1

Winning the Global Race for Solar Silicon

David Lynch

Professor of Materials Science & Engineering, University of Arizona

dclynch@email.arizona.edu&

Chief Technical Officer, Solar Technology Research Corporation, Tucson, AZ

2

Topics

• Market Conditions for Photovoltaics

• New Routes for Producing Low Cost Solar Silicon

• The Solar Technology Research Corporation’s Approach to Refining Silicon

3C & E News, Feb. 2003

The Future for Renewable Energy Generation

4

ARCO Magazine 1989 - cells produced by NAPS

Photovoltaics are not only modular, they are also mobile.

5

Although the role of silicon in photovoltaics is expected to decline, it is still expected to be a growth industry.

61992 1996 2000 2004 2008Year

0

2000

4000

6000

8000

Inst

alle

d P

V C

apac

ity

(MW

)

World Installed Capacityin IEA Reporting Countries

1992 1996 2000 2004 20080

20

40

60

% A

nn

ual

Gro

wth

Rat

eo

f In

stal

led

Cap

acit

y

30%Predicted Installed PV Capacity

2008 10.4 GW

2009 13.1 GW

PV Capacity 2005

Germany 1.42 GW

Japan 1.41 GW

USA 0.48 GW

7Impurity content (ppmw)

metallurgical

"high purity"

solar

electronic

2N 5N4N3N 6N 7N 8N 9N

8

Photovoltaics: The Production Path

Cryst Wafers Etch.TCS Dist CVDSiO2 Red. Silgrain

Metallurgical Si - $ 3 per kg Electronic Si - $ 40-60 per kg

Solar CellsWafersRemelt/Cryst

Missing Link

Solar GradeAim $ 10-14 per kg

Scrap Si - $ 25 per kg

Metallurgical SiSi Purification

SiemensIngot/Wafers Cells/Panels

$ 2 $ 25 $ 40 $ 100

9

Is there a pot of gold at the end of the race for solar silicon?

1/1/04 1/1/05 1/1/06 1/1/07 1/1/08

Date (mm/dd/yy)

0

10

20

30

40

Sto

ck V

alu

e (C

$ p

er s

har

e)

Stock Value of Timminco

Prior to March 2007 Timmincoproduced metallurgical-grade silicon, and magnesium.

March 2007, Timminco announcesit has low cost process for producingsolar-grade silicon, and has signed a purchase and sales agreement.

10

Only if you get it right!

1/1/04 1/1/05 1/1/06 1/1/07 1/1/08 1/1/09

Date (mm/dd/yy)

0

10

20

30

40

Sto

ck V

alu

e (C

$ p

er s

har

e)

Stock Value of TimmincoPrior to March 2007 Timmincoproduced metallurgical-grade silicon, and magnesium.

March 2007, Timminco announcesit has low cost process for producingsolar-grade silicon, and has signed a purchase and sales agreement.

11

Tech. Company Tech. Company

Integrated Reduction of Silica to s-Si

Solsilk/FESILSolarvalue ???Elkem Solar

Siemens TCS

Silane

Hemlock, Wacker, Tokuyama, MEMC, & numerous new entrantsREC Silicon

Up-grade m-Si

Elkem Solar,Dow Corning,Solarvalue, JFE, Nippon Steel. Timminco, STRC & others

Fluid Bed TCSSilane

Hemlock & WackerREC Silicon

Some of the Players in the Race

12

General Approach of Metallurgical Processes for Producing Solar Silicon

Silicon Submerged Arc Furnace

RefiningAcid Treatment

(Silgrain)

Directional Solidification

C SiO2

m-Si

hp-Si

hp-Si

hp-SiB & P

reduced-Si

solar - Si

The primary issues is removal of B and P before directional solidification.

Solsilk/FESILSolarValueElkem SolarDow CorningNippon SteelJFETimmincoSTRC

13

14

W Mo Zr B V Co Ni Fe Sc Cr Sn Cu Al Mn Ba Pb Sb Bi Ca Sr Mg Zn Na Cd K Se As S Hg P

Impurity Elements

0

2000

4000

6000B

oilin

g P

oin

t Tem

p. (

K)

Crater TemperatureTop Bed Temperature

Impurity Element Expected to Report to Product

ExpectedVariation inDistribution of Trace Element

Filter Temperature

Impurity Elements Expected to Report to Silica Fume

Off-Gas

W Mo Zr B V Co Ni Fe Sc Cr Sn Cu Al Mn Ba Pb Sb Bi Ca Sr Mg Zn Na Cd K Se As S Hg P

Impurity Elements

0

25

50

75

100

Per

cen

t

product

silica fume

filtered off-gas

15

2000 2100 2200 2300 2400 2500T (K)

0.1

1.0

10.0

p SiO

(bar

)

SiC + 2SiO2= 3SiO(g) + CO(g)

SiO(g) + SiC = 2Si + CO(g)

P T=

1 ba

rP T

= 10

bar

Si + SiO 2

= 2SiO(g)

ke

d

f

g

m n

pSiO: pCO 3:1h q

T = 2273 K

T = 2188 K

l

i

o

p

j

T = 2084 K

≅

≅≅

2:11:1

Division BetweenOuter Reaction Zone

Inner Reaction Zone

Ratios for PT= 1 bar

electrode

cavity

SiC Siarc

bed(silica& coke/coal)

Outer Reaction Zone

SiO(g) + 2C = SiC + CO(g)

Inner Reaction Zone

SiO2 +SiC = Si+SiO(g) + CO(g)

2CO(g) 2C SiO2

Si

SiO(g)

& CO(g)SiC

16

Oxide Slag

Timminco

Elkem Solar

Solarvalue

Oxi-Nitride Slag

STRC

17

1.2 1.6 2.0 2.4 2.8CaO/SiO2 molar ratio

0.0

0.2

0.4

0.6

0.8

1.0

Xi

or

L B

XSiO2

XO2-

LB

(K8' )1/4assumed = 6

orthosilicate composition

Refining with an Oxide Slag, B Removal

( ) )(34)(4)(6)(3 33

22 lSislagBOlBslagOslagSiO +=++ −−

181.2 1.6 2.0 2.4 2.8

CaO/SiO2 molar ratio

0.0

0.2

0.4

0.6

0.8

1.0

Xi

or

L B

XSiO2

XO2-

LB

(K8' )1/4assumed = 6

orthosilicate composition

Refining with an Oxide Slag, B Removal

ProcessSlag: Si

Mass Ratio

Batch 17

Counter Current

1.4

0 1 2 3 4 5CaO/SiO2 molar ratio

0

0.4

0.8

1.2

1.6

2

Dis

trib

uti

on

Co

effi

cien

t fo

r B

orthosilicate

acid base

Suzuki & Sano10th E.C. PV SolarEnergy Conference,April 1991

3SiO2 + 6O2- +4B = 4BO33- + 3Si

19

-19 -18 -17 -16 -15Log PO2

-2

-1

0

1

2

3

4

Lo

g (

%P

O43-

)/(P

P2)1/

2 o

r L

og

(%

P3-)/

(PP

2)1/2

CaO(sat.) - CaF2 Melts

T = 1773 K

maximum pO2

Si(l) - SiO2(l)equilibrium

aSiO2 = 1

aSiO2 = 0.1

aSiO2 = 0.01

(P 3-)

(PO 4

3- )

-20

232

4

3)()(

2

3OPOP +=+ −−

)()(4

5)(

2

3 342

2 −− =++ POgOOP

After Tabuchi and Sano 1984

Refining with an Oxi - Fluoride Slag, B Removal

Phosphide

Phosphate

20

STRC Approach

Technical Approach• Remove B and P by

replacing those elements with other impurities that can be easily removed down stream.

• Oxidative refining

• Silgrain

• Directional Solidification

• Patent Application, US 2007/0245854 A1

Business Approach• Modular Units at 600 metric

tons per year

• PV producer financed for guarantied price

• Estimated production cost: $12-$14 per kg (pre directional solidification)

• With projected process improvement, cost estimated at $8 to $10 per kg.

21

STRC / UA Research Laboratory

22

0 0.01 0.02 0.03 0.04

γBN

0

0.2

0.4

0.6

0.8

mas

s o

xi-n

itri

de

mel

t to

mas

s o

f S

i ref

ined

target

aSi3N4 = 1

a Si 3N 4

= 0

.01

a Si 3N4 = 0.1

(Si3N4) + 4B = 4(BN) + 3Si(l)

T = 1700 K

30 ppma to 0.1 ppma B

aSi = 1

minimum anticipated value for γBN

STRC Technical Approach to Refining for B

Liquid

Si3N

4 AlN

Al2O

3SiO2

Liq. + βss

Si2N

2O

βss

L + Si2N

2O + βss

% equivalent Al

% e

quiv

alen

t O

A

B

C

D

23

0 0.05 0.1 0.15 0.2 0.25mass slag / mass Si refined

0

5

10

15

20

25

pp

mw

B in

Si

as received m-Si

Starter Slag

27.7% Al2O3

8.5% MgO 17% SiO2

16% CaO 15.8% CaF2

15% Si3N4

targ

et

bench mark 1 ppmw

0 0.01 0.02 0.03 0.04

γBN

0

0.2

0.4

0.6

0.8

mas

s o

xi-n

itri

de

mel

t to

mas

s o

f S

i ref

ined

target

aSi3N4 = 1

a Si 3N 4

= 0

.01

a Si 3N4 = 0.1

(Si3N4) + 4B = 4(BN) + 3Si(l)

T = 1700 K

30 ppma to 0.1 ppma B

aSi = 1

minimum anticipated value for γBN

STRC Technical Approach to Refining for B

Liquid

Si3N4 AlN

Al2O

3SiO2

Liq. + βss

Si2N

2O

βss

L + Si2N

2O + βss

% equivalent Al

% e

quiv

alen

t O

A

B

C

D

24

0 0.004 0.008 0.012 0.016

γAlP

0

0.2

0.4

0.6

mas

s o

xi-n

itri

de

mel

t to

mas

s o

f S

i ref

ined

a Al 2

O3 /

a SiO

2 =

3

a Al2O3 / aSiO2

= 60

a Al 2O 3 / a

SiO 2 =

30

a Al 2

O3 /

a SiO

2 =

6

(Al2O3) + 2P + 1.5Si(l) = 2(AlP) + 1.5(SiO2)

T = 1700 K

30 ppma to 0.1 ppma P

aSi = 1

target

minimum anticipated value for γAlP

Liquid

Si3N

4 AlN

Al2O

3SiO2

Liq. + βss

Si2N

2O

βss

L + Si2N

2O + βss

% equivalent Al

% e

quiv

alen

t O

A

B

C

D

STRC Technical Approach to Refining for P

25

Conclusions

• For Si photovoltaics to compete in the market there must be cost compression all along the cost chain.

• Opportunity exists to significantly reduce the cost of silicon for photovoltaics through metallurgical refining practice.

• Si Photovoltaics are expected to remain a growth area, even though market share will decline.