life prediction for cigs solar modules · life prediction for cigs solar modules d.j. coyle, ......

1DJ Coyle

GE Global Research 16-February 2011

Life Prediction for CIGS Solar Modules D.J. Coyle, H.A. Blaydes, R.S. Northey, J.E. Pickett, K.R. Nagarkar,

R.A. Zhao, and J.O. Gardner

GE Global Research

Niskayuna, NY, 12309

DOE PV Module Reliability WorkshopDenver, CO

February 16-17, 2011

One can use accelerated testing data to predict real-world module failures only if the degradation mechanisms are known and their dependence on environmental factors measured. The moisture-induced degradation rate of flexible CIGS solar cells has been measured as a function of temperature and humidity and fit to a kinetic rate expression. This expression is coupled to a model of moisture diffusion into a package and typical meteorological input data to create a cumulative damage model to predict lifetime of packaged cells versus outdoor exposure and package construction. Estimated acceleration factors for damp heat (85C/85%RH) vs. Miami range from 15X to 50X, depending on the package, since diffusion through the package is accelerated differently than the cell degradation kinetics. The degradation rates are strongly dependent on the transparent conductive oxide used for the window layer and the electrically-conductive adhesive used for the contacts. The dependence of degradation on encapsulant materials is fundamentally different than is often assumed in the literature.

2DJ Coyle


Prototype Flexible CIGS Module

3DJ Coyle


Top Sheet• Transparent•UV Stable, UV block• Electrical insulation•Mechanical (cut, hail)•Moisture barrier

Flexible Thin-Film PV Package

CdSCIGS

TCO

MolybdenumFlexible Substrate(Stainless Steel)

Transparent Moisture BarrierTransparent Polymer Film

Moisture BarrierPolymer Film

Encapsulant

Encapsulant

Edge Seal

Encapsulant• Low Tg compliant “glue”• Transparent• Adhesion

Backsheet• Electrical insulation•Mechanical (cut)•Moisture barrier

PV Cell

Edge Seal• Barrier• Adhesion

4DJ Coyle


2 µm

ZnO/ITOCdS

Cu(InGa)Se2

Mo

Cu(InGa)Se2 Device Environmental Stability

(W. Shafarman & L. Stolt, Handbook of Photovoltaic Science and Engineering, Ed. A. Luque & S. Hegedus, Wiley, 2003)

Increased Series Resistance Rs

Decreased Shunt Resistance Rsh

Increased Recombination1. Quasi-neutral region2. Depletion region

Moisture-induced degradation

Increased Series Resistance Rs

i-ZnO Buffer Layer

5DJ Coyle


Test Cells:1. Global Solar Test Cells (ITO)2. AZO

~ 36 x 46 mm exposedStainless steel foilMo coatingECA – Tabs/ribbons

Efficiency ~ 12 – 13.5%Voc ~ 600 - 610 mVJsc ~ 33-36 mA/cm2

FF ~ 60 - 62%A ~ 16.5 cm2

GSE test celltabbed and encapsulated

6DJ Coyle


1. Cell construction• ITO vs AZO window layer• Type of ECA for interconnect• Other

2. Exposure• Accelerated testing (ovens with various temp, RH)• Real-world exposure (Miami, Phoenix, …)

3. Package• Barrier properties of topsheet and backsheet• Encapsulant• Edge seals• other

Factors for prediction of lifetime (moisture degradation):

7DJ Coyle


Life Model – Moisture Sensitivity1. CIGS Degradation Kinetics - Measure

• Degradation rate vs. Temp, humidity• ITO vs AZO• ECA - Interconnect degradation can play a role

2. Moisture Diffusion into Package - Model• Meteorological data – TMY3 from NSRDB

‐ Hourly irradiance, air temp, ground temp, humidity, wind speed• Heat transfer model of module

‐ Radiation, free & forced convection• Diffusion through barrier film, Saturation of encapsulant, no edge effects

3. Coupled Model - Predict• Cumulative degradation and average life vs. location and package design• Tradeoffs between CIGS sensitivity and package design/cost• Interpretation of accelerated tests results

8DJ Coyle


Degradation Data - Examples

y = -0.0077xR² = 0.6866

-35

-30

-25

-20

-15

-10

-5

0

5

0 500 1000 1500 2000 2500

% C

hang

e Ef

f

Time (hrs)

% Change Eff - 85C/50% - ECA1

y = -0.027xR² = 0.9017

-80

-70

-60

-50

-40

-30

-20

-10

00 500 1000 1500

% C

hang

e Ef

f

Time (hrs)

% Change Eff - 95/50% - ECA1

• High temperature, humidity faster• Driven by FF loss due to Roc and some shunting

-80

-70

-60

-50

-40

-30

-20

-10

00 500 1000 1500

% C

hang

e Ef

f

Time (hrs)


y=-0.0213xR2=0.86

-80

-70

-60

-50

-40

-30

-20

-10

00 200 400 600 800

% C

hang

e Ef

f

Time (hrs)


y = -0.0523xR2 = 0.86

9DJ Coyle


Scaling with partial pressure water

Negative activation energy!Need to scale with % saturation (RH)

RHSC

SC

b

b

a

a ==Interfacial equilibrium (Henry’s Law)

0

20

40

60

80

100

120

0 10 20 30 40 50 60 70

RD

(%/1

000h

rs)

PH2O (KPa)

22oC 85oC

75oC65oC

55oC35oC

ITO ECA0

αpRTEkR a

D

−= exp0

Ea = -50 kcal/mole

10DJ Coyle


0

10

20

30

40

50

60

70

80

0% 20% 40% 60% 80% 100%

RH

RD (%

/100

0hrs

)

22oC

85oC75oC

65oC

55oC

95oC

CIGS Degradation Kinetics (Global Solar test cells)

(Klinger, D. J., “Humidity acceleration factor for plastic packaged electronic devices”, Quality and Reliability Engineering International. Vol. 7, 965-3711, 1991).

• For every Temp & RH, fit data to linear degradation rate (1st 20% of degradation)• Fit rate of degradation vs Temp, RH to kinetic model

+−

=

−

εcell

cellRTE

oDeg RHRHekR

a

1

deg,

Dashed Curve - ECA0

• Strong RH dependence at high RH• ECA affects temperature dependence

11DJ Coyle


CIGS Degradation - AZO vs ITO

• AZO ~ 25X ITO• Comparable to published data

0

1

10

100

1000

10000

0% 20% 40% 60% 80% 100%

RD

(%/1

000h

rs)

RH

22oC

85oC75oC 65oC

55oC

95oC

AZO 85oC

AZO 22oC

Olsen 2008

Britt 2006Westin 2006

12DJ Coyle


Arrhenius Plot

-3

-1

1

3

5

7

0.0027 0.0028 0.0029 0.0030 0.0031 0.0032 0.0033 0.0034

Ln(R

D)+

Ln(R

H/(1

-RH

+ε))

1/T (1/K)

AZO

ITO ECA0

ITO ECA1

Olsen 2008

Britt2006

Westin2006

• ECA0 very low activation energy

13DJ Coyle


Package Diffusion Model

−−=

c

E

E ttRHSC e /

1

If initially dry:

maxWVTRSLt EE

c =

(Kempe, M.D., “Modeling of rates of moisture ingress into photovoltaic modules,” Solar Energy Materials & Solar Cells, Vol 90 (2006) pp. 2720–2738).

Mass Balance, Interfacial Equilibrium,Fickian Diffusion, Dbarrier<<Dencapsulant

RH

cellRH

( )0=xCB

( )BB LxC =

EC

RH

cellRH

( )0=xCB

( )BB LxC =

EC

encapsulantbarrier cell

Integrate moisture ingress with hourly weather data (TMY3)

c

EEE

tCRHS

tC −

=∂∂

14DJ Coyle


1

10

100

1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01

WVTR (g/m2/day)

Life

yrs

(20%

deg

rade

)

San FransiscoMiami

Golden CO

Phoenix

Life vs. Barrier: ITO-ECA0

Need ~ 4x10-5 g/m2/day package ~ 20 yr life

15DJ Coyle


1

10

100

1000

1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01

WVTR at 25oC (g/m2/day)

Life

yrs

(20%

deg

rade

)San Fransisco

Miami

Golden CO

PhoenixECA1 - solid linesECA0 - dashed lines

Life vs. Barrier: ITO-ECA1

Need ~ 8x10-3 g/m2/day package ~ 20 yr life

ECA0

ECA1

16DJ Coyle


Life vs Barrier – ITO vs AZO

0

1

10

100

1000

1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01


Life

yrs

(20%

deg

rade

) San Fransisco

Miami

Golden CO

PhoenixITO/ECA1AZO

ITO Life 5-25x AZO Life

AZO

ITO

17DJ Coyle


AcceleratedTesting• Nonlinear relationship• No simple scaling• Depends on details of

kinetics and package

,

0

1

10

100

1000

1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01


Life

yrs

(20%

deg

rade

)

Miami, FL

AZO

ITO/ECA0

ITO/ECA1

,

0

1

10

100

1000

1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01


Life

yrs

(20%

deg

rade

)

Miami, FL

AZO

ITO/ECA0

ITO/ECA1

1

10

100

1000

10000

1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01


Hou

rs 8

5o C 8

5% R

H to

10%

Deg

rade

Damp Heat

AZO

ITO/ECA0

ITO/ECA1

1

10

100

1000

10000

1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01


Hou

rs 8

5o C 8

5% R

H to

10%

Deg

rade

Damp Heat

AZO

ITO/ECA0

ITO/ECA1

~10,000 hrs~4,000 hrs~2,500 hrs

18DJ Coyle


FL, AZ Testing

Results as expected after 3 months

• ITO/ECA1 no measureable degradation

• ITO/ECA0 ~ 5% down as expected

Calc Pmax

changeMeasured

Pmax change

Phoenix ECA1 -0.1% -1% +-2%

Miami ECA1 -0.5% -1% +-2%Miami ECA0 -4.1% -5% +-2%

19DJ Coyle


Encapsulant Effect

• Low solubility encapsulants become saturated faster => BAD!!

maxWVTRSLt EE

c =

−−=

c

E

E ttRHSC e /

1

1

10

100

1.E-04 1.E-02 1.E+00 1.E+02

Life

Mia

mi -

Yrs

20%

deg

rada

tion

Diffusion Half-Time @25oC (yrs)

1

10

100

1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01

Life

Mia

mi -

Yrs

20%

deg

rada

tion

WVTR @25oC (g/m2day)

Urethane 2mm

Urethane 0.5mm

Silicone 2mm

Silicone 0.5mm

20DJ Coyle


-4

-3

-2

-1

0.0027 0.0029 0.0031 0.0033

Solu

bilit

y (g

/cm

3 )

1/T (1/oK)

Silicone

Urethane

EVA

Polyolefin

10

10

10

10

100

1,000

10,000

1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03


Hou

rs L

ife 8

5o C 8

5% R

H (2

0% d

egra

de)

TPU -6x14mil_2x14mil

Silicone_2x19mil

GSE cellsECA1

ACLAR

PC 5,7mil

PET 5,7mil

3M

2mil7.8mil

Polyolefin

Barrier Film

Aclar®Urethane

Encapsulant Experimental Plan (85oC, 85%RH)

All encapsulants the same

Thick, high SEencapsulants better

Thickness (mm)

Solubility @85C g/cm3

t1/2 Hrs @85C

Urethane 0.355 3.0E-02 89Urethane 1.065 3.0E-02 268Silicone 0.480 3.1E-03 12Polyolefin 0.400 3.6E-03 12

WVTR @85C g/m2day

Aclar® 0.307 2.0

Material

Encapsulant

Barrier

10x

21DJ Coyle


Encapsulant Confirmation Experiment

• EVA outlier – unusually fast degradation

• Urethane (TPU), silicone, polyolefin same, latter two even with Aclar®, as expected

• Barrier (Aclar®) makes no difference for silicone & polyolefin, as expected

• Barrier (Aclar®) improves TPU life, especially thicker, as expected

-80.

-70.

-60.

-50.

-40.

-30.

-20.

-10.

0.

0 200 400 600 800 1000

Pmax

Cha

nge

(%)

Time (hrs)

TPU/Aclar®

3xTPU/Aclar®

Silicone/Aclar®

1x,3xTPU/PC

PO/Aclar® PO/PCEVA/PC

22DJ Coyle


Comparison of Experiment & Model

Saturation

Degradation

-30

-25

-20

-15

-10

-5

0

0 200 400 600 800 1000

% C

hang

e Pm

ax

Time (Hours @85oC/85%RH)

0

10

20

30

40

50

60

70

80

90

100

0 200 400 600 800 1000

% E

ncap

sula

nt S

atur

atio

n (1

00C

E/S)

Time (Hours @85oC/85%RH)

PC

3xTPU/Aclar®

TPU/Aclar®

PC

3xTPU/Aclar®

TPU/Aclar®

23DJ Coyle


Conclusions1. Life model and accelerated test scaling developed

– Relative humidity and % saturation of encapsulant are key

2. Moisture sensitivity of CIGS almost independent of encapsulant type

3. Module lifetime longer for thick encapsulants with high water solubility

4. AZO vs ITO CIGS degradation kinetics quantified ~25X

5. ECA can also be a strong factor in degradation

6. Diffusion-controlled: Life ~ (tc/RD)1/2 ~ (diffusion-time*degrade-time)1/2

7. Significant moisture barriers required for 20 yr life – even for ITO

8. Acceleration factor for damp heat smaller than assumed, highly nonlinear!

9. Methodology can predict life for any moisture-sensitive module (once kinetic constants are measured)

24DJ Coyle


Acknowledgements

Mike Kempe NRELJeff Britt Global Solar EnergySteve Hegedus IEC

This work was supported by:U.S. Department of Energy Award DE-FC36-07GO17045

GE Global ResearchGE Energy

life prediction for cigs solar modules · life prediction for cigs solar modules d.j. coyle, ......

Documents