Dr. Joan Haysom Solar Project Manager SUNLAB Solar Research Group - - PowerPoint PPT Presentation

• Dr. Joan Haysom

Solar Project Manager SUNLAB Solar Research Group School of Electrical Eng. and Comp. Science University of Ottawa 1

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 2

• 1. Semiconductor Materials
• 2. Single junction and 3. Multi-

junction solar cells

• 4. Epitaxial Growth
• 5. Market Applications of III/Vs
• 6. Advanced III/V cells

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 3

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 4

VI III/V II/VI

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 5

Si Si Si Si Si Si

Covalent bonding between group IV (e.g. Si) atoms or group

III+V atoms (e.g. Ga + As) results in octets of shared electrons and a filled valence band

Figures adapted from PVDCROM

h+ e- Eg fermi level

Ga Ga Ga As As As

bonds can be broken without too much energy, promoting a

free electron into the empty conduction energy band and leaving a free hole in the valence band

Temperature, dopants, photon absorption all ↑

↑ ↑ ↑ # free carriers

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 6

Si Si Si Si Si Si

Dopants – provide an extra electron or hole carrier, shifts fermi level n-dopants

Si (IV) replacing Ga (III) resulting in extra valence electron Sn (IV) also used

p-dopants

C (IV) replacing As (IV) resulting in missing valence electron Zn or Be (II) replacing Ga(III) also used

Figures adapted from PVDCROM

e- Eg fermi level

Ga Si Ga As As As

n-type doping shown

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 7

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 8

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 9

Pout = I x V

Vmax is set by bandgap of the material (Eg/q) I is set by # photons absorbed

(can only absorb above the bandgap)

Conduction Band (empty excited state) Valence Band (full

• f electrons)

Eg

photon Energy Band Diagram of a Semiconductor thermalization thermalization

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 10

Jsc photons

Energy Band Diagram at equilibrium

Jsc Energy Distance from bottom

Device requires a p-n junction for separating electron from hole, and thus extracting current

http://en.wikipedia.org/wiki/File:Pn_junction_equilibrium.svg

e-

h+

e-

h+

Eip Ein Ef

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 11 Source: Renewable and Efficient Electric Power Systems, Gilbert Masters, Wiley Interscience, 2004.

Decreasing Eg ----->

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 12

Spectrum losses of 49%

20% below bandgap of Si above-bandgap photons lose on

average 30% of their energy as heat

Recombination

Need high quality crystals for long

carrier diffusion lengths

Lower hole mobility than electron

mobility in Si, 10% ↓ ↓ ↓ ↓ from theoretical

Blackbody Radiation

Warm cell re-radiates ~ 7% of total

incoming

http://en.wikipedia.org/wiki/Shockley-Queisser_limit

Max theoretical efficiency

• f Si is 33%

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 13

• The maximum efficiency single

junction cell is with a material with a band gap of ~1.4 eV (=GaAs)

• Theory 33.7% efficient
• Record 28% achieved by Alta.
• Production 23%, thin cell with

substrate reuse 1.4eV GaAs 1.1eV Si

• Si is cheap and abundant with a

band gap in a good range (1.1 eV), and with efficiencies of:

• Theory 30%
• Record 24%
• Commercial ~ 15-20%

V ∝ Eg I ∝ 1/Eg

• Additionally, GaAs has the

following advantages over Si:

• Direct bandgap
• high carrier mobilities
• Radiation hardness
• Lower temperature coefficient
• f bandgap shift

Included in η

http://en.wikipedia.org/wiki/Shockley-Queisser_limit

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 14

In energy (E) versus crystal momentum (k) space, the lowest point

in the conduction band does not align with the top point of the valence band

To conserve momentum, a lattice vibration (phonon) is required,

resulting it a three particle process, which is less efficient and blurs the turn-on around the bandgap

Eg=1.1 eV Eg=1.4 eV

http://physicsarchives.com/index.php/courses/967 & J. Nelson Physics of Solar Cells, Imperial College Press, 2003

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 15

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 16 Source: Renewable and Efficient Electric Power Systems, Gilbert Masters, Wiley Interscience, 2004.

Decreasing Eg ----->

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 17

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 280 780 1280 1780 2280 2780

Radiance (a.u.) Wavelength (nm)

• Bandgaps of each subcell’s III-V semiconductors can be selected

for improved absorption achieving a wider spectral response for improved η

AM1.5D, 900 W/m2

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 18

Three subcells in series => voltages add, which is great! Each subcell generates different current determined by

its bandgap and the solar spectrum, BUT current through series connected stack is limited by weakest performer => design challenge

1 2 3

I I I = =

1 2 3 tot

V V V V = + +

(I1, V1) (I2, V2) (I3, V3)

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 19

Because the length of the depletion region is narrow and very high doping is used, carriers can tunnel. Because the length of the depletion region is narrow and very high doping is used, carriers can tunnel. n+ n p p+ n+ n p p+ n n p Band Structure including TJ Band Structure including TJ

Jsc

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 20

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 21

Epitaxy: deposition of a crystalline overlayer on a crystalline substrate. The overlayer is called an epitaxial film or epitaxial layer. comes from the Greek roots epi, meaning "above", and taxis, meaning "in ordered manner“ Substrate acts as a seed crystal – it locks down the lattice constant and lattice structure Homoepitaxy – same alloy as substrate Heteroepitaxy – different alloy grown above

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 22

Substrates are polished to smooth finish, and must be clean and

• xide free – “epi

ready”. Yet, some impurites at initial interface => buffer layer Typical Ge substrates used are miscut ~6o => Provide sufficient # steps for Step-flow growth

• Layer by layer growth on perfectly

smooth surface might be preferred, but small imperfections in crystal lead to preferred locations for nucleation.

• Which growth mode depends on
• Substrate Temperature
• Substrate surface morphology
• Saturation of incoming elements

(Exaggerated step heights)

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 23

Zinc blende crystal structure Ge and GaAs have same lattice constant, Ge substrates are much cheaper and provide low λ absorption

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 24

• A. Metal Organic Chemical Vapour Deposition

(MOCVD)

• B. Molecular Beam Epitaxy (MBE)
• C. Chemical Beam Epitaxy (CBE)

Also called MO-MBE…a blend of the two (not covered here)

Veeco,

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 25

High vacuum 10-6 torr Solid sources heated in

Knudsen cell deliver direct beams of elemental molecules/atoms sent towards substrate

High vac allows in-situ

monitoring: RHEED (reflection high energy electron diffraction)

Can also have Gas Source

MBE for group-Vs (in particular P)

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 26

substrate ~ 400oC (lower

than MOCVD for good sticking

• f atoms)

Atoms adsorb and diffuse

along surface

Find lowest energy positions along

step edges

Atoms naturally arrange

themselves into ordered crystal structure

Veeco, http://www.veeco.com/markets/materials-science/molecular-beam-epitaxy.aspx

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 27 Gen III Veeco MBE at U. Delaware

Very high purity possible at

low growth rates

Great for Al containing

materials (low O environment)

Very good control,

monitoring, and modeling,

• f growth

Qdots developed on MBE Great for research Has some in-roads in

production too

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 28 http://www.tu-ilmenau.de/pv/forschung/

Low vacuum 100-10-2 torr, substrate heated to ~600-750oC to enable

chemical reactions

III and V sources fed as gases (mixed with carrier gas H2 or N2) into

chamber, react on or in vicinity of substrate’s surface

High gas flows, high growth rates, dynamic gas mixing effects MO by-products pumped to scrubbers and exhaust

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 29

Group III sources are metal-organics, such as:

Trimethyl galium=TMGa=Ga(CH3)3 Triethyl galium = TEGa = Ga(C2H5)3 TMI, TMAl,

Group V sources are hydrides:

Arsine=AsH3 Phosphine=PH3

Dopants: DMZn (dimethyl zinc), DETe, CBr4

Veeco,

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 30

planetary rotation of produce ~ spatially uniform wafers no vacuum pumping, high throughput tools Toxic outputs to manage

Adomaitis, Journal of Process Control, Volume 18, Issue 10, December 2008 Aixtron, http://optics.org/news/1/5/22/AIXlarge

Aixtron 15x4” planetary chamber chamber Three gas flow designs

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 31

Slide 31

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 32

1954 –Si solar cell with η=6% developed by Bell Labs 1958 –Vanguard I, the first solar powered satellite, was launched with a 0.1W Si solar panel to major success, solar cells quickly added to many future satellites 1970 - Ioffe Institute (Alferov) develops a heteroface GaAlAs/GaAs cell 1973 - single junction GaAs demonstrated by IBM η=13% 1981 –first concentrator array deployed using Si cell 1994 - NREL develops a GaInP/GaAs two-junction cell, at 180 suns concentration, first solar cell with η> 30% 1997 –2J cells on satellites 1999 –first triple junction cells demonstrated by NREL/Spectrolab, become widely used in space applications in 2000s 2006 - early use of triple junction cells in concentrator systems

Wikipedia

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 33

Slide 33

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 34

Biggest market for III/V MJ cells is still space

Satellites (and rovers) Emcore, Spectrolab, Azure Space

Spectrolab: 51’, 16 panel solar array NASA's Mars Opportunity Rover

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 35

New - Flexible cells via an etched substrate

Contour fit, lighter Also for unmanned aerial Microlink, Sharp

2J InGaP/GaAs solar cells in flexible sheet made by Sharp with η=24.4%; H. Yamaguchi, IEEE 2008 Microlink epitaxial lift-off of 1J, 2J and 3J wafers

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 36

1 sun AM1.0 spectra

Radiation

High energy particles cause

crystal damage => deep recombination centres => lower minority carrier lifetimes

Particles

Micro-meteorites

Thermal environment

Earth’s orbit -80 to 55°C Mercury 140°C Jupiter -125°C

Power-to-weight ratio (W/kg)

is key, cost ($/W) much less important

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 37

Slide 37

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 38

Due to epitaxy and high materials costs, MJ solar cells are expensive

Solution:

Couple MJSC with inexpensive concentrating optics to significantly reduce

the solar cell size.

Array of (optic + cell) = module

Solar Cell Concentrating Lens Typically 500X

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 39

Max Power Current density Voltage Jsc Voc Concentration

Efficiency increases with the

log of concentration!

Current density Voltage Jsc Voc Temperature

Efficiency does decrease

with increased T, but much less than Si

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 40

Typical 3MJ cell and Si cell performance

Higueras, Renewable&Sustainable Energy Reviews, vol. 15, p.1810, 2011

At 500X, the current

density can be very high

At high currents, internal

series resistance starts to ↓ η

Most III/V HCPV now

• perating at 500X

There are also LCPV

systems with Si cells in the 10-100X concentration

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 41

5.5mm x 5.5mm cell becoming the standard size Cell on carrier

• ptimized thermal properties to manage high heat flows

Optimized solder or epoxy attach of cell with no voids bond wires capable of high currents (2A)

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 42 Morgan Solar Inc. www.morgansolar.com

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 43

No matter the design, at

high concentration all optics have only a small acceptance angle (~ +/- 1o)

⇒ Must point directly at the

sun, can harvest only the Direct Normal Incidence (DNI) sunlight

⇒ High precision, dual axis

trackers are required

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 44

polycentric.csupomona.edu news story on lyle center 2010

Amonix: MegaModule design uses Fresnel lenses Soitec: Concentrix system use fresnel lenses with 500X concentration

http://www.soitec.com/en/technologies/concen trix/components/

SolFocus: reflective design

http://www.solfocus.com/en /solutions/

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 45

Innovative optic using total

internal reflection

Lightweight, compact for

shipping,

Low manufacturing cost Innovative tracker with very

low install costs

Pictures courtesy of Morgan Solar Inc.

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 46

Rehnu: mirrored dish Skyline Solar – low concentration troughs using Si cells Solar Systems: large mirrored dish focus onto an array of solar cells

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 47

APECS test site on uO campus :

• Morgan Solar’s Sun Simba total internal

reflection based optic (GenII shown)

• n a commercially bought carousel tracker

(2 trackers, ~ 10kW capacity

• Spectroradiometer & Pyrheliometer (DNI)
• Custom IV curves of individual cells under
• perating conditions and other data logging

SUNRISE test site at NRC Montreal Rd

• Opel Inc fresnel lens based optic
• n a pole-mount Feina tracker
• Pyrheliometer (DNI), global and

diffuse pyronometers

• Data logging every 2 min

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 48

Cells Optics Balance of System (Tracker, inverters, electrical) Install (site prep, regulatory, connection) ⇒Industry approaching $3/W installed ⇒Competitive with $2-4/W installed for flat plate PV

20% 30% 25% 30%

Very rough values – uO to publish cost model review in next year

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 49

Combining the requirement for high performance

with the low cost is a formidable challenge.

cells must be efficient and not too expensive current extraction must be effective and reliable the optics must be low cost, yet provide highly

accurate focussing with good spatial uniformity, high

• ptical efficiency

The tracking structure must be inexpensive yet

accurate and reliable

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 50

50

EPRI, CPV Consortium

Megawatts of CPV Power Deployed

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 51

SW US, N. Mexico Chile

• S. Africa

Australia

• N. Africa, Arabia, India

Spain, Italy

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 52

Solar has become grid competitive in high sun regions,

which tend to have:

• A. high DNI, low diffuse
• B. Hot temperatures
• C. Low water availability

A and B may tip the scale towards CPV over PV in sunny

locations

C may tip the scales toward CPV over CSP Additionally, CPV is easily scalable Materials are readily available, 97% recyclable

(primarily acrylic, steal, Al, glass)

CPV Consortium

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 53

Project size and project location will determine the most cost

effective solar technology solution

More significant decrease in costs (via efficiency gains of cells,

maturity) may increase market space for CPV over PV

53

Source: CPV Consortium, “CPV Industry Overview” Nancy Hartsoch 2011

CPV Si CSP CdTe

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 54

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 55

10 20 30 40 50 60 70 80 90 100 300 500 700 900 1100 1300 1500 1700

Quantum Efficiency [%] Wavelength [nm] AlGaInP (Top) InGaAs (Middle) Ge (Bottom) Quantum Dot Absorption

Cyrium Technologies provided these QE measurements within uO SUNRISE collaboration (www.cyriumtechnologies.com)

Add quantum dots (Cyrium) or

quantum wells (JDSU via acquisition of QuantaSol) into middle sub-cell

gain more current in middle

subcell from bottom sub-cell (which usually has excess)

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 56

Careful control of crystal structure/defects while moving beyond

lattice matching

ηtheory≈59%, top and/or middle junction can be MM

Can combine with 4J and/or inverted design Spectrolabs, Fraunhofer, NREL

Spectrolabs: R. R. King, D. C. Law, K. M. Edmondson, C. M. Fetzer, G. S. Kinsey, H. Yoon, R. A. Sherif, and

• N. H. Karam; Appl. Phys. Lett. 90, 183516 (2007)

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 57

Higher voltage, lower current Very careful current balancing required, diminishing returns Improved radiation hardness with thinner cells (shorter minority

carrier diffusion lengths)

78% max theoretical possibly, but space vs CPV designs different Spectrolab, Fraunhofer, etc

Dimroth, Fraunhofer, IEEE Photovoltaic Specialists Conference 02/2005. Higueras, Renewable&Sustainable Energy Reviews, vol. 15, p.1810, 2011

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 58

Semprius Inc uses a microprinting

technology to handle & transfer thinned 600 µm square cells

↓ heating, can operate at higher C

with higher η

Small, inexpensive optics They currently hold the record for

CPV module efficiency, reaching 33.9%

http://www.micromanufacturing.com/content/micro-technology-powers-low-cost-solar-energy http://www.technologyreview.com/energy/24504/page1/

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 59

3J 43.5%

Organics, dyes Silicon III/V under concentation Thin film

2012 NREL, http://www.nrel.gov/ncpv/images/efficiency_chart.jpg

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 60

have a winning combination of properties that allow for very high efficiency solar cells:

Direct bandgap -> high absorption High mobilities Low temperature dependent shift in efficiency GaAs ideal bandgap for 1J cell Can grow multiple alloy layers-> multiple sub-cells to

absorb wider band of sun’s spectra and more efficiently convert energy using 3+ junctions

>43% efficiency presently achieved, perhaps 50% efficiency

will be achieved by 2020?

3J cells in common use today for space solar arrays and CPV

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 61

Epitaxy growth, smaller more expensive substrates mean

expensive cells ~$5/cm2 (vs 0.02/cm2 for Si)

combined with inexpensive optics for high concentration,

CPV systems offer an attractive overall cost, can provide the grid parity & lowest $/kWh in high DNI areas today

Growth in high sun regions will be significant in the next 5

years, where CPV has a winning proposition

Accurate measurement and modeling of CPV systems are

essential to reach its full potential

Complex interdependent system (angles, wavelengths,

current matching in subcells, concentration, temperature, spectral accuracy, DNI, …) creates lots of great innovation

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 62

SUNLAB

Dr Jeff Wheeldon Dr Richard Beal Dr Joan Haysom Alex Walker Mark Yandt Olivier Thériault Pratibha Sharma Matthew Wilkins Jafaru Mohammed Aaron Muron Frederic Asselin-Guay Matt Wilkins Ahmed Gabr

• Prof. Karin Hinzer
• Prof. Trevor Hall
• Prof. Henry Schriemer
• Master teacher Gilbert Arbez
• Facilities manager Francine Proulx

Thanks to SUNLAB and other colleagues who provided material for this slidepack

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 63

Industrial

Cyrium Technologies Inc. (commercial CPV cells) Morgan Solar (commercial CPV systems) OPEL Solar Inc. (commercial CPV systems) Spectra Solaris (PV testing) Neptec (space applications) Quadra Solar (CPV+thermal system design)

Academic / Research Institutes

Université de Sherbrooke National Research Council of Canada McMaster University McGill University Universidad Politecnica de Madrid

• Dr. Joan Haysom

Solar Project Manager SUNLAB Solar Research Group School of Electrical Eng. and Comp. Science University of Ottawa

1st Canadian PV Grad School May 16-18th, 2012, University of Ottawa, Slide 65

Silicon accounts for ~90% of all PV sales in the world But Si crystals are expensive and slow to grow. Si requires the greatest thickness to absorb sunlight, due to

weak absorption because of its indirect band gap

Thicker semiconductor material means higher material

volume but also a higher quality material because of the longer paths for carriers

All this leads to relatively high material cost $/W Thin Films and Organics try to reduce thickness and input

materials cost, but generally have lower efficiency η

III/Vs can provide the highest η