R. Egan, M. Keevers, U. Schubert, T. Young, R. Evans, S. Partlin, M. Wolf, J. Schneider, D. Hogg,

CSG SolarB. Eggleston, M. Green,

UNSW, Sydney

CSG Minimodules by Electron Beam Deposition

UNSW, SydneyF. Falk, A. Gawlik, G. Andrä

Institute of Photonic Technology (IPHT), Jena M. Werner, C. Hagendorf

Fraunhofer - Centre for Silicon Photovoltaics, HalleT. Sontheimer, P. Dogan, S. Gall

Helmholtz Zentrum Berlin für Materialien und Energie

Page 2

The CSG Solar Technology*• Crystalline Silicon On Glass

– Deposited as amorphous – Crystallised through thermal processing

– Rear-side, point contact patterning

• Benefits of Crystalline Silicon on Glass– Conductivity of crystalline silicon, no TCO needed– Durable (no TCO, no wafer breakage, no metal fatigue)– Stable performance– Doesn’t require H in the deposition

*Green et al. Solar Energy 77, (2004) 857

Page 3

• In manufacturing– Over 110,000 modules produced

The CSG Product

• 10.4%* at minimodule level – 492 mV/cell, 29.5 mA/cm2, 72.1% FF– record for single junction, thin film silicon

minimodule – 94cm2, 20 cell minimodule

* Measured FhISE

– Over 110,000 modules produced– >6MW fielded– Average power 85W (6.7%)– Best modules >100W (7.8%)– Factory capacity 13MW, against planned

capacity of 20MW• Power penalty in pattern alignment

(cost 15% capacity)• Throughput penalty on PECVD tool

(cost 25% capacity)

Page 4

Electron Beam Evaporation

� Industrial scales� Deposition rates >1um/min� High throughput� High Vacuum (not UHV)

(1e-7 to 1e-6mbar)

� No toxic gases

substrate

� No toxic gases� No abatement

requirements

? But can we deposit PV grade silicon ?

� End 2007 2% reported *� Initial target for CSG >5%

crucible

boron effusion cell phosphorus

effusion cell

Sie-gun

* Aberle et al. 22nd EUPVSEC, Milan

Page 5

Collaborative Partners IPHT, HZB, UNSW

SiN/Si depositionSPCRTA

Hydrogenation

Glass texture

Dev

ice

Si

Gla

ss

SiN/Si depositionSPCRTA

Hydrogenation

Glass texture

Nov 2007IPHT, Jena

n+/p-/p+

April 2008HZB, Berlin

p-/p+

PECVD SiN

Glass substrate (texture)

PECVD SiN/n+

Nov 2008UNSW, Sydney

n+/p-/p+

PECVD SiN/(n+)

Cell isolationPattern

Dev

ice

Si

Gla

ss

• 10cm by 10 cm minimodules => 35 cm2, 12 cells• Characterisation with Fraunhofer CSP,

Halle (SiThinSolar)

n+/p-/p+p-/p+(SPC)

SPCRTA

Hydrogenation

n+/p-/p+(SPC)

PatternMetallisation

Measure

Page 6

Efficiency Progress - Planar

� Met initial target 5%

� Matched PECVD equivalent at 1.3um

4

5

6

7

8E

ffici

ency

(%

)

CSG pecvd

IPHT e-beam

6.1%

* Showing only the best result on each attempt

*

0

1

2

3

4

Nov

-07

Feb

-08

May

-08

Aug

-08

Nov

-08

Feb

-09

May

-09

Aug

-09

Date Measured

Effi

cien

cy (

%)

IPHT e-beam

Page 7

Efficiency Progress - Planar

� Met initial target 5%

� Matched PECVD equivalent at 1.3um from two different sources

4

5

6

7

8E

ffici

ency

(%

) CSG pecvd

IPHT e-beam

6.1% 6.1%

0

1

2

3

4

Nov

-07

Feb

-08

May

-08

Aug

-08

Nov

-08

Feb

-09

May

-09

Aug

-09

Date Measured

Effi

cien

cy (

%)

HZB e-beam

UNSW e-beam

* Showing only the best result on each attempt

*

Page 8

nid

5

5.5

6

6.5

7

1E+15 1E+16 1E+17

Eff

icie

ncy

(%)

300 nm/min600 nm/min

22 500

• Optimal base doping; 5e15 to 1.5e16 cm-3*

• 6.7 % efficiency at deposition rates up to 600nm/min, 20-30 times faster than PECVD

Planar Optimisation : Base Doping and Thickness

1.5um base

1E+15 1E+16 1E+17

Dopant Concentration (cm-3)

18

18.5

19

19.5

20

20.5

21

21.5

22

1.6 1.8 2 2.2

Base Thickness (um)

Jsc

(mA

/cm

2)

400

420

440

460

480

500

Voc

(mV

)

Jsc

Voc

• Optimal thickness >2.0um*

*See also T. Sontheimer et al, (HZB) EUPVSEC 24, Hamburg

1.5e16 cm-3 base300nm/min

Page 9

Planar Optimisation: Summary 6.7%

Equivalent performance in side by side comparisons• From different tools (HZB, IPHT)

• At 300 and 600 nm/min

• Sputtered and PECVD SiN

• Stationary or with rotation

• Continuous or with breaks in deposition

5

10

15

20

25

Cu

rre

nt

De

ns

ity

(m

A/c

m2

)

P E C V D T extured

IP HT P lanar

Planarall equivalent

Texture*

AND

we know we would do better

with light trapping.

* M Keevers etal EUPVSEC 22, Milan

6.73

0

1

2

3

4

5

6

7

8

1/11

/200

7

1/02

/200

8

1/05

/200

8

1/08

/200

8

1/11

/200

8

1/02

/200

9

1/05

/200

9

1/08

/200

9

Date Measured

Eff

icie

ncy

(%)

Planar PECVDPlanar (IPHT)Planar (HZB)Planar (UNSW)Beads Abrade

0

5

0 1 2 3 4 5 6

Voltag e (V)

Cu

rre

nt

De

ns

ity

(m

A/c

m2

)

IP HT P lanar

HZ B P lanar

P E C V D P lanar

Page 10

Standard CSG Bead Texture

PECVD silicon : conformal

glass substrate

silicon

bead

Image courtesy of MFA, Budapest. www.high-ef.eu

E-beam silicon : non conformal

glass substrate

silicon

bead

Page 11

Characterisation – Bead Texture

EBIC of p+ contact point

10 um

LIT of completed minimoduleoverlayed on optical micrograph

SEM of p+ contact point

480 um

10 um

2 um See also M. Werner et al, EUPVSEC 24, HamburgFraunhofer, Center für Silizium-Photovoltaik CSP

beads

planar

Page 12

Characterisation – Bead Texture

SiOx

Si

SiO2

TEM cross section of typical bead relatedStructures in post-RTA e-beam silicon

EDX analysis of species in

silicon

See also M. Werner et al, EUPVSEC 24, HamburgFraunhofer, Center für Silizium-Photovoltaik CSP

Page 13

• An oxygen rich Si microstructure forms in crystallised silicon at steep edges (bead edges, abrade ridges) when the normal to the substrate αis > 35o to the incoming silicon.

Mechanism for Shunting in Bead and Abrade Textures

α α

SiSi

• Process developments in e-beam deposition• Smoother glass textures• Silicon surface textures

Solutions

Bead texture

Page 14

2

3

4

5

6

7

8

Effi

cien

cy (

%)

Low Profile Glass Texture

• First attempt: 6.7%• 7% boost in current over planar control• Negligible shunting• No oxygen rich silicon

� Create a low profile (smoother) glass texture

6.7%

0

1

Nov

-07

Feb

-08

May

-08

Aug

-08

Nov

-08

Feb

-09

May

-09

Aug

-09

Date Measured

Beads

Abrade

Low Profile Glass Etch

AFMz +/-2 um

30 um x 30 um

• No oxygen rich silicon

Page 15

2

3

4

5

6

7

8

Effi

cien

cy (

%)

Silicon Surface Texture

� Deposit extra silicon on planar substrateand etch back to create texture*

• 17% boost in current over planar control• 7.4% new benchmark for e-beam Si

7.4%

* Edited TEM to illustrate, not a real sample

0

1

Nov

-07

Feb

-08

May

-08

Aug

-08

Nov

-08

Feb

-09

May

-09

Aug

-09

Date Measured

Beads

Abrade

Si Surface Texture

AFMz +/-1 um

20 um x 20um

• 7.4% new benchmark for e-beam Si

Page 16

2

3

4

5

6

7

8

Effi

cien

cy (

%)

Combined Glass and Silicon Texture

Low Profile Glass Texture

� Combined the two

AFMz +/-2 um

7.4%

7.1%

• 24% boost in current over planar control• 7.1% in first attempt

0

1

Nov

-07

Feb

-08

May

-08

Aug

-08

Nov

-08

Feb

-09

May

-09

Aug

-09

Date Measured

Beads

Abrade

Si Surface Texture

Low Profile Glass Etch

Low Profile Glass Etch + Si Surface Texture

Silicon Surface Texture

AFMz +/-1 um

20 um x 20um

z +/-2 um30 um x 30 um

Page 17

Summary: E-beam deposition for CSG Minimodules

• Planar silicon by e-beam – Minimodule efficiency 6.7%, currents of >20mA/cm2– Equivalent to PECVD, but at 600nm/min (20-30 times faster)– Robust: Equivalent results from two different sources

• Textured silicon by e-beam– Shunt cause identified – Alternatives developed ;

• Low profile glass textures • Silicon surface textures

– Minimodule efficiency 7.4%, currents >23mA/cm2

Page 18

Thank You• Colleagues at CSG Solar

• ARC Centre for Photovoltaics, UNSW

• Institute of Photonic Technology (IPHT), Jena • Fraunhofer - Centre für Silizium-Photovoltaik CSP , Halle

• Helmholtz Zentrum Berlin for Materialien und Energie

3.14 MW CSG + 3.3 MW mc-Si

Solar Valley, Germany

2. Photovoltaik-SymposiumWolfen, 05.Nov.2009

Jens SchneiderSenior Process Dev. EngineerCSG Solar AG