May 27, 2004 Photovoltaics Laboratory Chalcogenide Solar Cells: Choosing the Window Colorado State University Funding: US National Renewable Energy Laboratory.

May 27, 2004 Photovoltaics Laboratory

Chalcogenide Solar Cells: Choosing the Window

Colorado State University

Funding: US National Renewable Energy Laboratory (NREL) Japanese New Energy Development Organization (NEDO)

Special thanks to Markus Gloeckler for assistance with figures

Jim Sites

Markus Gloeckler, Alex Pudov, and Ana Kanevce (CSU) Falah Hasoon and Miguel Contreras (NREL)

Hans Schock (IPE) and Tokio Nakada (AGU)

Collaborators:

European Materials Research Society – Spring 2004

Approach


(1) Device-physics approach to the selection of window layers for fabricating high-performance solar cells with CdTe and CIGS absorbers.

[Device physics no means the whole story, but may give useful direction even when material structure or other factors play a major role]

(2) Large range of possible band gaps will be considered.

(3) Attempt to be quantitative.

(4) Focus on two areas: (a) Window absorption: how much of an effect? (b) Conduction-band offset: what happens when it changes?

Choosing the Window: Outline


(1) Photon considerations: Window absorption.

(2) Conduction-band offset problem I: Big spikes (and their “red-kink” precursor) that limit current.

(3) Conduction-band offset problem II: The cliff problem that limits voltage.

(4) How much slack does one get in choosing the window?

(5) Conclusions.

Short-Wavelength Current: CdS Windows on CdTe


Granata, Sites, Contreras-Puente and Compaan, IEEE PVSC-25, 853 (1996)

Same current loss should apply for CI(G)S cells.

Short-Wavelength Collection

Current Loss with Alternative Windows


Absorption spectra based on that of CdS, but shifted in energy.

Calculated Values

Fractional Current Loss


For 100-nm window layer

Larger fraction with smaller current from larger-gap absorber.

Efficiency Contours


Record CIGS Cell

Parameters for record CIGS cells EC effects neglected

100 nm window VOC = Eg – 550 meV Fill-factor = 80%







(5) Conclusions.

Sign Convention for EC


EC < 0

position [m]0.1 0.2 0.3 0.4 0.5

-2

-1

0

1EC > 0

position [m]0.1 0.2 0.3 0.4 0.5

-2

-1

0

1

ZnOCdS

CIGS

ZnOCdS

CIGS

"spike" "cliff"

Smaller Gap Absorber

Larger Gap Absorber

Some consensus on EC magnitudes between theory, experiment, and numerical simulations of J-V curves

Spike can impede photoelectrons (may be bad) Cliff slows forward electrons in interfacial-recombination region (also may be bad)

Earlier “Red-Kink” (Solarex Cells)


Also seen In cells from NREL, Boeing, and Siemens/Shell

Dark and Red-light J-V Curves

Producing a “Red” Spectrum


Wavelength [nm]

200 400 600 800 10000

20

40

60

80

100 filter trans-

mittance [%]AM1.5

AM1.5 X filter

[mA/cm2/m]

Wavelength [nm]300 400 500 600 700 800 900 1000 1100

Tra

nsm

ittan

ce [%

]

0

20

40

60

80

100

600-nm high-pass filter Series of high-pass filters with different-wavelength cut-offs

Use a high-pass filter

Red kink with CdS occurs when no photons are above 2.4 eV

The Red Kink in CdS/CIS


Position [m]0.2 0.4 0.6

Ene

rgy

[eV

]

-0.5

0.0

0.5

1.0

1.5

CdS

with "blue"photons

without "blue"photons

Ec

CISZnO

Voltage [V]

-0.5 0.0 0.5 1.0C

urre

nt D

ensi

ty [m

A/c

m2]

-40

-20

0

20

40

Well-behavedCollection

ImpededCollection

Red & DarkConverge

dark

white

red

NREL CdS/CIS J-VConduction Band at V = 0

(light/dark difference exaggerated)

CdS barrier impedes electron transport; blue photons may generate sufficient electron-hole pairs in CdS to alter trap occupation and mitigate the effect.

Can be a serious problem if no blue photons present.

Usually not a problem with white light, but small “kink” sometimes seen.

Compensated CdS

Kink Depends on CdS Thickness (Simulation)


Position [m]0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

Ene

rgy

[eV

]

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

variations in CdS thickness

w. light

r. light, dark

Voltage [V]-0.6 -0.3 0.0 0.3 0.6 0.9

Cur

rent

Den

sity

[mA

/cm

2 ]

-40

-20

0

20

40

NA = 1017 [cm-3]

n = 6 x 1016 [cm-3]

redwhite

10075 50 25

CdS parameters

noCdS

thickness (nm):

Weaker kink with thinner CdS. (Also seen experimentally)

More generally: strength of kink varies with the carrier densities of CdS and TCO, and with the CdS defect density.

Conduction Band. Impact of barrier increases with CdS thickness.

Kink Disappears at Higher Eg (NREL Cells)


22.9% GaEg = 1.14 eV

Voltage [V]

-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Cur

rent

Den

sity

[mA

/cm

2 ]

-40

-20

0

20

40

darkredwhite

17.8% GaEg = 1.12 eV

Voltage [V]-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Cur

rent

Den

sity

[mA

/cm

2 ]

-40

-20

0

20

40

darkredwhite

time

no GaEg = 1eV

Voltage [V]

-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Cur

rent

Den

sity

[mA

/cm

2]

-40

-20

0

20

40

darkredwhite

time

35.8% GaEg = 1.19 eV

Voltage [V]

-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Cur

rent

Den

sity

[mA

/cm

2 ]

-40

-20

0

20

40

darkredwhite

62.4% GaEg = 1.45 eV

Voltage [V]

-0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2

Cur

rent

Den

sity

[mA

/cm

2 ]

-40

-20

0

20

40

darkredwhite

Eg = 1.11 eV

Eg = 1.40 eVEg = 1.22 eV

Conduction-band offset decreases; changes from spike to cliff







(5) Conclusions.

Effect of Interfacial Recombination on VOC


CdS Window

Vary EC by expanding Eg (simulated)

See Poster P3.9 (Gloeckler)

Lack of spike allows significant interfacial recombination

Effect of EC at constant Eg discussed by several groups

CdS or Alternative Windows?


Vary the window and hence the offset

But, kink can return!


In(OH,S)C1591-23Bonly n-ZnO

Voltage [V]

-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Cu

rre

nt

De

nsi

ty [

mA

/cm

2]

-40

-20

0

20

40

dark0.01 sun0.1 sun1 sun

Voltage [V]

-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Cu

rre

nt

De

nsi

ty [

mA

/cm

2 ]-40

-20

0

20

40

darkwhite lightred light

CSU Photovoltaic Lab

whi

te li

ght

CdSULS345-8-3Standard ZnO

Voltage [V]

-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Cu

rren

t D

ensi

ty [

mA

/cm

2 ]

-40

-20

0

20

40

dark0.01 sun0.1 sun1 sun

Voltage [V]

-0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

Cu

rren

t D

ensi

ty [

mA

/cm

2 ]

-40

-20

0

20

40

darkwhite lightred light

whi

te li

ght

CSU Photovoltaic Lab

CdS Window (IPE) InS(O,OH) Window (IPE)

CIGS Absorber (Eg = 1.15 eV)

“Red” Cut-off 2.4 eV “Red” Cut-off 2.8 eV

See Poster P3.8 (Pudov)

Note: ZnS(O,OH) from AGU yields similar curves

Good Superposition







(5) Conclusions.

Efficiency Picture


Vary the offset independently of Eg

Choosing the Window Material


Big Spike

Small Spike or Cliff

Offset Values from Zhang,Wei, and Zunger, JAP 83, 3192 (1998)

Match absorber and window materials so EC is in optimal range

Conclusions


(1) From a device-physics perspective, the optimal choice of window material for chalcogenide solar cells varies with the band gap of the absorber.

(2) A general problem for CdS windows is low blue response.

(3) A problem for CdS on low-gap absorbers (CIS) is a big spike that impedes current. Mitigated by thin, high-carrier-density, or photoconductive CdS.

(4) A problem for CdS on high-gap absorbers (CdTe or CGS) is the lack of a barrier to inhibit interfacial recombination.

(5) At room temperature, a single window material is optimal over an approximate 300-meV range of absorber band gap.

May 27, 2004 Photovoltaics Laboratory Chalcogenide Solar Cells: Choosing the Window Colorado State University Funding: US National Renewable Energy Laboratory.

Documents

window absorption

bad slide

conductionband offset

smaller current

nm window v oc

cliff problem

selection of window

b conductionband offset