31
7/23/2019 MLE4208-Lecture-7A.pdf http://slidepdf.com/reader/full/mle4208-lecture-7apdf 1/31 Thin Film Solar Cells (I) Lecture 7 References: 1. Physics of Solar Cells. Jenny Nelson. Imperial College Press, 2003. 2. Photovoltaic Materials, Series on Properties of Semiconductor Materials, Vol.1, Richard H. Bube, Imperial College Press, 1998. 3. Handbook of Photovoltaic Science and Engineering . Antonio Luque, Steven Hegedus. Wiley, 2003. 4. Photovoltaic Solar Energy Generation. Adolf Goetzberger, Volker U. Hoffmann. Springer, 2005. 5. Wikipedia (http://en.wikipedia.org/wiki/Main_Page). 1

MLE4208-Lecture-7A.pdf

  • Upload
    hikol

  • View
    216

  • Download
    0

Embed Size (px)

Citation preview

Page 1: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 1/31

Thin Film Solar Cells (I)

Lecture 7 

References:1. Physics of Solar Cells. Jenny Nelson. Imperial College Press, 2003.2. Photovoltaic Materials, Series on Properties of Semiconductor

Materials, Vol.1, Richard H. Bube, Imperial College Press, 1998.3. Handbook of Photovoltaic Science and Engineering . Antonio Luque,

Steven Hegedus. Wiley, 2003.

4. Photovoltaic Solar Energy Generation. Adolf Goetzberger, Volker U.Hoffmann. Springer, 2005.

5. Wikipedia (http://en.wikipedia.org/wiki/Main_Page).

1

Page 2: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 2/31

 As minority carrier diffusion lengths exceeds absorption depths, p-n junctions

make efficient photoconverters. However, monocrystalline solar cells areexpensive to produce and are used in very limited areas.

Why Thin Film Solar Cells

For PV to provide the full level of C-free energy required for electricity andfuel, solar power cost needs to be ~5 cents/kWh ($1.00 /Wp). Thin film solarcells provide cheap alternatives.

EC

EV

EF

300 !m

0.3 !m

Distance

Energy

base

p-type

emitter

n-typee-

e-

h+

2

Page 3: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 3/31

Challenges of Thin Film Solar Cells

Polycrystalline and amorphous semiconductors contain intrinsic defects

increasing the density of traps and recombination centers. As a result,• Diffusion lengths are shorter , so the material needs to be a strong

optical absorber. Alternatively, multiple junctions must be used to makethe device optically thick. It may be necessary to use extended built-inelectric fields to aid carrier collection. e.g. p-i-n structures in amorphoussilicon cell.

• Losses in the layer close to the front surface are greater. It is better toreplace the emitter with a wider band gap window material.

• The presence of defect states in the band gap makes the materialdifficult to dope, and limits the built-in bias available from a junctionthrough Fermi level pinning.

• The presence of grain boundaries and other intrinsic defects increases

the resistivity of the films particularly at low doping densities.• The presence of defects makes the minority carrier lifetime and diffusion

rate are carrier density dependent.

3

Page 4: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 4/31

Materials for Thin Film Solar Cells

• They should be low cost, non-toxic, robust and stable.

• They should absorb light more strongly than silicon. Higher absorptionreduces the cell thickness and so relaxes the requirement for longminority carrier diffusion lengths, allowing less pure polycrystalline oramorphous materials to be used.

Requirements:

4

Page 5: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 5/31

• Silicon is the only elemental semiconductor which has a suitable band

gap. Compound semiconductors greatly extend the range of availablematerials and of these II-VI binary compounds and I-III-VI ternarycompounds have been used for thin film photovoltaics. Many of themare direct band gap semiconductors with high optical absorption.

•  A number of materials have been identified of which the best developedat present are amorphous silicon (a-Si), polycrystalline cadmium

telluride (CdTe), polycrystalline copper indium diselenide (CuInSe2 or CuInGaSe2), microcrystalline thin film silicon (µ-Si), andmolecular electronic materials.

• Usually produced by physical or chemical deposition techniqueswhich can be applied to large areas and fast throughput.

• Note that the term “thin film” refers more to solar cell technologies with

mass production possibilities rather than the film thickness.

Materials for Thin Film Solar Cells

5

Page 6: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 6/31

Laboratory Thin Films Cell Efficiencies

Efficiency is the Best Metric to Gauge Progress

6

Page 7: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 7/31

Amorphous Silicon

Amorphous silicon (a-Si) is the best developed thin

film material and has been in commercial productionsince 1980. As a material for photovoltaics, it has theadvantages of relatively cheap, low temperature(<300oC) deposition and possibility of growing on avariety of substrates.

Defects in a-Si

The lattice of a-Si contains a range of bond lengthsand orientations, as well as unsaturated “dangling”bonds. The long range order of the crystal is gone.

In a-Si, Si atoms are arranged in an approximatedtetrahedral lattice but with a variation of up to 10o in

the bond angle. The loss of order results that theselection rules for photon absorption are relaxed,and a-Si behaves like a direct gap material.

7

Page 8: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 8/31

Defects in a-Si

The variation of Si-Si distance andorientation gives rise to a spreadingin the electron energy levels, whichappears as a tail in the density ofstates at the top of VB and thebottom of CB, called Urbach tail.

Amorphous Silicon

The defects can be created uniformly throughout the material by irradiatingthe material by heating, sudden cooling, or by extrinsic doping. They aredeep in the band gap.

Dangling bond states are due tounsaturated Si atoms with one valanceorbital not involved in bonding. Theymay be positively charged (D+), neutral(D0) or negatively charged (D-). Anexcess of D- gives rise to n type a-Si

while excess D+ gives rise to p type.

8

Page 9: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 9/31

Absorption

The loss of crystal order renders theabsorption coefficient is about an orderof magnitude higher than in crystallinesilicon at visible wavelengths.

There is no well defined E-k relationshipand no requirement to conserve crystal

momentum. Absorption depends simplyon the availability of photons and thedensity of states in VB and CB, much asit does in direct gap crystal.

a-Si used for solar cells typically has aband gap of 1.7eV, as a result of

passivation of a-Si with hydrogen.

Amorphous Silicon

The band gap is not optimum for solar energy conversion, while materialwith lower concentration of hydrogen is unstable due to poor doping andtransport properties.

9

Page 10: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 10/31

Doping

In unpassivated a-Si, the defect density is so high (>1016 cm-3) that thematerial cannot normally be doped. The extra carriers which would beintroduced into the CB or VB by donor or acceptor impurities are capturedby dangling bond defects.

Amorphous Silicon

 

P0" P

+

+ e#

 

e"+  D

0#  D

"

Since D- are recombination centers for holes, this decreases the minoritycarrier lifetime.

Energy

CBVB

DOS

before passivationafter passivation

The background density of defect states may be reduced by saturating

the dangling bonds with atomic hydrogen. Hydrogen forms a bond withthe unpaired electron in a D0 defect and removes the capacity of thatdefect to trap an electron or hole. Passivation with 5-10% hydrogenreduces the density of dangling bonds to ~1015 cm-3, from which workablep-n junctions can be made.

10

Page 11: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 11/31

Amorphous Silicon

Doping

donor level 

defect level 

n type

acceptor level 

defect level 

p type

TransportThe defect states which remain after hydrogen passivation act both ascharge traps and recombination centers, and dominate charge transport ina-Si.

The distribution of tail states below CB and above VB edges act as trapsfor mobile carriers. Charge carriers in these states move by a sequence of

thermal activation and re-trapping events.The distribution in energy leads to a distribution in the time constant for thetransport. So the usual transport parameters of mobility, lifetime, anddiffusion constant are density dependent.

11

Page 12: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 12/31

Amorphous Silicon

Energy

CBVB

DOS

before passivation

after passivation

Charge transport in such a defective medium is sometimes calleddispersive transport .

 

g E ( ) = N t 

k  BT 0exp

  E  "  E C 

k  BT 0

$ % 

' ( 

The density of tail states in the CB is oftenmodeled as

Transport

The minority carrier lifetime in undoped hydrogenated amorphous silicon(a-Si:H) is typically 10-20 µs and the diffusion length around 0.1 µm.Electron mobility is rather better than hole mobility, probably due to an

asymmetric trap distribution.With doping the defect density increases and diffusion lengths are muchreduced. So the carrier collection in a p-n junction would be extremely poor,and consequently p-i-n structures are used.

12

Page 13: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 13/31

Amorphous Silicon

Stability

 Amorphous silicon suffers from light-induced degradation known as theStaebler Wronski effect --- The defect density in a-Si:H increases withlight exposure, over a time scale of months, to cause an increase in therecombination current and reduction in efficiency.

13

Page 14: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 14/31

Amorphous Silicon Solar Cell Design

Amorphous Silicon p-i-n Structures

The basic a-Si solar cell is a p-i-n  junction. As the diffusion lengths are soshort in doped a-Si, the central intrinsic(undoped) region is needed to extendthe thickness over which photons maybe effectively absorbed.

The built-in bias is dropped across the width of the i-region, creating anelectric field which drives charge separation. In the p-i-n structurephotocarriers are collected primarily by drift rather than by diffusion.

The thickness of the i  region should be optimized for maximum currentgeneration.

So the cell should be designed so that the depletion width is greater thanthe thickness of the i  region at operating bias. In practice , it is ~0.5 µm.

EC

EVEF

EVAC

!n

!p qVbi

p i n

Distance

14

Page 15: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 15/31

 A typical RF PECVDchamber and relatedparts. A silicon-

containing gas such asa mixture of SiH4 and H2 flows into the chamber.

Preparation of Amorphous Silicon

The a-Si is usually deposited by plasma deposition of silane or “glow

discharge”, but a number of other deposition methods such as sputteringand “hot wire” CVD are being investigated.

Hot Wire CVD

15

Page 16: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 16/31

Amorphous Silicon Solar Cell Design

p-i-n Solar Cell Device Physics

• If the charged background doping in the i region is too high, at biases

above some threshold the i region will not be completely depleted. Theremaining neutral part of the i  layer is a “dead layer” and does notcontribute to the photocurrent.

i  type n type p type

low backgrounddoping or low bias

high backgrounddoping or high bias

Because the doping level in the p and n regions are so much higher than inthe intrinsic region, the depletion widths within the p and n regions are verysmall and can be neglected. Provided that the background doping is lowenough, the space charge region thickness is equal to that of the i-region.

16

Page 17: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 17/31

Amorphous Silicon Solar Cell Design

p-i-n Solar Cell Device Physics

i n p

h"

EC

EF

EV

near dark 

h"

EC

EFn

EV

EFp

illuminated

•  At high injection levels (e.g. highillumination), the free carriers in the iregion becomes significant and willarrange themselves to minimize theelectrostatic potential energy. In veryhigh injection conditions the fieldvanishes throughout most of the iregion and n!p.

 

 Da=

 Dn D p

 Dn + D p

The current is now driven by diffusionrather than drift. The diffusion rate is given

by an ambipolar diffusion.• The electron and hole lifetimes are generally not constant. At low light

intensity charge carriers are likely to be trapped in tail states, extendingtheir lifetime. The dark current is therefore intensity dependent.

17

Page 18: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 18/3118

Page 19: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 19/31

a Ag/ZnO textured back reflector

Fabrication of a-Si Solar Cells

Superstrate design

Layers of TCO, p-type, undoped and n-type a-Si are deposited in sequence onglass substrate. ZnO is deposited onto then-layer followed by a metal (Al) for the rearcontact.

Substrate design

Layers of TCO, p-i-n structures aredeposited on a metal substrate (steel) forthe back contact. The p-layer can be verythin and textured.

19

Page 20: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 20/31

Strategies to Improve a-Si Cells Performance

Light Induced Degradation

The Staebler Wronski effect is the most important barrier to the widespreaduse of a-Si solar cells. Light-induced degradation is stronger in materialswith a higher hydrogen content, yet a high hydrogen content is neededfor suitable doping and transport properties.

If a-Si could be produced with a lower original dangling bond density, lesshydrogen would be needed for passivation. One possibility is the “hot wire”

CVD technique which appears to produce good a-Si when saturated withonly 1% hydrogen.

Improvement of VOC

The VOC in a-Si solar cells is substantially less than the optical band gap(0.89 V compared to 1.7 eV). It can be increased by the use of either (i) a

wider band gap emitter  such as a-SiC:H or (ii) a polycrystalline Si emitter,in which degenerate doping is possible.

20

Page 21: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 21/31

Strategies to Improve a-Si Cells Performance

Improvement of JSC

Improve the blue light response in the p-layer. This could be achieved byreplacing the a-Si p-layer with a wider band gap a-SiC:H window, or by agraded a-SiC:H-a-Si:H layer.

Improve the long wavelengths light response in the i-region. This could beachieved with light trapping techniques to increase the optical pathlength within the cell, or with the use of multilayer a-Si cells.

Date Design Efficiency

1977 Schottky diode 6%

1980 a-Si:H p-i-n 6%

1982 a-SiC:H/a-Si:H p-i-n heterojunction 8%

1982 textured substrates 10%1987 grading of p-i interface 12%

1990s multi-gap design 13%

21

Page 22: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 22/31

Structure of triple-junction p-

i-n substrate-type solar cells

Improvement of Limiting Efficiency

Strategies to Improve a-Si Cells Performance

Multi-gap cell designs are possible using a-SiC:H as the material for a wider gap celland a-SiGe:H for a narrower gap cell.

Two terminal tandem design

The limiting efficiency for three cell devices is calculated to be 33%. Themain problems have been the poorer quality of the alloy relative to a-Siand incorporating large area tunnel junction.

22

Page 23: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 23/31

Polycrystalline Thin Film Materials

 A polycrystalline material is composed of microcrystallites or “grains” of the

semiconductor arranged at random orientations to each other.The material is crystalline over the width of a grain,being typically the order of one micron. As the grainsare large in quantum mechanical terms, the bandstructure and therefore the absorption coefficient, isvirtually identical to that of the single crystal

material. However, the transport and recombinationproperties are strongly affected by the presence ofthe interface or grain boundary.

The different orientations of neighboring crystal grains give rise todislocations, misplaced atoms (interstitials), vacancies, distortedbond angles and bond distances at the interfaces. In addition, extrinsic

impurity atoms are likely to concentrate at the grain boundaries.The various types of defect introduce extra electronic states and act as“intra-band-gap states” or “trap states”. Deep defect levels may act asrecombination centers.

23

Page 24: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 24/31

EC

EV

EVAC

EF

EC

EV

EVAC

EF

"0

n-sc n-scinterface

EC

EV

EF

EVAC

n-sc n-scinterface

- --- ++++

Polycrystalline Thin Film Materials

Band Profile in the Presence of Grain Boundary

We consider the case of a grain boundary in n type material. As the localneutrality level of the defects #0<EF, the states will be acceptor-like andtrap electrons. This gives rise to a plane of fixed negative charge at theinterface, and a layer of positive space charge on either side where the ntype material had been depleted.

The electrostatic force sets up a potential barrier  opposing further

majority carrier migration, while the minority carriers see a potential well at the grain boundary and are pulled towards it and the probability ofrecombining with a trapped majority carrier is high.

24

Page 25: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 25/31

Grain Boundaries and Transport

The effect of grain boundaries on charge transport

depends on whether they lie normal to or parallelto the direction of current flow.

When current is flowing across a grain boundary, thepotential barriers slow down the transport of majoritycarriers, limiting its mobility, while the potential welldrive minority carriers towards recombination centersat the grain boundary, reducing its diffusion length

and lifetime.

Increasing the trap density increases the space charge stored at thegrain boundary. So the barrier height ", conductivity # and recombination ".

Increasing the doping first increases the barrier height, but at higherdoping levels the traps become saturated, the space charge region beginsto contract and the barrier is reduced.

Grain boundary lying parallel to the direction of current flow principallyaffect minority carriers. Majority carriers traveling parallel to the grainboundary are not affected. But minority carriers are still likely to be trapped.

25

Page 26: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 26/31

Grain Boundaries and Transport

The resistivity of a polycrystalline material is high

at low doping levels, falling rapidly at someintermediate doping levels until finally it compareswith crystalline values at high Nd.

However, high doping tends to increase the lossesof minority carriers via recombination at grainboundaries. As with crystalline materials, the best

material is a compromise between high lifetimeand high conductivity.

EB low light intensity

EB high light intensity

Increasing the density of free carriers by illuminationreduces the net charge stored at the grain boundaryand hence the barrier height.

 As EB is reduced under increasing illumination themajority carrier mobility, and hence the conductivityare increased.

Intensity dependent mobilities are commonlyobserved in polycrystalline semiconductors.

26

Page 27: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 27/31

Grain Boundaries and Recombination

To evaluate the overall effect on minority carrier  transport characteristics,

1

" eff 

=1

" SRH 

+1

" gb

+1

"  Auger

+1

" rad 

Grain boundary effects influence the I-V characteristics of polycrystallinesolar cells in several important ways:

• The effect of grain boundaries in reducing majority carrier mobility mayincrease series resistance.

• The enhancement of minority carrier recombination at grain boundariesreduces lifetimes and increases dark current.

• Because minority carrier lifetimes and diffusion lengths are carrierdensity dependent, the photocurrent may be bias dependent, and notwell approximated by JSC, indicating simple diode equation descriptions

are not appropriate.

27

Page 28: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 28/31

Thin Film Silicon Solar Cells

Thin film microcrystalline silicon (µ-Si) is characterized by grain sizes of

~1 µm. It has the optical properties of crystalline silicon, while its electronicproperties are dominated by the grain boundaries.

Defect states at grain boundaries include dangling bonds typical of a-Si aswell as extrinsic impurities introduced during growth.

µ-Si can be prepared by the same techniques as multi-crystalline Si,normally by casting of molten silicon into aggregates or sheets. A range of

other techniques have been investigated for thin film polycrystallinematerial, including liquid phase epitaxy, chemical vapor deposition andthe crystallization of amorphous silicon.

The goal is to fabricate cells by depositing the µ-Si on cheap ceramic or foilsubstrates, or on glass.

28

Page 29: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 29/31

Thin Film Silicon Solar Cells

 As in any polycrystalline material, the diffusion length in µ-Si is effectivelylimited by the grain size, which limiting the cell thickness or effectivecarrier collection of a few microns or tens of microns.

However, the low absorption coefficient means that a layer of µ-Si of afew microns thick harvests less than half of the available photons.

Two basic approaches to improving performances:

• Light trapping in thin film Si solar cells. Light trapping techniques

have to be used to increase the optical path length inside the cell.Texturing front and back surfaces could increase the optical depth by afactor of ~20. This relaxes the need for a long diffusion length, andallows higher base doping increasing VOC.

• Parallel multijunction thin film Si solar cells. The cell is composed ofconsecutive micron-thick layers of p and n type µ-Si. Layers of similarpolarity are connected together to give a set of p-n junctions connectedin parallel.

29

Page 30: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 30/31

Thin Film Silicon Solar Cells

Structure of parallel multijunction

thin film Si solar cells

Further reading:

30

Page 31: MLE4208-Lecture-7A.pdf

7/23/2019 MLE4208-Lecture-7A.pdf

http://slidepdf.com/reader/full/mle4208-lecture-7apdf 31/31

Summary