Reliability Issues for Photovoltaic Modules (Presentation) · Funded by Department of Energy...
Transcript of Reliability Issues for Photovoltaic Modules (Presentation) · Funded by Department of Energy...
Reliability Issues for PhotovoltaicReliability Issues for Photovoltaic ModulesSarah Kurtz
National Renewable Energy LaboratoryNational Renewable Energy LaboratoryGolden, CO
Funded by Department of EnergyNREL/PR-520-46481
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 1October 15, 2009
NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.
Outline
• Solar – a huge success, but still a long way to go• Key approaches to solar electricityy y
• Solar thermal• Crystalline silicon• Thin film – amorphous silicon CdTe CIGS• Thin film – amorphous silicon, CdTe, CIGS• Concentrator – low- and high-concentration approaches
• Importance of reliability to success of solar• Reliability issues specific to each approach
• Silicon – strong performance; continuous improvement; quantitative predictionsquantitative predictions
• Thin film – uniform, large-area deposition for product development and sensitivity to moisture; metastabilities
• Concentrator product development; simultaneous
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 2October 15, 2009
• Concentrator – product development; simultaneous optimization of multiple components
Growth of photovoltaic (PV) industry
Tons of Si passes
Area of Si passes
Tons of Si passes microelectronics
2006
microelectronics2001
0.01%-0.1% of electricity now comes from PV - extrapolates to > 5% in 2020
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 3October 15, 2009
competitive with conventional electricity for 0.1% - 1% of market; more in futureRogol, PHOTON International August 2007, p 112.
3
Growth of PV industry
Annual replacement of electricity capacity for 20 yr cycle
100
ped
Annual new electricity capacity 1996-2006*
10
of P
V s
hip
If we can maintain the
1
GW
If we can maintain the currentgrowth rate, PV will reach majormilestones in less than 10 yrs
If we can maintain the current growth rate, PV will reach major milestones in < 10 yrs
20202015201020052000Y
milestones in less than 10 yrsmilestones in < 10 yrs
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 4October 15, 2009
Year*www.eia.doe.gov/emeu/international/electricitycapacity.html (4012-2981 GW)/10 yr
Solar energy is abundantConvenient truth: small area can supply our energy needsConvenient truth: small area can supply our energy needs
5-6 kWh/sq m/dayAt 10%
efficiency, y,area needed
for US electricity
>10 kWh/sq m/day>10 kWh/sq m/day
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 5October 15, 2009
Sunlight reaching earth in 1 hour is enough to power the world for 1 year
Solar thermal electricParabolic trough is the primary technology todayResurgence of interest~ 400 MW installedCurrently generates ~ 0.01% of US electricityyCan generate electricity into the evening & use fuel into the night
80
60
r in
US
Solar thermalParabolic trough
40
20com
plet
ed/y
r20
0
MW
c
20052000199519901985
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 6October 15, 2009
Year4 GW planned in US by 2014
64 MW S l i P b li64 MW Solargenix Parabolic Trough Plant in Nevada -
2007
1-MW Arizona Trough Plant – near Tucson AZ 2006
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 7October 15, 2009
Tucson, AZ - 2006
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 8October 15, 2009
8
Three key approaches to photovoltaic (PV) panels
Two strategies to reduce semiconductor material
Conventional approach
FrontSolar cell
2. Thin film
Back 1. Silicon
Reduce cost by reducing use
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 9October 15, 2009
3. Concentratorof semiconductor
Many technology choices
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 10October 15, 2009
10
One “winner” or many technologies?
Nickel cadmium Nickel metal hydrideAlkaline Nickel metal hydridea e
Lithi i
Lead acid
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 11October 15, 2009
Lithium ion
LithiumDifferent technologies for different applications
Cost of electricity: two or three parts2 O ti d i t
6
5
1. Initial price (estimates*) 2. Operation and maintenance- Fuel cost (Coal PV - Operation (Coal PV - Maintenance (Coal PV s
($/W
)
5
4
3st ($
/W)
e es
timat
es
PV is already competitive for peak power in some locations3
2
1
Initi
al c
os
3. Total electricity generated- Capacity factor (Coal PV es
of p
rice
1
0Old coal New coal Clean coal PV
p y ( Coal ~100%; PV ~ 25%)- Life of plant (Coal PV
*Fortnightly’s SPARK, p. 10, May 2008
Exa
mpl
e
Upfront costs for PV and coal plants are converging Ongoing costs are less for PV
o g y s S , p 0, ay 008
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 12October 15, 2009
Operation only during daylight hours increases cost by ~X4 Key remaining question is life of PV plant (30 years?)
Importance of reliability & durability1 01.0
0.8
cost
Performance of good system
0.6
man
ce o
r c Performance of systems with problems
0.4
ativ
e pe
rfor
O ti d i t t O&M t0.2
0.0
Rel
a Operation and maintenance costs for systems with problems
O&M costs for good system
Cost of solar electricity depends on degradation, lifetime, & ongoing costs
2520151050
Year
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 13October 15, 2009
Those paying for upfront investment want guarantee of system life
How to satisfy the investor?
Historically, degradation & failure mechanisms have been found in the field that were not found in accelerated testing
Predictive models need to be validated with field data
Big challenge: How can we give 30-year predictions for degradation & failure rates when the product has only been in the field for 1-2 years?
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 14October 15, 2009
Silicon modules
SiliconSGlass
Si module cross section
EVA
EVA
Backsheet
Silicon cell Tab
Common encapsulation materialsEVA Ethylene vinyl acetateEVA - Ethylene vinyl acetatePET - polyethylene terephthalatePVF - poly vinyl fluoride
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 15October 15, 2009
History of Si module qualification test: JPL (Jet Propulsion Lab) Block buys
Test I II III IV V
Year 1975 1976 1977 1978 1981
Th l 100 l 50 l 50 l 50 l 200 lThermal Cycle (°C)
100 cycles-40 to +90
50 cycles-40 to +90
50 cycles-40 to +90
50 cycles-40 to +90
200 cycles-40 to +90
Humidity 70 C, 90%RH, 68 hr
5 cycles40 C 90%RH
5 cycles40 C 90%RH
5 cycles54 C 90%RH
10 cycles85 C 85%RH68 hr 40 C, 90%RH
to 23 C40 C, 90%RH to 23 C
54 C, 90%RH to 23 C
85 C, 85%RH to -40 C
Hot spots - - - - 3 cells, 100 hrshrs
Mechanical load
- 100 cycles ± 2400 Pa
100 cycles ± 2400 Pa
10000 cyc. ± 2400 P
10000 cyc. ± 2400 Pa
H il 9 i t 10 i tHail - - - 9 impacts 3/4” - 45 mph
10 impacts 1” - 52 mph
NOCT - - - Yes Yes
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 16October 15, 2009
High pot - < 15 µA 1500 V
< 50 µA 1500 V
< 50 µA 1500 V
< 50 µA 2*Vs+1000
JPL Block buys led to dramatic improvementsO t d l i d (Whi l 1993)• One study claimed (Whipple, 1993):• Pre-Block V: 45% module failure rate
P t Bl k V 0 1% d l f il t• Post-Block V: <0.1% module failure rate• Studies of c-Si modules show that module failures
ll (i t d i t h t i l )are small (inverters dominate when cost is low)Unscheduled maintenance costs
69%
Currently, most reports imply that c-Si module failures are
dominated by improperInverters
dominated by improper installation, lightning strikes,
critters, etc. 21%
System
PV
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 17October 15, 2009
(Prog. PV 2008; 16:249) 10%
Today’s qualification standards are similar to JPLare similar to JPL
IEC 61215 - Crystalline silicon design qualification includes 18 test procedures pThermal cycling - 200 cycles -40°C to +85°CHumidity freeze - 10 cycles +85°C, 85% RH to -40°CD h t 1000 h t +85°C 85% RHDamp heat - 1000 hrs at +85°C, 85% RHWet leakage current - Wet insulation resistance X area >
40 MΩm2 at 500 V or system voltageRequirement is typically to retain 95% of original power
production
IEC 61646 (thin film) and IEC62108 (CPV) IEC 61646 (thin film) and IEC62108 (CPV) are similar
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 18October 15, 2009
www.iec.ch
Silicon modules – remaining challenges
New materials (reduce cost, improve performance) – will these have same reliability?y
Continued quality assurance (e.g. impurities in Si give light-induced degradation)
Arcing, grounding, power conditioning, other system-related problems
Confident, long-term, quantitative predictions
Typical degradation rates are 0-1%/yr (difficult to measure); field failure rates are often < 0.1%/yr
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 19October 15, 2009
Proven module does not guarantee new products
Past success does not guarantee future success
C-Si
Thin filmThin film70%
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 20October 15, 2009
20About two-thirds of degradation rates are measured as < 1%/yr33rd PVSC, TamizhMani, 2008
Measurement of degradation rates takes years
12001.0
0 9-0.14%/yrDC power
1000
g (W
)
0.9
0.8
Inverte
inverterreplacedinverterreplaced
p
800
ower
ratin
g
0.7
er efficiency
inverterefficiency
600
Po
0.6
y
ASE Americas Si modules
400 0.52003 2005 2007
Xantrex Suntie 2500
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 21October 15, 2009
Need precise measurement of irradiance, temperature, etc. 21
Time
Thin-film approaches on the market
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 22October 15, 2009
CuIn(Ga)Se CdTe Amorphous silicon
Thin-film approach
• First Solar (Toledo, OH) - 5th biggest PV company in world in 2007p y
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 23October 15, 2009
Thin-film structures
Glass
CdTe uses superstrate
Glass for protection
CuInGaSe uses substrate
Glass
ITO or TCO
Glass for protection
CdS
CdT
ZnO or TCO
CdSNot to scaleCdTe
MetalCuInGaSe
scale
Molybdenum
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 24October 15, 2009
Glass for strength Glass
Monolithic module integration
5 x 120 µm
Conductor
Device
Conductor
ContactCell
( )
ContactCell
(I ti )
Active Cell
+(Inactive)(Inactive)
–w
Glass
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 25October 15, 2009
Hurdle for thin films: uniformity of deposition
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 26October 15, 2009
Hurdle for thin films: uniformity of interconnections
5 x 120 µm
Conductor
Device
Conductor
ContactCell
( )
ContactCell
(I ti )
Active Cell
+(Inactive)(Inactive)
–w
Glass
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 27October 15, 2009
Module metastability: Reversible vs irreversible changes
• Amorphous silicon degrades in light and recovers when annealed in the darkCIGS d CdT h t i t• CIGS and CdTe can show transients
9Rated efficiency
9
8
%)
June 2003
JulyJuly
Jan 20090.2 kWh/m
2Jan 2009
0.2 kWh/m2
0.25 kWh/m2
start test0.25 kWh/m
2
start test
7
6Effi
cien
cy (%
July2008
Dec2008Dec
2008
July2008 2.75 kWh/m
2
tested same day2.75 kWh/m
2
tested same day
3.15 kWh/m2
tested next day3.15 kWh/m
2
tested next day
50°Covernight
50°Covernight
0°C0°C As high as 6
5
tested next daytested next dayovernight
0 Covernight
CIGS d l CIGS no exposure
30% metastability
has been observed
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 28October 15, 2009
4CIGS modulesCIGS 3-year exposure in Florida
Damp heat can cause degradation ZnO (and other transparent conductors) react
with moisture, causing increase in series i tresistance
CuInGaSe may react with moistureT t C d ti O idTransparent Conductive Oxides
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 29October 15, 2009
Many possible failure mechanisms
• CdTe has shown instability of back contact (diffusion?)( )
• Edge seal may allow water into glass/glass module• Partial shunts or conducting diodes may be seen at
scribe lines or other defected areas• Adhesion to glass can be problem
Role of sodium is important in CuInGaSe modules• Role of sodium is important in CuInGaSe modules, but sodium can move
• Currently, the biggest effort with CuInGaSe is to try to y, gg yput it on a flexible substrate – requires excellent barrier coating unless cell can be hardened to moisture
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 30October 15, 2009
moisture
Range of concentrator approaches
Amonix JX Crystals
High concentration Low concentration
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 31October 15, 2009
• 35% - 40% cells• 400X – 1500 X
• 15% - 25% cells• 2X – 100 X
CPV challenges – many interactions
CPV can fail in many ways, but it can be difficult to understand where the problem is and to fix the pproblem without creating a new problem
• Tracking – optics must be aligned with the sun• Optics – durability can be problem, soiling; optics
affect rest of system• Cell – must be encapsulated, but not affected by
UV; size of cell affects rest of system• Heat sink – must be electrically isolated, but
excellent thermal contact
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 32October 15, 2009
• Modularity may be benefit!
UV transmission
Analysis of transmitted optical spectrum enabling accelerated testing of CPV designs
SPIE 2009 David Miller, et al
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 33October 15, 2009
Bonds to heat sink and optics
Optic Optical bond with 100% Transmission50 W/sq cmElectrical
Cell
H t i k
contactSmall T (<10°C)Electrical isolation
Heat sink No voidsT cycle OK
• Borrowing experience from power electronics and DBC (direct bonded copper) makes this a smaller
IR image of void in die attach
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 34October 15, 2009
issue• Intense UV may be a substantial problem, but optics may not transmit UV
Bosco 34th PVSC
Concentrators – reliability challenges
• Wide variety of designs• Qualification test is not well established• Qualification test is not well established• Companies spend time developing their own
l t d t t t d d taccelerated tests to speed product development cycles
• Very few companies have heritage with field testing
• Everyone wants to bring a product to market immediately
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 35October 15, 2009
y
Concentrator technology
CreativeCreative optical designs?designs?
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 36October 15, 2009
Summary
• Solar is growing rapidly and could become a i ifi t f l t i it ithi 10 20significant source of electricity within 10-20 yrs
• Silicon modules are performing well in the field; reliability testing of new designs is still importantreliability testing of new designs is still important
• CdTe and CuInGaSe modules are sensitive to moisture, so must be carefully sealed; only
h Si d l il bl i fl ibl famorphous Si modules are available in flexible form• Concentrator PV is in product development stage, but
is benefiting from expertise in other industriesis benefiting from expertise in other industries• In general, PV industry can benefit from the reliability
testing experience of the microelectronics industry
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 37October 15, 2009
Let’s work together to help PV grow!
Planet powered by renewable energygy
By year 2100 or before?Thank you for your attention!y y
Thank you to :yDave AlbinGlenn AlersNick BoscoJoe del CuetoChris Deline
GEd GelakSteve GlickPeter HackeDirk JordanMike KempeTom MoriconeTom MoriconeBill MarionDavid MillerMatt MullerJohn PernJose Rodriguez
ASTR 2009 Oct 7 – Oct 9, Jersey City, NJSarah KurtzPV Reliability Page 38October 15, 2009
gBill SekulicRyan SmithKent TerwilligerDavid Trudell