided by he mal ac i i y, elec ical exci a ion, o ligh exci a ion. Opposi e o his
gene a ion he e is he ecombina ion o such ca ie s, in which he annihila ion o
elec ons and holes is assis ed by se e al mechanisms. The a e a which his elec ons
and holes a e annihila ed, i.e., he ecombina ion a e, is de ined by he ela ion:
τ
n
U
∆
=
(1.1)
18
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
whe e ∆n is he densi y o ca ie in excess pe uni olume (o excess ca ie densi y)
and τ is he li e ime o such ca ie s.
I is an objec i e o he sola cell de elopmen o minimize all ecombina ion
p ocesses, so ha he ligh gene a ed ca ie s inc ease hei con ibu ion o he
pho ocu en . In o he wo ds, he con e sion e iciency o he sola cell is inc eased when
he ecombina ion o pho ogene a ed ca ie s dec eases. Such con e sion e iciency
depends s ongly on he spec um, illumina ion in ensi y and he de ice empe a u e.
The e o e, i is e y use ul o de ine he measu emen condi ions in o de o compa e he
p ope ies o di e en cells. The s anda d es condi ions a e de ined by he AM1.5D o
AM1.5G sola spec a wi h a de ice empe a u e o 25 °C. Spec um AM1.5 e e s o he
illumina ion p o ided by he sunligh a he sea le el unde speci ic condi ions [5]. The
index D indica es ha only he di ec ligh coming om he sun is used, while he index G
s ands o he spec um aking in o accoun he di use ligh , oo. The AM1.5G spec um
is also known as 1 sun illumina ion and i will be used in his hesis equen ly.
Be ween he h ee undamen al ecombina ion p ocesses p esen in semiconduc o s,
wo o hem a e in insic and he e o e hey a e una oidable. The hi d one is ela ed o
he p esence o de ec s, p esen majo ly a he su ace o he ma e ial. Su ace passi a ion
will be p esen ed in his hesis as he s a egy o educe ecombina ion a he su ace o
ha e a la ge posi i e impac o he sola cells pe o mance.
In his Chap e we e iew he ecombina ion mechanisms in silicon and he s a egies
o su ace passi a ion. Then, me hods o li e ime measu emen s will be in oduced as a
powe ul ool o cha ac e iza ion o ecombina ion and su ace passi a ion in c ys alline
silicon.
1.2 Recombina ion mechanisms in silicon
Th ee undamen al ecombina ion mechanisms occu in semiconduc o s:
(1) Band o band adia i e ecombina ion
(2) Auge ecombina ion
(3) Recombina ion h ough de ec s in he bandgap
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
19
The wo i s mechanisms depend only on he concen a ion o ee ca ie s (elec ons
and holes) p esen in he bulk silicon, while he hi d one is explici ly dependen on he
numbe o de ec s. The e o e, a gi en doping densi y and illumina ion le el mechanisms
(1) and (2) a e inhe en o he silicon p ope ies, and he only way o educe he o al
ecombina ion is by educing he numbe o de ec s du ing he manu ac u e p ocess. The
ee ca ie concen a ion a e gi en by:
nNnnn D
∆
+
≈
∆
+
≡
0
o n- ype ma e ial (1.2.a)
pNppp A
∆
+
≈
∆
+
≡
0
o p- ype ma e ial (1.2.b)
whe e n
0
, p
0
a e he elec on and hole concen a ions a equilib ium, N
D
, N
A
, he dono
and accep o densi ies, and ∆n, ∆p, he excess ca ie densi ies (also called injec ion
densi ies, o injec ion le el) o elec on and holes, espec i ely. In he absence o
apping e ec s o any elec ic ield bo h excess ca ie densi ies, o elec on and holes,
a e equal, since gene a ion in ol es he c ea ion o elec on-hole pai s, i.e. ∆n = ∆p.
1.2.1 Radia i e ecombina ion
Radia i e ecombina ion is he in e se p ocess o op ical gene a ion. The annihila ion
o an elec on-hole pai leads o he c ea ion o a pho on wi h an ene gy close o he
bandgap ene gy (see Fig 1.1). I depends di ec ly on he a ailabili y o elec ons and holes
and i is gi en by:
(
)
2
i ad
npnBU −=
(1.3)
whe e B is he coe icien o adia i e ecombina ion and n
i
is he in insic ca ie
concen a ion. Since silicon is an indi ec bandgap semiconduc o , he p ocess mus be
assis ed by a phonon o simul aneously conse e ene gy and momen um. The e o e he
p ocess in ol es ou pa icles ( wo cha ge ca ie s, one pho on and one phonon) and has
a low p obabili y o occu ing. This causes a low alue o B, which is calcula ed o be
2 × 10
-15
cm
3
s
-1
by [6,7] a 300 K. Howe e , expe imen al de e mina ions indica e a
signi ican highe alue, B = 9.5 × 10
-15
cm
3
s
-1
[8] o 1.1 × 10
-14
cm
3
s
-1
[9]. The
disc epancy may be a ibu ed o he use o a alue o n
i
highe han accep ed in he
calcula ions [10]. Also exci onic e ec s a e no aken in o accoun in such calcula ions
and a e hough o enhance he ecombina ion a e, being his mani es ed in a highe alue
o B.
20
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
-
+
EC
EV
pho on
Figu e 1.1. Radia i e ecombina ion
Ano he issue o ake in o accoun is ha he adia ion emi ed du ing ecombina ion
can be eabso bed be o e escaping om he silicon, hus he adia i e ecombina ion is
sligh ly in e io han p edic ed. This e ec , called pho on ecycling, is especially
mani es ed in di ec bandgap semiconduc o s and p o okes a dec ease o he adia i e
coe icien B. In silicon, i is commonly accep ed o use he alue B = 2 × 10
-15
cm
3
s
-1
[4].
Combining equa ions (1.1) o (1.3), i is possible o de ine he adia i e li e ime o
each doping ype a low and high injec ion condi ions, espec i ely:
dop
li ad NB
1
,=
τ
(1.4.a)
nB
hi ad
∆
=1
,
τ
(1.4.b)
whe e N
dop
e e s o he densi y o dono (N
D
) o accep o (N
A
) a oms o n- ype o p- ype
silicon. Hence, adia i e li e ime is kep cons an a low injec ion densi y and is in e sely
p opo ional o he excess ca ie densi y a high injec ion le el.
Despi e he analysis o sola cells is usually pe o med a 300 K, i is impo an o no e
ha he adia i e ecombina ion is empe a u e depending h ough he coe icien B. Such
dependence has been explo ed widely in he li e a u e [8,11-13]. I is also well known
ha B is enhanced by he Coulombic a ac ion be ween elec ons and holes, and
dec eases wi h bo h he injec ion and he dopan densi y. Recen ly, Al e ma e al.
[14]
ha e p o ided heo e ical calcula ions o such empe a u e and ca ie densi ies
dependencies. In hei model, he Debye po en ial was inco po a ed o desc ibe he
sc eening e ec s be ween elec ons and holes. Band o band adia i e ecombina ion is
he weakes mechanism in silicon and can be masked by he o he wo, especially Auge
ecombina ion a high injec ion condi ions. Fo ou calcula ions, i is a good
app oxima ion o use a cons an alue o B = 2 × 10
-15
cm
3
s
-1
.
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
21
1.2.2 Auge ecombina ion
1.2.2.1 Basic model o Auge ecombina ion
Auge ecombina ion is a p ocess in ol ing h ee pa icles, one elec on and wo
holes, o ice- e sa. The excess ene gy esul ing om he band o band ecombina ion
be ween an elec on and a hole is ans e ed o a hi d ca ie [15], as shown in Fig 1.2.
In i s simples analysis, he ca ie s a e assumed o be non-in e ac i e ee pa icles [16].
The wo ecombina ion p ocesses (elec on-elec on-hole and elec on-hole-hole) a e
p opo ional o he ca ie densi ies:
(
)
0
2
0
2
pnpnCU
neeh
−= (1.5.a)
(
)
2
00
2
pnnpCU
pehh
−= (1.5.b)
being he o al Auge ecombina ion:
(
)
(
)
2
00
2
0
2
0
2
pnnpCpnpnCUUU
pnehheehAuge
−+−=+= (1.6)
whe e C
n
and C
p
a e he so-called Auge coe icien s. Neglec ing n
02
p
0
and n
0
p
02
he
co esponding li e ime can be w i en as:
[ ]
nnNCnNC
DpDn
Aug
∆∆++∆+
=)()(
1
2
τ
o n- ype c-Si (1.7.a)
[ ]
2
)()(
1
nNCnnNC
ApAn
Aug
∆++∆∆+
=
τ
o p- ype c-Si (1.7.b)
A low injec ion le el densi ies he Auge li e ime depends on he squa e doping densi y
as (τ
Auge
~ 1/N
dop2
):
2
,
1
Dn
lowAuge
NC
=
τ
o n- ype c-Si (1.8.a)
2
,
1
Ap
lowAuge
NC
=
τ
o p- ype c-Si (1.8.b)
bu i is independen on he excess ca ie densi y. On he o he hand, a high injec ion
le el ange li e ime is only dependen on he squa e o he excess ca ie densi y (τ
Auge
~
1/∆n
2
), being independen on he doping densi y:
2
,
1
nC
a
highAuge
∆
=
τ
o n- and p- ype c-Si (1.9)
whe e C
a
is he so-called ambipola coe icien . Unde he assump ion o non in e ac i e
pa icles he ambipola coe icien is de ined as C
a
= C
n
+ C
p
. The mos commonly
employed alues o he Auge coe icien s a e hose de e mined by Dziewio and

22
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
-
+
E
C
E
V
-
-
+
+
Figu e 1.2. Auge ecombina ion
Schmid [17], C
n
=
2.8 10
-31
cm
6
s
-1
and C
p
= 0.99 10
-31
cm
6
s
-1
, esul ing in C
a
= 3.79 10
-31
cm
6
s
-1
. These alues we e aken om he i ings o li e ime measu emen s (τ ~ 1/N
dop2
)
o highly doped n and p- ype wa e s a low injec ion densi ies, and a e alid o
N
dop
> 5 × 10
18
cm
-3
. O he au ho s also epo an in e se quad a ic dependence be ween
Auge li e ime and doping concen a ion o n- ype [18] and p- ype [19] silicon, alid o
N
dop
> 10
17
cm
-3
.
1.2.2.2 Coulomb enhanced Auge ecombina ion
Fo doping densi ies lowe han 10
17
cm
-3
he measu ed ecombina ion li e ime a low
injec ion is lowe han expec ed. Hanglei e and Häcke [15] elaxed he assump ion o
h ee non-in e ac ing, quasi- ee pa icles by conside ing Coulombic in e ac ion be ween
elec on and holes, in which exci ons a e o med. The Auge ecombina ion a e is
s ongly enhanced because o an inc eased densi y o elec ons (holes) in he icini y o a
hole (elec on). As he concen a ion o majo i y ca ie s inc eases his Coulomb-
enhanced Auge ecombina ion is weakened due o he sc eening be ween elec on-hole
in e ac ions. The basic co ec ion consis s o inco po a ing he enhancemen ac o s g
eeh
and g
ehh
a low injec ion densi ies:
2
*
2
,
11
DnDneeh
liAuge
NCNCg ≡=
τ
o n- ype c-Si (1.10.a)
2
*
2
,
11
ApDpehh
liAuge
NCNCg ≡=
τ
o p- ype c-Si (1.10.b)
Using he Auge coe icien s o Dziewio and Schmid [17], Al e ma e al. [20]
de e mined he g
eeh
and g
ehh
enhancemen ac o s as a unc ion o he doping densi ies a
low le el injec ion:
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
23
( )


























⋅
−+=
−
34.0
316
105
anh1441 cm
N
Ng
D
Deeh
(1.11.a)
( )


























⋅
−+=
−
29.0
316
105
anh1441 cm
N
Ng
A
Aehh
(1.11.b)
Wi h he enhancemen ac o s g
eeh
and g
ehh
included in he model, he Auge li e ime does
no ollow he quad a ic dependence wi h he doping densi y. Ac ually, i s o
expe imen al da a o n- ype silicon a low injec ion densi ies show ha he exponen in
N
D
is signi ican ly lowe [21]:
65.124
,
109.1
1
D
liAuge
N
−
⋅
=
τ
(1.12)
A high injec ion densi ies he e a e also some disc epancies be ween he 1/n
2
dependence
p edic ed and he expe imen al e idences. Fi s ly, he measu ed alue o he ambipola
coe icien C
a
has been ound o be la ge han he sum o he low injec ion coe icien s,
C
n
+ C
p
. Values o measu ed C
a
ange be ween 2.8 × 10
-30
cm
6
s
-1
and 1.9 × 10
-30
cm
6
s
-1
[22][23]. Secondly, C
a
seems o dec ease as he injec ion le el inc eases [24].
1.2.2.3 Gene al models o Auge li e ime
Wi h he aim o p o ide a gene al de e mina ion o Auge li e ime alid o all
doping densi ies a any injec ion le el, some au ho s ha e sugges ed di e en models
based on he Coulomb-enhanced e ec :
Schmid e al. [25] and Al e ma e al. [23] used he enhancemen ac o s g
eeh
and
g
ehh
. These ac o s we e assumed o be dependen only on he sc eening beha iou o he
cha ge ca ie s. This allowed hem o eplace he doping densi y in equa ion (1.11) by
n+p. A good ag eemen is ound o in e media ely doped p- ype silicon, hough he
Auge ecombina ion a e is o e es ima ed a low le el injec ion o highly doped silicon.
Mo eo e , his model has no been es ed in n- ype silicon.
The simula ion ool PC1D [26] inco po a es a model ha ixes he Auge coe icien s
C
n
and C
p
o low injec ion condi ions and he C
a
o high injec ion condi ions. Then,
e ec i e Auge coe icien s a e de ined by weigh ing C
a
, C
n
and C
p
wi h he doping
densi ies and ca ie densi ies o be employed in equa ion (1.8) and (1.9):
24
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon








+
+








+
=pN
pC
pN
N
CC
D
a
D
D
llin
e
n
2
,
(1.13.a)








+
+








+
=nN
nC
nN
N
CC
A
a
A
A
llip
e
p
2
,
(1.13.b)
Wi h he coe icien s p o ided by he PC1D he Auge ecombina ion a e ends o be
unde es ima ed. Howe e , he use has he possibili y o modi y C
n,lli
, C
p,lli
and C
a
. In
Chap e 5 we employ his op ion o he analysis o su ace ecombina ion eloci y in n-
ype emi e s.
F om he model p o ided by PC1D, Glunz e al. inco po a e co esponding
enhancemen ac o s and eplace C
n,lli
, C
p,lli
wi h g
eeh
C
n,lli
, g
ehh
C
p,lli
. The model ends
howe e o unde es ima e he ecombina ion a e a mid and high injec ion le els and
o e es ima es i a e y high injec ion le el.
Up o now, he mos comple e pa ame e isa ion o he Auge ecombina ion a e o
all he injec ion le el ange, doping ype, and doping le el is ha p o ided by Ke and
Cue as [27,28]. The model is alid a 300 K. On he one hand, o he low injec ion le el
ange, he highes expe imen al li e ime da a om he li e a u e a e i ed wi h 1/n
1.65
dependence o he doping densi y, wi h di e en i ing pa ame e s o n- and p- ype
silicon. In his case, adia i e ecombina ion is assumed o be negligible, so ha he
li e ime is di ec ly a ibu ed o he Auge ecombina ion. On he o he hand, o high
injec ion le el ange, he adia i e ecombina ion canno be neglec ed, as i s 1/n
dependence becomes s onge as he excess ca ie densi y inc eases. The e o e, his las
is sub ac ed (equa ion (1.3) wi h B = 9.5 × 10
-15
cm
3
s
-1
) om he expe imen al li e ime
da a. Then, he pa ame e isa ion o he Auge li e ime a high injec ion densi ies is
ob ained, wi h 1/n
1.8
dependence wi h he excess ca ie densi y and wi h he same
pa ame e o bo h ma e ial ypes. To p o ide a physical basis o he model, a comple e
pa ame e isa ion ha conside s a h ee-pa icle in e ac ion, cha ac e is ic o Auge , is
sugges ed:
(
)
(
)
(
)
[
]
nnCppCnnCnpU
Auge
∆
∆
+
+
=
3002001
(1.14)
F om he a o emen ioned da a i ing a high and low injec ion le el densi ies, his
au oma ically leads o
[
]
8.02765.0
0
2565.0
0
24
103106108.1 npnnpU
Auge
∆×+×+×=
−−−
(1.15)
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
25
As Auge and adia i e ecombina ion a e in insic p ocesses, i can be assumed ha he
o al ecombina ion a e in silicon is p o ided by he sum equa ions (1.3) and (1.15):
=+=
adAuge in insic
UUU
[
]
158.02765.0
0
2565.0
0
24 105.9103106108.1
−−−−
×+∆×+×+×= npnnp
(1.16)
In such model he adia i e ecombina ion is simply p o ided by a cons an
coe icien , no accoun ing o he Coulombic in e ac ion ha also occu s o his p ocess.
Ne e heless, he dependence is inco po a ed on he o he e ms o equa ion (1.16). The
comple e model p o ides an excellen de e mina ion o he in insic li e ime and i s
alidi y was es ed ca e ully.
The in insic (Auge plus adia i e) li e ime used in his hesis is ha p o ided by
Ke pa ame e isa ion. The easies way o implemen i in any simula ion ool is by he
li e ime de ini ion. Combining equa ions (1.1), (1.2) and (1.16), assuming ha ∆n = ∆p
and conside ing ha he mino i y ca ie densi y is equal o he excess ca ie densi y, i
ollows
( )
×∆++= nNN
AD
in insic
τ
1
[
]
158.02765.02565.024
102103106108.1
−−−−
×+∆×+×+×× nNN
AD
(1.17)
This equa ion can be used o bo h wa e ypes by simply se ing he co esponding
doping densi y and lea ing he o he a ze o. No e ha we ha e changed he alue o he
las e m o equa ion (1.17), which accoun s he adia i e ecombina ion, om 9.5 × 10
-15
in (1.16) o 2 × 10
-15
. Doing his we ake in o accoun he pho on ecycling e ec
explained in sec ion 1.2.1.
1.2.3 Recombina ion h ough de ec s. Shockley Read Hall
heo y
Impe ec ions in he c ys alline silicon due o impu i ies, o c ys allog aphic de ec s,
such as acancies and disloca ions, p oduce a de e mined numbe o s a es wi hin he
bandgap ha ac as ca ie aps o ee elec ons o holes. A ee ca ie apped in he
de ec eleases i s excess ene gy by a mul iphonon emission p ocess, and can be ei he
emi ed again o i s o iginal band o ecombine wi h an opposi ely cha ged ca ie . While
he i s p ocess does no con ibu e o ca ie ecombina ion, he second one is a
dominan mechanism in indi ec bandgap semiconduc o s.
32
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
In SiO
2
/Si in e aces da a o abso p ion c oss sec ion and densi y o s a es we e measu ed
h ough Deep Le el T ansien Spec oscopy (DLTS) in Me al Insula o Semiconduc o
(MIS) capaci o s and used o e alua e equa ion (1.25) [37]. In o he silicon in e aces
ab ica ion o MIS s uc u es is di icul o un iable. Howe e , as a i s app oxima ion is
su icien o conside a single dominan de ec a he midgap o e alua e su ace
ecombina ion.
In o de o cha ac e ize he su ace ecombina ion a e, i is use ul o ela e his
pa ame e wi h he di usion e m o he cu en densi y equa ion (see o example [29]
p. 50). Fo a p- ype, homogeneously doped silicon, and in he absence o any elec ic
ield, he bounda y condi ion o his equa ion a he su ace holds:
( )
s
x
n
qUxn
dx
d
qD =∆
=0
(1.26)
whe e D
n
is he di usion cons an o elec ons and x = 0 deno es he su ace posi ion.
Su ace ecombina ion can be hen cha ac e ized by a su ace ecombina ion eloci y,
S(0), de ined as:
(
)
(
)
(
)
ss nSnSU
∆
≡
∆
=
000
(1.27)
whe e ∆n(0) ≡ ∆n
s
is he excess mino i y ca ie densi y a he su ace. Due o he
dimensions o U
s
(cm
-2
s
-1
) he su ace ecombina ion is desc ibed by a su ace
ecombina ion eloci y measu ed in cm s
-1
, ins ead o a ecombina ion li e ime.
An impo an di e ence be ween bulk ecombina ion ia de ec s and su ace
ecombina ion is ha an elec ic ield is usually ound a he semiconduc o su ace. In
his case, ∆n
s
is a away om ∆p
s
, since he elec ic ield c ea es la ge di e ences
be ween n
s
and p
s
. Hence, i is use ul o in oduce a i ual su ace a he edge o he
space cha ge egion, and de ine an e ec i e su ace ecombina ion eloci y, S
e
, as
[37,38]:
nSU
e s
∆=
(1.28)
whe e ∆n = ∆p is he mino i y ca ie densi y a he limi o he space cha ge egion
c ea ed a he su ace. In con as o ∆n
s
, ∆n can be easily measu ed and con olled by
changing he illumina ion le el.
Unde la band condi ions, he excess ca ie densi y o elec ons and holes a he
su ace, ∆n
s
and ∆p
s
espec i ely, a e equal. In addi ion, assuming uni o m gene a ion o

Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
33
ca ie s h oughou he wa e (p o ided o example by IR ligh ) and low- ecombining
su aces, he excess ca ie densi ies a e cons an h ough he ull hickness o he sample.
Hence, ∆n
s
= ∆p
s
= ∆n. Unde his pa icula case, equa ion (1.24), which ep esen s he
ecombina ion a e a he su ace wi h one de ec loca ed a E
, can be exp essed as
ollows:
0
10
0
10
2
00
))((
np
i
s
S
ppn
S
nnn
npnnn
U++∆
+
++∆
−+∆+∆
=
(1.29)
This equa ion shows ha he su ace ecombina ion eloci y, S (see equa ion (1.27)),
depends on he p ope ies o he su ace s a es (ene gy le el, S
n0
and S
p0
) and also on he
injec ion le el ∆n and he doping densi y. The dependence on he ene gy le el o he
de ec is he same o he Shockley-Read-Hall li e ime (see Figu e 1.4). The main
conclusion is he dis inc ion be ween deep le els, which a e close o midgap ene gies and
e y ac i e om he poin o iew o ecombina ion, and shallow de ec s, which a e
loca ed nea conduc ion o alence band and a e mo e likely o beha e as aps. This is
he eason why o simplici y in he modelling o li e ime cu es we conside a single
de ec loca ed a he midgap posi ion.
The dependence o su ace ecombina ion eloci y, S, on he injec ion le el, ∆n, is
simila o he one desc ibed o SRH bulk li e ime. Tha means ha he su ace
ecombina ion eloci y is cons an unde low and high injec ion condi ions. Rega ding
low injec ion, o a p- ype c-Si wi h a doping densi y o N
A
su ace ecombina ion is
desc ibed by:
0
0
1
0
0
1
0
1))((
1
1
1
n
A
n
p
n
A
nlow
S
N
K
S
n
S
S
p
N
SS ≤
+
=












++
=
(1.30)
whe e K is a posi i e cons an . I can be seen ha S
n0
becomes a undamen al uppe limi
o S
low
in a p- ype ma e ial. This is ela ed o he ac ha unde low-injec ion condi ions
he a ailabili y o mino i y ca ie s, elec ons in a p- ype c-Si, is he limi ing mechanism
o su ace ecombina ion. In case o high-injec ion condi ions, S
high
is cons an and equal
o bo h n- and p- ype c-Si:
0
0
0
00
00
1
p
n
n
np
np
high
S
S
S
SS
SS
S
+
=
+
=
(1.31)
34
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
1.2.4 Recombina ion in hea ily doped (emi e s) egions
Hea ily doped silicon egions a e used as emi e s in sola cells o p o ide sepa a ion
o he pho ogene a ed elec on-hole pai s, achie ed h ough a s ong elec ic ield p esen
a he space cha ge egion o he p-n junc ion. Consequen ly, a ol age appea s be ween
he wo egions. Unde open ci cui condi ions, such open ci cui ol age depends on he
pho ogene a ed ca ie s a he edge o he space cha ge egion as [39]:
( ) ( )
2
0 0 exp
oc
dop i
qV
np n N n n
k T
 
 
≈ ∆ + ∆ =
 
 
 
(1.32)
whe e N
dop
is he base doping and ∆n(0) is he excess ca ie densi y a he edge o he
space cha ge egion. I becomes clea ha o achie e high open ci cui ol ages i is
necessa y o use high doping o bo h base and emi e egions. In p ac ical de elopmen
o sola cells i is common o use p- ype bases o 1 Ω cm and n- ype emi e s wi h shee
esis ances be ween 40 and 100 Ω/sq. This ep esen s a ade-o be ween high V
oc
and
low ecombina ion losses.
Emi e ecombina ion is no a undamen al mechanism, bu he esul o he o he
mechanisms a special gi en condi ions. In he bulk o such hea ily doped egions he
main con ibu ion o ecombina ion is Auge (as i s componen goes wi h 1/n
3
), wi h
some no iceable e ec o he adia i e ecombina ion ( ha beha es like 1/n). The SRH
ecombina ion can in p inciple be neglec ed. The whole ecombina ion in a p-n junc ion is
desc ibed by a model including wo diodes, no mally called he di usion diode and he
ecombina ion diode. The di usion diode akes in o accoun he ecombina ion in he
emi e bulk (mainly Auge ) while he ecombina ion diode accoun s o he losses in he
Space Cha ge Region (SCR) due o a SRH p ocess. The calcula ions o hei cu en -
ol age (I-V) cha ac e is ics gi e simple exp ession o he di usion and ecombina ion
cu en densi ies [40]:






=kT
qV
JJ
edi
exp
0
(1.33.a)








=kTn
qV
JJ
ec
ec ec
exp
0
(1.33.b)
whe e J
0e
and J
0 ec
a e he co esponding mino i y sa u a ion cu en densi ies ha
cha ac e ize he wo diodes, q is he elec onic cha ge, k he Bol zmann cons an , T he
ac ual empe a u e and V he ex e nal ol age applied. In c ys alline silicon, when a single
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
35
ap le el loca ed a he in insic le el, E
i
, and iden ical undamen al SRH li e imes a e
conside ed (τ
n0
= τ
n0
), he ideali y ac o akes he alue n
ec
= 2. Fo o he dis ibu ion o
aps along he bandgap his alue may change.
I is in e es ing o ex ac a ela ionship be ween he emi e li e ime and he mino i y
sa u a ion cu en densi ies as a unc ion o he injec ion densi y [41]. We i s s a
assuming symme ical n
+
-p-n
+
o p
+
-n-p
+
emi e s loca ed a bo h side o a base wi h a
hickness W. Then, unde open ci cui condi ions he o al ecombina ion pe uni su ace
a he emi e is desc ibed by:
0 0
1 1
oc
oc
ec
V
Vn qkT
qkT
emi e e ec
qU J e J e
 
 
= − + −
 
 
 
 
(1.34)
This equa ion akes in o accoun all he ecombina ion occu ing be ween he edge o he
space cha ge egion a he base h oughou all he emi e , up o he su ace. On he o he
hand, he ecombina ion in he whole s uc u e is he sum o all ecombina ion p ocesses,
being in his case he bulk and he wo emi e con ibu ions. Taking in o accoun
equa ion (1.1) his gi es:
2
emi e
e bulk
n W n W U
τ τ
∆ ∆
= +
(1.35)
I is assumed in his equa ion ha ∆n has a cons an alue h oughou he wa e , his is
essen ially ue when he di usion leng h o he ca ie s is la ge han he wa e hickness,
W. Combining equa ions (1.29) (1.31) and (1.32) we inally each o:
(
)
2
0
2
1
i
dope
di
nWq
nNJ ∆+
=
τ
(1.36.a)
(
)
ec
n
i
dop
ec
ec
n
nnN
nWq
J
1
2
0
21






∆∆+
∆
=
τ
(1.36.b)
Unde high-injec ion condi ions, he dominan e m is he one ela ed o J
0e
which educes
τ
e
ollowing a 1/∆n dependence. Ne e heless, a low-injec ion he e m ela ed o J
ec
educes τ
e
om he cons an alue expec ed whe e only he di usion e m is conside ed.
The a e o his educ ion depends on he alue o he ideali y ac o . Then, τ
e
is
dec eased a low and high-injec ion. As a consequence, a maximum in τ
e
loca ed a
medium injec ion should be expec ed. An example o emi e ecombina ion is gi en in
Figu e 1.6.
36
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
1,E-05
1,E-04
1,E-03
1,E-02
1E+12 1E+13 1E+14 1E+15 1E+16 1E+17
Injec ed ca ie s (cm
-3
)
E . Li e ime (s)
Auge + adia i e
ec in SCR
di usion
e ec i e
Figu e 1.6. Recombina ion p ocesses in a p- ype silicon wa e
(N
A
= 1.6 × 10
16
cm
-3
) wi h n- ype emi e s loca ed a he wo sides.
Assump ions: J
0e
= 5
×
10
-14
A cm
-2
, J
0 ec
= 5
×
10
-9
A cm
-2
, n
ec
= 2, W = 300 µm.
1.3 Su ace passi a ion echniques
The high numbe o de ec s a he ba e silicon su aces makes su ace ecombina ion
he dominan mechanism in silicon. The educ ion o such su ace ecombina ion is called
su ace passi a ion and is a equisi e o main ain he mino i y ca ie densi y high and o
achie e high e iciency sola cells.
As s a ed in equa ion (1.32), he open ci cui ol age is di ec ly ela ed o he
p esence o ca ie densi y a he edge o he space cha ge egion o he p-n junc ion.
When hickness o he silicon wa e is dec eased o educe cos s, he p obabili ies o
pho ogene a ed ca ie s o each he ea su ace inc ease. Figu e 1.7 shows simula ed
e iciencies as a unc ion o he wa e hickness epo ed by Abe le [42]. The e we e
conside ed di e en su ace ecombina ion eloci y alues and wo di e en ma e ials,
one wi h good quali y and hence a high di usion leng h and he o he wi h low di usion
leng h. I is e iden ha e iciency will inc ease mono onically wi h inc easing di usions
leng h and by lowe ing su ace ecombina ion eloci y. The mos ema kable ac is ha
o low quali y ma e ials (di usion leng h equal o 200 µm) and su icien ly low alues o
su ace ecombina ion eloci y he e is a hickness, a a ound 50 µm, ha op imizes he
e iciency. The e o e, he educ ion o he su ace ecombina ion eloci y pa ame e , i.e.
su ace passi a ion, is c ucial o he de elopmen o high e iciency sola cells a low
cos s. Physically, his su ace passi a ion is achie ed by wo e ec s: he ield e ec
passi a ion and he di ec sa u a ion o de ec s.
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
37
Figu e 1.7. Simula ions o 1 sun e iciencies o an n
+
-p c-Si sola cell as a
unc ion o he wa e hickness epo ed in e e ence [42].
1.3.1 Field e ec passi a ion
Since wo ypes o ca ie s, elec ons and holes, a e needed o comple e he p ocess,
ecombina ion is maximized when hei popula ion is equal. On he o he hand, i he
popula ion o elec ons and holes is unbalanced, he ecombina ion a e is s ongly
educed. The ield e ec passi a ion consis s o p oducing a band bending a he silicon
su ace, hus c ea ing an elec ic ield ( he e he name) and educing one ype o ca ie s
a he in e ace.
Technically, he elec ic ield can be c ea ed in se e al ways, by deposi ing a cha ged
ilm o by c ea ing a hea ily doped egion. Dielec ic ilms s o ing a high cha ge densi y,
like amo phous silicon ni ide, ha e al eady been success ully applied in sola cell
indus y. I may also be possible o deposi o g ow dielec ic laye s wi h a ela i ely low
cha ge densi y, like he mally g own silicon dioxide, and apply an ex e nal ol age o an
elec os a ic cha ge a he su ace by a co ona cha ging ins umen o p o ide he ield
e ec . In he case o hea ily doped egions, hey can be ei he high-low junc ions wi h he
same ype o impu i ies (p
+
-p o n
+
-n) o p-n junc ions wi h opposi e doping ypes. The
p
+
-p combina ion is commonly employed a ea side o con en ional sola cells using
aluminium, con ac ing he base and achie ing a he same ime wha is called as back

38
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
su ace ield. On he con a y, he p-n junc ions a e usually con ac ed and loca ed a he
on side o he sola cell, hough hey can also be no con ac ed (in ha case hey a e
called loa ing junc ions).
1.3.2 Sa u a ion o de ec s
The high amoun o de ec s a he su ace can be educed by sa u a ing he emaining
silicon dangling bonds. Acco ding o he SRH heo y, his would ep esen a di ec
educ ion o he undamen al ecombina ion eloci ies o elec ons (S
n0
) and holes (S
p0
)
desc ibed by equa ion (1.24). Technologically, his s a egy is usually ela ed o a
passi a ing laye deposi ed o g own o e he c-Si su ace which dec eases N
s
, since
h
is
a cons an o he semiconduc o ma e ial and σ
n
and σ
n
a e pa ame e s in insically ela ed
o he ype o de ec . This s a egy is much mo e sensi i e o he p esence o de ec s a he
silicon su ace han he ield e ec passi a ion, because he ca ie s a e eaching he
in e ace. In his sense, he cleaning s eps be o e deposi ion o g ow h o he passi a ing
ilm a e o pa amoun impo ance o educe N
s
. In he las s ep o RCA [43] o o he
chemical cleaning me hods he wa e is dipped in o dilu ed HF solu ion in o de o
elimina e he na i e oxide a c-Si su ace. Mo eo e , his imme sion ies o achie e a
pe ec co e age o he c-Si su ace dangling bonds by a omic hyd ogen (leading o e y
low S
e
alues [44]). O he app oaches o wa e cleaning ha a e economically less
expensi e han RCA and he e o e mo e sui able o la ge a ea co e age a e d y cleaning
me hods. They no mally consis o subjec ing he wa e o plasma wi h an e ching gas
[45,46]. The me hod is also use ul o oughen he su ace, hus p o iding a ligh apping
scheme. Finally, some passi a ion echniques o educe N
s
a e based on dangling bond
sa u a ion by a omic hyd ogen. Fo ins ance, i is common o add molecula hyd ogen
(H
2
) in o he p ecu so gases when a passi a ing laye is deposi ed by PECVD [47]. The
same idea is exploi ed in annealing he samples wi hin a N
2
/H
2
a mosphe e (Fo ming
Gas).
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
39
1.4 The e ec i e li e ime as a cha ac e iza ion ool
o su ace passi a ion
1.4.1 Gene al ela ionship be ween e ec i e li e ime and
su ace ecombina ion eloci y
I would be desi able o s udy each ecombina ion p ocess sepa a ely. Howe e , all
he ecombina ion p ocesses desc ibed in sec ion 1.2 can occu simul aneously wi hin a
silicon wa e . They can be ea ed as an e ec i e ecombina ion wi h he co esponding
e ec i e li e ime, his las being he measu able pa ame e .
The ecombina ion mechanisms can be conside ed o occu independen ly, so ha he
o al ecombina ion is simply he sum o all p ocesses. I all he mechanisms
a o emen ioned occu , hen i ollows:
su aceemi e SRHAuge adia i ee ec i e
UUUUUU
+
+
+
+
=
(1.37)
Taking in o accoun equa ion (1.1), and assuming cons an excess ca ie densi y along
he whole silicon wa e , he de ini ion o e ec i e li e ime a ises as:
su aceemi e SRHbulkAuge adia i ee
ττττττ
111111 ++++=
(1.38)
In a p ac ical case, he emi e is loca ed a he su ace and hence he su ace
ecombina ion con ibu ion is inco po a ed in he sa u a ion cu en densi y, J
0e
, being
he e o e unnecessa y he inclusion o he su ace e m in he equa ion. In Chap e 5 a
deepe analysis o emi e ecombina ion and passi a ion is pe o med, and i is also
discussed how o sepa a e bulk emi e and su ace emi e con ibu ions.
On he o he hand, he bulk ecombina ion p ocesses a e no mally g ouped oge he :
SRHbulkAuge adia i eb
ττττ
1111 ++=
(1.39)
In all he expe imen s pe o med in his hesis Floa Zone silicon wa e s o e y high
quali y a e employed. The e o e, he p esence o de ec s wi hin he bulk is ex emely low
and he co esponding SRH li e ime will be assumed o be in ini e. On he con a y, SRH
model is employed o analyze he ecombina ion p esen a he in e ace.
40
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
x
y
z
W/2
-
W/2
Figu e 1.8. Schema ic o he calcula ion o he li e ime exp ession in a silicon
wa e illumina ed by he on side.
In he ollowing, we de elop an exp ession o ela e he e ec i e li e ime o he
su ace ecombina ion eloci y, S
e
. A schema ic o he sys em o be analyzed is shown in
Fig 1.8, wi h a silicon wa e wi h hickness W and wi h he wo sides loca ed a he W/2
and –W/2 posi ions, espec i ely. The ollowing easonable assump ions a e used in
o de o simpli y his calcula ion:
• The y- and z- dimensions o he wa e a e much longe han x-dimension.
• τ
b
is uni o m wi hin he wa e and bo h su aces ha e he same S
e
alue.
• The pho ogene a ion a e wi hin he wa e , G
ex
( ), is cons an h ough he whole
hickness leading o a symme ical p o ile o ∆n(x).
The s a ing poin is equa ion (1.1), modi ied in he ollowing way o accoun o all
he ecombina ion p ocesses occu ing in he wa e and inco po a ing he e ec i e
li e ime. Assuming a ecombina ion p ocess pe uni a ea, i holds:
∑
∆
=
ie
a
i
Wn
U
τ
(1.40)
whe e ∆n
a
is he a e age excess ca ie densi y wi hin he wa e de ined as:
∫
−
∆=∆
2/
2/
)(
1
W
W
a
dxxn
W
n
(1.41)
and W is he hickness o he c-Si wa e . Di e en p o iles o ∆n(x) can be ound
depending on he wa eleng h o he ligh sou ce and he su ace ecombina ion eloci ies
a on and back su aces. A high simpli ica ion is achie ed in he equa ion when
uni o m p o iles o ∆n(x) a e conside ed. Fo su icien high e ec i e li e ime alues ∆n
a
can be assumed o be cons an ega dless wha is he illumina ed side and he wa eleng h.
This is because he di usion ime o ca ie is sho e han he ecombina ion ime. Fo
low e ec i e li e ime alues ( o example due o bad passi a ed su aces o low bulk
li e ime alues as mul ic ys alline silicon wa e s) he ecombina ion ime is sho e han
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
41
he di usion ime and he e o e he ca ie densi y is no uni o m. In such cases i is
con enien he use o in a ed ligh o ca ie gene a ion, as he ligh abso p ion is mo e
uni o m along he wa e . Sepa a ing he bulk and he su ace con ibu ions we ha e:
∑∫
−
++=
i
W
W
backs on sbulki
UUdxUU
2/
2/
,,
(1.42)
F om he de ini ion o e e y ecombina ion a e de eloped in sec ion 1.2:
)2/()2/(
)(
,,
2/
2/
WnSWnS
dxxn
U
backe on e
ib
W
W
i
−∆+∆+
∆
=
∑∫
−
τ
(1.43)
We a e assuming o his las exp ession ha τ
b
is uni o m h oughou he wa e , i.e. he
wa e is homogeneous. When iden ical e ec i e su ace ecombina ion eloci ies a bo h
su aces S
e , on
= S
e ,back
= S
e
and a uni o m pho ogene a ion a e conside ed, he ∆n(x)
p o ile is symme ical and hen ∆n(W/2) = ∆n(-W/2). Applying his and he de ini ion o
∆n
a
p esen ed in equa ion (1.41), i ollows:
( )
2/2 WnS
WnWn
e
b
a
e
a
∆+
∆
=
∆
ττ
(1.44)
Finally, τ
e
can be isola ed om he abo e equa ion and de ined as:
(
)
a
e
be
n
Wn
W
S
∆
∆
+= 2/
2
11
ττ
(1.45)
Then, om he measu ed τ
e
alues, di e en ecombina ion pa ame e s o he sample
can be deduced depending on he p eponde an ecombina ion p ocess. Special a en ion
mus be paid o he ela ion be ween ∆n(W/2) and ∆n
a
in o de o accu a ely de e mine
S
e
om τ
e
da a.
1.4.2 Gene al ela ionship be ween e ec i e li e ime and
excess ca ie densi y
In all he echniques he de e mina ion o he li e ime is pe o med h ough
measu emen s o he excess ca ie densi y. A he end o his sec ion we will ha e
de eloped an exp ession ela ing he e ec i e li e ime, he a e age excess ca ie densi y
and he gene a ion a e o ca ie s in he wa e . I is e y con enien o achie e a uni o m
alue o he excess ca ie densi y along he wa e o simpli ica ion o he ela ed
equa ions, as well as equal excess ca ie densi y o elec ons and holes (∆n = ∆p).
48
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
To de e mine he li e ime om equa ions (1.67) o (1.68) he ca ie mobili ies need
o be known. The mobili y, µ, is de ined as he ela ion be ween he mean eloci y o a
ca ie and he elec ic ield which d i s i . The e a e ou e ms con ibu ing o he o al
mobili y, which a e he sca e ing due o he c s alline la ice, µ
L
, he dono densi y, µ
D
,
he accep o densi y, µ
A
, and he elec on-hole sca e ing, µ
eh
o µ
he
. The e o e, mobili y is
dependen on he excess ca ie densi y, ∆n. The e ec i e mobili y is gi en by he sum o
he ou in e se e ms:
11111
−−−−−
+++=
eheAeDeLe
µµµµµ
(1.69.a)
11111
−−−−−
+++=
hehAhDhLh
µµµµµ
(1.69.b)
whe e he sub index e s ands o elec ons and h s ands o holes. Masse i e al. [51]
de eloped an empi ical exp ession o he majo i y ca ie mobili y depending on he
doping densi y which is widely used in de ice simula o s. Howe e , we a e in e es ed in
mobili y o mino i y ca ie s. Tha means aking in o accoun he las e m in equa ion
(1.69). The mos eliable da a abou his pa ame e has been measu ed by Dannhäuse
[52] and K ause [53] who measu ed he sum o elec on and hole mobili ies as a unc ion
o ca ie concen a ion in he in insic egion o a PIN diode. A mo e comple e model
was sugges ed by Klaassen [54]. Ano he issue ha can be aken in o accoun i a e y
ine analysis wan s o be done is ha mobili y o he ca ie s nea he su ace is s ongly
educed due o su ace sca e ing. This e ec has been ex ensi ely s udied o Si/SiO
2
in e aces, showing ha he 2D mobili y o elec ons in an in e sion laye is g ea ly
educed due o in e ace oughness sca e ing [55]. In con as , he li e a u e on his opic
o dielec ic passi a ion ilms is e y sca ce. The only e e ence is he wo k o Elmige
and Kuns [56] ha epo ed elec on mobili ies o abou 250 cm
2
V
-1
s
-1
in highly in e ed
and accumula ed Si/SiN
x
in e aces wi h a cha ge densi y in he dielec ic laye o
Q
= 3 × 10
12
cm
-2
.
1.5.1 Pho o Conduc ance Decay (PCD)
The Pho o Conduc ance Decay (PCD) me hod was i s de eloped in 1955 [57]
whe e he pho oconduc ance o he sample was es ima ed by con ac ing i and measu ing
he I-V cu e. In his me hod he ca ie li e ime is de e mined om he pho oconduc ance
o he sample when he e is no elec on-hole pai s gene a ion by he ex e nal ligh sou ce,
i.e. he sample is in ansien condi ions. The lamp gene a es a sho pulse o ligh which
c ea es an excess ca ie densi y wi hin he wa e . This excess o ca ie s ecombines

Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
49
depending on ecombina ion pa ame e s o he bulk and su aces. Then, wi h no G
ex
( ),
equa ion (1.60) is educed o:
Wq
Wq
n
n
np
np
a
a
e
δ
σδ
σ
δ
µµ
σ
δ
µµ
σ
δ
δ
τ
∆
∆
=






+
∆
+
∆
−=
∆
∆
−=
)(
)(
(1.70)
Hence, in his me hod he only magni ude needed is he pho oconduc ance o he
sample. This can be an ad an age, since any e o in he calib a ion o he ins umen is
cancelled ou in equa ion (1.70). Howe e , a high signal- o-noise a io a he
pho oconduc ance signal is needed o educe he e o in he es ima ion o he de i a i e
e m. As a consequence, ela i ely high τ
e
alues a e needed o ob ain eliable alues.
1.5.2 Quasi-S eady S a e Pho o Conduc ance (QSS-PC)
decay
In his s a e, he ime cons an o he illumina ion sou ce is a ied slowly, so ha a
high illumina ion le els he de i a i e e m in equa ion (1.60) can be neglec ed.
Ne e heless, he de ia ions om he eal li e ime obse ed sugges o keep his e m o
he analysis. The Quasi-S eady S a e is e y use ul om high o low injec ion le el
anges. I is necessa y, howe e , o know he ex e nal gene a ion a he same ime ha he
pho oconduc ance is eco ded.
1.5.3 Pho oconduc ance measu emen echniques
In his sec ion we p esen wo commonly employed me hods o de e mine he
conduc ance o he illumina ed sample wi h a con ac less me hod.
1.5.3.1 Mic owa e de ec ed
This echnique is widely used o de e mine he ecombina ion pa ame e s o c-Si
wa e s [58-60]. In Figu e 1.11, we show a ypical block diag am o a µW-PCD sys em.
The pho oconduc ance, gene a ed by a pulsed ligh sou ce, is es ima ed h ough he
e lec ion o a mic owa e signal which is also applied on he sample. The pulsed ligh
sou ce can be a lase wi h a wa eleng h be ween 900 nm and 1100 nm, a led a ay o a
50
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
s oboscopic ligh o xenon. The IR ligh ag ees wi h he assump ion o a cons an G
ex
( )
wi hin he whole sample hickness. The ising and alling ime o he ligh pulses a e e y
as , since hese imes limi he τ
e
measu emen ange; o ins ance using a lase diode, a
ypical alue is 100 ns.
The mic owa e signal is usually gene a ed by a Gunn diode wi h a ypical equency
o 10 GHz. A ci cula o add esses his powe o he an enna and inally o he sample.
The mic owa e signal pene a es in o he sample and a pa o i is e lec ed depending on
i s conduc i i y. Then, he e lec ed powe a ies linea ly wi h he pho oconduc ance o
he sample. The an enna ecei es he e lec ed powe which is measu ed by he de ec o
h ough he ci cula o .
This sys em needs an accu a e calib a ion p ocedu e which akes in o accoun he
mic owa e powe loses, he e iciency o he an enna, he dis ance be ween he an enna,
he sample and he back e lec o , e c. Ano he impo an limi a ion is he pene a ion
dep h o he mic owa e signal in o he silicon and i s dependence on he esis i i y o he
sample. I is well-known ha lowe esis i i y leads o lowe pene a ion dep h. Fo
ins ance, wi h a mic owa e signal o 10 GHz, he pene a ion dep h anges om 350 µm
o a esis i i y o 0.5 Ω cm o 2200 µm wi h 10 Ω cm. The mic owa e signal only gi es
in o ma ion abou he excess ca ie concen a ion a hose zones whe e i has pene a ed.
Then, i he pene a ion dep h is lowe han he sample hickness, he measu emen can be
e oneous. On he o he hand, i he pene a ion dep h is much longe han he sample
hickness, a mul iple e lec ion phenomenon can occu . This phenomenon can be
calib a ed by changing he dis ance be ween he sample and he back e lec o sea ching a
maximum o he e lec ing signal. Howe e , due o he di e en pene a ion dep hs, he
a ailable injec ion le els a e e y limi ed. Du ing a ypical measu emen wi h a high
powe o inciden ligh , he sample mo es om e y high injec ion condi ions o almos
he modynamical equilib ium. F om he poin o iew o he e lec ion o he mic owa e
signal, he pene a ion dep h s ongly a ies leading o e o s in he measu emen . This is
he eason why his sys em is used only wi h low ligh in ensi ies which gene a e a low
excess ca ie densi y wi hin he sample. In addi ion o his limi a ion, i S
e
has a s ong
dependence on he injec ion le el, non monoexponen ial conduc ance signals a e ob ained
as epo ed when c-Si su ace is passi a ed by he mally g own SiO
2
[37,61], SiN
x
[62] o
e en wi h chemical me hods [63,64].
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
51
Figu e 1.11. Block diag am o a a ypical µ
µµ
µW-PCD measu emen sys em.
To a oid all hese limi a ions, Baso e and Hansen [65] in 1990 in oduced a second
ligh sou ce, known as bias ligh , which gene a es a cons an backg ound ca ie
concen a ion. The lase pulse in ensi y was adjus ed o modi y he excess ca ie
concen a ion in less han a 10%. In his case, he pene a ion dep h o he mic owa e
signal as well as he es o he physical pa ame e s o he sample a e p ac ically cons an
h oughou he eco ded PCD signal. Hence, by changing he bias ligh in ensi y,
ecombina ion pa ame e s o he sample can be de e mined o a wide ange o excess
ca ie densi ies. Howe e , ecen ly B endel [66] ealized ha he es ima ed τ
e
om he
ligh -biased PCD measu emen s is no he ac ual τ
e
bu a di e en ial e ec i e li e ime.
Hence, in o de o ob ain he τ
e
alues, an in eg a ion om he measu ed alues a e y
low injec ion le el is needed [42,64,66,67]
1.5.3.2 Induc i e coupling
This me hod was i s ly applied in Quasi-S eady S a e by Sin on and Cue as in 1996
[68] and cu en ly a come cial equipmen wi h ins umen a ion and so wa e is a ailable
a Sin on Consul ing Inc. company [69]. Since i allowed aking measu emen s in a e y
wide injec ion ange, he unde s anding and modelling o li e ime cu es imp o ed
apidly and i has been e y help ul o analyse su ace and ecombina ion pa ame e s, as
well as de ec s in low-cos sola -g ade ma e ials. Measu emen s o he li e ime cu es a
di e en empe a u es, he so called Tempe a u e and Injec ion Dependence Li e ime
Spec oscopy (TIDLS) [70] allowed he cha ac e iza ion o he ene gy le el o such
de ec s.
52
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
The sys em is based on a RF b idge wi h an induc i e coil ha gene a es elec o-
magne ic ields wi hin he wa e . Va ia ions in he conduc i i y o he wa e modi y hese
ields leading o a a ia ion o he e ec i e induc i e alue, hus changing he ou pu
ol age o he b idge. Such ou pu ol age and he pho oconduc ance ha e a pa abolic
dependence ha is acu ually close o be linea . A e e ence wa e o known conduc ance
is no mally used o pola ize he b idge, and he ol age-conduc ance cu e depends on he
p ope ies o his e e ence wa e . The e o e, i is possible o ha e di e en se s o
ol age-conduc ance cu es o di e en e e ence wa e s. The way o ob ain his
ol age-conduc ance calib a ion is o employ a se o samples wi h know da k
conduc ances. See an example o his in Figu e 1.12. Once he b idge is pola ized, he
da k and ligh conduc ances o he sample o be analyzed a e a ied along he calib a ed
cu e.
A lash lamp is used as exci a ion sou ce, wi h a ime cons an o 2.3 ms o he QSS
mode. To de e mine he ex e nal gene a ion o ca ie s, G
ex
( ), i is necessa y o measu e
he illumina ion le el wi h a e e ence sola cell. Because he ac ion o abso bed ligh
wi hin he wa e will depend on he sample cha ac e is ics (kind o passi a ing laye ,
wa e hickness, and so on) i is impo an o es ima e his quan i y, called op ical ac o .
The easies way o de e mine he op ical ac o is doing a PCD measu en , ha is, se ing
a sho lash lamp ime cons an (abou 20 µs) and chose an op ical ac o un il bo h PCD
and QSS coincide. This me hod is alid o su icien ly high alues o he li e ime. Figu e
1.13 shows he ag eemen be ween a PCD and QSS-PC measu emen .
This sys em has clea ad an ages compa ed o he µW-PCD sys em desc ibed abo e.
The main one is ha once he ligh sou ce is o , he sys em co ec ly de ec s he
pho oconduc ance un il he sample eaches almos he he modynamical equilib ium.
Then, he a ailable ange o excess ca ie densi y has no in insic limi . I he ligh
sou ce is powe ul and he ecombina ion wi hin he sample is low, excess ca ie
densi ies as high as 10
16
-10
17
cm
-3
can be measu ed. On he o he hand, signal- o-noise
a io o he sys em will de e mine he lowe limi , which lay ypically a
∆n = 10
12
-10
13
cm
-3
.
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
53
0,00 0,05 0,10 0,15 0,20
0
1
2
3
4
5
6
V = -13.4 S
2
+ 28.6 S + 0.0085
∆
V (V)
∆
S (S)
Figu e 1.12. Calib a ion cu e o he conduc ance o ol age signal con e sion
o he e e ence wa e . A pa abolic i ing is used o es ima e he conduc ance
o he es sample.
10
11
10
12
10
13
10
14
10
15
10
16
10
17
10
-5
10
-4
10
-3
10
-2
QSS-PC
PCD
E . li e ime
τe
(s)
Excess ca ie densi y
∆
n (cm
-3
)
Figu e 1.13. E ec i e li e ime o a FZ p- ype 6 Ω cm silicon wa e passi a ed
by phospho us doped amo phous silicon ca bide ilms. PCD me hod is used o
de e mine he op ical ac o used in he QSS-PC echnique.
1.5.4 A i ac s in pho oconduc ance based me hods
As we discussed in he p e ious sec ion, equa ions (1.45) and (1.60) can be applied in
mos o he cases wi hou loosing much gene ali y. The condi ions o hose equa ions o
be alid a e epea ed he ea e :
1. equal numbe o elec ons and holes pho ogene a ed ∆n = ∆p
2. open-ci cui condi ions
3. no elec ic ield in he bulk wa e
4. uni o m bulk li e ime

54
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
5. symme ic alue o he mino i y ca ie densi y
6. equal su ace ecombina ion eloci ies a bo h sides
Wi h hese condi ions equa ions (1.45) and (1.60) a e alid. I we add he ollowing
condi ion:
7. low su ace ecombina ion alues o ensu es uni o m alues o ∆n
Then equa ion (1.64) is also alid.
In some cases, condi ions 1, 3 and 7 a e no accomplished, o example nea he
su aces when a ield e ec passi a ion is p esen , o when shallow de ec s ( aps) a e
p esen . The i s case can be skipped i an e ec i e su ace ecombina ion eloci y, S
e
,
is conside ed, as we ha e al eady explained in sec ion 1.4.2.3. Ne e heless, when he
pho oconduc ance echnique is used o measu e he li e ime and condi ion 1 is no
accomplished he e a e wo e ec s ha lead o o e es ima ion o he li e ime a low
injec ion densi ies. These wo e ec s a e called apping o mino i y ca ie s and
Deple ion Region Modula ion (DRM), and hey will be explained below.
1.5.4.1 T apping o mino i y ca ie s
Some de ec s loca ed wi hin he bandgap o he c ys alline silicon ac as aps o
mino i y ca ie s wi hou con ibu ing o he ecombina ion. This is because he ap only
allows he ca ie o e u n o he o iginal ene gy band om which i was eleased. This
phenomenon has been widely s udied since mid-1950's [71,72] and i has been ecen ly
epo ed in QSS-PC measu emen s mainly in low-quali y c-Si subs a es [73,74].
Assuming a p- ype c-Si wa e unde he modynamic equilib ium, he neu ali y o cha ge
equa ion is as ollows:
000
NNnp
TA
+
+
=
(1.71)
Whe e p
0
(n
0
) is he hole (elec on) densi y unde he mal equilib ium, N
A
he accep o
densi y, N
T
he ap densi y and
0
he p obabili y o being occupied which ollows he
Fe mi-Di ac dis ibu ion and depends on he Fe mi le el posi ion. I he sample is
illumina ed, he neu ali y o cha ge equa ion is s ill alid:
NNnp
TA
+
+
=
(1.72)
In his case depends on he quasi-Fe mi le el o elec ons. We a e in e es ed in ∆n, hen
sub ac ing equa ions (1.71) and (1.72):
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
55
Nnp
T
∆
+
∆
=
∆
(1.73)
Then, he pho oconduc ance wi h aps, ∆σ
, is as ollows:
N Nnqpnq
TpTppnpn
∆+∆=∆+∆+=∆+∆=∆
µσµµµµµσ
)()(
(1.74)
This magni ude is used o es ima e τ
e
. Then, he e ec o mino i y ca ie apped is an
inc ease in he pho oconduc ance o he sample and, hence, an o e es ima ion o τ
e
. This
e ec is only isible unde low injec ion condi ions, since a high injec ion ∆n >> N
T
∆
and he excess o holes is equal o he excess o elec ons.
A way o pa ially co ec he apping a i ac s was i s ly sugges ed by R. A. Sin on
[75], and analysed by Macdonald e al [76]. I consis s o applying a bias ligh o keep he
aps cons an ly illed. The pho oconduc i i y is enhanced by a cons an ac o which
would lead o an enhanced appa en ca ie densi y. Once his cons an ac o is
sub ac ed a mo e ealis ic excess ca ie densi y and hence mo e ealis ic li e ime a e
de e mined. The echnique makes i possible o lowe by up o abou one o de o
magni ude he minimum injec ion le el a which he appa en li e ime is co ec o wi hin
an e o o less han 30%. I is no possible howe e o de e mine he low injec ion limi
o he appa en li e ime wi h his me hod.
1.5.4.2 Deple ion Region Modula ion (DRM)
When elec ic ields a e p esen in a ce ain loca ion o he silicon wa e he e
appea s a deple ion egion ha unbalances he popula ion o pho ogene a ed ca ie s (∆n
≠ ∆p). This happens in he p esence o p-n junc ions, high-low junc ions and in cha ged
insula o -semiconduc o in e aces. The wid h o he deple ion egion is dependen on he
ca ie concen a ion, and he e o e i is modula ed wi h he illumina ion le el. As a
consequence, an o e es ima ion o he excess ca ie densi y is encoun e ed i equa ion
(1.68) is applied o he calculus, wi h he co esponding o e es ima ion o he li e ime. A
kind o inc easing “ ail” is obse ed when ∆n dec eases (see Figu e 1.14), some imes
o e coming e en he in insic Auge limi , and clea ly indica ing i s a i icial na u e. This
e ec , called Deple ion Region Modula ion (DRM), was iden i ied o occu especially a
low injec ion densi ies [77,78] and modelled in he p esence o p-n junc ions [79] and
cha ged insula o s [80]. These wo cases will be analysed in sec ion 1.7.
56
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
1.6 Pho oluminescence based me hods
Wi h he aim o ge id o he apping and DRM a i ac s caused by he
pho oconduc ance echnique, T upke and Ba dos e al, om Uni e si y o New Sou h
Wales, de eloped a me hod o analyse he mino i y ca ie e ec i e li e ime om he
pho oluminescence (PL) o he silicon wa e [81,82].
Basically, his echnique consis o de ec ing he pho oluminescence (PL) decay a
a ound 1.12 eV employing a pho ode ec o a e ligh exci a ion by means o a LED
wo king a 870 nm. The pho oluminescence signal is di ec ly ela ed o he adia i e
ecombina ion, so ha he li e ime can be de e mined h ough equa ions (1.1) and (1.3).
Since he adia i e li e ime depends only on ca ie concen a ions and he pa ame e B,
he echnique is ee o conduc ance a i ac s.
In he case o apping a i ac s, he e o in he echnique is e y small [83]. In he
PL li e ime echnique, he measu ed PL in ensi y, I
PL
, is p opo ional o he p oduc o n
and p gi ing
(
)
(
)
(
)
A A APL
NnnNnnnNpnI
+
∆
≈
+
+
∆
∆
=
+
∆
∆
∝
(1.75)
o he p- ype wa e conside ed he e, whe e n
is he densi y o aps and he las line
holds o low le el injec ion condi ions i.e., o ∆n << N
A
and ∆n << n
. Impo an ly, I
PL
is only a ec ed by mino i y ca ie apping in p opo ion o he a io o he np p oduc
wi h and wi hou apping, i.e, in p opo ion o N
A
+ n
unde low injec ion condi ions and
an e en smalle a io o highe injec ion condi ions. Fo ypical sola cells o low
esis i i y he appa en li e ime ep esen s an e o o less han 1%. In he case o DRM
e ec , he e is no e o in conside ing ha he eal li e ime is measu ed, since he
pho oconduc ance is no used o calcula e he mino i y ca ie densi ies.
The e o e, i is possible o measu e he eal li e ime o mul ic ys alline o DRM
a ec ed wa e s. Howe e , he PL signal can only be measu ed in a bi a y uni s, and a
QSS-PC measu emen a mid injec ion le el ange is equi ed o de e mine he eal alue
o he li e ime and ma ch i using a calib a ing ac o wi h he adia i e li e ime.
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
57
10
10
10
11
10
12
10
13
10
14
10
15
10
16
10
17
10
-5
10
-4
10
-3
10
-2
0.85
Ω
cm
E . li e ime
τ
e
(s)
Excess ca ie densi y
∆
n (cm
-3
)
4.5
Ω
cm
Figu e 1.14. Li e ime measu emen s pe o med a UNSW h ough QSS-PC
(symbols) and QSS-PL (lines) me hods o wo silicon p- ype wa e s passi a ed
wi h amo phous silicon ca bide. No e he p esence o he DRM e ec o he
QSS-PC me hod.
In Figu e 1.14 we plo he QSS-PC and QSS-PL measu ed li e imes o a silicon
wa e passi a ed by amo phous silicon ca bide ilms. The measu emen s we e aken a
he Uni e si y o New Sou h Wales using o exci a ion subbandgap illumina ion (870
nm) wi h LED and mul iple a e aging. This allowed eaching e y low injec ion densi ies
o bo h PC and PL echniques. In sec ion 1.7 we discuss he applica ion o DRM e ec
o model li e ime cu es and ex ac impo an pa ame e s, such as he undamen al
ecombina ion eloci y o elec ons o holes and he cha ge densi y p esen a he
in e ace o in he passi a ing ilms.
1.7 Simula ion o li e ime cu es o passi a ed
wa e s
I has al eady been s a ed ha bulk, su ace and emi e ecombina ion can be p esen
a he same ime in a silicon wa e . All ecombina ion mechanisms ha e hei own
heo e ical models o p edic he co esponding li e ime. Thus, he measu ed e ec i e
li e ime can be i ed o ex ac impo an pa ame e s. Fo high quali y ma e ials he bulk
li e ime is a ibu ed only o Auge plus adia i e ecombina ion p ocesses, which a e
well known o e e y doping densi y. On he o he hand, su ace ecombina ion is based
64
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
he wa e conduc ance assuming la bands along he whole wa e , and σ
bb
he de ia ion
o he conduc ance om la band condi ion due o he su ace po en ial (band bending)
ψ
s
. We include a ac o 2 in equa ion (1.87) o ake in o accoun wo symme ical
su aces. The conduc ance e m σ
bb
is de e mined by sol ing he Poisson equa ions nea
he su ace. Fo con enience, he po en ial ψ
s
is used as he in eg a ion a iable, ins ead
o he posi ion x, esul ing in [77]:
( )
(
)
(
)
(
)
(
)
( )
∫
ΨΨΨ
∆Ψ
−∆++−∆+
±=Ψ
s
bulk
pbulknbulk
sbb
nF
enpenn
q
0
00
,
11
2
µµ
β
σ
ββ
(1.88.a)
wi h he auxilia y unc ion F de ined as:
(
)
=
∆
Ψ
bulk
nF ,
(
)
(
)
(
)
(
)
(
)
DApbulknbulk
NNenpenn −Ψ+−∆++−∆+=
ΨΨ
βµµ
ββ
11
00
(1.88.b)
whe e β is q/kT. P io o he calcula ion o σ
bb
, he su ace po en ial Ψ
s
is de e mined by
balancing he cha ge wi hin he semiconduc o and he ixed cha ge Q
, which is loca ed
a he insula o /semiconduc o in e ace.
Fi s , we calcula e he su ace ecombina ion a e by applying he ex ended Shockley-
Read-Hall o malism wi h inpu pa ame e s Q
and ∆n
bulk
. In his calcula ion we assume a
single ecombina ion cen e a he c-Si midgap wi h undamen al ecombina ion eloci ies
o elec ons and holes, S
n0
and S
p0
. Fo he bulk ecombina ion we assume in ini e SRH
li e ime and he Auge conside ed in equa ion (1.17). The explained calcula ion o wa e
conduc i i y is pe o med unde da k and illumina ion condi ions. The o me is
calcula ed wi h ∆n
bulk
equal o ze o, whe eas he la e is pe o med o a ce ain ∆n
bulk
.
This las alue is also used o calcula ing he co esponding pho ogene a ion a e. Thus,
o gene a e a comple e τ
e
(∆n
a
) cu e we scan a gi en ange o ∆n
bulk
calcula ing he
co esponding ∆n
a
and τ
e
alues. No ice ha he inal ∆n
a
scanned ange can s ongly
a y depending on he DRM e ec and, consequen ly, on he Q
alue.
Analy ically, we can ew i e he QSS-PC excess mino i y ca ie densi y using
equa ions (1.86) o (1.88) as
bb ba
nnn ∆+∆=∆
2
(1.89)
whe e
( )
pn
da k bligh b
b
qW
n
µµ
σ
σ
+
−
=∆
,,
(1.90)
is he excess mino i y ca ie densi y conside ing la ene gy bands along he wa e and

Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
65
( )
pn
da kbbligh bb
bb
qW
n
µµ
σ
σ
+
−
=∆
,,
(1.92)
is he excess mino i y ca ie densi y e m due o he su ace band bending. The e m ∆n
b
is wha we expec o ob ain in he ideal case, whe eas ∆n
bb
is a pa asi ic e m ha depends
on he a ia ion o σ
bb
wi h he illumina ion, gi ing ise o he DRM e ec . In o de o
e ec i ely see a ail in τ
e
, ∆n
bb
mus be compa able o ∆n
b
, which means ha ligh mus
in oduce a signi ican a ia ion in ψ
s
. Such a a ia ion is only possible o a highly
in e ed su ace, whe e he spli ing o he quasi-Fe mi le els d ags he su ace po en ial.
Ac ually, he su ace does no need o be in e ed unde da k condi ions, bu i mus
ce ainly be in e ed unde illumina ion. In sligh ly in e ed/deple ed o accumula ed
su aces, ligh can no induce a signi ican a ia ion in ψ
s
and he e ec disappea s.
To show how he DRM is sensi i e o he cha ge densi y, we pe o m in Figu e 1.17
se e al simula ions o a p- ype wa e a ying he pa ame e Q om 10
10
o 10
12
cm
-2
. The
ail s a s o be shown a in a e y na ow ange o Q, be ween 2 and 3 × 10
11
cm
-2
. A
3 × 10
11
cm
-2
he e is a maximum ha in some occasions has been measu ed (Figu e 1.14)
and hen i apidly disappea s a 5 × 10
11
cm
-2
and abo e. The DRM e ec is also e y
sensi i e o he wa e esis i i y. Figu e 1.18 plo s he dependence o he li e ime cu es
on he accep o densi y o p- ype wa e s. Again he e is a e y na ow ange o N
A
in
which a maximum is obse ed, being in he p esen case be ween 1 and 2 × 10
16
cm
-3
.
1.8 Chap e conclusions
In his Chap e we ha e desc ibed all he ecombina ion p ocesses p esen in
c ys alline silicon sola cells. Models ha e been p esen ed o each mechanism in o de
o sepa a e he di e en con ibu ions. The in insic Auge plus adia i e ecombina ion
has been ea ed wi h up- o-da e mos comple e model a 300 K o Ke and Cue as [27].
The ecombina ion h ough de ec s has been desc ibed by he Shockley Read Hall (SRH)
heo y. When wo king wi h high quali y silicon wa e s a negligible SRH ecombina ion
can be assumed. On he con a y, su ace ecombina ion has been ex ensi ely analyzed by
he ex ended SRH heo y. This conside s a dielec ic ilm loca ed a he su ace ha is
able o pe o m su ace passi a ion ei he by sa u a ing he de ec s and diminishing
undamen al ecombina ion eloci ies o by es ablishing a ield e ec passi a ion.
66
Chap e 1: Recombina ion and su ace passi a ion in c ys alline silicon
10
9
10
10
10
11
10
12
10
13
10
14
10
15
10
16
10
17
10
18
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
1x10
10
1x10
11
2x10
11
3x10
11
5x10
11
E . li e ime
τ
e
(s)
Excess ca ie densi y
∆
n (cm
-3
)
Q = 10
12
cm
-2
Figu e 1.17. Simula ions o li e ime cu es including DRM e ec in a insula o /
semiconduc o s uc u e. Assump ions: N
A
= 1.6 × 10
16
cm
-3
, W = 300 µm,
S
p0
= 100 cm s
-1
, S
n0
= 10
4
cm s
-1
.
10
9
10
10
10
11
10
12
10
13
10
14
10
15
10
16
10
17
10
18
10
-7
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
4 x1016
3 x1016
2 x1016
1,6 x1016
1015
E . li e ime
τ
e (s)
Excess ca ie densi y
∆
n (cm-3)
NA = 1014 cm-3
Figu e 1.18. Simula ions o li e ime cu es including DRM e ec in a insula o /
semiconduc o s uc u e. Assump ions: Q
= 3 × 10
11
cm
-2
, W = 300 µm,
S
p0
= 100 cm s
-1
, S
n0
= 10
4
cm s
-1
.
Li e ime measu emen s by Quasi S eady S a e Pho oconduc ance echnique (QSS-
PC) has been p esen ed as a powe ul ool o cha ac e ize he su ace ecombina ion
eloci y and hence he abili y o he ilms o pe o m c ys alline silicon su ace
passi a ion. An a i ac p esen in such measu emen s, Deple ion Region Modula ion
(DRM) has been iden i ied and explo ed ex ensi ely o imp o e he models and ha e
be e de e mina ion o he ecombiana ion pa ame e s, especially he amoun o cha ge in
he dielec ic ilm ha lead o he ield e ec passi a ion.
CHAPTER 2
S a e o he a in su ace
passi a ion
2.1 In oduc ion
An o e iew o he mos impo an ma e ials and echniques o c ys alline silicon
su ace passi a ion is p esen ed in his Chap e . Reasons o choosing a pa icula one
depend on se e al ac o s. Fo example, when applied a he on side o he sola cell we
equi e anspa ency o he ilm and an app op ia e hickness in o de o achie e
an i e lec i e p ope ies. On he o he hand, high e ac i e indexes (i.e. like in me als)
a e desi ed in he ea side in o de o e lec he long wa eleng hs ha ha e eached ha
side. O he issues o ake in o accoun a e he quali y o he silicon wa e used and he
p oduc ion cos s compa ed o he cell e iciency achie ed.
The li e a u e p o ides many de ails abou passi a ion on p- ype bases and
n
+
-emi e s, bu esul s on n- ype silicon bases o
p
+
-emi e s a e a he sca ce. The eason
68 Chap e 2: S a e o he a in su ace passi a ion
is ha adi ionally he sola indus y has ocused on p- ype silicon wa e s due o he
exis ence o a highe eeds ock o his ma e ial om he mic oelec onic indus y. In e es
in n- ype wa e s, and he e o e in i s su ace passi a ion, has ecen ly a isen. The
u iliza ion o n- ype bases has some ad an ages, o example ha passi a ion o n- ype
su aces a a gi en esis i i y is easie han in p- ype bases. Also, n- ype subs a es a e
ee o highly ecombina i e B-Fe o B-O cen es, which appea when low-cos sola -
g ade silicon is used.
2.2 The mally g own silicon dioxide (SiO
2
)
Silicon dioxide (SiO
2
) is one o he mos impo an ma e ials in semiconduc o
manu ac u ing, ha ing played a c ucial ole in he de elopmen o semiconduc o plana
p ocessing. The o ma ion o SiO
2
on a silicon su ace is mos o en accomplished
h ough he mal oxida ion.
The he mal oxida ion o silicon consis s o exposing he silicon subs a e o an
oxidizing en i onmen a ele a ed empe a u e (usually be ween 700-1300 ºC), p oducing
oxide ilms. This p ocess can be done in ube u naces o in Rapid The mal Anneal
(RTA) u naces. When he en i onmen is O
2
he p ocess is called d y he mal oxida ion
while he p ocess in ol ing H
2
O is called we he mal oxida ion. The co esponding
chemical eac ions in ol ed a e exp essed by:
2 2
Si O SiO
+ →
(2.1)
2 2 2
2 2
Si H O SiO H
+ → +
(2.2)
Oxida ion o silicon is no di icul as long as an oxidizing ambien is p esen . The
ele a ed empe a u e used in he mal oxida ion he e o e se es p ima ily as an
accele a o o he oxida ion p ocess, esul ing in hicke oxide laye s pe uni o ime.
Acco ding o he model p edic ed by Deal and G o e [88] he hickness ob ained depends
on he oo squa e o he ime, he p ocess empe a u e, he o ien a ion o he c ys alline
su ace, and he a mosphe e ype. In d y a mosphe e he g owing a es a e e y low. Fo
example, o g ow a 110 nm hick laye (adequa e o an i e lec i e laye s on Si) we need
o expose he wa e 70 min a 1100 ºC. In we a mosphe e he g owing a e is much as e
bu he elec onic p ope ies o he Si – SiO
2
in e ace a e poo e han hose p o ided by
d y oxida ion. The e o e, in o de o achie e good in e ace p ope ies and hick SiO
2
Chap e 2: S a e o he a in su ace passi a ion 69
laye s in a easonably as p ocess i is common o combine d y and we he mal
oxida ion.
F om he poin o iew o su ace passi a ion, silicon dioxide wo ks mainly by
educing he densi y o s a es a he in e ace. Analysis o he cha ge densi y e eal ha
he e is a posi i e ixed cha ge densi y wi hin he dielec ic laye wi h a alue a ound
3 × 10
11
cm
-2
[89], combined wi h a densi y o s a es ha a e nega i ely cha ged.
Al oge he causes a weak ield e ec passi a ion, leading o simila popula ions o
elec ons and holes in excess a he in e ace and he e o e o inc easing SRH
ecombina ion assis ed by de ec s. Fo his eason, wa e cleaning s ep p io in oducing
he wa e in he u nace is a c ucial p ocess. The s anda d p ocedu e in mic oelec onics
o wa e cleaning is he so-called RCA cleaning, which las s ep is a HF dip o emo e
he na i e oxide.
The alue o he ixed cha ge densi y in he SiO
2
is no in insic o he ma e ial. Sai e
al. s udied he cha ge alues o di e en g owing condi ions [90], measu ing low alues
o d y oxida ion (2 - 10 × 10
10
cm
-2
), signi ican ly highe alues o we oxida ion
(2 - 4 × 10
11
cm
-2
), and eally high alues o Chemical Vapou Deposi ion (> 10
12
cm
-2
).
Rega dless he echnique used, he alue o he ixed cha ge esul ed o be app oxima ely
p opo ional o he densi y o s a es a midgap posi ion, D
i ,midgap
. Howe e , highe Q
alues, which would cause highe ield e ec passi a ion, a e no bene icial o he su ace
passi a ion. On he con a y, su ace passi a ion ge s wo se as Q
inc eases due o he
inc emen o D
i ,midgap
. This is he eason why he lowes e ec i e su ace ecombina ion
alues a e ob ained o d y he mal oxida ion.
To imp o e u he su ace passi a ion so-called Fo ming Gas Anneal (FGA)
ea men is applied a e oxida ion is comple e. This ea men consis o exposing he
passi a ed wa e o a H
2
(5%)/N
2
(95%) a mosphe e a 400 °C o a conc e e pe iod o
ime, no mally 30 minu es. The hyd ogen p esen in his a mosphe e di uses easily
owa ds he in e ace and sa u a es mos o he emaining dangling bonds, hus leading o
a educ ion o he in e ace s a es densi y. A u he and almos comple e sa u a ion o
dangling bonds is pe o med by a omic hyd ogen wi h wha is called he Al-anneal
ea men . This consis s o e apo a ing aluminium ilm o e he SiO
2
ollowed by an
annealing a 400 ºC o abou 20 minu es. Du ing his anneal, a omic hyd ogen is c ea ed
om he oxida ion o he aluminium by wa e molecules o med du ing he SiO
2
g ow h.

70 Chap e 2: S a e o he a in su ace passi a ion
The bes esul s o he mally g own su ace passi a ion allow S
e
alues below 10
cm s
-1
o bo h n- and p- ype high esis i i y wa e s (> 100 Ω cm) [91,92]. Fo lowe
esis i e wa e s his excellen su ace passi a ion is s ill p o iding e y good esul s (on
p- ype silicon S
e
< 20 cm s
-1
o 14 Ω cm [90] and S
e
= 41 cm s
-1
o 0.7 Ω cm [93] a e
achie ed). In such esis i i y ange silicon is used as base in elec onic de ices. The
su ace passi a ion p ope ies o silicon dioxide in hese cases ha e been o e come by
ma e ials based on amo phous silicon hanks o ield e ec passi a ion (see nex sec ion).
The con a y is occu ing when hea ily doped egions a e in ol ed o be used as emi e s.
In hese egions he impu i ies a e di used in he wa e s o ming a laye a ound 1 µm
hick wi h doping densi ies exceeding 10
18
cm
-3
. A p-n o a high-low (p-p
+
n-n
+
) junc ion
is hen buil , p o iding al eady a ield e ec passi a ion. The e o e, he ield e ec
passi a ion p o ided by he passi a ing laye has a mino ole and he way o u he
educe ecombina ion is by educing in e ace s a e densi y. The mally g own silicon
dioxide is so a he bes ma e ial o p o ide such educ ion o de ec s. In Figu e 2.1 we
plo he su ace ecombina ion eloci y o phospho us doped bases and emi e s as a
unc ion o he doping densi y a he in e ace o silicon dioxide wi h FGA and Al-anneal
ea men s. The esul s ha e been ex ac ed om e e ence [28].
The end shown in Figu e 2.1 is ac ually a gene al end o all su ace passi a ion
s a egies, i.e. he su ace ecombina ion eloci y inc eases wi h doping densi y due o a
highe p esence o de ec s in oduced a he in e ace.
The s udy o bo on p
+
- ype emi e passi a ion is mo e ecen and wi h sca ce esul s,
since un il he p esen days he use o such emi e s was no as sp ead as he use o
n
+
- ype emi e s in he pho o ol aic ield. One o he mos ele an esul s is ound again
in e e ence [28]. Bo on emi e s wi h di e en esis i i ies ecei ed he mal oxida ion,
some o hem wi h ichlo oe hane (TCA) gas in he oxidizing a mosphe e. Excellen
passi a ion was achie ed, wi h low alues o he mino i y sa u a ion cu en densi y ( o
example o 100 Ω/sq J
0e
was a ound 30 A cm
-2
). Howe e , a e s o ing he samples in
he da k o wo yea s [94], J
0e
inc eased se e ely, bu could be lowe ed o he ini ial
alues by a FGA a 400 ºC. The samples ha ecei ed no TCA had simila ini ial
passi a ion quali y as wi h a TCA, bu i was no possible o eco e he deg ada ion wi h
he FGA ea men .
Chap e 2: S a e o he a in su ace passi a ion 71
10
13
10
14
10
15
10
16
10
17
10
18
10
19
10
20
10
21
10
-1
10
0
10
1
10
2
10
3
10
4
10
5
n-bases
SiO
2
+ FGA
Al-anneal SiO
2
S
e
a low injec ion (cm s
-1
)
Doping densi y a in e ace, N
s
(cm
-3
)
n
+
-emi e s
Figu e 2.1. Su ace ecombina ion eloci y o phospho us doped bases and
emi e s as a unc ion o he doping densi y a he in e ace p o ided by
he mally g own silicon dioxide. Ex ac ed om e e ence [28].
The p incipal d awback o he mal oxida ion o be applied in sola indus y is he
high cos implied, as he empe a u es in ol ed ange be ween 900 – 1100 ºC. The
empe a u e i sel can be an incon enien o low cos mul ic ys alline subs a es, which
can su e om deg ada ion. Ano he p oblem is he low g owing a es. Rapid The mal
Oxida ion (RTO) o silicon dioxide g own by plasma equipmen s help o diminish hese
cos s, bu usually o e poo e su ace passi a ion p ope ies. In hese sense, p omising
esul s ha e been ob ained ecen ly by Hoex e al. [95]. They de eloped SiO
2
ilms
deposi ed by means o Expanding The mal Plasma (ETP) echnique a high deposi ion
a es in he ange o 0.4 – 1.4 µm min
-1
using an a gon/oxygen/oc ame hylcyclo e a-
siloxane (OMCTS) gas mix u e. These plasma-deposi ed SiO
2
ilms yielded e ec i e
su ace ecombina ion eloci ies as low as 54 cm s
-1
on 1.3 Ω cm n- ype silicon a e a 15
min Fo ming Gas Anneal a 600 °C.
F om he poin o iew o applica ion o wo king de ices SiO
2
p o ides he highes
e iciencies because he combina ion o op ical p ope ies and elec ical quali y op imize
he cell e iciency. I was e ec i ely used in he wo ld- eco d e iciency c-Si sola cell
p oduced by he Uni e si y o New Sou h Wales achie ing sola cell e iciencies o
24.0% [96] a s anda d es condi ions (AM 1.5 25 ºC) wi h he concep o PERL
(Passi a ed Emi e Rea Locally di used) cell. Wi h he same concep o cell he
72 Chap e 2: S a e o he a in su ace passi a ion
e iciency was la e imp o ed up o 24.7% [3], which is he p esen e iciency eco d o
c ys alline silicon sola cells wo king a 1 sun illumina ion.
2.3 Amo phous silicon-based compounds
Amo phous silicon-based compounds ha e been sugges ed as al e na i es o
he mally g own SiO
2
o su ace passi a ion, since hey o e low cos p oduc ion and
scalabili y o la ge a ea implemen a ions. The p esence o silicon helps o adjus he
la ice cons an , so ha he densi y o oids and hence he densi y o dangling bonds is
highly educed. Be ween such compounds we can ind amo phous silicon (a-Si), silicon
ni ide (a-SiN
x
), silicon ca bide (a-SiC
x
), and non- he mally g own silicon oxide (SiO
x
).
Thei s oichiome y a ies on he deposi ion condi ions, and a ec s s ongly he op ical
and elec onic p ope ies. Ma e ials wi h a high con en o silicon p esen s ong op ical
abso p ion in he isible-ul a iole ange, wi h low conduc i i y compa ed o c ys alline
silicon ( o example, in silicon ca bide conduc i i y is a ound 10
-6
S cm
-1
a oom
empe a u e [97]). When he con en o silicon is low, ligh abso p ion and conduc i i y
dec ease, hus exhibi ing anspa ency and e y good dielec ic p ope ies.
The me hods o p oduce such ilms a e no mally based in decomposi ion o
hyd ogena ed gases. This ensu es a high con en o hyd ogen ha con ibu es o imp o e
he su ace passi a ion by sa u a ing dangling bonds. And a he same ime his hyd ogen
is bene icial o he passi a ion o de ec s in he bulk o low-quali y sola -g ade silicon.
Plasma Enhanced Chemical Vapou Deposi ions (PECVD) is he mos s udied o hose
me hods, and i is also he echnique used in his hesis o g ow amo phous silicon
ca bide.
Plasma Enhanced Chemical Vapou Deposi ion (PECVD) consis s o he deposi ion
o a ilm om a p ecu so gas mix u e whose molecules a e b oken by means o an
elec ic ield. The gas exci a ion o p oduce he plasma is placed be ween he wo
elec odes ha cause such elec ic ield. Depending on he exci a ion sou ce we can
dis inguish be ween Radio F equency (RF) wo king a low equencies (be ween
10 - 500 kHz), high equency (no mally 13.56 MHz) o e en mic owa e equencies.
The ins umen s can also be di ided in di ec con igu a ions, in which he wa e is placed
in con ac wi h he plasma, o emo e con igu a ions, in which he plasma and he wa e s
Chap e 2: S a e o he a in su ace passi a ion 73
Figu e 2.2. (a) Di ec -plasma eac o exci ed h ough a RF sou ce and (b)
Remo e-plasma eac o using mic owa e exci a ion. Ex ac ed om e e ence
[98].
a e loca ed in di e en chambe s. A schema ic o bo h con igu a ions is shown in Figu e
2.2 (ex ac ed om Re [98]), using in his case silicon ni ide ilms. This ma e ial is
p oduced by he decomposi ion o silane (SiH
4
) and ammonia (NH
3
) gases. In di ec
con igu a ions bo h gases a e exci ed, while in emo e con igu a ions he silane is
in oduced di ec ly in he deposi ion chambe and eac s wi h he decomposed a oms o
ni ogen and hyd ogen coming om he exci a ion chambe . The plasma exci a ion
equency has a s ong impac on he elec onic p ope ies o he esul ing ilm-silicon
in e ace. The eason is ha below he so-called plasma equency (< 4 MHz) ions a e
able o ollow he plasma exci a ion equency and he e o e p oduce a s ong su ace
bomba dmen . Due o he esul ing su ace damage, ilms ab ica ed wi h low- equency
di ec PECVD only p o ide an in e media e-quali y su ace passi a ion on silicon
su aces. Fo una ely, his p oblem can be la gely elimina ed as o exci a ion equencies
abo e 4MHz he accele a ion pe iods a e oo sho o he ions o abso b a signi ican
amoun o ene gy. Hence, ilms p epa ed by di ec PECVD a high equency (13.56
MHz) p o ide a much be e su ace passi a ion han ilms p epa ed a low equency.
The in oduc ion o a emo e chambe imp o es e en mo e he quali y o su ace
passi a ion, as no ion bomba dmen is p oduced in he wa e .
O he ele an deposi ion echniques in de elopmen o passi a ion ilms a e
Spu e ing and Ho Wi e Chemical Vapou Deposi ion (Ho Wi e CVD). Spu e ing
echnique belongs o he Physical Vapou Deposi ion (PVD) amily. The p inciples o
deposi ion a e based on bomba ding a solid silicon a ge wi h ions coming om a plasma
exci ed gas. The p incipal ad an age o spu e ing in indus y is ha no silane gas is used.
This is e y con enien because silane is explosi e. I also has he bene i o p oducing
highly uni o m ilms in composi ion and hickness in e y la ge a eas. Ho Wi e CVD is
80 Chap e 2: S a e o he a in su ace passi a ion
0,0 0,5 1,0 1,5 2,0
1,4
1,6
1,8
2,0
2,2
E
g,op
E
g,op
(eV)
CH
4
/SiH
4
low a io
Figu e 2.5. Op ical gap o amo phous silicon ca bide he me hane o silane low
a io. Ex ac ed om e e ence [32].
Figu e 2.6. O e iew o he bes esul s wi h PECVD-SiC
x
on 90 Ω
ΩΩ
Ω/sq p
+
- ype
emi e s on 2.8 Ω
Ω Ω
Ω cm n- ype Cz o PECVD-SiN
x
and 10 nm hin he mal SiO
2
.
A ows indica e 1 sun illumina ion, numbe s a a ows display he
co esponding implied V
oc
. Ex ac ed om e e ence [117].
used o imp o e he ca bon inco po a ion in he ilms and o p oduce lowe mic o oids
densi y. Consis en ly wi h ou esul s [115,119], he passi a ion achie ed o in insic
ca bon ich ilms is ela i ely low (S
e
< 300 cm s
-1
). When deposi ing silicon ni ide
ilms on op o he SiC laye , o ming s acks, he passi a ion imp o ed due o he ex a
ield e ec p o ided by he high cha ge densi y in he ni ide.

Chap e 2: S a e o he a in su ace passi a ion 81
2.3 Aluminium back su ace ield
Back su ace ield (BSF) consis o c ea ing a high-low junc ion, i.e. p-p
+
o n-n
+
a
he ea side o he sola cell (depending on he doping ype o he subs a e) ha builds a
ield e ec passi a ion. I is common o di use bo on o phospho us o c ea e he p-p
+
o
n-n
+
junc ions, espec i ely. Howe e , o he p-p
+
junc ion he use o aluminium is much
widely sp ead.
Aluminium is a s anda d ma e ial o con ac ing silicon. Aluminium-silicon alloy is
p oduced a a ela i ely low empe a u e o 577 °C ( he eu ec ic empe a u e), wi h
aluminium ac ing as a p
+
accep o . This p ope y is used in o c ea e an al eady con ac ed
p-p
+
junc ion. The p ope ies o he aluminium BSF a e op imal o indus ial sola cells:
in a e y simple s ep i p o ides good con ac , ield e ec passi a ion and e lec ion a he
ea side, inc easing ligh abso p ion. Howe e , he maximum open ci cui ol age
p o ided by his s uc u e is a ound 630 mV, which is insu icien o each e y high
e icien sola cells.
2.4 Chap e conclusions
The p incipal ma e ials used o c ys alline silicon su ace passi a ion a e e iewed.
Silicon oxide and amo phous silicon base hei s a egy in di ec educ ion o in e ace
s a es densi y, silicon ni ide and he so-called aluminium back su ace ield a e based
mainly on ield e ec passi a ion, and inally silicon ca bide appea s o use bo h
s a egies in a simila le el. Rega dless he ma e ial used, he su ace ecombina ion
eloci y no mally inc eases wi h he su ace doping due p ecisely o a majo p esence o
de ec s a he in e ace.
Fo silicon ca bide, he op imum deposi ion pa ame e s used in ou g oup a e
summa ized and will be used in he ollowing Chap e s as s a ing poin o imp o e
su ace passi a ion.
CHAPTER 3
Single phospho us doped
a-SiC
x
(n):H ilms
3.1 In oduc ion
An ex ensi e wo k has al eady been ealized a he Elec onic Enginee ing
Depa men , UPC, de eloping amo phous silicon ca bide ilms o he su ace passi a ion
o c ys alline silicon. PECVD pa ame e s ( low gas a ios, empe a u e, and o al
p essu e) we e widely a ied o ob ain op imum deposi ion condi ions ha minimized he
su ace ecombina ion eloci y [32]. I was ound ha passi a ing p ope ies o
phospho us doped ilms we e much be e han he in insic ones, as men ioned in
Chap e 2. Howe e , u he in es iga ions a e equi ed o be e unde s anding o he
physical p ope ies o he a-SiC
x
/c-Si sys em, bu also o i i s applicabili y o he
indus ial p ocesses o sola cells manu ac u ing. Rela ed o he i s issue, i would be
impo an o de e mine he alue o he cha ge densi y, Q, alloca ed in he amo phous
ilms and i s dependence wi h impo an pa ame e s, as he ilm composi ion, ilm
hickness and wa e esis i i y. Fo he second concep , he deposi ed ilms should keep
84 Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms
high uni o mi y in la ge a eas, and s abili y unde high empe a u e s eps, no mally used
in he sola cell ab ica ion. Fo applica ions o he on side o he sola cells high
ansmission o he ilms is also equi ed. Ano he impo an issue conce ns he doping
o he wa e o be used. In sola cells i is con enien o use wa e s wi h low esis i i ies
o achie e high open ci cui ol ages. A he same ime, he esis i i y alues ha e o be
high enough o ensu e a su icien di usion leng h o ca ie s, so ha hey can each he
con ac s once hey a e gene a ed. The e o e, i is common o use wa e s wi h esis i i es
a ound 1 Ω cm o lowe . This in ol es a ela i ely high doping ha in oduces mo e
de ec s on he ne wo k, being mo e di icul o passi a e. Hence, he passi a ion ilms
ha e o demons a e hei capabili ies on such su aces.
In his Chap e we ocus ou a en ion on phospho us doped a-SiC
x
(n):H ilms o he
passi a ion o low esis i i y p- ype wa e s, s a ing a he poin o op imum composi ion
de e mined in e e ence [32]. Tha composi ion is ich in silicon, leading o ela i ely low
alues o he op ical bandgap (a ound 1.8 eV) and p esen ing a signi ican abso p ion o
he isible ligh . The e o e, he applicabili y o he ilms is ocused on ea side
passi a ion o sola cells. The main pa ame e s analyzed a e he o al p essu e in he
chambe , he adio equency (RF) powe and he ilm hickness. Finally, op imized
silicon ca bide ilms ha e been applied o wa e s o di e en esis i i ies. In all cases he
quali y o passi a ion has been de e mined h ough li e ime measu emen s by QSS-PC
echnique. The τ(∆n) cu es ha e been analyzed in o de o ex ac he ecombina ion
pa ame e s men ioned in Chap e 1.
3.2 Wa e cleaning
Con aminan s p esen on he su ace o silicon wa e s ha e o be emo ed in o de o
ob ain high pe o mance and high eliabili y semiconduc o de ices, and o p e en
con amina ion o p ocess equipmen . Along his hesis, wo cleaning me hods ha e been
employed:
- Pi anha cleaning.
I consis s o a 2:1 H
2
SO
4
:H
2
O
2
solu ion. This mix u e gene a es an exo he mal eac ion
which boils spon aneously. The silicon wa e is imme sed in i o 10 minu es, in which
he i s monolaye s a e oxidized. A e wa ds he wa e is insed unde wa e and
immedia ely dipped in a 20:1 H
2
O:HF solu ion o 10 seconds. The na i e oxide g own is
Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms 85
emo ed and he wa e comes ou d y because o he hyd ophobic cha ac e o he non-
oxidized su ace.
- RCA cleaning.
The RCA clean is he s anda d o emo ing con aminan s om wa e s in he
mic oelec onic indus y. Ke n and Puo inen de eloped he basic p ocedu e in 1965 while
wo king o RCA (Radio Co po a ion o Ame ica) [43]. I has ou s eps used
sequen ially:
I. O ganic Clean: Remo al o insoluble o ganic con aminan s wi h a 5:1:1
H
2
O:H
2
O
2
:NH
4
OH solu ion boiling o 10 minu es a 75 °C.
II. Oxide S ip: Remo al o a hin silicon dioxide laye whe e o ganic
con aminan s may accumula ed as a esul o I, dipping he wa e in a 40:1
H
2
O:HF solu ion.
III. Ionic Clean: Remo al o ionic and hea y me al a omic con aminan s using a
solu ion o 6:1:1 H
2
O:H
2
O
2
: HCl o 10 minu es a 75 °C.
IV. Oxide S ip: Remo al o a hin silicon dioxide laye whe e me allic
con aminan s may accumula ed as a esul o III, dipping he wa e in a 40:1
H
2
O:HF solu ion.
Du ing hese wo p ocesses silicon is no e ched, and only a e y hin laye o silicon
dioxide is emo ed du ing he HF dip. When inished, he su ace is comple ely d y and
wi hou esidues.
3.3 Thickness measu emen s by p o ilome y
Thickness cha ac e iza ion is o pa amoun impo ance o he deposi ed ilms. In he
pa icula case o silicon ca bide, su ace passi a ion is imp o ed wi h inc easing ilm
hickness. I is also desi able o con ol he deposi ion a e and he uni o mi y o he
applicabili y o indus ial p ocesses. Ellipsome y and p o ilome y a e employed in his
hesis o measu e a-SiC
x
ilms hickness.
Two ypes o p o ilome e s we e employed along his hesis: KLA Tenco and
Dek ak 3030 Sloan Inc. The hickness o he ilms de eloped lay below 200 nm. In such
small nanome ic scale, measu emen s would be di icul in a single s ep. The e o e,
silicon samples wi h e apo a ed Al s ips (a ound 500 nm high) sepa a ed 100 µm we e

86 Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms
Y (a.u.)
X (a.u.)
48 nm
aw da a
co ec ed da a
base line
Figu e 3.1. P o ilome y measu emen and base co ec ion o de e mina ion o
a-SiC
x
ilm hickness.
Table 3.1. Op imal PECVD pa ame e s o he deposi ion o a-SiC
x
(n):H ilms
applied a he passi a ion o 3.3 Ω cm wa e s in Re [32], and he s udies
pe o med in his Chap e o 0.85 Ω cm wa e s
SiH
4
+PH
3
(sccm)
CH
4
(sccm)
empe a u e
(ºC)
o al
p essu e
(mTo )
RF
powe
(W)
deposi ion
ime
(min)
s anda d p ocess
o 3.3 Ω cm 29.5 24.3 400 300 15 5
his wo k unal e ed
unal e ed unal e ed 250 – 450 15 – 90 1 – 12
in oduced in he PECVD eac o a e e y deposi ion. A e wa ds, he Al was s ipped
wi h HCl in a li -o p ocess, emaining only a-SiC
x
s ips. Due o he silicon wa e
cu a u e, i is necessa y o pe o m a base co ec ion be o e de e mining he hickness.
The de ail can be obse ed in Fig 3.1.
3.4 Su ace passi a ion o 0.85 Ω cm p- ype c-Si
The p esen sec ion is ocused on he su ace passi a ion o low esis i i y, plana
silicon wa e s. The subs a es used we e Floa Zone, <100> o ien ed, p- ype, 0.85 Ω cm,
double side polished, 400 µm hick. As he wa e s we e new and had al eady a RCA
cleaning s ep om he manu ac u ing company, a pi anha cleaning was conside ed o be
su icien o achie e good ini ial p epa a ion o he su ace.
Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms 87
The silicon ca bide ilms de eloped in his hesis a e deposi ed by a PECVD sys em
(Plasmalab DP-80 om Ox o d Ins umen s) wo king in di ec con igu a ion a high RF
equency (13.56 MHz). The pa ame e s o he deposi ed ilms can be selec ed by he use
o p oduce he desi ed e ec on he su ace passi a ion. These a e:
- Gas lows o he p ecu so gases: silane (SiH
4
), phosphine (PH
3
) and me hane
(CH
4
)
- To al p essu e o he chambe
- Subs a e empe a u e
- RF powe densi y
- Deposi ion ime
Symme ical PECVD deposi ions we e pe o med in o de o measu e he e ec i e
li e ime. Be o e he second deposi ion he wa e s we e dipped again in HF. To s a ou
analysis o su ace passi a ion on 0.85 Ω cm wa e s, we depa om op imum condi ions
al eady ound in e e ences [32] and [115] o 3.3 Ω cm. These p e ious expe imen s
indica e ha he silicon ich composi ion is c ucial o achie e good su ace passi a ion, as
well as a ela i ely high deposi ion empe a u e (400 ºC). The e o e, hese pa ame e s a e
kep unal e ed, while o al p essu e in he chambe , RF powe and deposi ion ime a e
sys ema ically a ied. Table 3.1 summa izes he p ocess de eloped in his Chap e .
3.4.1 Dependence on chambe p essu e
In his sec ion we ixed he RF powe a 15 W and he deposi ion ime a 7 minu es.
A a ia ion o he o al p essu e in he chambe was done. A e he second side
deposi ion, he wa e s we e immedia ely subjec ed o a Fo ming Gas Anneal ea men a
430 ºC o 30 minu es o imp o e he su ace passi a ion. F om li e ime measu emen s
we ex ac ed alues o e ec i e ecombina ion eloci y a 1 sun illumina ion, using he
equa ion:
W
S
e
bulke
2
11 +=
ττ
(3.1)
As explained in Chap e 1, he high bulk li e ime o he wa e s and he good su ace
passi a ion p o ided by he amo phous silicon ca bide ilms make his equa ion an
excellen app oxima ion o he eal alue o S
e
. We only conside ed Auge li e ime in he
de e mina ion o
τ
bulk
, since we used high-quali y Floa Zone (FZ) c-Si wa e s. The e o e,
ecombina ion SRH ecombina ion h ough de ec s in he bulk can be neglec ed.
88 Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms
250 300 350 400 450
5
10
100
650
660
670
680
690
700
710
720
S
e
(cm s
-1
)
chambe p essu e (mTo )
V
oc
(mV)
Figu e 3.2. Su ace ecombina ion eloci y (S
e
) and implied open ci cui
ol age (V
oc
) o p- ype c-Si, 0.85 Ω cm, 400 µm hick wa e s, passi a ed by
silicon ich a-SiC
x
(n):H ilms. The g owing condi ions a e speci ied in Table
3.1. An op imum poin is ound a 335 mTo , wi h S
e
= 7 ± 2 cm s
-1
and V
oc
up
o 709 mV.
Pa icula ly, we used Auge li e ime pa ame e iza ion sugges ed by Ke e al. [27],
which is desc ibed in equa ion (1.17).
Fig 3.2 plo s he alues o he su ace ecombina ion eloci y a 1 sun illumina ion
as a unc ion o he o al p essu e in he chambe , showing a minimum o S
e
= 7 ± 2
cm s
-1
a 335 mTo . The implied open ci cui ol age, V
oc
, a 1 sun illumina ion is also
p esen ed in his Figu e. This pa ame e is calcula ed om he QSS-PC da a and
ep esen s an uppe limi o he open ci cui ol age in a wo king sola cell. I is shown
ha e y high V
oc
alues, up o 709 mV, can be achie ed when combining low esis i i y
wa e s wi h silicon ca bide passi a ion. Despi e he quali y o passi a ion is good in all
cases (S
e
is kep below 100 cm s
-1
, while V
oc
is abo e 660 mV) he e is a na ow ange in
p essu e o ge eally low alues o he su ace ecombina ion eloci y.
In o de o ge undamen al in o ma ion o he ecombina ion and passi a ion
mechanisms, he co esponding li e ime cu es we e simula ed using he i ing ou ine
de eloped a UPC by M. Ga ín and I. Ma ín [80], which employs he Gi isch model o
ex ac he cha ge alloca ed in he laye and he undamen al ecombina ion o holes S
p0
[86]. The i ing ou ine includes he Deple ion Region Modula ion (DRM) e ec ,
a ou ing he accu acy in he de e mina ion o cha ge densi y. Op ionally, i can be
swi ched o o hose cases in which a cons an bias ligh is applied. As explained in
Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms 89
Chap e 1, he DRM e ec is pa ially masked unde hese condi ions. A he ime o
pe o ming he expe imen s a ying he p essu e, ou QSS-PC se -up was no co e ed, so
ha a bias ligh was cons an ly illumina ing he measu ed wa e s. The e o e, simula ions
a e p esen ed wi hou inco po a ing DRM e ec , educing sligh ly he accu acy in he
de e mina ion o he cha ge densi y. Figu e 3.3 shows an example o li e ime
measu emen o he bes a-SiC
x
(n) su ace passi a ion ob ained o his esis i i y,
oge he wi h he simula ed cu e and he pa ame e s ex ac ed. The co esponding bulk
li e ime a 1 sun illumina ion p o ided by he Auge and he adia i e limi is 1.4 ms. This
esul s in a e y low su ace ecombina ion alue S
e
= 7 ± 2 cm s
-1
demons a ing he
good le el o passi a ion achie ed by silicon ca bide. The supe io and mo e conse a i e
limi o he su ace ecombina ion eloci y is calcula ed by assuming in ini e bulk
li e ime, hus S
e
= W/2τ = 24 cm s
-1
. Fig 3.4 shows o he examples o expe imen al
li e ime cu es o he p essu e se ies, wi h he co esponding simula ed cu es. I can be
seen ha in some cases he ag eemen be ween expe imen al and simula ed cu es is no
comple e. Apa om expe imen al sou ces o e o s (miscalib a ions, mobili y models
and so on) ha would p oduce e o s in he li e ime alues, he causes could be a ibu ed
o he simplici y o he model. In o de o imp o e i , some mechanisms can be aken in o
accoun [89]:
- An in e ace s a e densi y (o densi y o s a es), causing a cha ge Q
i
ha is
a iable wi h he Fe mi le el. No mally, in silicon ni ide ilms his alue is
neglec ed agains he high posi i e ixed cha ge, Q
. To include i in he model i
would be necessa y o know he densi y o s a es, D
i
, and he cap u e c oss
sec ions, σ
n
σ
p
. This has been possible in good dielec ic ilms like silicon oxide,
in which C-V measu emen s allow de e mina ion o densi y o s a es and Deep
Le el T ansien Spec oscopy (DLTS) p o ides in o ma ion abou he cap u e
c oss sec ions. Howe e , hese wo echniques a e e y di icul o apply o
conduc i e ilms like silicon- ich a-SiC
x
.
- Recombina ion in he space cha ge egion c ea ed by Q
and Q
i
, signi ican a
low injec ion densi ies. I is usually modelled by a ecombina ion diode wi h he
co esponding sa u a ion cu en densi y (J
ec
).
- Shun unnels h ough he space cha ge egion. The mino i y ca ie s can unnel
h ough he po en ial ba ie and ecombine a he in e ace. I is modelled by a
pa allel esis ance R
p
.
96 Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms
ins ead o using C-V measu emen s, he analysis o he cha ge densi y was ca ied again
by simula ions o he li e ime cu es using he Gi isch model.
The expe imen al p ocess de eloped in his sec ion is he same as in he wo p e ious
sec ions. The RF powe was kep a 30 W o achie e homogenei y in a leas 4” wa e s,
he o al p essu e a 335 mTo and he deposi ion ime (labelled as
dep
) was a ied om
1 o 12 minu es. We measu ed he hickness in he ange o 10 – 100 nm, inding a linea
ela ionship be ween his pa ame e and he deposi ion ime. As usually in his Chap e , a
FGA a 430 ºC o 30 minu es was applied be o e li e ime es ing.
3.4.3.1 Analysis o cha ge densi y and undamen al ecombina ion
eloci y
Fig 3.10 shows he τ
e
alues a 1 sun illumina ion o
dep
anging om 1 o 12
minu es. As i can be obse ed, τ
e
inc eases wi h deposi ion ime up o
dep
= 5 min,
indica ing an imp o emen in su ace passi a ion. The co esponding e ec i e su ace
ecombina ion eloci ies a e also shown in Fig 3.10. S
e
alues when
dep
is highe han 5
minu es a e abou 24 – 27 cm s
-1
sugges ing ha 40 nm is he minimum ilm hickness o
ob ain an op imum su ace passi a ion.
In Fig 3.11, we show he expe imen al τ
e
(∆n) cu es oge he wi h he modelled
cu es o ou di e en hicknesses. F om he heo e ical cu es we de e mined an
almos cons an Q alue o 3.4 ± 0.2 × 10
11
cm
-2
. Mo eo e , he dependence o S
p0
on ilm
hickness also plo ed in Fig 3.10 ollows he dependence ound o τ
e
, S
e
sugges ing
ha S
p0
is he pa ame e esponsible o he su ace passi a ion imp o emen .
The esul s ob ained sugges wo possibili ies o he o igin o he cha ge densi y ha
causes he ield e ec passi a ion. The i s possibili y is ha i could be alloca ed in a
laye na owe han 10 nm ( his is he minimum hickness used in he expe imen ). The
second possibili y is ha i depends only on he in e ace s a es densi y, D
i
, being hen
a ibu ed o he illed numbe o s a es acco ding o he Fe mi le el posi ion. In o he
wo ds, i would be dependen only on he wa e esis i i y. This las hypo hesis is
ein o ced by he esul s ob ained in nex sec ion and in Appendix I, in which we analyze
he p ope ies o silicon ca bide passi a ion on wa e s wi h di e en esis i i ies. In case
his second possibili y was ue, i would ep esen an impo an di e ence compa ed o

Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms 97
amo phous silicon ni ide, wi h high cha ge densi y alues in he ange o 10
12
cm
-2
,
ega dless he esis i i y o he wa e . Al hough u he esea ch is needed o con i m he
Q alues o a-SiC
x
:H ilms h ough independen measu emen s, addi ional expe imen s
on su ace passi a ion o bo on-doped emi e s has also poin ed o he same di ec ion. I
is well-known ha he high posi i e Q alues o SiN
x
ha e led o poo esul s o
passi a ion o bo on-doped emi e s [28]. Expe imen s epo ed in Re [120] and [121]
yielded lowe S
e
alues o a-SiC
x
:H ilms on bo on-doped emi e s compa ed o hose
ob ained wi h SiN
x
. On he o he hand, he e olu ion o he S
p0
alues wi h ilm hickness
indica es a highe su ace quali y, p obably due o he sa u a ion o dangling bonds by
a omic and/o molecula hyd ogen.
In o de o cla i y he e olu ion o in e acial hyd ogen, we pe o med addi ional ilm
deposi ions o 1, 2 and 5 minu es (co esponding o hickness o abou 10, 20 and 40 nm
app oxima ely). Consecu i e anneals in o ming gas o 10 minu es a 430 ºC we e
pe o med measu ing he τ
e
alue a 1 sun illumina ion a e each s ep. Fig 3.10 shows
he e olu ion o he measu ed τ
e
alues as a unc ion o he annealing ime. The alues
a e no malized o he alue o he as-deposi ed samples. As a gene al end, he e ec i e
li e ime and, hence, su ace passi a ion imp o es o sho FGA while he e is decay o
long anneals. Conside ing ha he main sou ce o hyd ogen is he a-SiC
x
(n):H ilm, his
beha io could be due o a compe ing mechanism be ween hyd ogen-co e ing o he
in e ace dangling bonds and dehyd ogena ion o he ilm. Ac ually, iden ical anneal bu
in N
2
a mosphe e has led o simila imp o emen in su ace passi a ion indica ing ha he
H
2
p esen in he o ming gas plays a mino ole, i any.
The e ec o he inc easing FGA ime has been analyzed o he 10 nm hick sample
(
dep
= 1 min). Simula ions o he τ
e
(∆n) cu es indica e a cons an densi y o cha ge
equal o (3.7 ± 0.1) × 10
11
cm
-2
and S
p0
alues inc easing om 2550, o 4250 cm s
-1
o
annealing imes anging om 0 o 30 minu es, con i ming he hypo hesis o
dehyd ogena ion. The in e ace quali y and hence he S
p0
a e s ongly a ec ed a he same
ime by he ilm hickness and he annealing ime. Fu he mo e, i is clea om he esul s
in Fig 3.11 ha hicke ilms a e mo e obus and ha e be e p ope ies unde FGA
ea men .
98 Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms
0 2 4 6 8 10 12
0
100
200
300
400
500
600
0
500
1000
1500
2000
2500
3000
3500
4000
0 20 40 60 80 100
E . li e ime a 1 sun
τe
(
µ
s)
deposi ion ime
dep
(min)
hickness (nm)
S
e
= 24 cm s
-1
56
33
27
221
26
S
p0
(cm s
-1
)
Figu e 3.10.
Expe imen al (ci cles) τ
ττ
τ
e
and S
e
alues a 1 sun illumina ion s.
deposi ion ime,
dep
. Also shown a e S
p0
alues de e mined om he simula ions
(squa es). Co esponding ilm hickness is ep esen ed on he uppe axis. Solid
lines a e a guide o he eye.
10
12
10
13
10
14
10
15
10
16
10
17
10
-5
10
-4
10
-3
a 1 sun
E . li e ime
τ
e
(s)
Excess ca ie densi y
∆
n (cm
-3
)
dep
= 5 min - ~ 42 nm
3,5 min - ~ 30 nm
2 min - ~ 22 nm
1 min - ~ 11 nm
Figu e 3.11.
Expe imen al (symbols) and simula ed (lines) τ
ττ
τ
e
s ∆
∆∆
∆n cu es o
p- ype 0.85 Ω cm wa e s passi a ed wi h single a-SiC
x
(n):H ilms wi h di e en
ilm hicknesses. The co esponding deposi ion ime and ilm hickness a e
labeled. The DRM e ec is obse ed a low ∆n alues. Rhombuses indica e 1
sun illumina ion.
Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms 99
0 10 20 30 40 50 60 70
0,2
0,4
0,6
0,8
1,0
1,2
1,4
Rela i e change in τ
e
a 1 sun
FGA ime (min)
1min
2 min
dep
= 5 min
Figu e 3.12. No malized τ
e
alues a 1 sun illumina ion as a unc ion o FGA
ime o p- ype 0.85 Ω cm samples passi a ed wi h h ee di e en deposi ion
imes labelled in he g aph. Solid lines a e a guide o he eye.
0 10 20 30 40 50 60
10
10
10
11
10
12
10
0
10
1
10
2
10
3
10
4
10
5
40 3,7 1,6 0,95 0,46 0,32
Cha ge densi y, Q (cm
-2
)
N
A
(x 10
15
cm
-3
)
Q
S
p0
S
e
, S
p0
(cm s
-1
)
S
e
a-SiC(n) as deposi ed
ρ
(
Ω
cm)
Figu e 3.13. Expe imen al alues o su ace ecombina ion eloci y and
simula ion pa ame e s, Q and S
p0
, as a unc ion o he wa e doping.
100 Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms
3.5 Su ace passi a ion dependence on wa e
esis i i y
In o de o analyze he dependence on bulk doping, op imized silicon ca bide ilms
we e used o passi a e wa e s wi h di e en esis i i ies anging om 0.3 o 40 Ω cm (N
A
om 5.4 × 10
16
o 3.4 × 10
14
cm
-3
). The wa e s we e chemically e ched using a 1:3:3
HF:HNO
3
:CH
3
COOH
3
solu ion ( he so-called CP133 solu ion) o p o ide simila su ace
condi ions in all samples. A RCA sequence ended wi h an HF dip was applied be o e
PECVD deposi ions. As usual, li e ime measu emen s we e pe o med and i ed wi h he
Gi isch model.
Fig 3.13 plo s he measu ed alues o he su ace ecombina ion eloci y, S
e
, a
∆n = 10
15
cm
-3
and he wo simula ion pa ame e s, he cha ge densi y, Q, and he
undamen al ecombina ion o holes, S
p0
. I can be seen ha su ace passi a ion ge s
wo se as he doping inc eases, indica ed by inc easing alues o S
e
. Inc easing alues o
S
p0
and Q a e also ound wi h doping. The e o e, as he doping inc eases he numbe
sa u a ed dangling bonds ( ela ed o S
p0
) is lowe , while he ield e ec ( ela ed o Q) is
enhanced. The expe imen al S
e
alues e idence ha he i s mechanism domina es he
second. The inc easing alues o S
e
as he subs a e doping inc eases a e easily explained
by a highe densi y o s a es p o ided by he bo on impu i ies. Howe e , he s ong
dependence o he cha ge densi y on he doping is ela i ely anomalous and indica es
again ha he o igin o he ield e ec passi a ion is no p o ided by a ixed cha ge
densi y, bu p obably by a a iable in e ace densi y o s a es ha depends on he Fe mi
le el posi ion o he c ys alline silicon. Appendix I p o ides p elimina y esul s abou he
undamen al p ope ies o he cha ge densi y in a-SiC
x
ilms based on co ona cha ge
measu emen s. Howe e , a comple e heo y o he phenomenon equi es u he
in es iga ions.
3.6 Chap e conclusions
This Chap e has ocused on he su ace passi a ion o p- ype c ys alline silicon by
means o phospho us-doped silicon- ich amo phous silicon ca bide ilms a-SiC
x
(n):H
g own by PECVD. Wi h he aim o ind uni o m ilms wi h good su ace passi a ion,
Chap e 3: Single phospho us doped, silicon ich a-SiC
x
(n):H ilms 101
di e en deposi ion condi ions ha e been explo ed o low esis i i y wa e s (0.85 Ω cm):
o al p essu e in he chambe , RF powe and ilm hickness. Values o S
e
as low as
7 ± 2 cm s
-1
ha e been achie ed wi h ilms g own unde e y low RF powe , al hough he
hickness p o ile was no comple ely uni o m. Highe alues o S
e
= 24 ± 4 cm s
-1
ha e
been encoun e ed o uni o m ilms, s ill p o iding excellen su ace passi a ion.
De e mina ion o he cha ge densi y alloca ed in he laye h ough li e ime
simula ions e eals in e es ing esul s. Fo he se ies in which he chambe p essu e is
a ied, he cha ge densi y seems o be ela i ely independen o he deposi ion condi ions,
a abou 3.4 × 10
11
cm
-2
. Mo e accu a e simula ions, including he Deple ion Region
Modula ion (DRM) e ec also show a cha ge alue ha can be assumed as cons an o
he esis i i y used. In he se ies wi h he ilm hickness a ia ion he cha ge densi y
showed, su p isingly, a cons an alue o Q = 3.4 × 10
11
cm
-2
, which is e y simila o ha
ob ained in mos cases o he RF powe (3.2 × 10
11
cm
-2
). On he o he hand, a s ong
dependence o he cha ge on he wa e esis i i y expe imen s has been ound. This may
indica e ha he o igin o he ixed cha ge is no a ibu ed o a ixed cha ge densi y in he
ilm, bu depends on he whole a-SiC
x
/c-Si sys em.
The undamen al ecombina ion o holes, S
p0
, a ies wi h he deposi ion condi ions,
indica ing ha besides he ield e ec passi a ion he imp o emen o he su ace
passi a ion is due o he sa u a ion o dangling bonds, pa ly p o ided by he a omic o
molecula hyd ogen p esen in he ilms.
Finally, i mus be men ioned ha due o he high con en o silicon hese ilms
p esen high abso p ion o ligh . Op ical ansmission da a show ha o ilms a ound
80 nm hick only 50% o isible ligh (400 – 1000 nm) goes h ough a ilm/glass sys em.
In an a-SiC/c-Si in e ace he ansmission o he c-Si migh be highe due o some
in e e ence phenomena. Ne e heless, he po ion o ligh abso bed in he amo phous
ilm is no suscep ible o con ibu ing o he pho ocu en because he pho ogene a ed
ca ie s migh ecombine apidly, be o e any chance o each he p-n junc ion. The e o e,
he applica ions o hese ilms should be elega ed o he passi a ion o he ea side o a
sola cell unless hei abso p ion is minimized, o example by educing he hickness. In
his sense, nex Chap e is ocused on he de elopmen o ilms o on side passi a ion.

CHAPTER 4
S acks o Si ich/ C ich ilms:
passi a ion wi h an i e lec i e
p ope ies
4.1 In oduc ion
The p e ious Chap e demons a ed ha i is possible o achie e good su ace
passi a ion o p- ype wa e s by means o amo phous silicon ca bide ilms. I has al eady
been es ablished ha he op imum composi ion o achie e low alues o su ace
ecombina ion eloci y implied he use o silicon ich ilms, wi h a ela i ely low
bandgap (a ound 1.8 eV), hus p esen ing signi ican ligh abso p ion in he isible ange.
Widening he bandgap by in oducing mo e ca bon in he ilms has esul ed in much
poo e passi a ion. The e o e, he use o hese laye s in sola cells should be es ic ed o
passi a e he ea side.
104 Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies
In addi ion, as he p e ious Chap e showed, inc easing he ilm hickness dec eases
s ongly he su ace ecombina ion eloci y by educing he undamen al ecombina ion
eloci y o holes (S
p0
), while he amoun o cha ge c ea ed in he SiC
x
ilm, which builds
he so-called ield e ec passi a ion [42], seems o be independen o his hickness. This
can be explained in e ms o highe hyd ogen inco po a ion a he SiC
x
/Si in e ace
co ela ed wi h longe deposi ion ime, and sugges s he possibili y o deposi ing by
PECVD s acks o wo SiC
x
laye s wi h di e en composi ions. The i s laye would
consis o a e y hin silicon ich ilm, hus keeping he abso p ion o a minimum,
p o iding cha ge densi y and also a ce ain educ ion o he densi y o s a es. The second
laye would be a ca bon ich, hyd ogena ed coa ing wi h good an i e lec i e p ope ies,
i.e. an app op ia e e ac i e index (n ≈ 2) and hickness (abou 70 – 80 nm), which would
con ibu e o addi ional passi a ion by hyd ogen.
The p incipal applica ion o he s acked ilms is he passi a ion o he on side o
sola cells, in which he emi e is no mally loca ed. The e o e, he s acks ha e o p o e
ha good su ace passi a ion can be achie ed in hea ily doped egions. In ecen wo ks
we epo ed he passi a ion o phospho us emi e s, which is e iewed in he p esen
Chap e , and bo on emi e s [120,121], which was achie ed wi h he collabo a ion o he
Uni e si y o Kons anz. Apa om he use a he on side, he s acks p o ide an
impo an bene i when applied o he passi a ion o he ea side in PERC sola cells, as
we explain in sec ion 4.4.3.
This Chap e explo es he passi a ing p ope ies o SiC(n)/SiC(i) s acks o be applied
o p- ype bases (0.95 Ω cm) and n
+
- ype emi e s (20 o 500 Ω/sq) wi h di e en
hicknesses o he SiC(n) laye . Fo he p- ype bases, he alues o e ec i e
ecombina ion eloci y a e de e mined in all con igu a ions, while o he n
+
- ype
emi e s he mos ele an pa ame e calcula ed is he mino i y sa u a ion cu en densi y
J
0e
. An analysis o he op ical p ope ies p o ided by ellipsome y measu emen s is
p esen ed and will be used o de e mine op ical losses. Finally, sola cell e iciency limi s
o 50 Ω/sq and 90 Ω/sq emi e s a e heo e ically de e mined.
Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies 105
4.2 Deposi ion condi ions
F om he an i e lec ion poin o iew, he op imum e ac i e index o he
passi a ing ilms in he isible ange should be n ≈ 2 o non encapsula ed sola cells and
n ≈ 2.3 o glass encapsula ed cells [122], wi hou op ical abso p ion and wi h a hickness
o a ound 70 nm. The op ical and s uc u al p ope ies o amo phous semiconduc o s a e
highly dependen on sample p epa a ion, especially o hose in ol ing he p esence o
hyd ogen. Fo pu e hyd ogena ed amo phous silicon, a-Si:H, many di e en alues o he
e ac i e index and op ical bandgap ha e been epo ed, anging be ween n = 3.5 – 4.5
and E
g
= 1.4 – 1.8 eV [123-126]. Fo pu e hyd ogena ed amo phous ca bon, a-C:H, i is
possible o each bandgap alues o 3.8 eV wi h a e ac i e index equal o 1.5 [127], bu
also alues up o 4.1 eV ha e been epo ed [128].
Al hough ca bon ich ilms a e no sui able o achie e e y low alues o S
e
, he
con en o hyd ogen in he ilm can be c ucial o p o ide an ex a su ace passi a ion a e
a FGA ea men . Hyd ogen can di use owa ds he a-SiC/c-Si in e ace, hus educing
he Densi y O S a es (DOS) by sa u a ing hose dangling bonds s ill emaining a e he
PECVD deposi ion. Appa en ly, a minimum ini ial hyd ogen concen a ion in he ilms is
equi ed o imp o e su ace passi a ion wi h he FGA. Such concen a ion depends
s ongly on he deposi ion empe a u e, as shown in Fig 4.1. I co esponds o
in es iga ions p e ious o his wo k, made in ou g oup by M. Ve e e al. [129].
Measu emen s o Fou ie T ans o m In a Red (FTIR) spec oscopy we e pe o med on
in insic silicon ich ilms g own a di e en empe a u es, calcula ing he in eg a ed
abso p ion a 2000 – 2100 cm
-1
peak. This abso p ion is di ec ly ela ed o he hyd ogen
concen a ion, and hence he g aph shows ha i dec eases wi h he empe a u e.
The e o e, low empe a u es a e p e e ed o ob ain ca bon ich ilms. On he o he hand,
o phospho us-doped silicon- ich ilms, su ace passi a ion is op imum when he
deposi ion empe a u e is abo e 350 ºC [32]. Due o echnological easons, i is desi able
ha bo h ilms a e deposi ed a he same empe a u e o a leas as close as possible. The
op imum op ion ound has been o g ow he silicon ich ilms a 350 ºC and he ca bon
ich ilms a 300 ºC.
112 Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies
10
12
10
13
10
14
10
15
10
16
10
17
10
-5
10
-4
10
-3
E . Li e ime
τ
τ
τ
τ
e
(s)
Excess Ca ie Densi y
∆
∆∆
∆
n (cm
-3
)
0 s - 0 nm
30 s - 4 nm
60 s - 8 nm
PAS
deposi ed ime - hickness
90 s - 12 nm
Figu e 4.6. Expe imen al (symbols) and simula ed (lines) li e ime cu es o he
p- ype 0.95 Ω cm plana wa e s passi a ed by s acks and wi h an op imized
FGA ea men .
0
2000
4000
6000
8000
10000
0 30 60 90
Deposi ion ime (s)
Sp0 (cm s-1)
Q = 3.3 x 10
11
cm
-2
Figu e 4.7. Ex ac ed pa ame e s co esponding o li e ime i ings pe o med
in Figu e 4.6. A cons an ixed cha ge abou Q = 3.3 × 10
11
cm
-2
and a dec ease
o S
p0
.
he cha ge densi y wi h some sca e ing. Indeed, some mechanisms ha could lead o
misin e p e a ions o he alue o he cha ge ha e al eady been iden i ied in he p e ious
Chap e (sec ion 3.4.1).
An impo an issue a ises a his poin , since appa en ly he PAS laye is no c i ical
o build he ield e ec passi a ion, bu o minimize he densi y o s a es a he in e ace.
This p o ides ano he a gumen a ou ing he hypo hesis ha he o igin o he cha ge is
induced by he a-SiC/c-Si sys em a he han being a ixed cha ge in he ilm, as i is
clea , o example, a he SiO
2
/c-Si [89] o he a-SiN/c-Si sys ems [99].

Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies 113
4.4.3 Applica ions o s acks o he ea side o he sola
cells
A d awback o ilms wi h high posi i e cha ge when applied o MIS s uc u es ( o
example he ea side o a PERC sola cell) is ha he ield e ec passi a ion may anish
due o shun ing e ec s. La e al shun ing appea s when pe o a ing he dielec ic ilm o
apply he me al con ac s [132]. This shun ing can be pa ially p e en ed i a SiO
2
ba ie
is loca ed be ween he con ac and he ni ide, because he lowe cha ge o he oxide
c ea es a much weake in e sion laye . Ano he al e na i e is using ilms which
passi a ion s a egy is based mo e in he educ ion o he in e ace s a es densi y han he
ield e ec passi a ion, like silicon ca bide.
Howe e , he pe o mance o he inal de ice depends no only on he dielec ic ilm,
bu also on he me allic con ac s. Recen ly, ou g oup has in es iga ed he in luence o he
me al wo k unc ion on he open ci cui ol age o he sola cell [133] o h ee di e en
me als: Al (4.25 eV), Cu (4.65 eV), and Au (5.1 eV). The applica ion o he s acks
esul ed in highe V
oc
alues in Al and Au (642 and 658 mV, espec i ely) compa ed o
single Si ich ilms (619 and 628 mV).
Ano he bene i o using s acks ins ead o a single Si ich laye conce ns he ligh
e lec ion. The ca bon ich laye , wi h a lowe e ac i e index, helps o enhance ligh
e lec ion owa ds he bulk silicon, hus enhancing he cell pe o mance.
The h ee ea u es a o emen ioned (be e p e en ion o shun ing, good beha iou
wi h me al wo k unc ion, and highe ea side e lec ion) accomplished by silicon ca bide
s acks we e used o manu ac u e PERC sola cells wi h e iciencies abo e 20% in
collabo a ion wi h he F aunho e Ins i u e o Sola Ene gy Sys ems [134].
4.5 Su ace passi a ion o phospho us emi e s
Con en ional sc een-p in ed sola cells use phospho us doped n
+
-emi e s wi h low
esis i i y (be ween 40 and 60 Ω/sq). The ecombina ion wi hin hose hick emi e s
domina es o e he ecombina ion a he su ace. When only he mal oxida ion was
114 Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies
known, he passi a ion o ha side did no ep esen a signi ican imp o emen in
e iciency in on o manu ac u e expenses o indus ial sola cells. The e o e, he
su ace was usually co e ed wi h an an i e lec i e TiO
2
o wi h Ta
2
O
5
coa ing ha only
p o ided enhancemen o ligh abso p ion.
The on side su ace passi a ion became in e es ing wi h he i s s sola cells
passi a ed by silicon ni ide, hus p o iding passi a ion and good an i e lec i e
p ope ies. No only i keeps low manu ac u e cos s, bu also main ains compa ibili y wi h
he new sola -g ade silicon, as mul i-c ys alline (mc-Si) o ibbon silicon ( b-Si). The
mino i y ca ie bulk li e ime o such low quali y subs a es may su e om deg ada ion
when exposed o high empe a u es (>1000 ºC), as i was epo ed o he mally g own
SiO
2
on mc-Si [135]. We he mal oxida ion a lowe empe a u es (800 ºC) a oids he
p oblem o bulk li e ime deg ada ion [136,137], bu i s ill implies expensi e manu ac u e
s eps. As a consequence, lowe empe a u es a e p e e ed.
Silicon ni ide, deposi ed a 400 ºC has become he al e na i e o p oduce high
e iciency sola cells a easonably low cos s. I s ad an ages o he passi a ion o p- ype
su aces ha e al eady been discussed in Chap e 2. Now, hei applica ions o he
passi a ion o phospho us emi e s a e commen ed. The bes passi a ion scheme o
silicon ni ide has been ob ained by using a s ack o SiO
2
/SiN
x
, including a Rapid
The mal Oxida ion (RTO) p ocess [100] a 1050 ºC. O he ou s anding esul s ha e been
p esen ed whe e silicon ni ide is deposi ed a e he mal oxida ion [138] o a e a d i e-
in s ep [139,140], emo ing i s he oxide. In bo h cases he he mal s ep modi ies he
phospho us p o ile wi hin he emi e , educing he su ace doping and mo ing he
junc ion u he away om he su ace, he e o e enabling a mo e e icien su ace
passi a ion. On he o he hand, he same app oach sugges ed in his Chap e o silicon
ca bide, i.e., he use o s ack ilms, has al eady been success ul o silicon ni ide [141].
In his sec ion we explo e he capabili ies o silicon ca bide s acks o he passi a ion
o plana n
+
- ype emi e s o be applied a he on side o he sola cell. One o he main
objec i es is o demons a e excellen emi e su ace passi a ion wi hou equi ing ei he
a d i e-in o any o he high empe a u e s ep han phospho us p e-di usion. Resul s a e
compa ed wi h s a e-o - he a silicon ni ide passi a ion. The hickness o he PAS laye
is op imized, eaching a ade-o be ween he be e passi a ion achie ed o hicke
Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies 115
laye s and he inc eased ligh abso p ion wi hin he laye , which educed he
pho ocu en .
4.5.1 Emi e di usion wi h Plana Di usion Sou ces (PDS)
The solid Plana Di usion Sou ces (PDS)
®
[142] a e an al e na i e o he
con en ional liquid di usion using POCl
3
. These sou ces con ain he ac i e componen
CeP
5
O
14
, which is decomposed a empe a u es abo e 800 ºC, eleasing P
2
O
5
apou , as
shown in Fig 4.8. The apou is deposi ed on op o he wa e s c ea ing he phospho us
glass, which is he inal doping sou ce. This di usion echnique allows a high h oughpu ,
needing minimum p ocess equi emen s. These a e a h ee-zone di usion u nace wi h an
open qua z ube and a N
2
gas line. N
2
gas is no used as di usion ca ie , bu se es o
a oid back low o con aminan s in o he di usion ube.
A e a RCA and HF cleaning, he wa e s we e inse ed in o a ube u nace and
phospho us emi e s we e p e-di used simul aneously on bo h sides o 30 minu es using
PDS. The deposi ion p ocess is essen ially he same as in Re . [143], and he emi e
p o iles ob ained a e iden ical. The empe a u es employed anged om 920 o 800 ºC,
ob aining shee esis ances R
sh
be ween 20 and 500 Ω/sq. On he one hand, R
sh
was
measu ed using a ou p obe es e a e e ching he phospho us glass in HF (5%). On he
o he hand, he di usion p ocess pe o med in he clean oom is simula ed by SUPREM
so wa e, which calcula es he emi e p o ile once he inpu pa ame e s a e se . The ixed
pa ame e s a e he up and down amps, he di usion empe a u e and he di usion ime,
al oge he con o ming he empe a u e p o ile as a unc ion o ime. The a iable
pa ame e is he doping concen a ion a he phospho us glass, N
s
’, which is se o a alue
nea o he solubili y limi o phospho us in silicon a he p ocessed empe a u e. Doing
his SUPREM calcula es he ac ual phospho us concen a ion a he silicon su ace, N
s
,
and a alue o he emi e shee esis ance. Then, N
s
is a ied un il bo h shee esis ances,
measu ed and calcula ed, ma ch. The su ace doping concen a ions N
s
we e ound o lie
in a na ow ange be ween 0.9 and 2.7 × 10
20
cm
-3
, which a e consis en wi h he
echnological p ocess due o he absence o a d i e-in s ep.
The phospho us p o ile o he 50 Ω/sq emi e was measu ed by Seconda y Ion Mass
Spec oscopy (SIMS) echnique and compa ed wi h ha calcula ed by SUPREM (see Fig
4.9). In he absence o a d i e-in s ep, he phospho us is loca ed e y nea he su ace and
116 Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies
Figu e 4.8. Di usion concep using solid dopan sou ces. Ex ac ed om [143].
0,0 0,2 0,4 0,6 0,8 1,0
10
15
10
16
10
17
10
18
10
19
10
20
10
21
Phospho us concen a ion, N
D
(cm
-3
)
dep h (
µ
m)
R
sh
(
Ω/
sq)
500
430
180
90
50
20
Figu e 4.9. Emi e p o iles simula ed by SUPREM (lines), and he
co esponding SIMS measu emen (symbol) o R
shee
= 50Ω/sq.
he p o ile de e mina ion by SIMS is di icul . Howe e , he ag eemen be ween
expe imen al and simula ed da a is easonable.
A e ano he RCA cleaning s ep, ended wi h a HF dip o emo e any g own oxide,
he SiC s acks (PAS and ARC laye s) we e deposi ed symme ically on bo h sides o he
wa e s. The wa e s we e hen subjec ed o pos -deposi ion Fo ming Gas Anneals (FGA)
a 400 ºC o 20 minu es, which is he op imized p ocess o ba e wa e s in he p e ious
sec ion. The empe a u es used o bo h PECVD deposi ion and FGA a e low enough o
keep he phospho us dis ibu ion unal e ed. Li e ime measu emen s we e pe o med
be o e he FGA, and o e he ollowing i e mon h pe iod o check s abili y.
Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies 117
10
11
10
12
10
13
10
14
10
15
10
16
10
17
10
18
10
-5
10
-4
10
-3
10
-2
i (no DRM)
E . li e ime
τ
e
(s)
Excess ca ie densi y
∆
n (cm
-3
)
Auge + adia i e
di usion (J
Oe
)
i (DRM)
ec SCR (J
ec
)
Figu e 4.10. Example o ex ac ion o J
0e
om τ
e
(∆n) da a (symbols) o a n
+
-p-
n
+
symme ic s uc u e wi h 90 Ω/sq emi e s and simula ion showing all he
ecombina ion mechanisms in ol ed (lines). The model o he i ing includes
he di usion diode (J
Oe
= 72 A cm
-2
), he ecombina ion diode (n
ec
= 2, J
ec
=
2 × 10
-9
cm
-2
), Auge li e ime, and DRM e ec .
4.5.2 Li e ime measu emen s and modelling
The emi e sa u a ion cu en densi y, J
0e
, can be de e mined om he dependence o
he e ec i e mino i y ca ie li e ime, τ
e
, on he injec ion le el, ∆n. Fo low esis i i y
wa e s wi h signi ican bulk ecombina ion, i is impo an o use a comple e model ha
includes all he in ol ed ecombina ion mechanisms, in o de o ge an accu a e
es ima ion o J
0e
.
We measu ed he e ec i e li e ime (τ
e
) in a wide injec ion le el ange
(10
12
< ∆n < 10
17
cm
-3
). A e wa ds, he τ
e
(∆n) cu es o he symme ic s uc u e we e
i ed using a model ha inco po a es he ollowing ecombina ion mechanisms:
1. A di usion diode de ined by J
0e
wi h ideali y ac o n
di
= 1
2. A ecombina ion diode ha accoun s o he ecombina ion in he Space Cha ge
Region (SCR), de ined by J
ec
and ideali y ac o n
ec
= 2
3. The in insic Auge plus adia i e bulk ecombina ion [27] (eq. 1.17)
4. The Deple ion Region Modula ion (DRM) e ec [79]. In his sec ion, DRM was
modelled analy ically using he ab up junc ion app oxima ion wi h cons an doping
densi ies in he SCR.
Since he subs a es a e Floa Zone wa e s o high quali y, he model does no include he
Shockley-Read-Hall (SRH) ecombina ion in he bulk. Finally, i is impo an o men ion
ha he ex ac ion o pa ame e s was pe o med conside ing a empe a u e T = 25 ºC,
co esponding o an in insic concen a ion n
i
= 8.65 × 10
9
cm
-3
. Figu e 4.10 shows an

118 Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies
10
100
1000
10000
100000
0,01 0,1 1 10
Dep h junc ion, X
j
(µm)
J
0e
( A cm
-2
)
Ns = 2 x 10 20
Ns = 5 x 10 19
Ns = 2 x 10 19
SRV =
∞
SRV = 0
OPAQUE
EMITTER
Figu e 4.11. The asymp o es o he base cu en in bipola de ices o Gaussian
emi e s a gi en N
s
. Ex ac ed om e e ence [144]
10 10
2
10
10
2
10
3
10 10
2
10
3
FGA
J
0e
( A cm
-2
)
Emi e shee esis ance (
Ω/
sq)
as deposi ed
Only ARC
30 s PAS
60 s PAS
90 s PAS
Figu e 4.12. J
0e
alues o phospho us-di used, plana emi e s passi a ed by
PECVD SiC
x
s acks, be o e and a e he FGA ea men . The inc easing
deposi ion ime o he Si ich laye leads o be e alues o J
Oe
, while he
ela i e imp o emen a e FGA (400 ºC 20 min) is mo e no iceable o hinne
laye s. Lines a e a guide o he eye.
example o li e ime measu emen and simula ion co esponding o a 90 Ω/sq emi e
passi a ed wi h a SiC
x
s ack (90 s o PAS laye wi h FGA o 20 minu es), wi h good
ag eemen be ween measu ed and heo e ical cu es o e he whole injec ion le el ange.
Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies 119
4.5.3 J
0e
o plana emi e s passi a ed by SiC
x
s acks
The gene al beha iou o J
0e
wi h he emi e shee esis ance (o in e sely he
junc ion dep h) depends on he Su ace Recombina ion Veloci y, SRV [144,145]. The
asymp o es o he J
0e
a e summa ized in Fig 4.11 o Gaussian emi e s wi h di e en N
s
alues as a unc ion o he junc ion dep h. Fo hin emi e s (usually high shee esis ance)
wo opposi e ends a e ound depending on SRV. Ve y low alues o J
0e
a e ound o
well passi a ed emi e s (SRV ~ 0), while e y high alues a e ob ained o non-
passi a ed ones (SRV ~ 5 × 10
6
cm s
-1
). As he junc ion dep h inc eases ( he shee
esis ance diminishes) J
0e
inc eases o hose emi e s well passi a ed and dec eases o
non-passi a ed ones. Fo he so-called opaque o deep emi e s, wi h e y low shee
esis ance, J
0e
does no depend on he su ace ecombina ion eloci y because he
ecombina ion in he bulk o he emi e domina es o e he su ace ecombina ion.
Expe imen ally, hese gene al ends a e obse ed o emi e s passi a ed by he
silicon ca bide s acks. Fig 4.12 shows he measu ed J
0e
dependency on R
sh
o di e en
hicknesses o he PAS laye . Two clea endencies a e obse ed: he passi a ion
imp o es as he PAS laye hickness inc eases and i also imp o es due o he FGA
ea men . Be o e FGA, he samples wi h single ARC laye (wi hou PAS laye ) show
inc easing alues o J
0e
wi h R
sh
, indica ing a high su ace ecombina ion eloci y. On he
o he hand, o he hicke PAS laye he opposi e endency is ound, e ealing a low
SRV. The in e media e cases lay in be ween. A e he FGA, he ou s uc u es wi h
di e en PAS laye hickness show an impo an imp o emen , especially o samples
wi hou PAS laye . The passi a ion quali y was checked o e i e mon h pe iod a e he
FGA, p esen ing a s ong deg ada ion o he single ARC laye and o he hinnes (30 s)
PAS laye a e his pe iod. In pa icula , single ARC laye s showed a comple e de-
passi a ed beha iou , wi h J
0e
alues e en highe han he as-deposi ed case. Samples
wi h PAS laye deposi ed o 60 and 90 s we e ound o hold he ini ial J
0e
alues. The
co esponding J
0e
alues a e plo ed in Fig 4.13.
The e ec s a o emen ioned can be explained in e ms o hyd ogen passi a ion. The
ilms ob ained by decomposi ion o hyd ogena ed p ecu so gases con ain a ce ain
amoun o hyd ogen ha is pa ially esponsible o he su ace passi a ion, dec easing
he su ace ecombina ion eloci y. Ei he in a omic o in molecula o m, he hyd ogen
mobili y is high e en a empe a u es below 400 ºC, and i s dis ibu ion depends on he
120 Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies
10 10
2
10
3
10
10
2
10
3
106
2x104
5x103
105
Only ARC
30 s PAS
60 s PAS
90 s PAS
Se = 0
J
0e
( A cm
-2
)
Emi e shee esis ance (
Ω
/sq)
Se = 5 x 106 cm s-1
Figu e 4.13. J
0e
alues (symbols), 5 mon hs a e he FGA ea men .
Deg ada ion is obse ed o he hinnes PAS laye s (0 and 30 s deposi ed)
while abo e 60 s he passi a ion quali y is main ained. Theo e ical cu es (solid
lines) o gi en alues o SRV a e plo ed o compa ison.
he mal p ocesses applied. In he s acks wi h hickes PAS ilms (8 and 12 nm) he
sa u a ion o de ec s a he in e ace may come mainly om he Si-Si and Si-C bonds,
while in he hinnes PAS ilms (0 and 4 nm) i may come mainly om he high con en
o hyd ogen p esen in he ARC laye . This is why he FGA has a s ong impac on he
hinnes ilms, and i also explains why hose ilms deg ade as e .
I is clea ha a PAS laye wi h a minimum hickness o 8 nm, co esponding o 60 s
o deposi ion ime, is equi ed o achie e good su ace passi a ion. Wi h such PAS laye s
J
0e
as low as 160 – 100 A cm
-2
can be ob ained o deep emi e s (≈ 40 – 60 Ω/sq),
sui able o me al con ac s by sc een p in ing p ocess. In he ange o emi e s sui able o
e apo a ed con ac s (≈ 100 Ω/sq) alues a ound 70 A cm
-2
a e ob ained. In he case o
ligh ly doped emi e s, adequa e o no con ac ed egions in selec i e emi e s (≈ 200 –
500 Ω/sq), he bes alue ob ained is a ound 30 A cm
-2
.
4.5.4 Su ace ecombina ion eloci y
In o de o e alua e he passi a ion quali y, alues o su ace ecombina ion eloci y,
SRV, ha e o be calcula ed. To his end, i is necessa y o de e mine bo h phospho us
p o ile and he su ace doping concen a ion (N
s
), so ha i is possible o dis inguish he
ecombina ion wi hin he emi e om he ecombina ion a he su ace.
Chap e 4: S acks o Si ich/C ich ilms: passi a ion wi h an i e lec i e p ope ies 121
The p ocess o de e mina ion o su ace ecombina ion eloci y in a highly doped
egion (emi e ) is summa ized he ea e :
1. De e mina ion o emi e p o ile, i.e. he doping concen a ion as a unc ion o he
dep h. I is pe o med by SUPREM cha ac e iza ion ool, co ela ed wi h he
measu ed shee esis ance, and compa ed wi h SIMS measu emen s (sec ion
4.5.1).
2. Li e ime measu emen o he symme ic s uc u e and ex ac ion o he sa u a ion
cu en densi y, J
0e
, as desc ibed in sec ion 4.5.2.
3. PC1D simula ion [26] o a p-n junc ion and gene a ion o he co esponding da k
I-V cha ac e is ic. The s uc u e o simula ion is he ollowing:
• Ve y hin p- ype base (abou 2 µm hick) wi h e y high li e ime, abou
10 ms (high esis i i y and no SRH ecombina ion). In his way, he
ecombina ion in he base can be neglec ed.
• The emi e p o ile de e mined om SUPREM, a he on side.
• A andom alue o he on su ace ecombina ion, S
e , on
. We assumed
a de ec ene gy posi ion loca ed a he in insic le el E
i
and ha he
emi e is always wo king a Low Le el Injec ion (LLI). This implies ha
su ace ecombina ion is domina ed by he undamen al ecombina ion
eloci y o he mino i y ca ie s (S
p0
in ou case). The e o e we use
S
e , on
= S
n0
= S
p0
= S
e a LLI
as inpu pa ame e s.
• No ea su ace ecombina ion, S
e ,back
= 0.
Unde hese condi ions, he ecombina ion is only due o he on su ace and he
Auge plus adia i e in he bulk emi e . The sa u a ion cu en densi y o he I-V
cu e can he e o e be ully a ibu ed o he emi e , and he base componen can
be neglec ed:
ebe
JJJJ
0000
≈
+
=
(4.2)
4. Simula ion o he da k I-V and ex ac ion o pa ame e s. The model used o he
i ing is desc ibed by:
( )






−+






−=
11
00
T ecT
qVn
V
ec
qV
V
eIeIVI
(4.3)
Whe e V is he applied ol age, q is he elec on cha ge, V
T
= k T /q is he
he mal ol age, I is he o al cu en , I
0
and I
0 ec
he sa u a ion cu en s o he
di usion and ecombina ion diodes, espec i ely, n
ec
is he ideali y ac o o he
ecombina ion diode (no mally equal o 2). The co esponding sa u a ion cu en

CHAPTER 5
Phospho us doped silicon ca bon
ni ogen alloys, SiCN(n)
5.1 In oduc ion
In Chap e s 3 and 4 we ha e shown he applicabili y o phospho us doped silicon
ca bide ilms o pe o m su ace passi a ion o c ys alline silicon. I was al eady shown in
a p e ious wo k ha he in oduc ion o phospho us ep esen s a signi ican imp o emen
espec o in insic ilms [32]. We ha e been able o esol e pa ially he abso p i e
beha iou o he passi a ing ilms by educing hei hickness and comple e he
passi a ion wi h an i e lec i e laye s. Howe e , o p o ide enough amoun o hyd ogen
his an i e lec i e ilm had o be g own a a lowe empe a u e han he passi a ing ilm.
A p ocess like his, in ol ing wo di e en empe a u e p ocesses, is clea ly a om he
objec i es o simplici y and obus ness claimed o mass p oduc ion o silicon sola cells.
Fu he mo e, i has been ound ha when he passi a ing ilms a e 4 nm o hinne he
130 Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n)
su ace passi a ion deg ades a e a pe iod o ime a oom empe a u e. The e o e, i
would be p e e able o ind new g owing condi ions ha ensu e good su ace passi a ion,
an i e lec i e p ope ies and s abili y.
In he p esen Chap e , hese new g owing condi ions a e ocused on he in oduc ion
o ni ogen o he silicon ca bide ma e ial. I has al eady been demons a ed ha in insic
ca bon ni ogen alloys a-SiC
x
N
y
(i) ep esen a u he imp o emen compa ed o a-SiC(i)
o he passi a ion o n- ype wa e s [148], eaching alues o S
e
as low as 16 cm s
-1
.
Wi h he aim combining he bene i s o ni ogen and phospho us, we s udy phospho us-
doped hyd ogena ed-amo phous silicon ca bon ni ogen alloys, a-SiC
x
N
y
:H(n) as a new
op ion o su ace passi a ion. Hence, i ep esen s he mixing be ween silicon ni ide and
silicon ca bide.
The in oduc ion o a ilm wi h i e elemen s (silicon, ca bon, ni ogen, phospho us
and hyd ogen) could be ega ded as complica ed and he e o e useless o he
pho o ol aic indus y, which is in he sea ch o low-cos and simpli ied p ocesses o
mass p oduc ion, especially when he silicon ni ide has demons a ed excellen
capabili ies. Howe e , some di e en p ope ies be ween silicon ni ide and silicon
ca bide (o silicon ca bon ni ogen alloys) can be o in e es in ce ain ab ica ion
p ocesses. Fi s ly, silicon ni ide p o ides a s ong ield e ec passi a ion mainly due o a
high posi i e cha ge ha is calcula ed o be a ound 2.5 × 10
12
cm
-2
[99], while in silicon
ca bide i appea s o be one o de o magni ude lowe (see Chap e 3 o e e ence [149]).
A second p ope y is ha silicon ni ide can be e ched in HF, while silicon ca bide
canno . A hi d issue conce ns he in oduc ion o phospho us in he ilm. I is in e es ing
since i could se e as an emi e laye in a HIT s uc u e, wi h he bene i o mo e
anspa ency achie ed hanks o he ca bon/ni ogen inco po a ion, hus a oiding loss in
pho ocu en . Finally, a eco e y o su ace passi a ion a e long ime anneals a high
empe a u e has been ecen ly obse ed a e ini ial s ong deg ada ion a sho annealing
imes [150] (see Chap e 6). Ano he in e es ing p ope y o ni ogen in SiC sys ems is
ha i ac s as a shallow dono impu i y o mic oc ys alline silicon ca bide phases [151].
This concep has been used o ab ica e n-i-p amo phous silicon sola cells wi h a µc-SiC
n- ype egion ac ing as open window and hence allowing a highe sho ci cui cu en s
han in ypically amo phous silicon phases [152] . All his issues make in e es ing he
s udy o such e na y alloys.
Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n) 131
Wi h he aim o in es iga e he applicabili y o a-SiC
x
N
y
:H(n) ilms o all kind o
concep s o sola cells, he su ace passi a ion is es ed on low esis i i y p- ype and n-
ype silicon wa e s, as well as on n- ype emi e s. I would be e y in e es ing o s udy
also hei possibili ies in p- ype emi e s. Howe e , he echnical oppo uni ies o ou
labo a o y kep his op ion una ailable a he momen o pe o ming his wo k. Va ia ion
o he composi ion om silicon ich (highly abso p i e) o ca bon o ni ogen ich (highly
anspa en ) ilms is pe o med o ind op imum su ace passi a ion. As in Chap e 4,
s acks o di e en composi ions a e applied o combine excellen passi a ion and good
an i e lec i e p ope ies.
5.2 Deposi ion condi ions o SiCN(n) ilms
The pu pose o his expe imen was o es ablish a co ela ion be ween ilms
composi ion and su ace passi a ion. Addi ional in e es was ocused on inding
anspa en ilms ( ich in ca bon o ni ogen) wi h good su ace passi a ion.
Excep gas lows, all deposi ion condi ions we e kep cons an and conside ed o be
op imum. Tempe a u e and o al p essu e in he chambe we e se o 400 ºC and 300
mTo espec i ely, as hey we e op imized o he su ace passi a ion o n- ype wa e s
by means o in insic SiCN(i) ilms in a p e ious wo k [148]. RF powe was se o 30 W
(10% o sou ce) o achie e homogenei y in a leas 4” wa e s and he deposi ion ime was
se o 5 minu es. Acco ding o esul s ob ained in Chap e 3, deposi ion imes longe han
5 minu es would lead o hicke ilms wi h highe ligh abso p ion bu wi hou signi ican
imp o emen in su ace passi a ion. In his con igu a ion, he ilm hicknesses ob ained
we e a ound 40 nm o hose ilms ich in silicon, wi h he usual g owing a e o 8 nm
min
-1
ound o all silicon ich ilms in his hesis. As he con en o ni ogen o ca bon
inc eases he deposi ion a e dec eases mono onically, un il 4 nm s
-1
o ilms wi h high
con en o ca bon o ni ogen. In hose cases he expe imen s we e epea ed doubling
deposi ion ime o achie e again hickness o 40 nm. The SiCN(n) ilms a e ea ed in his
Chap e as e na y alloys p oduced by he decomposi ion o h ee gases:
- Silane mixed wi h 5% o phosphine, SiH
4
+ PH
3
- Me hane, CH
4
- Molecula ni ogen, N
2
132 Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n)
Phospine was no conside ed as a sepa a e gas since i was a ied by he same p opo ion
han silane. In u n, he amoun o hyd ogen inco po a ed in he ilms canno be
con olled. Ins ead o ni ogen, ammonia (NH
3
) could be used in a u he wo k o explo e
di e en al e na i es o he e na y alloy.
Gases lows we e a ied o achie e di e en ilm composi ions. In hese
expe imen s, we used ela i e low pa ame e s de ined as ollows:
(
)
[
]
3444
PHSiHCHCH
+
+
≡
Χ
(5.1.a)
(
)
[
]
23442
NPHSiHCHN
+
+
+
≡
Υ
(5.1.b)
The X pa ame e akes in o accoun he a io be ween me hane and silane lows and he
pa ame e Y indica es he ela i e ni ogen low o he o al gas low. Bo h a e no malized
o 1, hen ei he bo h X and Y alues close o 0 co espond o silicon- ich ilms, while
ca bon (ni ogen) ich ilms a e deposi ed wi h X (Y) ending o 1. Two scans we e
pe o med wi h he aim o inco po a e mo e ca bon o ni ogen o he ilms. In he i s
scan, we kep cons an he ni ogen low o 15.7 sccm and he o al gas low o 75 sccm
(equi alen o Y = 0.21). Then, X was a ied be ween 0.25 and 0.9. Fo he second scan, X
was kep cons an o X = 0.4 and Y a ied be ween 0.14 and 0.83.
5.3 Su ace passi a ion o on p- and n- ype wa e s.
As in he whole hesis, low esis i i y subs a es we e used o ocus on he
applicabili y o su ace passi a ion on silicon sola cells. Floa zone, <100> o ien ed,
plana silicon wa e s we e employed, wi h a esis i i y o 0.95 Ω cm o he p- ype wa e s
(accep o densi y, N
A
= 1.6 × 10
16
cm
-3
) and 1 Ω cm o he n- ype ones (dono densi y,
N
D
= 5 × 10
15
cm
-3
). They we e cleaned in a pi anha solu ion ended wi h a HF dip and
immedia ely in oduced in o he PECVD eac o o pe o m symme ical deposi ions o
he a-SiC
x
N
y
:H(n) ilms.
An example o li e ime measu emen o bo h wa e ypes wi h he co esponding
bulk li e ime limi s and he alues a 1 sun illumina ion indica ed by a ows is gi en in
Fig 5.1. Su ace ecombina ion eloci ies alues lowe han 10 cm s
-1
(5 cm s
-1
) a e
ob ained in a wide injec ion le el ange in his case. The li e ime alues we e con e ed
o su ace
Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n) 133
10
13
10
14
10
15
10
16
10
17
10
-4
10
-3
10
-2
10
-1
10
13
10
14
10
15
10
16
0
5
10
S
e
(cm s
-1
)
∆
n (cm
-3
)
p- ype
n- ype
Auge + ad. p- ype
Auge + ad. n- ype
E . li e ime
τ
e
(s)
Excess ca ie densi y
∆
n (cm
-3
)
Figu e 5.1. Example o li e ime cu es o p an n- ype wa e s passi a ed by
SiCN(n). The Auge plus adia i e bulk li e ime a e plo ed, and he a ows
indica e 1 sun illumina ion le els. The inse shows he co esponding S
e
alues.
ecombina ion eloci y alues employing equa ion (1.66) and a e p esen ed as an inse in
he Figu e.
5.3.1 Dependence on gas lows
Fig 5.2 shows he dependence o S
e
alues a 1 sun illumina ion on he e ac i e
index o he deposi ed ilm o bo h ypes o subs a es. The bes passi a ion is ob ained
by he silicon- iches ilm in bo h subs a es showing ex emely low S
e
alues o 3 cm s
-1
o p- ype wa e s and 2 cm s
-1
o n- ype wa e s. They co espond o open ci cui ol ages
alues o 721 and 697 mV, espec i ely This ep esen s an impo an imp o emen
compa ed wi h he passi a ion le el p e iously achie ed by ou silicon ca bide ma e ial
o p- ype (S
e
= 7 cm s
-1
wi h a-SiC
x
(n) in Chap e 3) and n- ype (S
e
= 16 cm s
-1
wi h
in insic a-SiC
x
N
y
:H [148]) silicon wa e s o simila esis i i y. Ac ually, hey o e
simila passi a ion le el as in insic amo phous silicon epo ed alues, o which in
p- ype 1.5 Ω cm silicon wa e s S
e
alues a ound 3 cm s
-1
we e ob ained [109]. The
incon enience o amo phous silicon is ha he low deposi ion empe a u es (a ound
200 ºC) equi ed o achie e good in e ace hyd ogena ion cause ins abili y unde high
empe a u e anneals, while ou silicon ca bide ilms, g own a 400 ºC, should be in
p inciple mo e s able. Nex Chap e ocuses in his issue.

134 Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n)
0 0,2 0,4 0,6 0,8 10 0,2 0,4 0,6 0,8 1
10
-1
10
0
10
1
10
2
10
3
Y = [N
2
] / [CH
4
+ (SiH
4
+PH
3
) +N
2
]
(X = 0.4)
scan 2
p- ype 0.95
Ω
cm
n- ype 1
Ω
cm
S
e
(cm s
-1
) a 1 sun
X = [CH
4
] / [CH
4
+ (SiH
4
+PH
3
)]
(Y = 0.21)
scan 1
Figu e 5.2. Su ace ecombina ion eloci y as a unc ion o he gas low a ios X
and Y a ied in wo scans.
5.3.2 Ellipsome y measu emen s, Tauc-Lo en z pa ame-
e s and co ela ion wi h passi a ion mechanisms
Spec oscopic ellipsome y measu emen s we e pe o med in a wide spec al ange
( om 200 o 1600 nm) wi h a phase-modula ed ellipsome e (UVISEL, Jobin Y on
Ho iba). The pseudo dielec ic unc ions measu ed we e i ed using he dispe sion law
based on he Lo en z oscilla o and he Tauc join densi y o s a es o amo phous
ma e ials [130]. In his model, he ene ge ic dependence o he imagina y pa o he
dielec ic unc ion (ε
2
) is gi en by:
( )
(
)
( )
g
g
EE
E
ECEE
EECEA
E>
+−
−
=,
1
22
2
2
0
2
2
0
2
ε
(5.2)
whe e E is he pho on ene gy, E
g
is he op ical bandgap, and A, C and E
0
a e he
ampli ude, he b oadening e m and he cen e e ms o he Lo en z oscilla o unc ion,
espec i ely. These ou pa ame e s allow he s uc u al cha ac e iza ion o he ma e ial
and hey a e plo ed in Figu e 5.3 as a unc ion o he lux gases a io, X o Y. The op ical
bandgap anges om 1.73 eV o 2.09 eV, he b oadening e m anges om 2.66 eV o
6.04 eV, he cen e e m ange om 3.47 o 6.72, and inally he ampli ude e m dec eases
om 183 o 70 eV as he ca bon/ni ogen con en inc eased.
Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n) 135
1.6
1.7
1.8
1.9
2.0
2.1
2.2
40
60
80
100
120
140
160
180
200
220
0.0 0.2 0.4 0.6 0.8 1.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
0.0 0.2 0.4 0.6 0.8 1.0
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
op ical bandgap (eV)
ampli ude e m, A (eV)
scan 1
scan 2
b oadening e m, C (eV)
X o Y
cen e e m, E
0
(eV)
X o Y
Figu e 5.3. Tauc-Lo en z pa ame e s de e mined om ellipsome y
200 400 600 800
0,0
0,4
0,8
1,2
1,6
2,0
2,4
2,8
3,2
n = 2.1 X = 0.9
n = 2.3 Y = 0.83
n = 2.7 Y = 0.66
n = 3.2 Y = 0.5
n = 3.6 X = 0.5
n = 3.7 common poin
scan 1
scan 2
ex ic ion coe icien , k
wa eleng h (nm)
Figu e 5.4. Ex inc ion coe icien o se e al a-SiCN(n) ilms. Labels indica e
he e ac i e index measu ed a 633 nm and he co esponding gas low a io.
The eal pa o he dielec ic unc ion, ε
1
, is ob ained om ε
2
using K ame s-K onig
in eg a ion, leading o di ec de e mina ion o he e ac i e index, n, and he ex inc ion
coe icien , k. This las alue is plo ed in Fig 5.4 o se e al composi ions used in he
s udy, labeling he e ac i e index a 633 nm and he ac ion o gases lows X o Y. The
common poin , which is also he op imum passi a ing poin in his s udy, p esen s s ong
abso p ion o ligh due o i s silicon ich composi ion. Fig 5.5 shows he ela ion be ween
e ac i e index, n, a 633 nm as a unc ion o he X o Y pa ame e s o he wo scans. As
he ca bon o ni ogen con en inc ease (inc easing X o Y) he e ac i e index dec eases
as expec ed.
136 Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n)
0,0 0,2 0,4 0,6 0,8 1,0
2,0
2,2
2,4
2,6
2,8
3,0
3,2
3,4
3,6
3,8
4,0
X o Y
scan 1, Y = 0.21
scan 2, X = 0.40
e ac i e index a 633 nm
Figu e 5.5. Re ac i e index o he a-SiCN(n) ilms measu ed a 633 nm.
5.3.3 S acks wi h SiCN(n) ilms
In o de o educe ligh abso p ion, s acks o silicon ich / ca bon ich ilms we e
al eady sugges ed o silicon ca bide ilms in Chap e 4, ob aining good passi a ing and
an i e lec i e p ope ies applied on n- ype emi e s. The same s a egy was used in he
p esen Chap e o a-SiC
x
N
y
(n) ilms, combining passi a ing ilms (n = 3.6, S
e
= 5 o
p- ype) and an i e lec i e ilms (n = 2.3, S
e
= 90 o p- ype). An impo an ad an age in
he p esen case espec o he SiC s acks de eloped in Chap e 4 is ha he p ocessing
empe a u e is he same o bo h laye s (400 ºC) and no FGA ea men was applied, hus
simpli ying he p ocess. The passi a ing ilm hickness was a ied om 8 o 24 nm, while
he an i e lec i e coa ing hickness was kep cons an a 70 nm. As he passi a ing ilm
hickness was educed, he su ace ecombina ion eloci y inc eased, as shown in Fig 5.6.
The co esponding e ec i e op ical cons an s n
e
, k
e
and hickness d
e
, a e calcula ed
om he ini ial alues using equa ions om he model explained in e e ence [153]:
( )
(
)
(
)
21
2
2
2
2
21
2
1
2
1
22
dd
dkndkn
kn
e e
+
−+−
=−
(5.3.a)
( )
(
)
(
)
21
222111
dd
dkndkn
kn
e e
+
+
=
(5.3.b)
21
ddd
e
+
=
(5.3.c)
The e ec i e op ical cons an s o he s acks used a e summa ized in Table 5.1.
Chap e 5: Phospho us doped silicon ca bon ni ogen alloys, SiCN(n) 137
0 10 20 30 40 50
1
10
100
p- ype
n- ype
S
e
(cm s
-1
) a 1 sun
Si ich ilm hickness (nm)
single Si ich
(no s ack)
Figu e 5.6. Su ace ecombina ion eloci y a 1 sun o low esis i i y wa e s
(~ 1 Ω cm) achie ed by SiCN(n) s acks o di e en hicknesses. The hickness o
he an i e lec i e coa ing is a ound 70 nm.
Table 5.1. E ec i e op ical cons an s and hickness a 633 nm o SiCN(n)
s acks used in Fig 5.6. The an i e lec i e coa ing hickness is 70 nm.
PAS hickness
n
e
k
e
d
e
(nm)
8 2.44 0.005 78
12 2.51 0.008 82
16 2.57 0.010 86
24 2.67 0.013 94
5.3.4 Su ace ecombina ion eloci y as a unc ion o he
e ac i e index
The S
e
alues de e mined in Fig 5.2 ( o single SiCN(n) ilms) and 5.7 ( o s acks)
a e epea ed in Fig 5.7 o p- ype wa e s and Fig 5.8 o n- ype wa e s as a unc ion o he
e ac i e index a 633 nm. As a compa ison, alues o S
e
a he same esis i i y wi h
pu e a-SiC(n) and pu e a-SiN a e plo ed, he la e coming om he decomposi ion o
ammonia (NH
3
) gas, a he han he ni ogen (N
2
) used in his wo k. Fo silicon ca bide
compounds, wi h o wi hou ni ogen, S
e
is minimized o ilms wi h e ac i e index
equal o 3.72. This alue is close o he e ac i e index o amo phous silicon, ( he
li e a u e sp eads alues be ween 3.5 and 4.5 a simila g own condi ions [123-
125,154,155]), indica ing ha composi ion o he op imum ilms is e y ich in silicon.
Howe e , he p esence o ni ogen in he PECVD chambe leads o an impo an
imp o emen o passi a ion espec o SiC(n), despi e hei e ac i e index a e iden ical.

CHAPTER 6
Time and he mal ime s abili y o
silicon ca bide ilms
6.1 In oduc ion
When p oducing elec ici y om sola ene gy, bo h economical cos s and he ene gy
used o manu ac u e he modules ha e o be aken in o accoun . In gene al, he e is a
common ag eemen ha he ene gy eco e y ime in silicon sola cells is nowadays less
han 3 yea s and depends on he ype o silicon used ( ibbon, mul ic ys alline o
monoc ys alline) [158,159], while he payback ime o economical cos is signi ican ly
highe . S ong e o s a e being done by he whole pho o ol aic communi y o educe
hese payback imes wi h se e al s a egies ( hinning wa e s, using low cos ma e ials,
e c.). In addi ion, sola cells need o keep a ce ain deg ee o s abili y in e iciency du ing
a minimum pe iod o 20 yea s wo king unde no mal ope a ion condi ions ( empe a u es
up o 60 – 70 °C unde illumina ion), which can be indeed e y di e en om s anda d
146 Chap e 6: The mal and ime s abili y o silicon ca bide ilms
es condi ions (25 °C wi h 1 sun-AM 1.5 spec um). Su ace passi a ion mus hen keep
i s p ope ies o a leas he same pe iod.
On he o he hand, he majo i y o indus ial p ocesses use he sc een-p in ing
echnique o apply he me allic con ac s. This is a e y obus and well es ablished
p ocess ha in ol es high empe a u e s eps a e su ace passi a ion. In such scheme
me allic pas es a e p in ed h ough a mesh mask o de ine he on and ea g ids. The n
+
-
emi e , loca ed a he on side, is usually p in ed wi h sil e pas e, while he ea side is
no mally p in ed by aluminium. Then, a co- i ing s ep a high empe a u es o a ew
seconds is pe o med o simul aneously achie e ohmic con ac s a bo h sides. The p ocess
is done in bel u naces, o in apid he mal anneal (RTA) u naces. Fi ing p o iles (peak
empe a u es, imes, numbe o pla eaus and up and down amps) ha e been ex ensi ely
s udied (see o ins ance Re . [160]), wi h a gene al ag eemen ha mo e han 700 °C
(and up o 900 °C) a e equi ed o he s anda d sil e pas es, while abou 600 °C a e
enough o good aluminium con ac ( he empe a u e ha o m eu ec ic be ween Al and Si
is 577 ºC). Despi e low empe a u e pas es a e ecen ly being unde in es iga ion, a he
momen su ace passi a ion echniques wi h s abili y unde high empe a u e p ocesses
a e needed. Silicon ni ide passi a ion has al eady demons a ed good s abili y a e i ing
p ocesses, ei he wi h single deposi ion o silicon ni ide [106,107] o in combina ion
wi h hin silicon oxide laye g own by Rapid The mal Oxida ion (RTO) [101].
While s abili y o silicon ni ide passi a ion has been ex ensi ely explo ed o many
yea s and by many esea ch g oups, applica ion o silicon ca bide o he passi a ion o
silicon sola cells is s ill a i s mos p ima y s a e. The e o e, his Chap e p esen s
p elimina y esea ch abou ime and he mal s abili y o su ace passi a ion by
amo phous silicon ca bide. Time s abili y es consis s o measu ing li e ime as a unc ion
o he ime passed a e ilm deposi ion. The mal s abili y is explo ed by i ing he
li e ime es samples a high empe a u es ( om 500 o 900 ºC) and a di e en annealing
imes. The passi a ion ilms a e hose p esen ed in Chap e s 3, 4 and 5. In some cases, he
ilms ha e been deposi ed on ex u ed wa e s. Tex u ing ea men is c ucial o enhance
ligh abso p ion, bu i p esen s he incon enien ha he inal su ace a ea is
app oxima ely 1.3 imes highe han in plana su aces, and he planes exposed a e <111>
o ien ed. Bo h ac o s can in luence on he su ace passi a ion and i s s abili y on ime
and unde high he mal p ocesses.
Chap e 6: The mal and ime s abili y o silicon ca bide ilms 147
6.2 Time s abili y
6.2.1 S abili y o di e en passi a ion schemes
The p oblem o li e ime deg ada ion on ime has al eady been obse ed in Chap e 4
when using s acks o silicon ca bide o passi a e n
+
- ype emi e s. When he silicon ich
ilm was 8 o 12 nm hick he passi a ion was s able du ing he ime o analysis
(5 mon hs), while when i was 4 nm o did no exis a all, as deg ada ion o li e ime
occu ed. In his sec ion we explo e he s abili y on ime p o ided by di e en
con igu a ions o silicon ca bide. Symme ically passi a ed wa e s we e simply s o ed in
da k condi ions and hei li e imes we e measu ed o e a long pe iod o ime. The
passi a ion ilms analysed a e he ollowing:
- In insic silicon ca bide ilms, silicon ich, SiC(i), [32]
- Phospho us doped silicon ca bide ilms, silicon ich, SiC(n)
- In insic silicon ca bon ni ogen alloys, SiCN(i), [32]
- Phospho us doped silicon ca bon ni ogen alloys, SiCN(n)
- S acks o Si ich / C ich ilms
All ilms analyzed a e ich in silicon, and wi h an op imized FGA ea men applied,
excep o he SiCN(n). The wa e s used we e Floa Zone wi h plana su aces. Doping
ype, esis i i y and hickness a e de ailed in Table 6.1, as well as S
e
alues. The s acks
used o his es ha e been applied on ligh ly doped emi e s (350 Ω/sq).
To show he deg ada ion o su ace passi a ion, Fig 6.1 plo s no malized li e ime
alues as a unc ion o ime o he s uc u es a o emen ioned. The labels co espond o
hose in Table 6.1. In some cases he li e ime alues inc ease wi h ime, which we assume
o be due o a decalib a ion o he QSS-PC ins umen du ing he long ime pe o mance
o he expe imen (up o 7 yea s). The i s di ec compa ison is be ween in insic and
phospho us doped SiC ilms. I is shown ha in insic ilms a e much mo e s able han
phospho us doped ilms. I has been clea ly s a ed ha phospho us doped SiC p oduces a
be e su ace passi a ion, o equi alen doping concen a ion, han in insic SiC [32],
and he cause has been a ibu ed o a highe inco po a ion o hyd ogen in he ilms [129],
hus p oducing mo e e ec i e sa u a ion o dangling bonds. Howe e , ei he in molecula
o a omic o m, hyd ogen is e y mobile and e uses easily om he in e ace [161].
148 Chap e 6: The mal and ime s abili y o silicon ca bide ilms
Table 6.1. Samples used o analysis o li e ime deg ada ion wi h ime. All o
hem we e subjec ed o FGA ea men a e SiC deposi ion, excep SiCN(n).
Film Wa e Ini ial condi ions
ype
ρ (Ω cm)
W (µm) τ (µs)
S
e
(cm s
-1
)
SiC(i) – Si ich p 3.3 350 214 ≤ 82
SiC(n) – Si ich p 0.85 400 825 12
SiCN(i) – Si ich n 1.9 365 1060
12
SiCN(n) – Si ich – wi hou FGA p 0.95 300 878 6
S ack 4 nm – SiC(n) Si ich / C ich p 0.95 300 41 N.A. (emi e )
S ack 8 nm – SiC(n) Si ich / C ich p 0.95 300 111 N.A. (emi e )
1 10 100 1000 10000 100000
0,2
0,4
0,6
0,8
1,0
1,2
S ack 4 nm
SiC(n)
SiCN(n)
SiC(i)
no malized li e ime
ime a e deposi ion (hou s)
SiCN(i)
S ack 8 nm
Figu e 6.1. Li e ime deg ada ion wi h ime in s uc u es passi a ed by silicon
ca bide ilms. De ails o he di e en laye s a e shown in Table 6.1.
Fo he phospho us doped SiC(n) passi a ed wa e he decay ime has been analysed
employing exponen ial decay beha iou (see Fig 6.2). The ime cons an is a ound 11000
hou s and he inal li e ime alues (a e in ini e ime a e he deposi ion) would be
a ound 250 µs. In his case i would ep esen , a 1 sun illumina ion, an implied open
ci cui ol age V
oc
≈ 670 mV, ha is s ill a easonably good alue o indus ial ype sola
cells ( he ini ial alue was 702 mV).
When analyzing e na y SiCN alloys, he same di ec compa ison be ween in insic
and phospho us doped ilms can be done, i.e. in insic ilms keep mo e s able su ace
passi a ion. Howe e , in he p esen case he deg ada ion in phospho us doped ilms is
no as s ong as in SiC(n) ilms. Clea ly, he in oduc ion o molecula ni ogen (N
2
) in
Chap e 6: The mal and ime s abili y o silicon ca bide ilms 149
0 5000 10000 15000 20000 25000 30000
200
300
400
500
600
700
800
900
τ
, e ec i e li ime (
µ
s)
, ime a e deposi ion (hou s)
SiC(n) on p- ype
τ
= A exp(- /T) +
τ
0
τ
0
= 250
±
40
µ
s
A = 550
±
40
µ
s
T = 11000
±
1700 hou s
Figu e 6.2. Tempo al decay o p- ype wa e (0.85 Ω cm) passi a ed by
phospho us doped amo phous silicon ca bide SiC(n).
he PECVD chambe o p omo e SiCN alloys has some bene i s wi h s ill unknown
causes. One o hem could be a highe amoun o cha ge densi y p o ided by ni ogen
( he epo ed alues o ixed cha ge in silicon ni ide sys em is 2.5 × 10
12
cm
-2
[99]) being
he e o e less sensi i e o he p esence o dangling bonds and educing he impo ance o
hyd ogen a he in e ace. Ano he e ec could be ela ed o he s uc u e o he
compound. Ac ually silicon ni ide deposi ed wi h dilu ed silane in molecula ni ogen
p oduces be e esul s in su ace passi a ion han ha p oduced only om decomposi ion
o silane and ammonia [28].
Finally, he s acks p esen a semi-s able beha iou when he silicon ich ilm
hickness is 8 nm, and a as deg ada ion when i is only 4 nm. This is again in ag eemen
wi h he e usion o hyd ogen, which plays a majo ole o passi a ion o s acks wi h hin
Si ich ilms and a mino ole in hick ones.
Despi e in insic silicon ca bide ilms o e excellen ime s abili y, he ini ial su ace
passi a ion is in gene al lowe han ha o e ed by he phospho us doped ilms once hey
a e deg aded. The e o e, he bes solu ion o achie e good ini ial condi ions and ime
s abili y would be by using silicon ca bide s acks (wi h o wi hou ni ogen) o in insic
silicon ca bon ni ogen alloys.

150 Chap e 6: The mal and ime s abili y o silicon ca bide ilms
6.2.2 In luence o FGA on ime s abili y
The o ming gas anneal (FGA) plays an impo an ole in su ace passi a ion by SiC.
In Chap e 4 and 5 we ha e analysed ex ensi ely his beha iou and shown ha he same
ea men in FGA can lead o e y di e en esul s depending on he s uc u e (single
laye s o s acks), hickness, and composi ion o he SiC ilms. In polished wa e s
passi a ed by silicon ich SiC(n) we ha e al eady shown ha li e ime can inc ease o a
ac o up 1.3 a e FGA. In he nex sec ion we will see ha his ac o can be much mo e
impo an in ex u ed samples. I would be in e es ing o ind ou i any delay be ween
deposi ion and FGA ea men a ec on he inal passi a ion le el o on he ime s abili y.
Fou p- ype polished wa e s (0.95 Ω cm) we e passi a ed by silicon ich SiC(n) ilms
g own a 400 ºC (simila condi ions han in Chap e 4). The li e imes we e measu ed 30
minu es a e deposi ion. A FGA ea men a 400 ºC o 20 minu es was applied o all
wa e s, bu a di e en imes:
Wa e 1: 1 hou a e deposi ion
Wa e 2: 1day a e deposi ion
Wa e 3: 1 week a e deposi ion
Wa e 4: no FGA
The li e ime was hen measu ed egula ly o see changes wi h he FGA and deg ada ion
e ec s. Figu e 6.3 is summa izing he whole p ocess, indica ing he li e ime a usual 1
sun illumina ion.
In all cases, he ini ial li e ime alues a e signi ican ly lowe han expec ed o his
ype o wa e s and passi a ing ilm, indica ing ha he expe imen is a ec ed by some
con amina ion. Wha is mo e, he ini ial li e ime alues a e no equal, indica ing small
non-homogenei ies in he p ocess. Howe e , wi h he p ecision o he expe imen , he es
is alid o see ha in all cases he deg ada ion is s a ing o be signi ican 1000 hou s a e
deposi ion and ha he ime in which he FGA is applied is no ele an o he su ace
passi a ion s abili y. The o de o magni ude o he ime cons an (1/e) is a ound 10
4
hou s (~ 400 days) in all cases, bu wi h conside ably high unce ain y (a ound 7000
hou s).
In he end, i can be concluded ha a ime delay be ween deposi ion and FGA
ea men does no a ec ei he he inal passi a ion le el o he ime s abili y.
Chap e 6: The mal and ime s abili y o silicon ca bide ilms 151
0,1 1 10 100 1000 10000 100000
250
300
350
400
450
1 week
Li e ime a 1 sun (
µ
s)
Time (hou s)
1 hou
1 day No FGA
Figu e 6.3. Tempo al decay o p- ype wa e (0.95 Ω cm) passi a ed by
phospho us doped amo phous silicon ca bide SiC(n) and wi h he FGA
ea men pe o med a di e en imes a e he PECVD deposi ion.
6.3 The mal s abili y a high empe a u es
In he p esen sec ion we p o ide ex ensi e analysis o su ace passi a ion
dependence unde high empe a u e (abo e 500 ºC) anneals. Di e en es samples and
es ilms a e used. In o de o cla i y he eade , a b ie summa y o e e y expe imen is
p esen ed he ea e :
Sec ion 6.3.1 s udies silicon ca bide ilms de eloped in Chap e s 3 and 4. The
p ocess is as ollows:
1. Sample p epa a ion
a. Subs a e
i. All wa e s a e ex u ed
ii. Doping: p- ype bases (40 Ω cm) and n
+
- ype emi e s (60 Ω/sq)
b. Films deposi ions
i. Single Si ich SiC(n)
ii. S acks o Si ich SiC(n) and C ich SiC(i)
c. Fo ming Gas Anneal (FGA) ea men o enhance li e ime be o e he mal
s ess. P e ious expe imen s a e done o choose op imum annealing ime.
152 Chap e 6: The mal and ime s abili y o silicon ca bide ilms
A he end o sample p epa a ion p ocess he e a e 4 di e en ypes o s uc u es
o be es ed: single ilms on p- ype wa e s, single ilms on n
+
- ype emi e , s acks on
p- ype wa e s, and s acks on n
+
- ype emi e s.
2. The mal s ess in RTA u nace
a. Tempe a u es 500, 600, 700 and 900 ºC
b. Annealing imes om 3 s o mo e han 3 h
Sec ion 6.3.2 s udies annealing e ec s on silicon ca bon ni ogen alloys de eloped in
Chap e 5:
1. Sample p epa a ion
a. Subs a e
i. All wa e s a e plana
ii. Doping: p- ype bases and n- ype bases, bo h 1 Ω cm, and n
+
- ype
emi e s (130 Ω/sq)
b. Films deposi ions
i. Single silicon ich ilms, SiCN(n)
ii. S ack o Si ich C ich ilms, SiCN(n)
c. No FGA applied
2. The mal s ess in bel u nace (equi alen o i ing a e sc een p in ing)
a. Tempe a u e 720 °C, o 35 s
b. Tempe a u e 900 °C, o 10 s
6.3.1 The mal s abili y o SiC(n), single and s acks
Films de eloped in Chap e 3 and 4 ha e been deposi ed on p- ype wa e s and n
+
-
ype emi e s, bo h wi h ex u ed su aces. Excep silicon ca bide deposi ions, samples
p epa a ion, he mal ea men s and cha ac e iza ion we e lead in he Enginee ing
Depa men a he Aus alian Na ional Uni e si y.
6.3.1.1 Sample p epa a ion
As s a ing ma e ial, we used <100> o ien ed, Floa Zone, p- ype, 40 Ω cm, c-Si
wa e s, wi h an ini ial size o 12.5 × 12.5 cm. The wa e s we e hen subjec ed o an
alkaline saw Damage Remo al E ch (DRE) o 13 min, which p oduced unca ed
Chap e 6: The mal and ime s abili y o silicon ca bide ilms 153
py amids, hus p o iding a pa ial ex u ing ea men and a inal hickness o 270 µm.
To p epa e n
+
-p-n
+
s uc u es, hal o he wa e s we e cleaned in a RCA solu ion and
inse ed in o a ube u nace o phospho us di usion a 855 ºC o 30 minu es using
POCl
3
as he dopan sou ce and esul ing in 60 Ω/sq emi e s. Then he phospho us glass
was s ipped in bu e ed HF.
A e ano he RCA cleaning s ep he sample p epa a ion ended wi h an HF dip o
emo e any g own oxide. Silicon ca bide deposi ion o ilms ollowed. To cla i y he
di e en s uc u es es ed we es ablish a nomencla u e as in Chap e 4. Then, he
phospho us-doped silicon- ich laye is labelled as PAS, ema king i s passi a ing
p ope ies, while he ca bon ich laye is labelled as ARC, indica ing ha i ac s as an
an i e lec i e coa ing. When only Si ich laye s a e p oduced he scheme will be called
single, while PAS + ARC s uc u e will be labelled as s ack. The deposi ion pa ame e s
a e e y simila o hose employed in Chap e 4. The di e ences a e he g owing
empe a u e o he Si ich ilm (400 ºC in he p esen case, p e iously 350 ºC) and he
hickness o his ilm in he s ack (16 nm in he p esen case, p e iously om 4 o 12
nm). All pa ame e s a e summa ized in Table 6.1. The e o e, ou di e en s uc u es a e
analysed: p- ype ba e wa e s and emi e s passi a ed by single laye s and s acked laye s.
Figu e 6.4 shows a schema ic o he ou di e en con igu a ions es ed. A e silicon
ca bide deposi ions he wa e s we e cu in o 2.5 × 2.5 cm samples, o p o ide possibili ies
o di e en es s.
Be o e s a ing he i ings a high empe a u e i would be con enien o maximize
he li e ime o he es samples, so ha any p obable deg ada ion caused by he he mal
s ess can be mo e easily obse ed. The e o e, se ies o anneals in Fo ming Gas a low
empe a u e we e pe o med a 400 ºC in s eps o 10 minu es.
Fig 6.4 plo s he e olu ion o li e ime as a unc ion o he FGA ime. In ba e p- ype
wa e s, he li e ime inc eases a e 10 minu es by a ac o abou 8 and 15 o single and
s ack schemes, espec i ely. Su p isingly, a e he i s FGA he s ack s uc u e gi es
much be e passi a ion quali y han he single laye , which in p inciple con ains he bes
composi ion o his end. Then o he PAS laye sa u a ion is eached and inally a small
deg ada ion a e 60 minu es is ound. This beha iou is sligh ly di e en om he
expe imen pe o med in Chap e 4 on plana su aces, in which he e olu ion is less