Solar cells made of crystalline silicon: successful mainstream thanks ... · Solar cells made of crystalline silicon: successful mainstream thanks to intense research and development

Pietro P. Altermatt Trina Solar Changzhou, China

Solar cells made of crystalline silicon:

successful mainstream thanks to

intense research and development

Presentation at the Oxford Energy Network, Oxford University, GB 7. Nov. 2017

2 Altermatt, Trina Solar

A common solar module

module

cell

4 bus bars

front metal fingers

contains 60 cells

15.4 15.4 cm2

delivers about 300 Watts


Photovoltaic (PV) principle

Two-level system



1 Photo generation

Two-level system



1

2

Photo generation

Transport

Two-level system



1

2 3

4

Photo generation

Transport Contact External circuit

Two-level system



1

2 3

4

5

Photo generation

Work


Two-level system



1

2 3

4

5

6 7

Photo generation

Work

Second contact


Two-level system



1

2 3

4

5

6 7

Photo generation

Work

Second contact


Two-level system



Two-level system

Metals not suitable (recombination)

Most solar cells are made of crystalline silicon (semiconductor)



Two-level system




1 Photo generation


Most insulators not suitable (insufficent absorption)



1 Photo generation



Bands



1

2

Photo generation

Transport

Two-level system



Thinner cells don’t need good transport

Silicon cell:

170 m thick

Si cell



1

2 3

4

Photo generation


Two-level system



Thin cells don’t need transport

Different carrier inside than outside possible



1

2 3

4

5

Photo generation

Work


Two-level system

Power = current voltage, (IV)





Too small gap gives too small voltage



1

2 3

4

5

Photo generation

Work


Two-level system









1 2

3 4

5

Photo generation

Work


Two-level system









1

2 3

4

5

6 7

Photo generation

Work

Second contact

Transport Contact

Two-level system

Homogeneous system: semiconductor is required, Si is a good choice


Overview

1. Mainstream Si solar cell production

2. Improvements from research development production

3. Conditions for research ideas or new inventions to enter mainstream


Overview





From quartz sand to silicon?

1. High-purity quartz sand needed.

2. Production of metallurgical Si mostly for steel/Al/silicone industry

SiO2 + C → SiO(gas) + CO

SiO + CO → Si + CO2

Electrically heated furnace

about 1900C

with carbon arc

Si melts

at 1414C


Silicon purified by distillation


2. Production of metallurgical Si mostly for steel/Al industry

3. Small part of it is being purified:

distillation via SiHCl3 (trichlorosilane, TCS)

Si + 3 HCl → HCl3Si + H2


Decomposition to solar grade silicon


2. Production of metallurgical Si mostly for steel/Al industry

3. Distillation via SiHCl3

4. Pyrolytic decomposition (Siemens process) or fluidized-bed deposition

5. Solar grade silicon

REC, Norway


Monocrystalline silicon

Czochralski pulling of monocrystalline ingots

(1 mm/minute) Courtesy: SolarWorld Courtesy: Fraunhofer CSP


Multicrystalline silicon

Cast silicon via

direcitonal

solidification

15 mm/h

cut into bricks

15.4 cm


Sawing into wafers

H. Lauvray et al,

New wire silicon slicing technology for solar cells,

3rd EU PV Conf. 1980, p. 603

Si ingot

wire

free abrasives, “slurry”

kerf loss

fixed abrasives

“diamond wire”

Multi-wire sawing


PERC

8 main fabrication steps


Saw damage etching, texturing

10 μm

A. Hauser et al, 19th EU PV Conf. 2004, p. 1094

30 µm

Courtesy: ISFH

KOH etch acidic etch

University of Constance


Texturing

“old” days without acidic texturing

multi

mono

nowadays with acidic texturing

University of Constance


Phosphorus diffusion

20 – 30 min, 750C

Courtesy: Centrotherm

p-type wafer

J. Mandelkorn et al., J El Chem Soc 109, 313 (1962)

n-type

(conductive

for electrons)

cross-section

of cell


Rear side etching

Courtesy: Schmid

Emitter


Emitter etch back

H. Haverkamp et al., 33rd IEEE PV Conf (2008), no 112

University of

Constance

Emitter


Front dielectric layer deposition

Palsma-enhanced chemical vapor deposition (PE-CVD)

SiH4 + NH3 + N2 SiNx + H2

Silane Silicon nitride

SiNx


Front passivation and ARC

ISFH,

University of Hanover

90’s

Photo generation Passivation

SiNx

Si surface

similar to metal


Front passivation and ARC

Courtesy: Meyer Burger

SiNx


Rear passivation with Al2O3

Al2O3

SiNx

stack

University of Eindhoven 2006

University of Erlangen

B. Hoex et al, Appl Phys Lett 89, 042112 (2006)


Rear passivation with Al2O3

Al2O3

SiNx

stack

University of Eindhoven 2006

University of Erlangen

Courtesy: Meyer Burger


Rear laser opening

R. Preu et al., 16th EU PV Conf. (2000) 1181

ISE Freiburg

Courtesy: ISE


Screen-printing of metal contacts

Paste

Screen

Squeegee

cell

Front silver fingers

Rear Al with Ag soldering pads

E.L. Ralph, 11th IEEE PV Specialists Conf. (1975), p. 315

Courtesy of www.gwent.org

Spectrolab (startup company)


Screen-printing of metal contacts

< 100 mg silver/cell


Firing-through ARC

IMEC, 1997

University of Constance, 2000 -> mass production

Conveyor belt furnace

Firing

2-3 seconds

800C


Firing: Al back surface field

Firing

2-3 seconds

800C

electron conduction

hole conduction

emitter

back surface field (BSF)


PERC

8 main fabrication steps


Module and systems


Overview





Experience curve


Experience curve


Silicon shortage 2003 – 2012

30

20

10

0

19

95

20

00

20

05

20

10

kt

by-products of electronic industry

solar grade silicon

demand

Andrews and Clarson, Silicon 7, 303 (2015)

R&D on a multitude of more direct

purification/crystallization methods

M. Heuer, Semic & Semimet 89, 77 (2013)


Only established methods survived in oversupply

R. Fu etal., IEEE J PV 5, 515 (2015)

Today:

electronic industry: 30 000 t

solar industry: 460 000 t

US$ 15/kg (spot price)

Siemens processes

Fluidized bed

Reality in 2013,

in oversupply


Established methods get optimized

Yadav et al., Ren & Sust En Rev 78, 1288 (2017)

Siemens process

(pyrolytic decompostion of e.g. SiHCl3)

Hard to compete against

optimized standard


Over- and under-supply

Fast growing industries go through

cycles of under and over-supply

In times of over-supply,

many ideas die

Glo

bal annual P

V p

roduction

[GW

]

Year


Experience curve: different interpretations


Production shifted to China

Mark

et

share

Year

China

US

Japan

Germany Taiwan

Malaysia Thailand

Vietnam

2017


Production shifted to China

Mark

et

share

Year

China Japan

Taiwan

Malaysia Thailand

Vietnam

2017

Antidumping duties: 27.3% – 64.9%

Added anti-subsidy duties: 11.5%

Combined duties should not be higher than

76.4% in future

EU Commission

Sept. 2017:

courtesy: Wikipedia

US

Germany


Innovation or economy of scale?

C. Zheng. D.M.Kammen, Energy Policy 67, 159 (2014)


Recent innovations

• Fluidized-bed deposition of Si (REC, Norway)

• Acidic texturization of multi Si (University Constance)

• Rear Al2O3 passivation (University of Eindhoven)

• Rear etching (mainly German tool companies)

• Rear laser ablation (ISE Freiburg)

• High-performance multi (National Taiwan University)

• Diamond wire sawing (companies in Japan and Taiwan)

• Texturing of diamond-wire sawn mutli wafers (start-ups in China)

• Improvements of screen-printing pastes (global paste companies)

Most recently:

Already mentioned:


Recent innovations

• Fluidized-bed deposition of Si (REC, Norway)

• Acidic texturization of multi Si (University Constance)

• Rear Al2O3 passivation (University of Eindhoven)

• Rear etching (mainly German tool companies)

• Rear laser ablation (ISE Freiburg)

• High-performance multi (National Taiwan University)

• Diamond wire sawing (companies in Japan and Taiwan)

• Texturing of diamond-wire sawn mutli wafers (start-ups in China)

• Improvements of screen-printing pastes (global paste companies)

Most recently:

Already mentioned:

European grant bodies often do not want

Chinese companies being involved

in a grant application

Buying and installing Chinese solar modules in the EU:

70% of revenue generated within EU, only 30% in China.


Contact difficult to understand and manipulate

metal

conduction

tunneling

through

glass layer

Ag colloid-assisted

tunneling

H. Mäckel, P.P. Altermatt, IEEE J. of PV 5, 1034 (2015)

Fastest improvements

during over-supply

Silicon Silicon


Innovation or economy of scale?

Yearly staff turnover in China: 60% in production, 40% in R&D (pollination)


Factory doubles every 2½ years


Factory doubles every 2½ years


Overview





Fabrication requirements

• Annual production of modules surpasses 100 GW.

• Silver used per cell is below 100 mg; still, PV industry is main silver consumer.

• Cells are produced in a one-second sequence, so each fabrication step must

not take longer than one second (e.g. P diffusion for 20 min on 1200 wafers)

• Warranty for modules given for 20 – 30 years.

Until then, the module power must decrease only by a very few percent.

• The energy pay-back time is about one year (including Si and Ag production).

• A Si wafer costs 40 pence (spot-price); fabrication costs 25 pence per cell;

60 cells in module makes 40 pounds; module fabrication cost 25 pounds.

Costs are falling 18% per year

Hence, improvement or alternatives must:

• grow to large output,

• be stable over 20 – 30 years,

• must not consume scarce materials or lots of energy,

• must be fabricated fast,

• and be cheap and cheaper.


Suggestions from previous slides

Entering mainstream solar cell production:

• Good collaboration between university and tool manufacturers

• Don’t keep your invention secret, you will stay alone

• Don’t grant an exclusive license of your patent, they will stay alone

• Stimulate a whole community to develop your invention to production


Experience curve for CdTe and CIGS

Y. Chen, Feng, Verlinden, 6th World Conf PV, 2014, paper 9WeO.5.5

CIGS

too flat


Where is mainstream heading to?

Colors:

various existing technologies

continuously being

developed further

B. Min, H. Wagner, M. Müller, H. Neuhaus, R. Brendel, P.P. Altermatt, 31st EU PV Conf. 2015 p. 473.

Device

modeling

PERC cells


Cells are getting more efficient

PERC cell efficiency

is expected to increase

by about 0.4%abs each year

Y. Chen, Feng, Verlinden, 6th World Conf PV, 2014, paper 9WeO.5.5


Better and cheaper mainstream cells

In about 8 years* PERC cells may come close to 23% – 24% cell efficiency,

but require every 2 years a new technological step

while module cost may go down to about half.

* if no major disruption or breakthroughs

P.P. Altermatt et al, 44th IEEE PV Conf. Wasington, July 2017

Currently: module fabrication cost 65 pounds


New PV material into mainstream?

Cell efficiency can be improved further

e.g. by adding an other layer of PV material

on top of PERC cells (hetero-emitter),

and/or by introducing passivated contacts

electron conduction

hole conduction

hetero emitter

(other than Si)

passivated contact

A. Louwen, Sol En Mat & Sol Cells 147,295 (2016)


Current research topics

Current research topics in silicon PV (cells, not modules):

• Passivated contacts

• n-type Si material

• Replacement of Al2O3 passivation

• Narrower front metal fingers

• Hetero-emitter (organic material?)

• Tandem cells


Categories of PV

1st generation

2nd generation

3rd generation

Standard silicon solar cells (e.g. PERC)

Thin-film cells (e.g. CdTe, other semiconductors)

Quantum-engineered material

(e.g. quantum dots, hetero-structures, perovskites)

M. Green, Prog in PV 9, 123 (2001)


Categories of PV

M. Green, Prog in PV 9, 123 (2001)

May cells, made of

other material than Si,

replace Si cells?


30% or 40% efficient cells?

Calculations: ISE Freiburg

If cell efficiency is lifted to

30% (or 40%) the cell

must cost only maximally

55% (70%) more to be

competitive.

Crucial consequences for basic research

to have a prospect

for being applied one day


30% or 40% efficient cells?

?


Mainstream PV is important for climate

2C

3C

3.5C

>4C

Paris:

increased

ambition

Paris:

continued

ambition

Low policy

No policy

Research & development which benefits mainstream silicon PV

has a direct and significant impact on climate

global warming:

[email protected]

www.trinasolar.com


Forecast in 2016


Forecast in 2016


Silver forecast

C. Muschelknautz, 4th Metallization Workshop, Constance, 2013


Silver forecast



Silver forecast



Early, forgotten work in Manchester

A.D. Haigh, 12th IEEE PV Conf (1976), 360

Ferranti Ltd, Manchester


Where is R&D done?

Tool manufacturers vs. PV companies

GCL

LONGi


Passivated contacts

Sentaurus simulation

Highly-doped silicon behind oxide enables:

minority tunneling hindered

(large ox barrier) and/or depletion

majority band edge lower behind

oxide than at the front of the oxide

Conditions:

minority carrier depletion at the Si/SiO2

interface to reduce recombination

(passivation)

majority carrier accumulation

to enhance tunneling

passivated majority contact

directly on lowly doped n-type base

SiO2

pc-Si c-Si

n-type

1 Ωcm

n-type

highly-

doped

MPP (600 mV)


Materials for passivated contacts

defect band

MoOx

Metal

Bullock et al, Appl Phys Lett 105, 232109 (2014)

Oxide layer can be made thicker

if there is a defect band


Solar cell materials for global mass production

CdTe, CIGS

1980 – 1998: 2008 Navigant Consulting report; 1999-2009: March issues of Photon International; 2010-2015: IHS 2015 Q4 market

report.


Where is mainstream heading to?

Colors:

various existing technologies

continuously being

developed further

B. Min, H. Wagner, M. Müller, H. Neuhaus, R. Brendel, P.P. Altermatt, 31st EU PV Conf. 2015 p. 473.

Device

modeling

PERC cells

Introduce these technologies

at a suitable

development stage

ineffective

effective


Transition from standard to PERC cell

standard cell PERC cell

Global production capacity of Si cells (GW)

PERC

standard

Expected from orders at tool manufactures Forecast


Price decay for roadmap (linear plot over time)

Documents

Solar cells made of crystalline silicon: successful mainstream thanks ... · Solar cells made of crystalline silicon: successful mainstream thanks to intense research and development