Slide 1
DERFDERFJune 2008June 2008
APCVD TiOAPCVD TiO22 vs PECVD SiN vs PECVD SiNxx::
The Battle forThe Battle forAntireflection SupremacyAntireflection Supremacy
(Findings from a 3-yr ARC Linkage Project (Findings from a 3-yr ARC Linkage Project between ANU, SierraTherm & SunPower)between ANU, SierraTherm & SunPower)
06-June-200806-June-2008
Slide 2
Project details• Contributions over three years (AUD):
Contributor Cash In kind
Agreed Actual Agreed Actual
ARC $340,000 $385,000 – –
SierraTherm $120,000 $120,000 $492,000 $270,000
SunPower – – $339,000 $138,000
• Direct Personnel: Research Fellow: Keith/Bryce (2005-08)Research Fellow: Keith/Bryce (2005-08) MPhil student: Luke (2005-07)MPhil student: Luke (2005-07) Part-time contributions fromPart-time contributions from
- Mihail Ionescu (ANSTO)- Mihail Ionescu (ANSTO)- Jeff Simpson, Bob Spekkart (SierraTherm)- Jeff Simpson, Bob Spekkart (SierraTherm)- Dick Swanson (SunPower)- Dick Swanson (SunPower)
• Indirect Personnel: 2 x PhD students: Andy (2006-09), Bijaya (2006-09)2 x PhD students: Andy (2006-09), Bijaya (2006-09) 3 x honours students: Amarjit (2006), Simeon (2008), Lachlan (2008).3 x honours students: Amarjit (2006), Simeon (2008), Lachlan (2008). 3 x foreign students: Efthimios (2005), Joan (2007), Maria (2008)3 x foreign students: Efthimios (2005), Joan (2007), Maria (2008) 2 x summer students: Yunki (2006), Teng (2007-08) 2 x summer students: Yunki (2006), Teng (2007-08)
Slide 5
Comparison at beginningProperty APCVD TiO2 PECVD SiNx
1. Anti-reflection Good Fine
2. Cost Low ~3 x higher
Slide 6
Comparison at beginningProperty APCVD TiO2 PECVD SiNx
1. Anti-reflection Good Fine
2. Cost Low ~3 x higher
3. Passivation ofbare Si
Terrible Good
Slide 7
Comparison at beginningProperty APCVD TiO2 PECVD SiNx
1. Anti-reflection Good Fine
2. Cost Low ~3 x higher
3. Passivation ofbare Si
Terrible Good
4. Passivation ofSiO2-Si
Good Excellent
Slide 8
Comparison at beginningProperty APCVD TiO2 PECVD SiNx
1. Anti-reflection Good Fine
2. Cost Low ~3 x higher
3. Passivation ofbare Si
Terrible Good
4. Passivation ofSiO2-Si
Good Excellent
5. Static charge Unknown ModeratePositive
Slide 9
Comparison at beginningProperty APCVD TiO2 PECVD SiNx
1. Anti-reflection Good Fine
2. Cost Low ~3 x higher
3. Passivation ofbare Si
Terrible Good
4. Passivation ofSiO2-Si
Good Excellent
5. Static charge Unknown ModeratePositive
6. Hydrogen content
Little, if any High
Slide 10
Comparison at beginningProperty APCVD TiO2 PECVD SiNx
1. Anti-reflection Good Fine
2. Cost Low ~3 x higher
3. Passivation ofbare Si
Terrible Good
4. Passivation ofSiO2-Si
Good Excellent
5. Static charge Unknown ModeratePositive
6. Hydrogen content
Little, if any High
7. Barrier towater vapour
No Yes
Slide 11
Comparison at beginningProperty APCVD TiO2 PECVD SiNx
1. Anti-reflection Good Fine
2. Cost Low ~3 x higher
3. Passivation ofbare Si
Terrible Good
4. Passivation ofSiO2-Si
Good Excellent
5. Static charge Unknown ModeratePositive
6. Hydrogen content
Little, if any High
7. Barrier towater vapour
No Yes
8. Reaction with EVA
Possibly No
Slide 12
Hope vs time
0
25,000
50,000
75,000
100,000
125,000
150,000
175,000
200,000
225,000
2005
2006
2007
2008
Ho
pe
(me)
Slide 13
Project goalAddress properties 4, 5, 6 & 7 to:
• Attain equivalent surface recombination (J0E)with TiO2/SiO2/Si compared to SiNx/SiO2/Sifor textured silicon after accelerated testing(damp-heat, UV).
TiO2 or SiNx
SiO2
n+ diffused silicon
n-type silicon
Slide 14
Mercury Maximum efficienty - 1D modeling with tauP=2e-3 s, J0e(Back) = 1e-13 A/cm2
20.0
20.5
21.0
21.5
22.0
22.5
0 20 40 60 80 100 120 140 160 180 200
FSF J0e [fA/cm2]
Max
imu
m E
ffic
ien
cy [
%]
Merc3.5
Merc3.7
Property 4: Passivation of SiO2-Si
Efficiency of rear-contact cells strongly affected by J0E
SiNx/SiO2
TiO2/SiO2
Slide 15
Causes for lower J0E (Andy)• Not contamination
Phew!Phew!
• Not moisture in reactants Good.Good.
• Loss of hydrogen during deposition at 400 oC Reversible with FGAReversible with FGA
Does not occur at lower 250 oCDoes not occur at lower 250 oC
• Stress on texture (?) HmmmHmmm
• No atomic H with APCVD, unlike PECVD Probably not a big deal – attain similar DProbably not a big deal – attain similar Ditit
• Static charge Need to add some!Need to add some!
Slide 16
Property 5: Static Charge
1E-15
1E-14
1E-13
1E-12
1E-11
1E-10
-1E+12 -5E+11 0 5E+11 1E+12
Insulator charge density (cm-2)
J0
Ee
ff (
A/c
m2
)
3.00E+12
1.00E+12
3.00E+11
1.00E+11
3.00E+10
1.00E+10
TiO2/SiO2 SiNx/SiO2
• 5 ohm-cm n-type
• All donor defects
• n = 1015
Slide 17
Add charge by corona? Adding charge by corona discharge has problems:
• our corona damages Si-SiO2 interface- will add gate (Tom/Klaus- will add gate (Tom/Klaus))
• charge is slowly removed by water vapour in atmosphere- add moisture barrier (Simeon/Maria),- add moisture barrier (Simeon/Maria),- anneal TiO- anneal TiO22 (Simeon/Maria) (Simeon/Maria)
Slide 18
Property 6: Hydrogen• Measurement
ERDA (Andy)ERDA (Andy)- Almost underway...- Almost underway...
SIMSSIMS RTA & PC (Teng)RTA & PC (Teng)
- increase/decrease in recomb as H evolves- increase/decrease in recomb as H evolves
• As deposited TiO2 is permeable to H2 and H+
HH22 lost during deposition at 400 oC lost during deposition at 400 oC post-deposition FGA replaces Hpost-deposition FGA replaces H22
FGA within N2 curtains replaces HFGA within N2 curtains replaces H22..
• Annealed TiO2 appears impermeable to H2 (Maria)
• H not bonded in significant concentration to TiO2 when FGA in N2 curtains. Will test FGA in separater NWill test FGA in separater N22 H known to alter crystal state but not necessarily bonded to Ti H known to alter crystal state but not necessarily bonded to Ti
and O.and O.
20 30 40 50 60 70 80 90 100 110 120 130 140 150
0
10
20
30
H Y
ield
[cou
nts]
Channel Number
2B 2C 2D 2B, 2C and 2D
Slide 19
Property 7: Moisture barrier
Occurs on* planar and texture (Thimios)* annealed, alnealed, RTA (Thimios)* n-type & p-type (Luke)* P & B diffusions (Joan)
Does it matter? Yes! Damp heat test degrade Si-SiO2.
1E-6
1E-5
1E-4
1E-3
1E-2
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5
Damp-heat exposure (mins)
Eff
ec
tiv
e li
feti
me
(s
)
Planar,FGA
Planar,Alneal
Planar,RTA
Textured,FGA
Texutred,Alneal
Textured,RTA
0
Slide 20
Property 7: Moisture barrier
Does APCVD TiO2 prevent it degradation No! (Maria, Luke)
0.E+00
1.E-14
2.E-14
3.E-14
4.E-14
5.E-14
6.E-14
7.E-14
8.E-14
9.E-14
0 2000 4000 6000 8000 10000 12000 14000 16000
Time in damp heat (mins)
J0E
(A/c
m2
)
2a SiO2 control
2b SiO2
7a TiO2/SiO2 control
7b TiO2/SiO2
Slide 21
Mercury Maximum efficienty - 1D modeling with tauP=2e-3 s, J0e(Back) = 1e-13 A/cm2
20.0
20.5
21.0
21.5
22.0
22.5
0 20 40 60 80 100 120 140 160 180 200
FSF J0e [fA/cm2]
Max
imu
m E
ffic
ien
cy [
%]
Merc3.5
Merc3.7
Passivation of SiO2-Si after damp heat
Efficiency of rear-contact cells strongly affected by J0E
SiNx/SiO2
TiO2/SiO2
TiO2/SiO2
after damp-heat
Slide 22
DH degradation: cause / solution• Cause is not
removal or deposition of surface charge (Joan)removal or deposition of surface charge (Joan) removal of H at interfaceremoval of H at interface
• Increase in J0E corresponds to increase in Dit (Luke). Possibly due to wet oxidation at 85 oC wet oxidation at 85 oC silicic acid in SiOsilicic acid in SiO22 creating stress at Si-SiO2 interface creating stress at Si-SiO2 interface
• Solution requires moisture barrier HMDS: no success so farHMDS: no success so far Annealed TiOAnnealed TiO22: anneal removes H: anneal removes H Module encapsulationModule encapsulation (PECVD Si(PECVD SixxNNyy))
Slide 23
Conclusion – the best hope...
1E-15
1E-14
1E-13
1E-12
1E-11
1E-10
-1E+12 -5E+11 0 5E+11 1E+12
Insulator charge density (cm-2)
J0
Ee
ff (
A/c
m2
)
3.00E+12
1.00E+12
3.00E+11
1.00E+11
3.00E+10
1.00E+10
SixNy/SiO2TiO2/SiO2
2) Embed high Q without damage (gated corona?)
3) Engineer moisture barrier to prevent Q loss & DH deg.
1) Deposit hydrogenated TiO2
3a) Anneal TiO2
3b) Add barrier
3c) Vapour-tight module