role of surfaces and interfaces in solar cell manufacturing · of surfaces and interfaces in solar...

thergyen-

theels.

s ofandThe

viaore

ght.ble

ringght.y ist inien-ringells

ll aseld

itsthatility

CIRP Annals - Manufacturing Technology 63 (2014) 797–819

wed

key

hree

and

d by

role

IRP.

Role of surfaces and interfaces in solar cell manufacturing

Abhijit Chandra (2)a,*, Grant Anderson a, Shreyes Melkote (2)b, Wei Gao (1)c,Han Haitjema (2)d, Konrad Wegener (2)e

a Department of Mechanical Engineering, Iowa State University, Ames, IA, USAb GWW School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA, USAc Department of Nanomechanics, Tohoku University, Sendai, Japand Mitutoyo Research Center Europe, Best, The Netherlandse Department of Mechanical and Process Engineering, ETH Zurich, Zurich, Germany

1. Introduction

Solar energy is abundantly available. The amount of solar energythat hits the earth in 40 min is equivalent to the global energyconsumed in one year. Also, converting just 2.5% of the sunlight thathits land available for solar installation in the southwestern UnitedStates could produce the nation’s total energy consumption in 2006[95]. However, collecting this energy requires a large number of solarpanels, which have been too expensive to produce and deploy.Ongoing efforts are attempting to reduce this cost.

With carbon dioxide levels rising and renewable sources ofenergy receiving more attention, the solar industry has expandeddramatically over the past 10 years. Wind energy, hydroelectricpower, and solar cells have all received significant attention aspotential modes of producing sustainable energy. Current eco-nomic advantages of fossil fuels are limited in many ways. Coal andoil produce toxins that harm the environment, the public fearsnuclear power and its dangers, and extraction of natural gas canharm the environment through fracking and additional heating ofthe atmosphere caused by pipe leaks and ventilation of shale gaswells [123]. Even though solar is currently one of the moreexpensive energy sources according to 2010 data (nuclear and coalcost 4 US cents/kWh, hydroelectric costs 3 cents/kWh, natural gasand oil cost 10 cents/kWh, and solar costs 8.3 cents/kWh infavorable environments [82,88,89,118,125,83,159]), it also has the

According to numerous international panels [20,28,35,166]three key parameters for commercial success of solar eneharvesting are: efficiency, reliability, and cost. Significant

hancement in efficiency and reliability are warranted to lowercost of producing a kWh, and make it competitive to fossil fuManufacturing solar cells is the art of creating large scalefunctional surfaces and interfaces; surfaces for incoming light

interfaces for the separation of the majority load carriers.

efficiency enhancement of solar cells is currently being pursuedthe dual means of: (i) trapping more light and (ii) producing melectricity from a unit of energy contained in the trapped liThese two characteristics are primarily realized by suitamaterial choices in photovoltaic (PV) devices and by engineethe surface morphology to trap a greater fraction of incident liUnfortunately, the conversion of photonic energy to electricitassociated with a waste stream of heat. Production of heaphotovoltaic devices can degrade their energy conversion efficcy and damage the interfaces between cell layers, thereby lowelife expectancy. Thus, reliability enhancement in photovoltaic chinges on the effective management of heat production as wemaking interfaces robust against thermal cycling under fiservice conditions.

This paper starts with a schematic of a photovoltaic cell andoperations. This is followed by discussion of current avenues

are being pursued for efficiency enhancement. Issues of reliab

A R T I C L E I N F O

Keywords:

Sustainable development

Surface

Surface integrity

Surface modification

Semiconductor

A B S T R A C T

Solar cells gained significant interest recently due to the rapid increase in fossil fuel costs and rene

attention to sustainability. The effectiveness of a solar cell in energy harvesting hinges on three

characteristics: efficiency of energy conversion, reliability and life expectancy, and cost. All t

attributes are critically influenced by surface and interface properties inherent in the design

manufacture of these devices. This paper starts with an exposition of solar cell manufacturing, followe

discussion of efficiency enhancement, reliability issues, and cost and energy footprint reduction. The

of surface and interface modifications in realizing such improvements is assessed.

� 2014 C

Contents lists available at ScienceDirect

CIRP Annals - Manufacturing Technology

journal homepage: http: / /ees.elsevier.com/cirp/default .asp

areuesothlost

most potential. Unlike fossil fuel reserves, solar energy is notprojected to run out anytime soon. Scale-up in solar energy isfurther facilitated by the vast supply of landmass currentlyavailable for solar installations.

des beent* Corresponding author.

http://dx.doi.org/10.1016/j.cirp.2014.05.008

0007-8506/� 2014 CIRP.

and life expectancy of surfaces and interfaces within solar cellsconsidered, and testing protocols as well as analytical techniqfor life prediction are discussed. Next, cost related aspects – bshort term capital costs and long term operational as well as

opportunity costs – are considered. Finally, the paper concluwith a discussion of possible futuristic endeavors that canpursued to combine the strengths of efficiency enhancem

http://crossmark.crossref.org/dialog/?doi=10.1016/j.cirp.2014.05.008&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1016/j.cirp.2014.05.008&domain=pdf


http://www.sciencedirect.com/science/journal/00078506


avening

play

2. T

Teffecwithsmano egenewithhighphoand

do nblueconse.g. 6cellsabsovalethe

enhaA

solathe

abouof thbandsemThe

tablIn thwhiprevforcvice

WelecpasselecretuThe

frombandlackan ean ewithprim

Iwhi

F

A. Chandra et al. / CIRP Annals - Manufacturing Technology 63 (2014) 797–819798

ues with measures for reliability improvement, while avoid-their weaknesses. In all these aspects, surface and interfaces

an essential role.

ypes of solar cells

he premise of photovoltaic cells is based on the photovoltaict, discovered by Edward Becquerel in 1839 while experimenting

an electrolytic cell. He found that certain materials generated all current when exposed to light [165,167]. However, there wasxplanation as to why only certain wavelengths of light canrate an emission of electrons from an irradiated metal; blue light

little intensity can produce a current of emitted electrons, but intensity red light cannot. In 1905, Einstein discovered the

toelectric effect, which used the viewpoint of light as a particleallowed light to be classified as individual photons. Red photonsot have enough energy to knock electrons out of metal atoms, but

photons do if the energy is higher than hc0/l (h is Planck’stant, c0 is the velocity of light, l is the wavelength of blue light00 nm) [127,167]. The internal photoelectric effect used in solar

takes place only in semiconductors and describes the process ofrption of the photons in order to excite electrons from thence to the conduction band. The counterpart of the electron inconduction band is the electron hole in the valence band. Thisnces the conductivity of the material.

p-n junction is needed in semiconductors to make them ar cell. This p-n junction is created through dopants. One side ofsemiconductor is doped with another material at a ratio oft 1 dopant to 1 million original atoms [153]. The ‘n-type’ sidee semiconductor has dopants with one electron in the valence

that it wants to shed, and the ‘p-type’ side of theiconductor has dopants that want to incorporate an electron.‘n-type’ dopant can be a V or VI column material in the periodice, and the ‘p-type’ dopant is usually a II or III column material.e ‘n’ region, the conductivity of electrons is strongly increased,

le in the ‘p’ region, the conductivity of electron holes isalent. This leads to an electrochemical potential (Fig. 2.1) that

es electrons from the ‘p-type’ region to the ‘n-type’ region and versa for the holes in the ‘p-type’ region.

hen light hits the semiconductor, photons add energy for antron to pass from the ‘p’ side to the ‘n’ side and allow a hole to

in the opposite direction. With an external current path, thetron flows outside the cell, generates electricity, and then

with each other in an array. These cells are covered and protectedby glass panes. Thus, modules consist of the cells, electricalcontacts and wiring between cells and to other modules, andprotective coverings, with numerous interfaces.

2.1. Crystalline silicon

The most common type of solar cell is made of crystallinesilicon. This is because silicon is abundant, non-toxic, has anappropriate bandgap of 1.1 eV, and respective process technologyhas been intensively investigated and developed for microelec-tronics. The major bottleneck of silicon is its relatively weakabsorption of light, forcing silicon to be used in a relatively highthickness in order to absorb sufficiently. Silicon is generally dopedwith boron or phosphorus, but this must be done in a cleanenvironment to control impurities. Also, silicon is a brittle material,which limits the lower bound on its thickness. Because of thesethickness limitations, the material cost of silicon solar cells is highand can be lowered only to a certain extent.

Both single crystal and multicrystalline silicon materials areused for silicon based solar cells. Single crystal silicon (sc-Si),usually boron-doped p-type silicon, is grown using the Czochralski(CZ) crystal growing approach. Single crystal silicon based solarcells typically have 5–6% higher energy conversion efficienciesthan thin film or multicrystalline Si cells due to their inherentlylow defect density [85]. However, the cylindrical sc-Si ingots areground to have pseudo-square cross sections and are then slicedinto pseudo-square wafers with a wire sawing technique, whichresults in material loss and increases defect density [41,157,158].The material loss is the kerf width plus 20 mm that are etched. Inaddition, p-type sc-Si solar cells also suffer from light-induceddegradation of efficiency, leading to further reduction in costeffectiveness [138].

Compared to sc-Si, multicrystalline silicon (mc-Si) is lessexpensive. However, mc-Si based solar cells typically have lowerefficiencies compared to sc-Si based devices due to defectstructures that develop during directional solidification or ingotcasting used for crystal growth [97,131]. The structural defectslead to premature recombination of electrons and holes (minoritycarrier recombination) and consequent efficiency loss at thesedefect sites (e.g., dislocations, grain boundaries, metallic impurityprecipitates [22,64,98,99,130,132,133,140,148,175]). Though de-fect engineering has been successful in reducing the number ofdefects, multi-crystallinity still limits efficiency since the mosteffective low cost texturing method can only effectively be appliedon sc-Si silicon wafers.

To eliminate waste during wafer slicing, continuous ribbongrowth technologies such as Edge-defined Film-fed Growth (EFG)and String Ribbon processes were developed as alternative mc-Siwafer production methods. The EFG technique is based on thegrowth of octagonal mc-Si tubes from silicon melt. The tubes aregrown continuously and silicon wafers are cut from the faces of thetube with less than 10% material loss [33,66,133,134,115]. Ribbonmaterials generally contain more point defects and higherdislocation density due to larger temperature gradients duringgrowth compared to CZ and ingot casting approaches [115,152].Thus, while these silicon wafer production techniques have a costadvantage over traditional approaches, the efficiency of solar cellsmade from these materials is compromised by high process

ig. 2.1. Operation of the semiconductor in a photovoltaic cell [141,153].

rns to the ‘p’ side, completing the circuit and refilling the hole.minimum amount of energy required to bump the electron

the ‘p’ side to the ‘n’ side of the semiconductor (known as thegap) corresponds to the depletion zone [141]. A photon

ing the same or more energy than the bandgap does not excitelectron. A photon with more energy than the bandgap exciteslectron, and the rest of the energy turns into heat. This, along

the absorption of light in all other layers of the cell, is theary loss mechanism in a photovoltaic cell.

n photovoltaics, the cell is made from one contacted wafer,le the module is a panel containing a number of cells wired

induced defect density. Efficiencies of 18.2% for EFG and 17.8% forstring ribbon based solar cells have been reported for specificprocesses [137], much lower than the highest efficiencies of single-and multi-crystalline silicon. The highest overall panel efficiencyachieved is 23.5% [24].

2.2. Thin film cells

Thin films are typically classified as the next generation of solarcells that will replace expensive crystalline silicon cells. Fig. 2.2shows a clear trend of increasing shipments of both crystalline

tedandces.tiontaic

ed thergers of

ofthehis

theTheost.c-Sitiontion to ayeroutage

A. Chandra et al. / CIRP Annals - Manufacturing Technology 63 (2014) 797–819 799

silicon cells/modules and thin film cells/modules in recent yearsand into the future. The total market share of c-Si cells/modules,including single or monocrystalline, multicrystalline, and ribbonSi, has decreased from a peak of 95% in 2004 and 2005 to 87% in2010. However, in the near future, crystalline cells are projected tokeep a market share of about 80% due to their matureinfrastructure and growth in China and other countries (Fig.2.2). On the other hand, the growth of newer thin film technologiesas well as their decrease in cost will continue to shape the future ofphotovoltaics [150].

Thin film solar cells are made of materials such as amorphoussilicon, Cadmium Telluride (CdTe), and Copper Indium GalliumSelenide (CIGS), but there are many more combinations ofsemiconductors that have been made and are still beingexperimented with. These cells are typically deposited on glasssubstrates, but can vary depending on the type of thin film cell. Thesmall amount of material used in the semiconductors and electricalcontacts in thin film cells make them cost effective.

A thin film solar cell is composed of several layers of differentelements. In addition to the absorbing layer there is a transparentconducting oxide (TCO) layer, a window layer, and rear metalcontact layer, etc. The properties of each of these layers, thestructure of the solar cell and the properties of the interfacesbetween different layers are important factors that influence theactual efficiency of the thin film cell. The physics and properties ofthe TCO layers are treated separately in Section 4.1.

Fig. 2.3 shows the configurations of a single junction

changing the material from classical a-Si to hydrogenaamorphous silicon (a-Si:H), which is suitable for doping

alloying with other materials and fabrication of junction deviFig. 2.4 shows the relationship between the optical absorp

coefficient and the photon energy for different photovolmaterials. The minimum thickness of the absorbing layer formby the photovoltaic material in a solar cell is of the order ofinverse of the optical absorption coefficient. Thus, a laabsorption coefficient results in a thinner minimum thicknesthe absorbing layer. For incident light with photon energy1.25 eV, which corresponds to a wavelength of 1 mm,

absorption coefficient of c-Si is approximately 100 cm�1. Tmeans that 63% of the light energy is absorbed by c-Si whenlight has propagated 1/100 cm (100 mm) into the material.

other 37% of light energy will pass through the material and is lThus, the thickness of the crystal silicon substrate used in a

solar cell ranges from 150 to 300 mm, without taking reflecfrom the rear into account [142]. On the other hand, the absorpcoefficients of the materials used for thin film solar cells are tenhundred times larger than that of c-Si, reducing the absorbing lathickness of the thin film solar cell to less than 10 mm withsacrificing light energy absorption. This is a significant advant

Fig. 2.2. Previous and estimated future PV module production capacity (MW) until

2017 [150].

Fig. 2.3. Configurations of single junction amorphous silicon solar cell.

Fig. 2.4. The relationship between the absorption coefficient and the photon energy

for different PV materials [142].

amorphous silicon (a-Si) solar cell, which is the most widely usedthin film solar cell. There are two types of configurations – thesuperstrate and the substrate. The main difference between theseis the substrate, which provides a layer of support to the thin filmcell. The superstrate type employs a transparent glass substrateand the substrate type employs a non-transparent substrate,which can be a metal or metallic coating on a glass/plastic.

In recent years, the conversion efficiency of thin film a-Si solarcells has been improved by a number of technologies. The earliestamorphous silicon solar cells were used in calculators and digitalwatches. Practical power PV modules were then made possible by

fromviewcellstaic

highand

Csolafilmeffic

Fig. 2modu


the material consumption and solar cell manufacturing costpoint, however, thin film cells have lower efficiency than c-Si. In addition to the large absorption coefficient, the photovol-material of a thin solar cell should have the characteristics of

quantum efficiency of excited carriers, long diffusion lengthlow recombination velocity.

dTe and CIGS solar cells are types of commercialized thin filmr cells. Fig. 2.5 shows a schematic of the structure of a CdTe thin solar cell. CdTe has an excellent theoretical energy conversioniency of 29% (seen in Fig. 2.7) due to its bandgap of 1.5 eV [90].

Because the optical absorption coefficient is 105 cm�1 (Fig. 2.5), aCdTe layer with a thickness of several mm can absorb 90% ofincident photons, making it attractive for use in a thin film solarcell. The superstrate configuration and heterodyne-junctionstructure are typically employed in the CdTe thin film solar cell.It consists of a soda lime glass superstrate as the supportingmaterial, a TCO layer, a cadmium sulphide (CdS) window layerwith a thickness of 100 nm, a CdTe absorber layer with a thicknessof several mm, and a back contact. The polycrystalline semicon-ductor materials of CdS and CdTe have the advantage of highchemical stability. Up to 16.5% energy conversion efficiency hasbeen achieved for a laboratory CdTe thin film solar cell [90]. Theefficiencies of commercial CdTe thin film modules are typically 10–11%. A major advantage of the CdTe thin film solar cell over the a-Siand CIGS cells is the simple and low cost deposition processrequired for fabrication of the cell elements. The main issue of theCdTe cell is the toxicity of cadmium, which can cause environ-mental problems.

As seen in Fig. 2.4, Copper Indium di-Selenide (CuInSe2) orCopper Indium Gallium di-Selenide (CIGS) has the best opticalabsorption coefficient spectrum. Due to its large absorptioncoefficient of up to 105 cm–1 and wide bandwidth, a CIS or CIGSlayer with a thickness of 1 or 2 mm can absorb most of the photonsover almost the entire sun light spectrum. Fig. 2.6 shows aschematic of a CIS/CIGS thin film solar cell. Differing from the a-Siand CdTe cells, the CIS/CIGS cell employs a substrate configuration.This provides CIS/CIGS cells with flexibility in selection of substratematerials because there is no requirement on the light permeabil-ity of the substrate. For this reason, low-cost substrates can beemployed for CIS/CIGS cells, which contribute to cost effectivenessof the CIS/CIGS cell. The CIS/CIGS cell also employs the heterodyne-junction structure, which consists of a substrate, a molybdenum(Mo) back contact film, a CIS/CIGS absorber p-type layer with athickness of 1 mm, a CdS n-type layer with a thickness of 50 nm,and a TCO layer. The CIS/CIGS absorber p-type layer and a CdS n-type layer form the heterodyne junction. Soda-lime glass is oftenused as the substrate for the CIS/CIGS cell. In addition to its costeffectiveness, the sodium atoms in the soda-lime glass diffusingfrom the glass into the CIS/CIGS layer improve the dopingconcentration in the CIS/CIGS absorber layer during the fabricationprocess. The best energy conversion efficiency of a CIGS cell isreported to be 20.8%, and that of a CIGS module is 17.4% [2]. Sincethe CIS/CIGS cell is composed of much more material elementsthan a-Si and CdTe cells, more attention toward process control isnecessary during their fabrication. Another drawback for CIS/CIGScells is the limited amount of Indium and Gallium available onEarth.

The energy conversion efficiency of a solar cell is mainlydetermined by the properties of the PV material. Fig. 2.7 shows thetheoretical maximum energy conversion efficiencies of a numberof PV materials. Among the thin film PV materials mentioned, CdTehas the highest efficiency of about 29%, followed by a-Si (about26%) and CIS (about 25%) [2]. It can be seen that a large portion ofthe overall light energy is lost instead of being converted toelectrical energy. There are several reasons for such energy loss. Asseen in Fig. 2.4, each of the PV materials is suited only for a specificspectral bandwidth. The photons with energy outside theabsorption band pass through the PV material and cannot beabsorbed. Such radiant energy is converted into heat rather than

Gla ss S uperstr ate

Front Co ntact (I TO/ZnO)

CdS (Cadmium S ulfide) n -type l ayer

CdTe p-type l ayer

Back Co ntact

Reflected photon

Absor bed pho tons

Fig. 2.5. Schematic of a CdTe thin film solar cell [90].

ZnO tr ansparent oxi de

CdS bu ffer la yer (or Indium S ulfide (I nS))

CIGS (a bsorber)

Mo con tact la yer

Gla ss

Reflected photon

Absor bed pho tons

Fig. 2.6. Schematic of a CIGS thin film solar cell [90].

.7. Theoretical energy conversion efficiencies of commercial thin film solar

les [2,7].

electric current, resulting in a loss of up to 25% of the incident lightenergy. Recombination and resistive heating are the main factorsfor further drop in the energy conversion coefficient of a solar cell[163].

The rest of the losses in the cell are due to module inefficiencies.The largest portion of this loss, about 35% of the incident lightenergy, is caused by optical reasons, such as scattering andreflection at the surface of the solar cell. Other major losses includeelectrical resistance losses in the PV material and the electricalinterconnects, as well as losses due to shadowing of bus bars andfingers, which all together account for about 15% of the loss. As a

The be

heirese

ughme

and

theintoting

theass

ates Fort beothellslls),86].ingfterf PV

byringinsi or

is auchcks,andells.ring

to

pedype


result, only about 25% of the incident light energy can be convertedinto electrical energy.

Although thin film PV materials have theoretical conversioncoefficients comparable to crystalline silicon, thin film solar cellstend to have lower actual conversion efficiencies because of thethinner layer of light absorbing PV material in thin film solar cells.As seen in Fig. 2.8, the energy conversion efficiencies of commercialthin film solar modules range from 10% to 13% while those ofcommercial crystal silicon solar modules typically range from 17%to 22% [85].

Projected worldwide thin film solar cell production through theyear 2013 is seen in Fig. 2.9. Although the amount of thin film cellsis growing, there is a downside to the growth of some of the cells.The materials used in some thin film solar cells, such as Gallium,Indium, and Tellurium, are rare-earth materials and thereforeexpensive. If these cells become popular, the prices will rapidlyincrease because of the diminishing amounts of these rarematerials. Continuing research on using more common materialsin thin film cells and improving their conversion efficiencies willallow these cells to become more efficient and less expensive.

a current [85]. Quantum dot cells use a similar approach.

quantum dots are made of semiconductor materials and cangrown to any size. The size of the quantum dots determines tbandgap, which allows the cells to be easily tunable [173]. Thcells are relatively cheap, but their efficiency is not yet high enoto be competitive. However, they have the potential to becomore efficient than silicon cells through better light trapping

material selection.

3. Manufacturing

Manufacturing of solar modules requires controlling

physical effects active in the conversion of light energy

electric energy with the least possible losses. The key to preventhese losses is to create cohesive surfaces and interfaces withincells. Manufacturing techniques also need to be amenable to mproduction. A 30 GWp production per year worldwide necessitthe manufacturing of 150–200 square kilometers of PV panels.encapsulation, both the front and back sides of the panels musmanufactured and processed with care because they bcontribute to the efficiency of energy conversion [80]. Solar cmade in labs can achieve 45% efficiencies (for multijunction cebut mass produced cells have efficiencies of only 24% or less [This implies that scale-up plays an important role in transitionthe efficiency gained in the lab to commercial products. Adesign and structure are determined, the relevant properties ocells – namely efficiency and life expectancy – are influencedthe manufacturing process chain. The generic manufactuprocess chain is shown in Fig. 3.1. In c-Si cells, this chain begwith the wafering and slicing of mono or polycrystalline CZ SUMG Si ingots into wafers. Creation of surfaces and interfaceskey ingredient in solar cell manufacturing process chain. Sprocess elements naturally introduce defects, e.g. scratches, craand dislocation pile ups that degrade the surface integrity

ultimately affect both performance and reliability of the cAccordingly, care must be taken in the solar cell manufactuprocess chain to minimize such defects and their propagationlarger (from micro- to macro-) scales.

The emitter in a solar cell is typically formed by a heavily dolayer of opposite polarity, n doped in p-type Si and p doped in n-t

Fig. 2.8. Energy conversion efficiencies of various solar cell types and their

respective modules [85].

Fig. 3.1. Manufacturing process chain.

theacergeingncel is thecell,ncepedand

Fig. 2.9. Major types of thin film PV cells/modules [24].

Other types of photovoltaics include dye-sensitized andquantum dot solar cells. Dye-sensitized cells utilize semiconductorcrystals over a thin layer of titanium dioxide (TiO2). A light-absorbing dye is spread over the TiO2, and the excited electronsfrom the semiconductor crystals flow through the TiO2, resulting in

Si, separated by an area of neutral polarity. On both sides ofmetallurgical junction (which has neutral polarity) there is the spcharge area, which is characterized by depletion of mobile chacarriers and only contains fixed charges due to the ionized dopatoms P+ and B�, which exert an electric field and counterbalathe diffusive force on the mobile charge carriers. The celcompleted by using metallic fingers as contacts on the sun face ofcell (front) and a metallic surface on the dark face (back) of the

which needs to be attached with lowest possible ohmic resista(Fig. 3.2). Back and front contacts need to reduce leakage of traplight, since long wavelength light has a long absorption length

needwith

Tprocsignsilicintedevewidparawhicutt18020 mstre[51,

Benerany

etchmaypyramonredumulligh(100geneGasetechSi nrefle250proption

Timpshowof eapplfronat 8concconcthe

mayperfredubarremienhamai

Fig. 3conta


s multiple reflections internally. The number of reflectionsin the cell increases as the cell gets thinner [31].he silicon used in the emitter is typically produced by slicingesses, which tend to have high material loss. The wafering stepificantly impacts the surface and subsurface integrity of theon substrate. Recent research is aiming to enhance the surfacegrity while reducing material loss due to wafering. Theselopments include lower kerf width, reduced influenced zone

th, and enhanced surface integrity by optimizing cuttingmeters. The disturbed zone must be removed by etching,

ch is done after a cleaning step to remove residues of theing fluid and debris. Average etching processes start with

mm (after wire sawing) and are reduced in thickness bym. The etching process increases the four point bending

ngth by removing cracks that create stress concentrations81,120,164,168,126,129,135,147].are Si surfaces reflect more than 30% of the light spectrum withgies higher than the bandgap of Si [145]. During removal ofscratches and chip remnants left by the saw, the surface ised to create a light trapping surface structure. Etching agents be acidic or alkaline. The front 5 mm is etched to generate amid or inverse pyramid structure by either self-masking forocrystalline cells or printed masking as in [160]. The pyramidsce the overall reflectivity of external light by generating

tiple reflections on the pyramidal surfaces and trapping thet that already entered the cell. This allows longer wavelengths0–2000 nm) to be absorbed by the cell. Structuring can alsorate a porous surface layer through anodic HF etching.ous techniques like RIE plasma may soon replace wetnology [38], thus reducing handling efforts. Etching of randomano tips has been reported to reduce reflectivity to 1%ctance losses for a wavelength regime between 200 and

0 nm [86]. Besides geometrical enhancement of reflectionerties, physical–chemical reduction of reflection (antireflec-

) on the surface is necessary (see Section 4.1).he next step after texturing is the formation of the mostortant functional layer, the emitter, of which the principle is

n in Fig. 3.3. There are a number of very different possibilitiesmitter formations available in industry. The most commonlyied formation involves the diffusion of dopant atoms into thet surface. Diffusion of phosphorus into the surface of p-type Si

At diffusion temperatures all simple P-donors are gaseous,which is the reason why phosphosilicate glass (PSG) is used as aprecursor for dopant infiltration. It is formed by a chemical vapordeposition and surface reaction process, where part of the silicon isconsumed in the oxygen and phosphorous rich atmosphere. In P-diffusion, the diffusivity changes with concentration at differenttransition points, which gives a distribution profile that is non-Gaussian and provokes dead layer formation [70]. Multi plateautime temperature curves are used in [52] to manipulate the P-profile. However, liquid based phosphorus sources like sol–gel ordiluted phosphoric acid may be spun onto the surface. PSG isformed by baking after coating. Typically the doping is performedthrough a batch process in a quartz tube furnace to avoid metalliccontamination, but research is directed toward inline emitterjunction formation as shown in Fig. 3.4. P doping allows veryheavily doped surfaces to be exploited because lower resistivitycontacting of metallic fingers is possible. This is the basis forselective emitter formation, where stronger doping is appliedunderneath the fingers but not on the front surface to avoidincreased recombination losses.

Diffusion of boron into n-type Si is performed in a similar

.2. Principle of a solar cell showing essential interfaces and surfaces. The front

ct is made of metallic ‘‘fingers’’ across the cell [31].

Fig. 3.3. Principle of the emitter.

Fig. 3.4. Inline phosphorous emitter junction formation with sprayed phosphorus

deposition and diffusion in a belt furnace [38].

00–900 8C is the dominant industrial doping process. If theentration of the dopant becomes so high that the chemical P-entration is higher than the electrically active concentration,

minority charge carrier lifetimes will be greatly reduced. This lead to dead layers that can severely deteriorate cellormance. Dead layers are removed by etching or avoided bycing the dopant source strength or by diffusion through aier layer. As high concentrations of P distort the Si-lattice, thetter may act as a sink for impurities. This gettering effect maynce the cell performance if the emitter recombination is

nly governed by Auger recombination [70].

manner to phosphorus doping. The diffusion temperature is 100 8Chigher than P-diffusion. No inactive B-atoms are present, so nodead layer exists. Gettering is also much weaker than with P, so ahigher quality material with less contamination in the diffusionprocess is required. PSG and BSG must be removed by a cleaningstep involving HF or KCN-Clean [132].

Other methods of creating a p-n junction in Si include ionimplantation, epitaxial growth, layer inversion, and PECVD coating.Ion implantation is performed by accelerating a kV ion beam of thedopant onto the substrate surface, which generates a delta profilewith very high ion concentration at the surface. Co-diffusion can be

VDear

iO2/g a

bycesthe

gas. theure1%

red

ostthatnic

temlineges

toivi-

asteent,

fine by

thee tohile

ofas a

by, is

areing,ckelent


used with ion implantation to uniformly distribute the implanteddopant and avoid dead layers, which indicate a total separation ofthe deposition and drive-in process. Advantages of ion implantationinclude excellent controllability and repeatability along with thepossibility of creating selective emitters by directing the ion beam. Amain disadvantage is the necessity to anneal surface defects due toimplantation, which can be made within a co-diffusion oxidationprocess. Because of high temperature annealing and the absence ofgettering, implantation is not used with multicrystalline material.

Epitaxial growth of the emitter onto the surface of the Sisubstrate has an advantage of very fast growth in high temperatureCVD. The dopant gas is directly added to the Si precursor. Normally,epitaxially grown emitters are used for epitaxially grown thin filmbase material. The epitaxial growth also allows for a sequencechange, as shown by [171], one of the first groups to grow anepitaxial emitter, texture the surface, and grow a highly dopedepitaxial top layer.

The upcoming heterojunction technology is based on PECVDdeposition of an amorphous p-type silicon layer on a typically n-type monocrystalline Si substrate. Utilization of an intrinsic(undoped) amorphous Si layer has been introduced by [66] withenhanced passivation as well as open circuit voltage and cellefficiency. HIT cells (Fig. 3.5) combine heterojunction and intrinsiclayers together into one cell. HIT cells are completed by a PECVDgrown intrinsic layer and amorphous n-type layer on the back side,which serves as a back surface field (BSF) to reduce surfacerecombination velocity [154]. Transparent conductive oxide layers(TCO), also PECVD deposited, cover the front and back side toenhance conductance to the electrode grid. The TCO layers haveexcellent antireflection properties (see Section 4.1). Heterojunc-tion technology advantageously requires only low temperatureprocesses, as no more than 300 8C is required throughout thewhole process. They also have low temperature coefficients ofvoltage, which perform well at high ambient temperatures. Bifacialcell design is possible, which reduces stress on the wafer andallows for bifacial modules. Disadvantages introduced throughtemperature processing involve very careful surface preparationand low temperature contacting, which today requires highersilver content [146]. Current research is investigating coppersubstitutes for silver in contact printing pastes.

As indicated before and as already discussed for HIT cells,antireflection coatings are essential for efficient solar cells.Monolayer and double layer antireflection coatings on the basisof quarter wavelength coatings to reduce reflectivity only work in alimited bandwidth region. The task of the top layer is to increaseconductivity to fingers, reduce reflectivity, and reduce surface

as it is a source of additional defects. Passivation of PECdeposition of amorphous Si and SiO2 has been successful for rside passivation. The double layer antireflection coating SSiNx is manufactured by first oxidizing the Si surface, then usinPECVD coating of SiNx. The quality of SiNx is better if depositedhigh frequency PECVD. The final coating with SiNx introduproblems for the contacting step, as it hinders etching for

local contacts. SiNx contains hydrogen from the precursor

The hydrogen can be released during firing and diffuse intobulk, thus passivating recombination sites in the bulk. Futresearch involves seeking AR coatings with less than

reflectivity. Today, screen printed contacts cannot be fithrough those AR coatings.

The final interface to be manufactured is the contact. In mcases, the contact is manufactured by screen printing pastes

are fired to create a metallization layer. The pastes contain orgacompounds and a glass frit (small shattered glass particles) systhat enables contacting on the Si, which allows the printed

created after firing to be porous and limits conductivity. Challeninclude low contact resistivity, achieving a high aspect ratioreduce the shaded region, and creating high electrode conductty. Choosing a material presents another challenge, as the pfiring in low temperature processes requires a high silver contwhich is rare and expensive [36].

A two-step process from [36] applies a seed layer through

line printing. In a second step, this printed line is reinforcedelectroplating of highly conductive and dense metal to achievedesired conductivity in the contact. The seed’s main tasks arallow low contact resistance and to form a very thin contact wconducting electricity for the plating process. The utilizationplated copper is possible if nickel, a diffusion inhibitor, is used

seed.Fine line printing by contact formation was originally made

structured photolithography, which, on an industrial scaleexpensive and error prone. On the industrial scale, fine linesrealized by fine line screen printing or thick film stencil printfollowed by a firing process. Besides printing, electroless niplating and high-rate inline evaporation processes of differ

Fig. 3.5. HIT cell with TCO and BSF [154].

Fig. 3.6. Simulated relative reflectance for antireflection coatings.

uiretion

tualtionrial

is aried. Intive

recombination by contact formation. These requirements create atradeoff between the different material properties.

Fig. 3.6 shows the results of a PC1D simulation, which is used tosimulate new device performance. The effectiveness of differentantireflection coatings on SiN, SiO2, SiN/SiO2, and SiN/MgF isdemonstrated in the figure. The superiority of double layercoatings in comparison to single layer coatings can easily be seen.Adjustment of the optimal thickness of both layers can be used toachieve broad band absorption. SiNx has the lowest reflectivity, butonly within a small bandwidth. SiO2 can be easily manufactured byfiring in an oxidizing atmosphere, but it needs a high temperature

metals may be used for manufacturing contacts, but both reqan additional step for structuring, which is done by laser ablaor ink jet masking [8].

For electroplating of the second step, which is the acelectrode, the deposition rate is of major importance. Deposirates that are too high create dendrite growth and porous mate(Fig. 3.7), which limits the production speed. Shown in Fig. 3.8cell architecture, where the contact with high aspect ratio is buwithin a groove made by laser ablation after SiO2 passivationthe grooves, deep diffusion is utilized to generate a selecemitter, where SiO2 is used as a mask.

Ationsubsprindepeexamtionalignsubsbakipoly

Rveryvery

Fig. 3elect


manufacturing technology that is gaining significant atten- is roll-to-roll processing. This process involves a long flexibletrate that moves over rollers and into machines that etch,

[16,18,79]. Roll-to-roll fabrication is geared toward cells that arefabricated on flexible substrates because they can easily be pulledaround rollers and moved with a lower chance of breaking.Microcrystalline and amorphous silicon are both materials used inroll-to-roll cells because they are very thin, making the cost lower,they are much more flexible than crystalline silicon wafers, andthey still contain the benefits of using silicon, such as highabsorption [16]. Also, roll-to-roll techniques that do not use anyrare materials exist, such as the one presented in [144].

It is not possible for grown cells to be exactly identical to eachother. However, this is possible for roll-to-roll fabricationtechniques, since the process induced errors are minimized viacomputer control. Thus, cells created through roll-to-roll opera-tions are almost always identical. This allows for easier reliabilitytesting, as the test results on a few cells will almost certainly bereplicated for all the cells.

4. Efficiency drivers: minimizing defects and associatedrecombination losses

The first step toward making solar cells more viable energyproducers is by improving the knowledge and technology associatedwith increasing efficiency while keeping costs down, thus masteringthe manufacturing of surfaces on the basis of understanding thephysics. This efficiency section gives an overview of the previous andcurrent antireflection methods used to improve efficiency in solarcells, and then describes three new technologies–plasmonics,negative index metamaterials, and multijunctions – that can bevery helpful in pursuing more efficient and cost effective cells.

4.1. Antireflection

The purpose of antireflection techniques is to trap more lightthan the cell originally would so that more electricity can begenerated. The surface of the solar cell generally reflects some ofthe light that hits it, eliminating some electricity that could beproduced. Etching or changing the material of the surface can allowmore light to be trapped in the cell. Most of these light trappingmethods are used for crystalline silicon cells, but there is ongoingresearch on antireflection technology in thin film cells.

4.1.1. Transparent conductive oxide coatings (TCO’s)

Next to the photon-absorbing material that creates voltagebased on the photoelectric effect, transparent conducting coatings(TCO’s) play an important role in certain kinds of solar cells. For thisreason they are separately discussed in this section. Dominantmarkets for TCO’s are in architectural applications and flat-paneldisplays [104]. Next to these, TCO’s are essential for PVapplications, as they can simultaneously act as antireflectioncoatings or as conductive electrodes that cover a solar cell.

The application of transparent conducting coatings in solar cellsis dominated by just a few materials–doped tin oxide, indiumoxide, and zinc oxide. These are wide bandgap, insulatingmaterials–and therefore transparent for visible light – that becomeconductive by natural impurities, and especially by doping thatgives them conductive properties. Most commonly known is tin-doped indium oxide (ITO), which can be conveniently produced atmoderate temperatures. Next to this, fluorine-doped tin oxide andAl-doped ZnO are most common. Furthermore, CdO, Cd2SnO4,

.7. Electrical contacts after different 2-step metallization and growth rates for

roplating [36].

Fig. 3.8. LGBC cell with contacts buried in laser groove [8].

t, dispense solvent, compress, and do many other functionsnding on the type of solar cell and how it is being made. Oneple of this is the ESSENCIAL process, which allows evapora-

of solvent through surface encapsulation and inducedment. The process involves a solvent dispensed on thetrate, the active layer pressed onto the solvent, a thermalng time, and then electrode deposition. This process created amer cell with 3.5% power conversion efficiency [108].oll-to-roll manufacturing allows cells to be fabricated with a

high throughput and a small number of workers. It can also easily be customized to the type of cell that is needed

MgIn2O4, CdSb2O6:Y and GaInO3 are examples of materials underinvestigation that exhibit these properties as stated by [48].

In order to illustrate some basic concepts of the transparent andconductive properties of these materials, a model that is roughlyvalid for the three major coating types mentioned is given. Theircomplex electric permittivity e(l) as a function of the wavelength lcan be approximated according to [52,70] as:

eðlÞ ¼ nðlÞ � ikðlÞð Þ2 ¼ e1 �ilag

ffiffiffiffiffiffi

e1p

2p� ile0rðlÞ2pc0

(1)

the

the the

hascialibletingext

ngs.ith

iredcost

ogyhisall

.1).thethe

rossogy

all cell

ichall

d orthencethyl

onidalanygies

with

in.


Here, n and k are real and complex parts of the refractive index,respectively, e1 is the limit of the electric permittivity, ag is thebandgap absorption coefficient, c0 is the speed of light, and r(l) isthe complex conductivity. Eq. (1) gives the relevant terms for thevisible/NIR optical region and the conductivity. Their relevance canbe summarized as follows:

� e1: This is the valence electron contribution that determines therefractive index in the visible region. It is a material constantdependent on the crystal structure, and slightly dependent on asomewhat smaller density when prepared as a coating. Its typicalvalue is 4 for TCO’s, giving a refractive index of about 2.� ag: This is the bandgap absorption coefficient that accounts for

the photon’s absorption near the bandgap energy of 4 eV,corresponding to l = 320 nm. When the material is doped, thisbandgap may shift to 4.5 eV (l = 285 nm) due to the Burstein-Moss shift [109]. As there is hardly any energy in the solarspectrum in this region, this term will not be discussed further.� r(l): This is the complex conductivity that makes the coating

conductive from an electronic point of view and causes atransition from transparent to reflective near the plasmawavelength lp to which it is related according to the Drude-theory [103] as following:

rðlÞ ¼ ð2pc0Þ2g þ ið1=2pc0lÞe0 21l p

(2)

The real part of r(l) for l = 21 is the DC-resistivity that can alsobe defined as r = (n_ � e � m), with n_ being the electron density,e the elementary charge and m the electrical mobility. g is arelaxation frequency that is related to the mobility m by:

g ¼ e

meffm(3)

Here e is the elementary charge, and meff is the effectiveelectron mass, in the 0.2–0.3 me range where me is the electronmass, for TCO materials [48,103]. For controlling conductive andtransparent properties, the doping concentration determines thefree electron density, and the other process dependent parameter,the mobility m, is determined by various scattering mechanismsand a proper crystal structure of the coating.

For having optimum conductive properties, both the dopingconcentration and mobility should be as high as possible. However,if doping is increased, the plasma frequency may become close tothe visible spectrum, and the coating might absorb solar radiationinstead of transmitting it into silicon. Once the doping concentra-tion is fixed, the mobility would ideally be as high as possible. It hasbeen shown by [110] that in optimum conditions, ionized impurityscattering is the mechanism that limits conductivity with atemperature dependent scattering by acoustic phonons as anotherimportant factor. Incomplete crystallization and grain boundariesmay further reduce the mobility, especially at low electrondensities where mobility is more sensitive to grain boundariesdue to a lower Fermi-energy [25]. Further, the layer conductance isproportional to coating thickness, so the coating should be as thickas possible for the electrical properties.

To obtain optimum antireflective properties at normal irradi-ance, the coating should have a refractive index of HnSi � 2.02,

doping concentration that keeps the plasma frequency outsidesolar spectrum.

Once the process parameters are optimized for giving

proper doping concentration and desired film thickness, andmobility approaches its theoretical maximum, the optimal TCObeen reached; as is the case for coatings available from commermanufacturers. Further optimization for solar cells is only possby either modifying the upper layer to increase its antireflecproperties, especially for non-normal irradiance [53] (see the nsection), or combining these TCO coatings with other coatiAlternative TCO materials are currently being investigated whigher maximum electron mobility being the most desproperty. Improvements are still necessary, as the future

and availability of indium is unknown.

4.1.2. Antireflective materials and surface texturing

Black silicon is a state of the art antireflection technolcurrently being used to reduce light reflection in solar cells. Tprocess uses a coating that is catalyzed on the cell to create a smlayer of gradually changing index of refraction (Fig. 4.1According to [45], reflection can be strongly suppressed if

nanostructure is comprised of structures smaller than

wavelength of incident light and the Si density is graded aca thickness more than half the wavelength of light. This technolhelps to increase light absorption. This is because very smchanges in refractive indices keep light moving further into theinstead of being reflected away [9].

A process used extensively today is pyramidal texturing, whetches away material on top of the solar cell to create very smpyramids. When a photon hits these pyramids, it is either absorbereflected to another pyramid, creating another possibility for

photon to be absorbed. This process effectively doubles the chathat a photon can be absorbed. [57] uses TMAH, or tetrameammonium hydroxide, to create reliable pyramidal structuresvarious silicon surfaces. This technique of creating pyramtexturing achieved 13% reflectance without the use of

antireflective coatings such as black silicon. Combining technolo

Fig. 4.1.1. Cross-sectional SEM images of multi-scaled black silicon etching,

the white bar representing 500 nm. This etch was achieved by [9] after 1.5 m

onD).tchingved

cesthatsedtterith

with the silicon refractive index nSi approximately equal to 4.09,and a thickness t of a quarter wavelength at the solar maximum atl = 550 nm; making t = 80 nm. The refractive index of TCO’s, about1.9 at l = 550 nm, is close to the ideal value for silicon. In practice, athickness t = 80 nm is often too thin to obtain reasonableconductivity, and used coatings are in the range of 300–1000 nm. However, coatings in this thickness range have beenprepared [121] and, in these cases, the TCO still forms a reasonableantireflective coating with a refractive index of approximately 2[32]. For minimum solar absorbance in the coating it remainsimportant to have a maximum electron mobility combined with a

creates reflectance of 2–3%. [93] utilize low density SiO2 depositedsilicon throughplasmaenhanced chemical vapor deposition (PECVThe SiO2 acts as a mask that solutions such as TMAH or KOH earound and create inverted pyramids on the silicon, allowincoming light to achieve 3–4 reflections. This method achie14% reflectance, similar to that of the TMAH chemical etching.

Combining black silicon and pyramidal texturing enhanabsorption of the solar cell even more. In 2011, [46] found

combining black silicon with pyramidal texturing greatly decreathe amount of blue light recombination as well as creating even beantireflection. The combination allows light that is not absorbed w

blacachisilicthe

Oeye

showeyesrefracan

surfcurrto crsubsor chcoupbecoviab

ASiO2

nanprovmandepoincr

Fig. 4micro

200 n

Fig. 4antir

struc


k silicon to be reflected by pyramids, or vice versa. This celleved 17.1% efficiency using 100 nm compared to 250 nm thickon wafers. Two weaknesses of this process are passivation anddifficulty with large scale application to solar cells.ther antireflection techniques being researched include moth-texturing and spiral light trapping. Moth-eye texturing, asn in Fig. 4.1.2, uses the idea of very narrow pillars seen in moth

, which may be arranged to have a similar gradually changingctive index effect on light as black silicon. Moth-eye texturinggreatly reduce reflection, and new techniques for etching a

ace could create a cheap method of production. However, it isently a costly process, pillars with high aspect ratios are difficulteate, and, according to [29], aligning pillars to fit on thin filmtrates is a difficult process. The second process – known as spiraliral etching – creates a fractal pattern that polarizes light andles it to the cell. This method is currently experimental, but mayme practical as cheaper nanoimprinting techniques that arele for thin films are discovered.

different approach to antireflection involves growing thinon triacetate polymers over bump- and sponge- like

ostructures. The SiO2 theoretically boosts performance whileiding a protective layer to the nanostructures. The perfor-ce of bump-like nanostructures was less than ideal, but thesited SiO2 film decreased reflectance as its thickness

eased, seen in Fig. 4.1.3. The sponge-like nanostructures by

themselves suppressed reflection to 1%, but adding the SiO2 layerdid not improve performance. Abrasion resistance of the SiO2 layerwas not changed in the bump structures, but it was improved inthe sponge-like nanostructures [91]. This is a cheap method ofimproving antireflection in polymer cells, but more research mustbe done to discover more polymer cell antireflection methods.

A method that has contributed to efficiency improvement of a thinfilm a-Si solar cell is using a large grain size SnO2 film instead of theearly ITO film. The SnO2 film has a rough surface texture, which iseffective in trapping light in the solar cell. Fig. 4.1.4 shows a schematicof light trapping by the SnO2 film. Fig. 4.1.5(a) shows an SEM imageof the SnO2 surface deposited on a glass plate by CVD. Fig. 4.1.5(b)shows a transmission electron microscopy (TEM) image of theinterface of the SnO2 layer and the a-Si layer fabricated on the SnO2

layer by capacitively-coupled RF glow discharge decomposition.

.1.2. (a) SEM image of an example of moth-eye texturing, with the white bar 1

meter. (b) Example of a silicon moth-eye array, with both d and h equaling

m [29].

Fig. 4.1.4. A schematic of light trapping by the SnO2 film [117].

.1.3. The average reflection of bump like nanostructures for cells without any

eflection method, cells with bump structures, and cells with the bump

ture and various SiO2 thicknesses. SEM images are also given [91]. Fig. 4.1.5. Images of the SnO2 surface and a-Si/SnO2 interface [111].

ells,tor.

ve a be

ake

ar-

ells

ating

ons,

170].

film


Polygonal-shaped textures formed on the base shapes of the SnO2

grain can be observed at the interface between the two layers,changing the directions of the light rays [111,117].

There are multiple techniques toward achieving very goodantireflection, as described above and in [63], but many also havesignificant weaknesses such as material waste, high cost, orcomplexity. These antireflection techniques are geared towardassisting light trapping in crystalline silicon solar cells. Technolo-gies with the potential to reduce reflectivity are currently an activearea of research and one of the most important approaches forimproving efficiency.

4.2. Plasmonics

Thin film solar cells are capable of being the dominant solar cell inthe future because they do not consume high quantities of siliconand will continue to get cheaper. Plasmonics presents a solution toone of the major problems with thin films – light trapping.Plasmonics are the collective oscillation of free electrons between ametal and a dielectric which create a resonant frequency.

A plasmon can be caused by the interaction of a metallic surface(non-localized) or metallic nanoparticles (localized) with light asshown in Fig. 4.2.1. This can also be described as the interaction ofcharges (conduction electrons) with electromagnetic waves [12].This oscillation, or ‘‘wave’’, that is created has a wavelength, andwhen the wavelength of incoming light converges to thewavelength of the plasmon, more photons are absorbed. Plasmonsact as miniature dipoles that have free charge carriers; as theradiation is scattered, the carriers pick up charge and carry it tomodes in the semiconductor beneath the nanoparticles. Fig. 4.2.2shows this ‘‘wave’’ as the oscillation of nanoparticles due to chargeinteraction. The surface plasmon wave depends on the size, shape,and dielectric properties of the medium. The ideal plasmonwaveguide does not have nodes in the semiconductor. Thesemiconductor must then be very thin, making thin film solarcells an ideal application for plasmonics [6,27,170]. Also, a thinnercell creates a greater voltage and less recombination, which alsobodes well for thin films.

permittivity, which is a better insulator and absorber. In solar cthe higher permittivity dielectric must be the semiconducSome benefits of using this first method include:

� Scattering light at the surface of the cell allows the cell to hahigher optical path length, allowing more time for light toabsorbed.� Repeated scattering by the nanoparticles allows light to m

several passes through the absorbing (active) layer.� The need for antireflection methods is eliminated, as nanop

ticles greatly reduce electron hole recombination [6,98].

The second method of incorporating nanoparticles in solar c

Fig. 4.2.2. Depiction of the physics behind a surface plasmon wave for propag

(a) and localized (b) waves. Solar cell research involve localized surface plasm

which use nanoparticles much smaller than the incident wavelength of light [

Fig. 4.2.3. Example of the first method of incorporating nanoparticles onto thin

solar cells. ITO = Indium Tin Oxide [6].

llerongthatthe

owsctlyciesctor

thehis

Fig. 4.2.1. Analogy of a plasmon. Inherently positive metallic nanoparticles interact

with electron clouds, creating an oscillation that produces an electromagnetic wave

[12].

Three methods are possible for incorporating plasmonicnanoparticles into solar cells. The first involves placing thenanoparticles on the surface of the cell (Fig. 4.2.3). Thesenanoparticles must be relatively larger, because they act as sub-wavelength scattering elements. The light scatters normallyaround the nanoparticles, and when the nanoparticles are closeto dielectric media, they scatter toward the dielectric with higher

involves placing them in the active semiconductor. Smananoparticles are used in this case because of their very strelectric fields that attract electrons. This property constitutes

the semiconductor must have a bandgap corresponding to

wavelength of light the nanoparticles attract. This method allsolar cells to be very thin, as the nanoparticles must direinteract with light. However, these cells have lower efficienbecause of the small bandgap being used, and the semiconducannot absorb all of the excited electrons.

The third and final method involves placing nanoparticles onmetallic back contact of the solar cell and is shown in Fig. 4.2.4. T

arrawhetravby tcontabsophothis

semcontwaveffec

Fnoblvisibscatwill

defeosciabsolight

Fig. 4micro

SEM

subst

Fig.

wave

(blue

energ


ngement creates surface plasmons on the back of the cell that,n excited, become polaritons, or plasmon waves. These wavesel parallel to the back contact until they dissipate or are absorbedhe semiconductor. Because the waves run parallel to the backact, there is a greater opportunity for the photons to berbed. Also, the metallic back contact can reflect dissipating

tons back into the cell for a chance to be re-absorbed. Typically instructure, green and blue light is already absorbed by the

iconductor, but red light is not. Therefore, plasmonic rearacts focus on red light so as to achieve better absorption at thoseelengths. This arrangement does not create any shadowingts, which may occur in the former two methods [27,43,84].or photovoltaics, the metals best suited for nanoparticles aree gas compounds because they resonate in the infrared andle frequencies. The size of these particles determines the

tering and absorption seen in the cell. A smaller particle (<60 nm)absorb more, but a particle that is too small will act as a pointct in the silicon. A large particle (>300 nm) will create multiplellations across the nanoparticle, thus reducing scattering andrption [43]. Also, particle size determines what wavelength of will be absorbed, as depicted in Fig. 4.2.5. Increasing

nanoparticle size changes the absorbed wavelength of light fromblue (700 nm) to red (400 nm).

Two problems with nanoparticles involve their fabrication andimplementation in solar cells. One method of creating nanopar-ticles is by using laser ablation. This process utilizes extremelyshort laser bursts of the order of pico- or femtoseconds to shape thenanoparticles out of sheets of metal [34]. This technique providesthe possibility of generating various sizes of defect-free nanopar-ticles [149]. A method of implanting the nanoparticles involves a 3-D ultra-low energy ion implantation (ULE-II). In the ULE-II process,thin silicon nitride membranes are fabricated onto silicon wafersby a soft lithography method. This ‘‘mask’’ controls where thenanoparticles are placed in the wafer. The amount of chargeapplied to the ions dictates the depth where ions will be implanted,and the mask controls whether they are implanted or placed on thesurface of the cell [37].

Another important property of plasmonics is that they have avery strong electromagnetic near field. This capability allows aplasmon to individually excite an electron so that it releases aphoton. However, the near field only reaches a distance of a fewtens of nanometers, so the solar cell must be compact, similar tothe second method of incorporating plasmonics discussed earlier.However, this property is not fully understood and requires testingto determine its capabilities. Overall, plasmonics is an effectivemethod for improving the surfaces and interfaces of a cell to createa more efficient solar cell.

4.3. Negative-index metamaterials

The negative index metamaterial (NIM) is an artificial materialthat demonstrates properties not seen in nature. It refracts light onthe same side of the normal to the surface (Fig. 4.3.1). Whenapplied to solar cells, more light could be absorbed rather thanreflected by the solar cell. Other properties unique to NIMs includebackwards Cerenkov radiation (the electromagnetic radiationproduced when an electron passes through a dielectric mediumfaster than the phase velocity of light), a reversed Doppler shift,and sub-wavelength diffraction [61]. Accordingly, Snell’s Lawpredicts that light rays will be refracted on the same side of thenormal for this class of materials. NIMs can be used on the surfacesof solar cells to improve light absorption and direct light towardthe center of the cell. This section will discuss various NIMstructures and their possible uses.

The NIM device proposed in [39] is an improvement to the

.2.4. Ag nanoparticles are formed on the ZnO:Al/Ag back contact within a

crystalline silicon (mc-Si:H) solar cell (a) Cross section of the entire cell. (b)

image of nanoparticles on a back contact in relation to the location of the

rate [43].

Fig. 4.3.1. Difference between positive and negative refractive index [1].

4.2.5. Schematic showing which size of nanoparticles absorbs certain

lengths of light. Smaller nanoparticles couple light with higher energy

) and thus, a higher bandgap, while larger nanoparticles couple red light (lower

y and bandgap) [105].

‘‘fishnet’’ structure, which, as seen in Fig. 4.3.2, is a mixture of across and a circle that works for arbitrary linear polarizations. The‘‘fishnet’’ can be seen to the far left of the picture as many of thecross-circles in an array. The current density is very high on thestructure, creating a high magnetic resonance. This structureachieves a large NIM bandwidth of 2.9 GHz with a negativerefractive index around �1. The structure was fabricated using asimple etching technique, and it created 20 � 20 unit cells withtotal dimensions of 10 cm � 10 cm. There are other experimentsthat vary the fishnet structure to achieve better absorption on theedges of the visible or infrared spectrum or to achieve stronger

orbrayate

ates �9e to

ayely

tivetiveore

ightl is94].olarhenriodgesof a.tionrnalusener

olarrgy.me

erals toPV

C:H. a-igh]. Alineousousconherc-Sie to


negative indexes [59,62,128]. This specific fishnet structure on asolar cell could potentially improve absorption of infrared and partof the visible light spectrum.

A possible NIM structure constructed by [122] creates a dual-band negative refractive index. This is achieved by magneticresonances of two frequencies introduced via a metallic ring and anembodied round plate (Fig. 4.3.3). If the dimensions of themetamaterial are in a precise range, both an electric and amagnetic field are created. The top of the structure achieves abandwidth of 11.1 GHz while the bottom is at 14.6 GHz. The cell isvery sensitive to geometry, as the refractive index becomespositive if the dimensions are slightly mismatched.

Other experiments [26,30] use an NIM structure to absvisible rather than microwave and infrared light. A hexagonal arof subwavelength coaxial waveguide structures is used to crethe NIM of Fig. 4.3.4. Out of the two waveguide modes, one crea positive refractive index while the other creates an index fromto 1 over 1.5 to 3.0 eV. This structure creates an NIM insensitivincidence angles (�50 to 50 degrees) and polarization, and malso be used with visible frequencies. NIM structures are extremsensitive to geometry, as a small change will cause the refracindex to become positive. However, with appropriate proteclayers, these structures can help thin film cells achieve mefficient light trapping and higher efficiencies.

4.4. Multijunctions

A major problem of the single junction a-Si:H solar cell is linduced degradation of material properties when the celexposed to illumination due to the Staebler-Wronski effect [This causes a decrease of energy conversion efficiency for the scell during the initial stage of operation. The solar cell tstabilizes at a lower efficiency after the initial degradation peof one to three months. The stabilized efficiency typically ranfrom 70% to 90% of initial efficiency. The stabilized efficiency

commercial single junction a-Si:H solar cell is typically 6–7%A possible solution to the problem of light induced degrada

is to use a thinner intrinsic layer for generating a higher inteelectric field and lower sensitivity to distortions, which cadegradation of material properties. On the other hand, the thinintrinsic layer will cause a reduction of light absorption in the scell, resulting in a decrease in the amount of absorbed eneAlthough the decreased efficiency can be compensated to soextent by the multiple p-i-n structures composed of sevamorphous silicon layers shown in Fig. 4.4.0, a better solution iemploy the heterodyne junction, in which two different

materials are used to form the p-i or i-n junction [112].

An early attempt at making a heterodyne cell was to use a-Sias the window layer instead of a-Si:H, as shown in Fig. 4.4.1SiC:H has the advantages of good light transmittance and hbuilt-in potential, which is suited for the window layer [113more effective technology combines a-Si:H with microcrystalSi (mc-Si) in the heterodyne junction. mc-Si is a form of porsilicon with an amorphous phase that is similar to amorphsilicon. mc-Si differs from a-Si by small grains of crystalline siliwithin the amorphous phase. Compared with a-Si, mc-Si has higelectron mobility due to the presence of silicon crystallites. malso has a higher absorption coefficient than single-crystal Si du

Fig. 4.3.2. A proposed negative index metamaterial that is a fishnet structure. The

dimensions are t = 790 mm, s = 17.5 mm, w = 1.5 mm, R = 2 mm, a (period) = 5 mm

[39].

Fig. 4.3.3. The current densities, strengths, and directions in the metallic ring and

embodied plate negative index structure [122].

Fig. 4.4.0. A multi-layered a-Si solar cell [112].

Fig. 4.3.4. A single layer NIM slab that consists of a close-packed hexagonal array of

Ag/GaP/Ag coaxial waveguides. r2 = 75 nm, r1 = 70 nm, and p = 165 nm [26]. Fig. 4.4.1. a-SiC:H/a-Si:H heterodyne junction solar cell [113,143].

the

highwhiwavschewhethatnecewithconv

HcomhetejuncSi/mhetebetwas il28.9

Sin tsembandenersingbandHowthe

cal ltop

usin


presence of amorphous components. More important, it has a absorption coefficient in the red and infrared wavelengths,

ch can compensate the low absorption coefficient of a-Si in thiselength range, as shown in Fig. 4.4.2. Fig. 4.4.3 shows amatic of a heterodyne junction composed of a-Si and mc-Si,re mc-Si acts as the light absorbing material. It should be noted

a relatively thick layer of mc-Si, on the order of 1.5–2 mm isssary for absorbing a sufficient amount of photons. A solar cell a single heterodyne junction of a-Si/mc-Si can reach aersion efficiency of 9–10%.

igher energy conversion efficiencies can be reached bybining the multi-layer structure shown in Fig. 4.4.0 with therodyne junction shown in Fig. 4.4.3, which is called the multi-tion or tandem structure. Fig. 4.4.4 shows a schematic of an a-c-Si tandem structure [136]. In addition to the a-Si/mc-Sirodyne junction, a TCO intermediate layer can be placedeen the top and bottom cells for trapping light in the structure

lustrated in Fig. 4.4.5. A conversion efficiency of higher than% has been reported with this type of solar cell.ingle bandgap solar cells have two power-loss mechanisms

In spite of the technologies shown, the best achievable efficiencyof commercial a-Si thin film solar cells was less than 10% (untilrecent improvements), which is too low for residential solar powersystems. The main approach to further improve conversionefficiency is the development of a triple junction tandem a-Si thinfilm solar cell. a-Si, mc-Si, and mc-Si/Ge are typical materialcombinations in triple junction structures. Fig. 4.4.6 shows thecollection efficiency (ratio of electron carriers received to totalavailable carriers) of a laboratory triple junction tandem solar cell,

Fig. 4.4.2. Bandwidths of a-Si and mc-Si [136].

Fig. 4.4.3. A heterodyne junction composed of a-Si and mc-Si [136].

Fig. 4.4.4. A tandem a-Si/mc-Si solar cell [143].

Fig. 4.4.5. Schematic of the interlayer [136].

Fig. 4.4.6. Quantum (collection) efficiency (QE) spectrum of a triplejunction tandem

solar cell [143].

hat they cannot absorb photons with energy below theiconductor’s bandgap, and photons with energy above thegap of the cell turns into heat. Almost half the incident light

gy is lost because of this. Most solar cells are in the category ofle bandgap cells, and they overcome this problem by using lowgap semiconductors and various antireflection methods.ever, exploiting more than one energy level (bandgap) avoids

Shockley-Quessier limit of 33.7% efficiency [65]. The theoreti-imit using multijunction cells, which consist of cells stacked onof one another, is 67% without concentrated light, or 86.8%g concentrated light.

ingl 4-d 6

orld 4-

ibleafertwosedareemers,ent

lt tolowandsivethe

atorum

areonsigh- byand

inenedereew

oflve

ightar-

be


which achieved a 16% energy conversion efficiency [106,143]. Such amultijunction cell works by using cells layered on top of each otherto capture the maximum possible light. The top sub-cell capturesblue light, and the next sub-cell captures a slightly lower energy lightthat passes through the first sub-cell because its energy is lower thanthe top sub-cell bandgap. This process could go on for many morecells, although as more sub-cells are added to a multijunction, eachincrease in efficiency provided by the new sub-cell will be less thanthe efficiency bump gained by the previous sub-cell. Thus, aneconomic limit exists for the number of sub-cells that have optimalefficiency compared to cost.

One method of creating a multijunction cell is explained by [114].A 2.0 eV silicon top layer and an intermediate layer were used, whichgreatly improved many properties of the cell, such as light trappingwithin the multijunction. By further optimizing the bandgap andthickness combinations used in the cell, Si1�xCx:H can be considereda possible top layer to a dual or triple junction solar cell. Otherexperiments, such as [55], have found the intermediate layer canalso be applied to organic tandem cells, which use conductiveorganic polymers or molecules for light absorption and chargetransport. Organic cells are receiving attention because they consistentirely of environmentally friendly materials and they have uniqueproperties. These unique properties include flexibility, transparencyfor use on windows in tall buildings, and they can be made fromsolutions coated on glass, allowing easy production.

An example of a triple-junction cell is shown in Fig. 4.4.7. This cellconsists of three sub-cells: 1.87 eV AlGaAs/1.65 eV AlGaAs/1.42 eVGaAs, and it achieves an efficiency of over 30%, which is similar tothat of the most efficient dual-junction cell. The triple-junction alsohas 2.25 times less power loss than the highest efficiency triple-junction, which allows higher voltages and generates less heat. Lighteasily travels from one sub-cell to another without scatteringbecause of the AlxGa1�xAs family of semiconductors. Thesesemiconductors have bandgaps ranging from 1.87 to 1.42 eV,antireflection coatings can easily be applied, and they can be grownon GaAs wafers using metal organic vapor phase epitaxy (MOVPE)without any extra steps. This cell is a viable option to becommercially made because it is much cheaper than a maximumefficiency cell, it has low losses, and it is fairly simple to make [139].

Higher junction (4, 5, or 6 layer) cells are currently beexplored to create the highest efficiencies possible. The ideajunction cell can theoretically reach 59% efficiency while 5 anjunction cells can reach over 60%. However, due to real weffects like shadowing and series resistance, the maximumjunction cell efficiency is 47% [23].

Wafer-bonding and layer transfer processes are possmethods of connecting multijunctions. The method of wbonding is described in Fig. 4.4.8. Wafer bonding requires the

materials being bonded to be atomically flat. They are then prestogether and heated, and, because of their planarity, atoms

allowed to diffuse between the materials, thus bonding thtogether [47]. Current easily passes through wafer bonded laywhich increases efficiency and lowers heat-generating currresistance. The created interfaces are seamless and very difficubreak, but the process is expensive. Layer transfer processes alcrystalline films to be grown on a typically expensive substrate

then moved to the multijunction. This allows the expensubstrate to be used again instead of trying to build

multijunction on the substrate.

The most applicable area for multijunctions is in concentrsolar plants, where there are not many clouds and the maximabsorbance of light is critical. If material and production costslowered, multijunctions can be used in a variety of applicatibecause of their high efficiencies. Commercialization of such hefficiency tandem a-Si thin film solar cells is expectedimproving material properties, film qualities, and high-rate

low-cost deposition technologies.

4.5. Combinations

Opportunities are available for creating solar cells that combthe new technologies described above. An example explaiearlier is combining black silicon with pyramidal texturing. This a greater capability of expanding this combination idea with ntechnologies presenting themselves as potential enhancersefficiency. The most feasible combinations typically invomultijunctions. Plasmonics may be used in tandem cells as ltrapping devices to improve absorption and efficiency. Nanopticles capable of trapping certain wavelengths of light can

Fig. 4.4.8. The process of wafer bonding [47].

ulds ofuldith

lowdex toallyolar

ithFig. 4.4.7. The various layers of a triple-junction cell [139].

inserted in each sub-cell. The appropriate particle sizes woassist each sub-cell in absorbing their respective wavelengthlight [100]. Antireflection and negative index metamaterials coalso be used in multijunctions instead of plasmonics to assist wlight trapping.

Consumers generally dismiss solar cells because of the

efficiency combined with high costs. Plasmonics, negative inmetamaterials, and multijunctions are excellent methodsimprove efficiencies while avoiding expensive materials typicused in solar cells. New technologies that can assist the sindustry will continue to emerge. They will create better cells w

highare

5. R

Ring

subjwellissuand

mayto realsoduriexamDegbecaenercell

cell

morintetestisolu

5.1.

FEner

1. Lo2. P

C3. A4. Fa

en5. Le6. O7. D

Trefefailu(betTherinstaconnmodslighinteabsodaminto

Fig. 5appa

modu


er productivity and lower cost. Other issues to be addressedhigh reliability and high life expectancy.

eliability

eliability and thus longevity are key parameters in determin-break-even characteristics of PV modules. As solar panels areect to substantial thermal changes and because PV modules as

as the cells consist of multiple layers and interfaces, reliabilityes need to be traced back to the manufacturing of the surfacesinterfaces. Cracks already initiated by the wafering process

lead to catastrophic failure if the subsequent etching step failsmove them. All the designed and manufactured interfaces are

prone to spallation and delamination from thermal cyclingng actual operations. Reliability testing of solar modules

ines the behavior of modules in environmental conditions.radation of system performance cannot be entirely eliminated,use heat generation at the cell level is a by-product of photonicgy conversion. This type of degradation occurs mostly in solarinterfaces, which leads to decreased performance and a shorterlife. For cells to become more reliable, the interfaces must bee robust against thermal cycling. To create a more reliablerface, the knowledge of failure modes, gained from rigorousng and modeling, is needed. This will enable innovativetions to enhance cell life and prevent catastrophic failure.

Failure modes

ollowing [42], the criteria that the U.S. National Renewablegy Laboratory (NREL) uses to define cell failure are:

ss of 50% of initial power output (x-Si).ower output less than 50% of the manufacturer’s rating (a-Si,dTe, CIGS).rcing in module circuitry or junction box.ilure of dielectric to withstand insulation resistance tests atd of test segment.akage current greater than 50 mA during biased DH exposure.

pen-circuit fault during forward-biased TC.evelopment of major visual defects.

hese guidelines give researchers and maintenance workersrences to determine how well a cell is performing. Many of theres are caused by overheating or large thermal variationsween night and day) mixed with high humidity (Fig. 5.1) [71].mal cycling and excess mechanical movement caused byllation, transportation, or severe weather can sever electricalections to the cell, junction box, and leads coming out of theule. Moisture seepage caused by rain and dew can get into anyt cracks, percolate into the cells, and cause delamination of

rfaces, which decreases light absorption, increases moisturerption, and critically damages the cell. In thin films, constantp heat increases series resistance due to the ingress of moisture

a cell [169]. Silicon cracking or fracturing can lead to severed

electrical connections or hot spots (places of high resistance,creating large amounts of heat) [69]. Multijunctions used inconcentrator systems are exposed to large amounts of heat andUV radiation, which may decrease the lifetime of these cells. All ofthese cell failures can be avoided through the use of durable, heatand water resistant materials, and ensuring that precautions aretaken to protect these materials during production. Also, under-standing the flexibility of a material, how thin that material can beuntil it cracks and how to address a cell when it is cracked are areasthat must be further researched to determine how basic materialproperties affect cell performance.

5.1.1. Effect of stresses and cracks

A significant portion (30–40%) of solar cell cost is attributable tothe production of silicon wafers [102]. The mechanical integrity ofSi wafers is of considerable importance in the solar cell productionprocess because fracture during processing can significantly raisecosts [76]. With the use of thinner wafers to decrease materialcosts, wafer breakage due to fracture during handling andfabrication of solar cells is a major issue. Fundamentally, breakageof Si wafers is due to microcracks and residual stresses in the waferand applied stresses arising from wafer handling/processingoperations that exceed a critical value necessary for fracturefailure. Knowledge of the microcrack population, location andgeometry as well as the distribution of applied stresses is requiredto predict and/or prevent wafer breakage. Monte Carlo simulationwas employed in [161] to study the fracture strength of Si waferswith surface, edge, and bulk flaws. In [49] and [50], computationalfluid dynamics and the finite element method are used to analyzethe effects of wafer handling stress caused by a Bernoulli waferhandling device on the breakage of Si wafers. Also, the influence ofmicrocrack length and geometry on the fracture behavior of waferswas examined.

The effect of residual stresses and cracks on the mechanicalfailure strength of Si wafers and solar cells is typically character-ized via different types of bending tests, including four-linebending and bi-axial flexure [21,50,77,78]. Knowledge of thelimiting stress level for failure is useful in optimizing the solar cellmanufacturing process.

Residual stresses generated by crystallization, wafer slicing,and solar cell processing, especially tensile residual stresses, cancompromise the mechanical strength of the wafer or cell. A nearinfrared polariscope technique has been successfully applied to theanalysis of maximum principal stresses in thin silicon wafers [73–75]. The polariscope method determines the stress inducedbirefringence through measurement of transmitted light intensi-ties in which the birefringence is manifested in the form of fringes.Residual stress is obtained from the light intensity equations byapplying the stress optic law [73]. It has also been shown that thereis a correlation between the stress, photoluminescence andelectron-hole lifetime of the Si wafer [75].

5.2. Basis for measurements and testing protocol

There are many different ways in which solar cells can fail.Attempts have been made to determine the effect of a micro-defector micro-failure on overall cell life based on its magnitude andlocation. A root-cause analysis is carried out to identify theunderlying mechanism and its relation to macro-scale failure

.1. Infrared scan after 1000 dynamic loading cycles and 25 thermal cycles. It is

rent from the dark spots that large areas of the cell are disconnected from the

le [71].

phenomena. New generation solar cells then avoid commonproblems that previous cells have encountered.

There are two reliability tests that can be applied to all solarcells. These methods are P-spice electric circuit simulations andreal time degradation tests. The P-spice tests are very simplecomputer programs that simulate a current running through asolar cell design and determine if there are any kinks or inflectionpoints where there is current loss [155]. This can be an early signthat something is wrong with the cell. Another similar methoduses a heat sensor to find any places in the cell where there is nocurrent flow. Real time degradation tests use solar cells placed

tely at

k atorkver,Ds,ter.celltest

heirressund

ears is

(4)

(5)

(6)

on- themalous

apor

er to


directly under the sun to work in a natural environment. The effectof naturally occurring rain, hail, changing temperatures, andhumidity on the cells is monitored to determine if cells or moduleshave any significant negative effects or are damaged in any way.These tests are the most accurate out of any reliability testsbecause cells are submitted to real world stresses at real worldlevels. However, these tests also take the most time because solarcells are made to last for 30–40 years without being damaged bythe environment around them.

Although the real world tests generate the best results, it isimpractical to test a cell in real time for 30 or 40 years. Acceleratedlife testing is the most common and diverse form of testing. Thissaves time and money, but also forces the tests to be highlyaccurate and precise. Previous weather patterns, future predictionson climate change, and potential city/pollution growth all must beaccounted for in determining what the conditions will be like forsolar cells over the next three decades. Typically, more rigoroustesting than necessary is used to ensure that cells will not fail.Several rules of thumb are practiced in industry. An example of thisis called the 10 8C rule, and it applies to both crystalline and thinfilm cells. It states that for every 10 8C increase in temperature, thetime to failure is cut in half [10].

The various measurements taken during the accelerated testsare compiled together and used to predict the lifetime of a solar cellor module and which failure in the cell will occur first. Each testgives a data point telling how long a cell will last until there is afailure. Altogether, these various data points come together to givean overall estimate of the lifetime of that cell.

Some examples of the measurements taken from these tests areaccording to [156]:

� Mechanical loading tests find how the cell(s) cope with largeamounts of quickly changing stress.� Damp heat exposure examines the effect of high moisture levels

mixed with a large amount of heat.� Thermal cycling determines how well the cell(s) withstand

continuously changing temperatures.� Humidity-freeze cycling determines how well the cell(s)

withstand going from an environment with a high amount ofwater vapor to one that is very cold, resembling the day and nightconditions of the desert.� The UV plus heat test simulates ultraviolet radiation and excess

heat.

Moisture barrier performance of thin film cells is a criticalparameter. If water finds a crack in a cell, multiple failures canoccur. One of these failures is known as hygroscopic swelling,which is caused by water pooling into small bubbles inside of thecell. The cell heats up when it is exposed to light, evaporating thewater and giving it more energy. If the cell gets hot enough, thebubbles will burst because of this hot vapor (Fig. 5.2).

The methodology in testing moisture barrier performanceinvolves placing an aluminum layer underneath the barrier, and ifwater gets through, the aluminum degrades and may be seenunder a microscope. Thin film solar cells need to have very thinmaterials that still protect all the electronics of the cells. Todecrease the chances of failure, a moisture barrier should not onlyprevent moisture from entering the cell, but also be semi-flexibleand prevent bursting from hygroscopic swelling [169].

identical at times. Also, the testing used for LEDs would definitest the capabilities of multijunctions because LEDs operatehigher stress conditions. It has been found that LEDs that worhigher temperatures have less room to dissipate heat and wwith higher current densities than multijunctions [11]. Howemultijunctions do have more degradation in the cell than LEwhich would necessitate more rigorous testing for that parameBecause the LED industry is almost as large as the solar

industry, there is a large amount of experience and methods to

reliability.

5.3. Modeling of life expectancy

Solar cells have a multitude of layered materials in telectronic package, and they all might delaminate at various stlevels. The maximum stresses in the materials are typically foat the junction between the two materials.

In [56], a thermal stress model gives the peeling and shstress caused by thermal expansion. The interfacial shear stres

tinterfacial ¼ sthermaleð�jxÞ

where the peeling stress is given as

speeling ¼ zsthermals0ðxÞ

with

sthermal ¼Ga

haj½ð1 þ y1Þa1 � ð1 þ y2Þa2�DT

where x denotes the spatial coordinate, x and z are ndimensional material parameters, E is Young’s modulus, G isshear modulus, y is Poisson’s ratio, a is the coefficient of therexpansion, and DT denotes the thermal cycling amplitude. Vari

Fig. 5.2. The vapor pressure in a cell over a period of time. The saturated v

pressure is reached around the 12 hour mark, eventually causing the cell lay

rupture [169].

ts a

cantoolion.ackk is) toear

row by

Multijunctions have very good absorption capabilities andefficiencies and are typically used for concentrator systems, wheremirrors direct the light to an array of solar cells on a tower. Thisarrangement creates much higher temperatures than in typicalsolar cells. Also, multijunctions incorporate multiple semiconduc-tors into each cell. Both of these indicate that a somewhat differentmethod of reliability testing than for crystalline or thin film cells isneeded. One way to test reliability in multijunctions that couldpotentially reduce costs is to use existing infrastructure from LightEmitting Diode (LED) manufacturers and researchers. LEDs andsolar cells both use semiconductors that are similar or even

temperatures can be used in this model to determine the effecchanging temperature has on the stresses in solar cells.

An initial defect density is assumed in the model. This

simply be the resolution limit of the metrology or inspection

that allows defects below a threshold to exist without detectBased on multi-mode damage evolution and fatigue crpropagation characteristics, it is envisioned that a micro-cracnucleated when neighboring defects (or voids) coalesce (v = 1form a line defect (Figs. 5.3 and 5.4). In Fig. 5.3, tensile and shstresses work the interfacial layer, causing voids to form and guntil they combine (v = 1). Fig. 5.4 shows this process in a step

stepsuchcyclcell.ampcyclto amag

6. C

Tcostdiffeinvo

Fig. 5homo

Fig. 5Voids

the p


manner. Finally, the number of cycles of DT needed to extend line defects into an unstable crack is estimated. The number of

es then determines the expected life of the interface in the PV It is observed and shown in Fig. 5.5 that the thermal cyclinglitude DT exerts a very strong influence on the number ofes to failure (v = 1). An increase in DT from 20 8C to 40 8C leads

drop in cycles-to-failure by more than three orders ofnitude for surface mount assemblies [56].

the cells, manufacturing, reliability testing, transportation, andinstallation of cells. The long term cost of solar cells is based on lifeexpectancy – it involves the cost of replacing the cells orperforming maintenance. Determining the lowest cost for solarcells without sacrificing efficiency or reliability is the key issue inthe solar cell industry. The roles that efficiency and reliability playin determining the short and long term costs of creating solar cellsare discussed in this section.

6.1. Short term cost

6.1.1. Materials

There is a general trend that higher efficiency results in a highercost per cell. Silicon has been the main material used in solar cellsfor the past two decades, but as the efficiency of these cells rise, theprice fluctuates up and down. It has decreased since 2008 [96] butis still volatile, as the price has doubled and halved again all in thepast 10 years.c-Si cells have reached efficiencies of 25% [87,92], andmulticrystalline has reached an efficiency of 20.4% [87,100].However, these efficiency peaks occurred years ago, while othernew cell types involving Si have grown. Microcrystalline Si hasreached an efficiency of 10.8% [87,124], and thin films that use a-Sihave attained 13.4% efficiency [4,87]. Other thin film cells that usecadmium, tellurium, gallium, indium, silver, and gold have all beentested and achieved efficiencies up to 28.8% [87,119,162].However, these cells are generally expensive and, more impor-tantly, they cannot be a long term solution because the materialsused in them are too rare [19,151]. According to [13], cadmium istoo toxic and tellurium and indium are too rare to make significantcontributions to the total energy production of the world.However, a material such as copper zinc tin sulfide (CZTS) ismade entirely out of cheap, non-toxic elements, has a favorablebandgap (1.45 eV), and a large absorption coefficient. An experi-ment eliminated the H2S anneal step from the growing process tocreate a reactively sputtered CZTS thin film, which saves time andmoney [58]. Other papers synthesized CZTS nanocrystals or inksfor use in low cost thin films [3,5,107]. A new efficiency of 12% hasbeen reached using CZTSS (selenium added to CZTS) [67,87]. Newefficiencies have recently been achieved with CIGS cells (19.8%),dye-sensitized minimodules (8.8%), and organic minimodules(8.5%) [68,87,101,116].

New efficiencies are continuously being achieved for variouscell types each year. However, the PV cell designer andmanufacturer must face a trade-off that balances the threepronged performance metrics of efficiency, reliability and cost.The ecologic footprint of the cells must also be considered witheach of these metrics, as cells produced with environmentallyharmful processes may harm more than help the environment.

6.1.2. Manufacturing

Manufacturing process chain selection for a solar cell cansignificantly influence its cost [77]. Different solar cells typicallyuse different manufacturing techniques. If the cells contain raremetals, a vacuum chamber is needed, especially for silicon becauseof the very high purity and ease at which it collects impurities. Forlarge scale manufacturing, that is not an easy task because of thedifficulty in keeping very large rooms extremely clean. However, ifthese measures are not taken, then the efficiency of the solar cells iscritically compromised.

.3. Schematic diagram of a solar cell interface with an isothermal and porous

genous interfacial layer.

Fig. 5.4. Process of crack nucleation.

.5. Effect of thermal cycling amplitude DT on the number of cycles to failure.

coalesce when v = 1. v is the ratio of porosity length over total length across

art [56].

ost

he efficiency and reliability sections are both governed by the of creating solar cells. The cost can be divided into tworent categories – short term and long term. The short term costlves everything that goes into buying the materials needed for

The method used to create silicon cells was described earlier inSection 3. The silicon manufacturing process has been usedextensively because of the popularity of crystalline silicon cells(they make up 90% of the current market [172]), however, it alsoremains quite expensive due to the larger number of sequences inthe process chain, as well as the sensitivity of each processing step toexternal variations and disturbances. Roll-to-roll fabrication tech-niques have been introduced in a small percentage of the market,however, the cell efficiencies reached via roll-to-roll processingremains much lower compared to other batch processing techniques[18]. Manufacturing multijunction cells takes more time and money,

tionredand

tion cellingliketenrgy

s onost

ver,.e ofls. Aniesls. A

Si,ize,

zed, 6.1 for

theost,ese

the in

n ofk ofhileolarefitcle.gasandCO2


although they have unparalleled efficiencies, such as the 38.8%efficient 5-junction made by [40,87].

6.1.3. Reliability testing

Finding materials that yield high efficiency at low cost isimportant, but the reliability of the solar module must also beaccounted for. Even if cells have very high efficiency, it isimpractical to install them if they do not have the reliabilityneeded to last many years. Crystalline silicon requires special carein terms of reliability. However, thin film cells are also subject tofilm delamination and spallation defects.

The equipment needed to test reliability is very large andexpensive. There are thermal generators, humidifiers, pressurechambers, and many more large pieces of equipment that arenecessary to generate data that helps in discovering failure modes.While such equipment may be affordable for a large corporation,many PV cell manufacturing start-ups (typically small businesses)find it difficult to gain access to such testing facilities. Developingcountries have recently attempted to resolve this dilemma byproviding a central facility that rents testing platforms, on demand,to small businesses [72].

6.1.4. Transportation and installation

Transportation costs are simple to calculate; the more thatneeds to be transported, the higher the cost. If an installer is usingthin film cells, but wants to produce the same amount of electricityas crystalline silicon cells, then more solar cells must be installed(and transported) to make up the difference in efficiencies, thusincreasing transportation costs. However, the thickness andweights of cells can differ, allowing thin, lightweight cells to havethe same or lower transportation costs than higher efficiency,thicker crystalline cells. Installing these thin, lightweight cellscosts more than crystalline cells because of the much largernumber and greater area covered.

The materials, manufacturing, transportation and installation ofsolar cells all add up to create the short term cost. The materialsand manufacturing influence each other; if a company is set onusing a certain material, there are a limited number ofmanufacturing methods to choose from. Also, the material choiceis often dictated by the chosen manufacturing process, and viceversa. These short term costs are all initially very important, butthey are not the only costs to examine.

6.2. Long term cost

Depending on the application, solar cells are expected to lastanywhere from 20 to 40 years [54]. However, there are multiplereliability issues that can create problems and add on to the shortterm cost to create a very large overall cost. These issues need to beexamined and addressed in the near term so that they do not createlong term liabilities.

The main factor that significantly affects the long term cost isthe life expectancy of the solar cell. Accelerated testing andassociated modeling efforts (Section 5.2) allow projections on thelife expectancy of a cell under ideal or desired conditions. However,service conditions, due to natural or man-made effects, may differsignificantly from designed or expected conditions. Such condi-tions may significantly influence cell degradation during service. Adegrading cell can lose efficiency over a period of years and

requirements warrant worst-case analyses and extreme conditesting. Results from such worst-case testing must be tempewith a risk assessment of their statistical significance

likelihood of occurring.

6.3. Energy balance and environmental foot print

A major argument against the growth of solar cell producand use is that more energy is required to manufacture a solarthan the cell produces in its lifetime. Another argument criticizsolar cell manufacturing is that it creates pollution just

burning fossil fuels. These arguments could not be refuted

years ago because there was not enough data describing the eneintake of solar cell manufacturers or the effects this product hathe environment. Finding such data is also very difficult, as mcompanies consider such information to be proprietary. Howein recent years, there have been a few public domain studies

The European Union funded CrystalClear project is an exampla large integrated project focusing on crystalline silicon solar celproduct of this project was a group of scientists and PV compawho authored a paper describing the energy payback of solar celLife Cycle Assessment (LCA) study was performed on ribbonmulti-Si, and mono-Si solar cells [14]. Cell and module sinstallation location, and the lamination type were all standardiand both framed and unframed modules were analyzed. Fig.shows the energy requirements for different process chainsmanufacturing solar cells and reveals that ribbon silicon usedleast amount of energy and mono-Si production used the mwhile multi-Si was in the middle. The environmental impact of ththree cells is also depicted in Fig. 6.1. Ribbon Si has about halfenvironmental impact as mono-Si, with multi-Si somewherebetween. The energy payback varies depending on the locatiosolar cell installation. Southern Europe had an energy paybac1.7–2.7 years (with solar insolation of 1700 kWh/m2/yr) wmiddle Europe had an energy payback of 2.8–4.6 years (with sinsolation of 1000 kWh/m2/yr) [174]. However, the huge benfrom solar cells is the amount of CO2 saved in their life cyCompared to coal combustion, and even combined cycle

combustion, the CO2 produced by solar cells is negligible

concentrated on the manufacturing phase [15]. To reduce their

footprint, cells must be reliable and last as long as possible.

Fig. 6.1. Top: Energy input for different silicon PV modules. The majority of energy

usage is in creating and wafering Si, while creating the cell is not energy intensive.

Bottom: Energy Pay-Back time (yr) for silicon PV modules tested in European

sunlight. The laminate can be seen as the most energy intensive operation. BOS –

Balance of System.

sometimes act as a current resistor to other less degraded cells inthe module [15]. This is a problem that may be difficult to avoid, asreliability testing methods cannot effectively determine if a cell isgoing to degrade over a long period of time, and if so, how much.Extra heat, particularly amplified thermal cycling amplitude is oneof the commonly cited root causes of accelerated degradation[69,156], and radiation caused by the intensity of the sun can causephotodegradation [11]. This failure is difficult to combat, butaccelerated tests can simulate the effects of a large amount ofsunlight over a period of time to acquire an estimate of how muchthe cell will degrade and how long it will take before it starts. Such

IcloscellsCrysemiwerenerincrthis

Cadmof ththanto [1genedispsolacostmanAccoto 2enersolaener

Astorgas

As locomprodthatComUnitmotcavegenestorcrea

Afor mexplcontmanand

reneenerlarg

Ethe ubetwand

typeenvithe eenviimpelimothethe

elimthus

7. C

Sbecoinitienorwithachiman


n [174], the emissions created by solar cell production are moreely examined. These emissions greatly depend on where the

were manufactured. Using the electricity data from thetalClear project (from Europe) yields less than half the

ssions that would be produced if the American electricity datae used. The amount of heavy metal pollutants emitted bygy sources is also greater in America than Europe, which

eases the environmental impact of solar cell production. Also,article finds that certain emissions are very source specific.

ium is almost exclusively produced by burning oil, as the reste cadmium producing energy sources combined produce less

a third of the cadmium released from burning oil. According5], at least 89% of air emissions associated with electricityration could be prevented if electricity from photovoltaics

laces electricity from the grid. The energy cost of producing ar cell might seem high, but studies have examined the energy of extracting oil, coal, and natural gas as well as the cost ofufacturing power plants that create energy from these sources.rding to a study from California [44], over a period from 1955

005, the ratio of energy produced from burning oil per MJ ofgy used to extract the oil was reduced from 6.5 to 3.5. Thus,r cells are a green technology that also has a much bettergy produced to energy used ratio than oil.

technology that has been discussed at great length is energyage. Energy is currently not stored because coal, oil, and naturalpower plants produce energy at a constant rate or on demand.ng as there is enough energy to fulfill customers’ needs, powerpanies do not care if the rest is wasted. Because solar energyuction depends on the weather, it is an intermittent source

can be very strong one day and almost gone the next.pressed air storage systems, such as those in Alabama in theed States and in Germany, allow unused energy to go through aor driving a compressor, which forces air into undergroundrns. When more energy is needed, the released air drives arator that produces the needed electricity. Other possible

age options include batteries or pumping water up a hill tote potential energy [60].

PV breeder that allows solar cells to create the energy neededanufacturing plants to create more solar cells has also been

ored [15]. Combining this system with energy storage andinually allowing it to expand would allow solar cellufacturers to quickly become energy independent from coalcreate a system that uses renewable energy to produce morewable energy sources. Energy storage can stabilize solargy production and make it a more viable system to generate a

e portion of a country’s energy needs.nergy payback is a very difficult metric to calculate because ofnderlying variance in data. The energy payback can greatly varyeen different solar cell types, installation location, reliability,manufacturing processes. Also, in order to compare differents of solar cells, the whole process must be standardized. Theronmental effects of solar cell manufacturing depend on hownergy was made. Energy produced by wind has much less of an

ronmental impact than that produced by coal [15]. Anotherortant caveat exists when examining energy payback. In order toinate the largest sources of CO2 and other hazardous pollutants,r renewable energy sources must be produced. Farther along infuture, once solar cells are installed and unclean energy isinated, more solar cells can be produced using clean energy,

technical product is as much defined by surfaces and interfaces as thePV panel. This begins with the initial creation of surfaces by wafering,where unwanted properties are first induced and must be dealt within subsequent process steps. The process chain is thus determined bycreating surfaces step by step. The physics and therefore physicallimits are known theoretically and experimentally explored underideal laboratory conditions. These ideal limits currently lie far aheadof what has been industrially achieved. Transferring the physicalunderstanding of large scale surface and interface creationtechnology to mass production is of crucial importance inapproaching these ideal performance limits for PV cells. However,the efficiency of solar cells is continuously rising; one to twopercentage points are gained every year. This is due to advancedarchitectures of solar cells, and more innovative usage of surfacesand interfaces in layered electronic circuits. Newly developedheterojunction technology, presented in 2011 [17], which is at thethreshold of commercialization today, improved heterojunctionefficiency by 2%. Other new technologies in the developmentpipeline include black silicon, plasmonics, and multijunction solarcells. Continual improvements build upon the architecture, driven bynew surface structuring and coating technology, and aided by newmaterials and processes. Introduction of innovative surface andinterface creation technologies via new materials and processes aremaking solar cells more efficient, reliable, cost effective andenvironmentally benign. Reliability continues to critically influencepayback characteristics and ecological footprint. A 30 year celllifetime is the standard assumption today, and indications tendtoward 40 year lifetimes, which strongly impact ecologic as well aseconomic balances. As life expectancy and efficiency improve,continual maintenance techniques must also improve to keep agingcells efficient, even after 30 years of service. Further, suchmaintenance needs to be anticipated and accounted for duringdesign and manufacturing of these cells. Introduction of predictivemodeling techniques can aid in this endeavor to reduce expensivetesting and predict cell behavior under severe loading. Acceleratedlife testing, moisture barrier performance, and hygroscopic swellingtests discover what materials and solar cells are strongest and findways to improve cell performance and reliability under severeoperating environments. Failure modes can be discovered inadvance, and remedied. Continuing to utilize new technologies thatcan make the manufacturing of photovoltaic cell surfaces andinterfaces more efficient, reliable, economic and ecologic will create amore favorable public perception of solar cells, which will lead to afavorable reception of solar technology by the community and,ultimately, a greener, solar driven world.

Acknowledgements

The authors are grateful to the CIRP community, particularlySTC-S colleagues for their many valuable and constructivecomments. In particular, we appreciate the efforts of Bert Huisin’t Veld and D.A. Roosen-Melsen. The valuable comments andcritical remarks of Ralf Preu are also gratefully acknowledged.Some of the results presented here are extensions of worksupported by the U.S. National Science Foundation under GrantNos. CMMI-0900093 and CMMI-1100066. Chandra and Andersongratefully acknowledge this support. Any opinions, conclusions orrecommendations expressed are those of the authors and do notnecessarily reflect views of the sponsoring agencies.

creating a sustainable energy system.

onclusion

olar cells still have not reached their enormous potential forming a large producer of energy for the world. Because the

al investment in solar cell research did not match themous potential of the solar cell industry, significant progressin the research and development of solar cells has only beeneved the past two decades. Solar cell manufacturing is theufacturing of surfaces and interfaces on a very large scale. No

References

[1] Photonic Metamaterials, Encyclopedia of Laser Physics and Technology, (2011),RP Photonics.

[2] Aberle AG (2009) Thin Film Solar Cells. Thin Solid Films 517(17):4706–4710.[3] Agrawal R, Birkmire RW, Guo Q, Hillhouse HW, Kar M, Kim SJ, Shafarman WN,

Stach EA (2008) Development of CuInSe2 Nanocrystal and Nanoring Inks forLow-Cost Solar Cells. Nano Letters 8(9):2982–2987.

[4] Ahn SW, Lee HM, Lee SE (2012) Toward Commercialization of Triple-JunctionThin-Film Silicon Solar Panel with >12% Efficiency. 27th European Photovol-taic Solar Energy Conference, Frankfurt, Germany.

http://refhub.elsevier.com/S0007-8506(14)00190-5/sbref0005












with

n In-plied

riza-onics250.

013)ions.

tacthical

tionlline

tzolds for5-1–

tion:

cingrado.cale-ting.

Film

arch

, andl of

llineSolar

an J,ng ofring?

itterional

meCellsltaics

duletaics:

n ST,Solar

tionronic

MAH(15):

s foralists

uasi-Effi-Cells

om-

HKature

Lens

an Xction


[5] Akhavan V, Goodfellow B, Koo B, Korgel BA, Panthani MG, Steinhagen C (2009)Synthesis of Cu2ZnSnS4 Nanocrystals for Use in Low-Cost Photovoltaics.Journal of the American Chemical Society 131(35):12554–12555.

[6] Akimov Y, Koh WS (2010) Design of Plasmonic Nanoparticles for EfficientSubwavelength Trapping in Thin-Film Solar Cells. Plasmonics 6(1):155–161.

[7] Albin DS, Asher S, Dehert C, Dhere RG, Dude A, Gessert TA, Wu X (2001) 16.5%-Efficient CdS/CdTe Polycrystalline Thin-Film Solar Cell, Seventeenth EuropeanPhotovoltaic Solar Energy Conference, Munich, Germany.

[8] Aleman N, Barucha D, Bay N, Biro D, Glunz SW, Knorz A, Preu R (2009)Advances in Electroless Nickel Plating for the Metallization of Silicon SolarCells Using Different Structuring Techniques for the Arc. 24th European PVSolar Energy Conference, Hamburg, Germany.

[9] Al-Jassim M, Branz HM, Jones K, Yan Y, Yost VE, Yuan HC (2010) Microstruc-ture and Surface Chemistry of Nanoporous Black Silicon for Photovoltaics.Photovoltaics Specialists Conference, Honolulu, Hawaii, 2255–2257.

[10] Algora C, Alvarez R, Barbero J, Cordero E, Diaz V, Galiana E, Gonzalez JR,Montalvo F, Nunez N, Rey-Stolle I, Vazquez M (2006) Strategy for CertifiedReliability Analysis of III–V High Concentration Solar Cells. IEEE WorldConference on Photovoltaic Energy Conversion, Waikoloa, Hawaii, 702–705.

[11] Algora C, Gonzalez JR, Nunez N, Rey-Stolle I, Vazquez M (2009) ReliabilityEvaluation of III–V Concentrator Solar Cells. Safety Reliability and Risk Analy-sis: Theory Methods and Applications 1949–1953.

[12] Ali M (2007) Plasmonics Part 2, Chemistry Blog.[13] Alivisatos AP, Kammen DM, Wadia C (2009) Materials Availability Expands

the Opportunity for Large-Scale Photovoltaics Deployment. EnvironmentalScience Technology 43(6):2072–2077.

[14] Alsema EA, Wild-Scholten MJ (2005) Environmental Impacts of CrystallineSilicon Photovoltaic Module Production, Materials Research Society, Boston,United States.

[15] Alsema E, Chul Kim H, Fthenakis VM (2008) Emissions from Photovoltaic LifeCycles. Environmental Science and Technology 42(6):2168–2174.

[16] Andersson BA, Azar C, Holmberg J, Karlsson S (1997) Material Constraints forThin-Film Solar Cells. Energy 23(5):407–411.

[17] Andrault Y, Andreetta L, Batzner DL, Buchel A, Frammelsberger W, Guerin C,Strahm B (2011) Properties of High Efficiency Silicon Heterojunction Cells.Energy Procedia 1(8):153–159.

[18] Andreu J, Ballif C, Fahland M, Fischer D, Haug FJ, Loffler J, Schlemm H, SoppeW, Terrazzoni-Daudrix V, Topic M, Wurz M (2007) Flexcellence: TowardsRoll-to-Roll Mass-Production of Low Cost Thin Film Silicon Solar Cells.European Photovoltaic Solar Energy Conference, Milano, Italy.

[19] Angmo D, Hosel M, Krebs C, Larsen-Olsen TT, Sondergaard R (2012) Roll-to-Roll Fabrication of Polymer Solar Cells. Materials Today 15(1–2):36–49.

[20] Anspach Oea (2010) International Technology Roadmap for Photovoltaics.[21] Apel M, Behnken H, Franke D (2003) Simulation of Mechanical Stress During

Bending Tests for Crystalline Wafers. Proceedings of the 3rd World Conferenceon Photovoltaic Energy Conversion.

[22] Arafune K, Kakimoto K, Kojima T, Sameshima T, Tachibana T (2012) Evalua-tion of Defects Generation in Crystalline Silicon Ingot Grown by Cast Tech-nique with Seed Crystal for Solar Cells. Applied Physics Letters 111:074505.

[23] Archer MJ, Atwater HA, Bhusari D, Boca A, Edmondson KM, Fetzer CM,Haddad M, Isshiki T, Karam NH, King RR, Law DC, Mesropian S, Sherif RA,Yen J, Yoon H (2010) Future Technology Pathways of Terrestrial III–V Multi-junction Solar Cells for Concentrator Photovoltaic Systems. Solar EnergyMaterials and Solar Cells 94(8):1314–1318.

[24] Ardani K, Margolis R (2010) 2010 Solar Technologies Market Report.[25] Ashcroft NW, Mermin ND (1976) Solid State Physics, Holt, Rinehart and

Winston, New York, NY.[26] Atwater HA, Burgos SP, Polman A, de Waele R (2010) A Single-Layer Wide-

Angle Negative-Index Metamaterial at Visible Frequencies. Nature Materials9:407–412.

[27] Atwater HA, Ferry VE, Munday JN (2010) Design Considerations for PlasmonicPhotovoltaics. Advanced Materials 22:4794–4808.

[28] Axmann HJ (2009) Technology Roadmap for Crystalline Silicon PV: First Resultsof German Working Group, ITRPV.

[29] Bagnall DM, Boden SA (2010) Optimization of Moth-Eye AntireflectionSchemes for Silicon Solar Cells. Progressive Photovoltaics: Research and Appli-cations 18(3):195–203.

[30] Bagnall DM, Boden SA, Papakostas A, Potts A, Zhang W (2011) NanostructuredOptical Metamaterials, University of Southampton, Southampton.

[31] Bailey S, Galipeau D, Joshi P, Qiao Q, Ropp M, Xie Y (2009) Dye-SensitizedSolar Cells Based on Low Cost Nanoscale Carbon/TiO2 Composite CounterElectrode. Energy and Environmental Science 2:426–429.

[32] Balestrieri M, Becker JP, Glunz SW, Hermle M, Pysch D, Warta W (2011)Characterization and Optimization of Indium Tin Oxide Films forHeterojunction Solar Cells. Solar Energy Materials and Solar Cells 95:2390–2399.

[33] Ballif C, Bidiville A, Michler J, Nasch PM, Van der Meer M, Wasmer K (2010)Mechanisms of Wafer Slicing and Impact on Wafer Properties. Progress in

Instruments and Methods in Physics Research Section B: Beam InteractionsMaterials and Atoms 272:214–217. (in press).

[38] Bentzen A, Holt A, Christensen JS, Svensson BG (2006) High ConcentratioDiffusion of Phosphorus in Si from a Spray-On Source. Journal of ApPhysics 99.

[39] Berroth M, Chen K, Vogel W, Zaoui WS (2012) Low Loss Broadband Polation Independent Fishnet Negative Index Metamaterial at 40 GHz. Photand Nanostructures – Fundamentals and Applications 10(3):245–(in press).

[40] Bhusari D, Boisvert J, Chiu PT, Karam NH, King RR, Law DC, Zakaria A (2Direct Semiconductor Bonded 5J Cell for Space and Terrestrial ApplicatIEEE Journal of Photovoltaics 4(1):493–497.

[41] Bierwisch C, Kleer G, Kubler R, Moseler M (2011) Modeling of ConRegimes in Wire Sawing with Dissipative Particle Dynamics. PhilosopTransactions of The Royal Society: A 369:2422–2430.

[42] Bosco N, Gelak E, Hacke P, Kurtz S, Terwilliger K, Trudell D (2009) Applicaof the NREL Test-to-Failure Protocol for PV Modules. Workshop on CrystaSilicon Solar Cells and Modules: Materials and Processes, Vail, Colorado.

[43] Bottler W, Carius R, Hagemann V, Meier M, Michaelis D, Moulin E, PaeUW, Rau U, Wachter C (2011) Plasmonic Reflection Grating Back ContactMicrocrystalline Silicon Solar Cells. Applied Physics Letters 99(18). 18110181105-3.

[44] Brandt AR (2011) Oil Depletion and the Energy Efficiency of Oil ProducThe Case of California. Sustainability 3:1833–1854.

[45] Branz H (2010) NREL’s Black Silicon Increases Solar Cell Efficiency by ReduReflected Sunlight, National Renewable Energy Laboratory, Golden, Colo

[46] Branz HM, Page MR, Toor F, Yuan HC (2011) 17.1%-Efficient Multi-STextured Black Silicon Solar Cells without Dielectric Antireflection CoaPhotovoltaic Specialists Conference, Seattle, Washington.

[47] Brennan A, Kornelsen K, Lai K, Stark JG (2012) Pressure Indicating

Characterization of Wafer-to-Wafer Bonding, Sensor Products Inc..[48] Bright C, Ginley DS (2000) Transparent Conducting Oxides. Materials Rese

Society Bulletin 15–18.[49] Brun XF, Melkote SN (2009) Modeling and Prediction of Flow, Pressure

Holding Force Generated by a Bernoulli Handling Device. JournaManufacturing Science & Engineering ASME Transaction 131:031018.

[50] Brun X, Melkote SN (2009) Analysis of Stresses and Breakage of CrystaSilicon Wafers During Handling and Transport. Solar Energy Materials &

Cells 93:1238–1247.[51] Bultman J, Hoornstra J, Komatsu Y, Romijn I, Stassen A, Tool K, Bultm

Hoornstra J, Komatsu Y, Romijn I, Stassen A, Tool K (2011) Inline ProcessiCrystalline Silicon Solar Cells: The Holy Grail for Large Scale ManufactuManufacturing the Solar Future the 2011 Production Annual.149–157.

[52] Bultman J, Cesar I, Geerligs B, Komatsu Y, Sinke W (2012) Methods of EmFormation for Crystalline Silicon Solar Cells. Photovoltaics Internat5:69–80.

[53] Cabarrocas PRi, Charpentier C, Courtois G, Francke L, Labrune M, Prod’HomP, Salomon A (2012) Effect of Annealing on Silicon Heterojunction Solar

with Textured ZnO:Al as Transparent Conductive Oxide. EPJ Photovo3:35002–35005.

[54] Canizo Cd, Coso Gd, Sinke WC (2009) Crystalline Silicon Solar MoTechnology: Toward the 1s per Watt-Peak Goal. Progress of PhotovolResearch and Applications 17:199–209.

[55] Carroll DL, Demir HV, Ke L, Kwong DL, Kyaw AKK, Li Y, Lo GQ, Sun XW, TaZhao DW (2011) Optimization of Inverted Tandem Organic Solar Cells.

Energy Materials and Solar Cells 95(3):921–926.[56] Chandra A, Fu G, Huang Y, Hu KX, Jiang ZQ (2000) A Model of Crack Nuclea

in Layered Electronic Assemblies Under Thermal Cycling. Journal of ElectPackaging 122:220–226.

[57] Chaumartin A (2006) Pyramidal Texturing of Silicon Solar Cell with TChemical Anisotropic Etching. Solar Energy Materials and Solar Cells 902319–2328.

[58] Chawla V, Clemens B (2010) Inexpensive, Abundant, Non-Toxic Thin FilmSolar Cell Applications Grown by Reactive Sputtering. Photovoltaics SpeciConference, Honolulu, Hawaii, 1902–1905.

[59] Chen L, Gu X, Guo K, Wang D, Yang D, Yu X (2012) Seed-Assisted Cast QSingle Crystalline Silicon for Photovoltaic Application: Towards High

ciency and Low Cost Silicon Solar Cells. Solar Energy Materials & Solar

101:95–101.[60] Cherry B (2014) Pumped Storage Projects, Federal Energy Regulatory C

mission (FERC).[61] Chettiar UK, Drachev VP, Kildishev AV, Ni X, Shalaev VM, Xiao S, Yuan

(2010) Loss-Free and Active Optical Negative-Index Metamaterials. N466:735–738.

[62] Chevalier CT, Downey AN, Schwartz ZD, Vaden KR, Wilson JD (2005) FlatFocusing Demonstrated with Left-Handed Metamaterial, NASA.

[63] Chhajed S, Cho J, Kim JK, Poxson DJ, Schubert EF, Sood AK, Welser RE, Y(2011) Nanostructured Multilayer Tailored-Refractive-Index Antirefle

. Ap-

008) TM

andora-

(11):

008)gress

Photovoltaics: Research and Applications 18:563–572.[34] Barberoglou M (2009) Generation of Al Nanoparticles via Ablation of Bulk Al

in Liquids with Short Laser Pulses. Optics Express 17(15):12650–12659.[35] Barnett A, Crowley L, Davis JM, Farris C, Forest H, Hamer G, Kimerling L, Little

R, Paranzino M, Roppenecker Ws, Schwartz R, Shimp H, Sklar S (2003) SolarElectric Power: The U.S. Photovoltaic Industry Roadmap.

[36] Bartsch J, Glunz SW, Gombert A, Horteis M, Kamp M, Reinecke H (2011)Effects of Seed Layer and Substrate Morphology on Solar Cell ContactsDeposited by Light-Inducing Plating. Journal of the Electrochemical Society158(6):H651–H658.

[37] BenAssayag G, Benzo P, Bonafos C, Carles R, Cattaneo L, Farcau C, Pecassou B,Zwick A (2012) 3D Patterning of Ag Nanoparticles by ULE Ion Implantationand Stencil Soft Lithography for Plasmonic Device Applications. Nuclear

Coating for Glass with Broadband and Omnidirectional Characteristicsplied Physics Express 4(5). 052503-1–052503-3.

[64] Clark R, Park Y, Rozgonyi G, Stoddard N, Wagener M, Witting I, Wu B (2Casting Single Crystal Silicon: Novel Defect Profiles from BP’s Mono2Wafers. Solid State Phenomena 131–133:1–8.

[65] Collins E, Dvorack M, Mahn J, Mundt M, Quintana M (2009) ReliabilityAvailability Analysis of A Fielded Photovoltaic System, Sandia National Labtories, Albuquerque, New Mexico.

[66] Conibeer G (2007) Third-Generation Photovoltaics. Materials Today 1042–50.

[67] Contreras MA, DeHart C, Egass B, Noufi R, Perkins CL, Repins I, To B (219.9%-Efficient ZnO/CdS/CuInGaSe2 Solar Cell with 81.2% Fill Factor. Proin Photovoltaics: Research and Applications 16:235–239.
































































































































































































[68]

[69]

[70]

[71]

[72]

[73]

[74]

[75]

[76]

[77]

[78]

[79]

[80]

[81]

[82]

[83]

[84]

[85]

[86]

[87]

[88]

[89]

[90]

[91]

[92]

[93]

[94]

[95]

[96][97]

[98]

[99]

[100]


Crotty G, Hanoka JI, Rohatgi A, Wallace RL (1997) Thin Silicon String Ribbon.Energy Materials & Solar Cells 48:179–186.

Cui Y, Fan S, Yu Z, Zhu J (2010) Nanostructured Photon Management for HighPerformance Solar Cells. Materials Science and Engineering: Report 70(3–6):330–340.

Cui Y, Narashimhan VK (2013) Silicon Nanowires with Controlled SidewallProfile and Roughness Fabricated by Thin-Film Dewetting and Metal-AssistedChemical Etching. Nanotechnology 24:225305.

Cunningham DW, Kelly GJ, Nguyen AM, Placer NV, Wohlgemuth JH (2008)The Effect of Cell Thickness on Module Reliability. Photovoltaics SpecialistsConference, BP Solar International, San Diego, California1–4.

Cunningham DW, Miller J, Monus P, Nguyen A, Wohlgemuth JH (2006) LongTerm Reliability of Photovoltaic Modules. Photovoltaic Energy Conference,Waikoloa, Hawaii, 2050–2053.

Danyluk S, He S, Zheng T (2004) Analysis and Determination of the Stress-Optic Coefficients of Thin Single Crystal Silicon Samples. Journal of AppliedPhysics 96:3103–3109.

Danyluk S, Garcia V, Li F (2006) Full Field Stress Measurements in Thin SiliconSheet. Proc. the 4th World Conference on Photovoltaic Energy Conversion, 1245.

Danyluk S, He S, Ostapenko S, Tarasov I (2006) Residual Stresses in Polycrys-talline Silicon Sheet and Their Relation to Electron-Hole Lifetime. AppliedPhysics Letters 89:111909.

Danyluk S, Li F, Skenes K (2011) Analysis of Residual Stress in Thin SiliconWafers with an NIR Polariscope. Proc. 26th European Photovoltaic Solar EnergyConference and Exhibition, 2082.

Danyluk S, Melkote SN, Wu H (2012) Effects of Carbide and Nitride Inclusions onDiamond Scribing of Multicrystalline Silicon for Solar Cells, Precision Engineering.

Danyluk S, Melkote SN, Wu H (2012) Mechanical Strength of Silicon WafersCut by Loose Abrasive Slurry and Fixed Abrasive Diamond Wire Sawing.Advanced Engineering Materials 14:342–348.

Degroot MW, Elowe PR, Feist R, George J, Wilbur J (2009) High ThroughputMethodology for Evaluation of the Moisture Barrier Performance of ThinFilms for PV Applications. Photovoltaics Specialists Conference, Dow SolarSolutions, Philadelphia, Pennsylvania, 450–455.

Desyatnikov AS, Fischer R, Kivshar YS, Krolikowski W, Lopez-Aguayo S,Neshev DN, Sukhorukov AA (2007) Light Localization in Azimuthally Modu-lated Bessel Photonic Lattices. Journal of Materials Science: Materials in Elec-tronics 18:S277–S283.

Devilee C, Dorenkamper M, Heijna MCR, Loffler J, Soppe WJ, Van Aken BB(2009) Roll to Roll Fabrication Process of Thin-Film Silicon Solar Cells on SteelFoil. European Photovoltaic Solar Energy Conference and Exhibition, Hamburg,Germany.

Doderer D, Jabbour GE (2010) Quantum Dot Solar Cells: The Best of BothWorlds. Nature Photonics 4:604–605.

Dodge D (2006) The Future of Wind Power, Illustrated History of Wind PowerDevelopment.

Duche D, Escoubas L, Flory F, Mathian G, Monestier F, Simon JJ, Torchio P(2009) Improving Light Absorption in Organic Solar Cells by PlasmonicContribution. Solar Energy Materials and Solar Cells 93(8):1377–1382.

DunlopED,EmeryK,GreenMA,HishikawaY,WartaW(2011)SolarCellEfficiencyTables. Progress in Photovoltaics: Research and Applications 20(1):12–20.

Dunlop ED, Emery K, Green MA, Hishikwa Y, Warta W (2012) Solar CellEfficiency Table (Version 39). Progress of Photovoltaics: Research and Applica-tions 20:12–20.

Dunlop ED, Emery K, Green MA, Hishikawa Y, Warta W (2014) Solar CellEfficiency Tables (Version 43). Progress in Photovoltaics: Research and Applica-tions 22:1–9.

EIA (2013) Levelized Cost of New Generation Resources in the Annual EnergyOutlook 2013.

EIA (2012) Electric Power Monthly: t.F. Table 5.3. Average Retail Price ofElectricity to Ultimate Customers: Total by End-Use Sector, efd., U.S. EnergyInformation and Administration, Washington, DC.

El Chaar L, Iamont LA, El Zein N (2011) Review of Photovoltaic Technologies.Renewable and Sustainable Energy Reviews 15(5):2165–2175.

Fahland M, Kaiser N, Munzert P, Schonberger W, Schulz U (2011) Thin FilmGrowth on Nanostructured Polymer Webs for Anti-Reflection Purposes.Surface and Coatings Technology 205(2):S498–S501.

Ferrazza F, Green MA, Wang A, Zhao J (1998) Novel 19.8% Efficient Honey-comb Textured Multicrystalline and 24.4% Monocrystalline Silicon SolarCells. Applied Physics Letters 73:1991–1993.

Fogel K, Inns D, Kim J, Sadana DK (2010) Surface Texturing of Single-Crystalline Silicon Solar Cells Using Low Density SiO2 Films as an AnisotropicEtch Mask. Solar Energy Materials and Solar Cells 94(12):2091–2093.

Fritzsche H (1995) Photo-Induced Structural Changes Associated with theStaebler-Wronski Effect in Hydrogenated Amorphous Silicon. Solid StateCommunications 94:953–955.

Fthenakis V, Mason J, Zweibel K (2008) By 2050 Solar Power Could End USDependence on Foreign Oil and Slash Greenhouse Gas Emissions. Scientific

[101] Gotanda T, Hayase R, Hosoya M, Nakano Y, Nakao H, Ooka H, Saito M (2012)Development of Organic Thin Film Photovoltaic Modules. 73rd Fall Meeting ofJapan Society of Applied Physics.

[102] Granata J, Kurtz S, Quintana M (2009) Photovoltaic-Reliability R&D Toward aSolar-Powered World. Society of Photographic Instrumentation Engineers(SPIE) Solar Energy and Technology Conference, San Diego, California.

[103] Granqvist CG, Hamberg I (2013) Theoretical Model for the Optical Propertiesof In2O3:Sn Films in the 0.3–50 mm Range. Proc. Optical Materials Technologyfor Energy Efficiency and Solar Energy Conversion IV.

[104] Granqvist CG, Hultaker A (2002) Transparent and Conducting ITO Films: NewDevelopments and Applications. Thin Solid Films 411:1–5.

[105] Green MA, Pillai S (2010) Plasmonics for Photovoltaic Applications. SolarEnergy Materials and Solar Cells 94(9):1481–1486.

[106] Guha S, Jiang C-S, Sivec L, Yan B, Yue G (2011) Innovative Dual Function nc-SiOx:H Layer Leading to a >16% Efficient Multi-Junction Thin-Film SiliconSolar Cell. Applied Physics Letters 99:113512.

[107] Gunawan O, Hovel HJ, Mitzi DB, Todorov TK, Wang W, Winkler MT (2013)Optical Designs that Improve the Efficiency of Cu2ZnSn(S,Se)4 Solar Cells.Energy and Environmental Science 7:1029–1036.

[108] Guo LJ, Ahn SH, Kang MG, Park HJ (2010) A Facile Route to Polymer Solar Cellswith Optimum Morphology Readily Applicable to a Roll-to-Roll Processwithout Sacrificing High Device Performances. Advanced Materials 22(35):E247–E253.

[109] Haitjema H (1989) Spectrally Selective Tinoxide and Indiumoxide Coatings,(Ph.D. Thesis) Delft University of Technology, Delft, Netherlands.

[110] Haitjema H, Elich JJP, Hoogendoorn CJ (1989) The Optical, Electrical andStructural Properties of Fluorine-Doped, Pyrolytically Sprayed TindioxideCoatings. Solar Energy Materials 18:283–297.

[111] Hama T, Ichikawa Y, Sakai H, Yoshida T (1990) Effects of Surface Morphologyof Transparent Electrode on the Open-Circuit Voltage in a-Si:H Solar Cells.Japanese Journal of Applied Physics 29:630–635.

[112] Hamakawa Y, Okamoto H, Nitta Y (1979) A New Type of Amorphous SiliconPhotovoltaic Cell Generating More Than 2.0V. Applied Physics Letters35(2):187–189.

[113] Hamakawa Y, Okamoto H, Tawada Y (1981) A-SiC:H-a-Si:H HeterojunctionSolar Cell Having More Than 7.1 Percent Conversion Efficiency. AppliedPhysics Letters 39(3):237–239.

[114] Hamashita D, Konagai M, Miyajima S, Nagashima H, Yunaz IA (2010) Wide-Gap a-Si1-xCx:H Solar Cells with High Light-Induced Stability for Multi-junction Structure Applications. Solar Energy Materials and Solar Cells 95(1):107–110.

[115] Hanoka JI (2001) An Overview of Silicon Ribbon Growth Technology. SolarEnergy Materials & Solar Cells 65:231–237.

[116] Haven VE, Keavney CJ, Vernon SM (1990) Emitter Structures in MOCVDInP Solar Cells. 21st IEEE Photovoltaic Specialists Conference, Kissimmee,FL, USA.

[117] Hayashi Y, Iida H, Ito A, Karasawa H, Mishuku T, Shiba N, Yamanaka M (1983)Efficiency of the a-Si:H Solar Cell and Grain Size of SnO2. TransparentConductive Film IEEE Electron Device Letters (EDL) 4(5):157–159.

[118] Hibberd B, Yates T (2012) Levelized Cost of Energy: Page 7 of 10. SolarPro, 5.3.[119] Higashi GS, Kayes BM, Kizilyalli IC, Nie H, Reinhardt F, Spruytte SG, Twist R

(2011) 27.6% Conversion Efficiency, A New Record for Single-Junction SolarCells Under 1 Sun Illumination. 37th IEEE Photovoltaic Specialists Conference,Seattle, WA, USA.

[120] Holland P, Hutchison K (2008) Crystalline Silicon Solar Cell ManufacturingRequires Vacuum-Based Solutions. Solid State Technology 51(2):55.

[121] Holt A, Marstein ES, Wright DN (2005) Double Layer Anti-Reflective Coatingsfor Silicon Solar Cells. Proc. the 31st IEEE Photovoltaic Specialists Conference,1237–1240.

[122] Hou DY, Li MH, Tian Y, Yang HL (2011) Dual Bands of Negative RefractiveIndexes in the Planar Left-Handed Metamaterials. Journal of Magnetism andMagnetic Materials 323(5):607–610.

[123] Howarth RW, Santoro R (2011) Methane and the Greenhouse-Gas Footprintof Natural Gas from Shale Formations. Climatic Change 106(4):679–690.

[124] Hozuki N, Kondo M, Sai H, Saito K (2013) Relationship Between the CellThickness and the Optimum Period of Textured Back Reflectors in Thin-Film Microcrystalline Silicon Solar Cells. Applied Physics Letters 102:053509.

[125] Hvistendahl M (2008) China’s Three Gorges Dam: An Environmental Catastro-phe? Scientific American..

[126] Ide D, Kondo Y, Matsuki T, Sakata I, Takato H, Watanabe N (2010) Character-ization of Polycrystalline Silicon Wafers for Solar Cells Sliced with NovelFixed-Abrasive Wire. Progress in Photovoltaics: Research and Applications18:485–490.

[127] Physics Institute (2007) Did You Know. . . Photoelectric Effect, Einstein Year.[128] Ji L, Varadan VV (2010) Optical Metastructures for Trapping Light in Thin Si

Solar Cells. Photovoltaics Specialists Conference, Department of Electrical Engi-neering, University of Arkansas, Honolulu, Hawaii.

American 298(1):64–73. Fuhs M (2013) Trina Solar: Stagnant Module Prices Until 2017, PV Magazine. Funke C, Moller HJ, Rinio M, Scholz S (2005) Multicrystalline Silicon for SolarCells. Thin Solar Films 487:179–187.

Ghitani HE, Martinuzzi S (1989) Influence of Dislocations on ElectricalProperties of Large Grained Polycrystalline Silicon Cells. I. Model. Journalof Applied Sciences 66:1717–1722.

Glunz S, Riepe S, Schultz O, Willeke G (2006) High Efficiency Solar Cells onPhosphorus Gettered Multicrystalline Silicon Substrates. Progress in Photo-voltaics: Research and Applications 14:711–719.

Glunz SW, Schultz O, Willeke GP (2004) Multicrystalline Silicon Solar CellsExceeding 20% Efficiency. Progress in Photovoltaics Research and Applications12:553–558.

[129] Jin R, Shi J, Wu S, Zhao H (2011) PDE-Constrained Gaussian Process Model onMaterial Removal Rate of Wire Slicing Process. Journal of ManufacturingScience and Engineering ASME Transaction 133(2):021012.

[130] Johnson S, Regnault W, Yoo K (1985) Lattice Defects Within Grain Volumesthat Affect the Electrical Quality of Cast Polycrystalline Silicon Solar CellMaterials. Journal of Applied Physics 57:2258–2266.

[131] Kaden T, Moller HJ, Scholz S, Wurzner S (2009) Improving Solar Grade Siliconby Controlling Extended Defect Generation and Foreign Atom Defect Inter-actions. Applied Physics A 96:207–220.

[132] Kalejs JP (2002) Silicon Ribbons and Foils – State of the Art. Solar EnergyMaterials & Solar Cells 72:139–153.

[133] Kalejs JP, Mackintosh BH, Surek T (1980) High Speed EFG of Wide SiliconRibbon. Journal of Crystal Growth 50:175–192.
































































































































































































s. inensa,

Side

ules.

ratedpart-

ctileate-

rittlewingSME

olar,

MAIEEE

nics

tion-

kyo.cess,

Solar

, andty of

ance67–

Film2.

t PVaris,

le at

011,

ells.


[134] Kalejs JP, Schmidt W, Woesten B (2002) Manufacturing Technology forRibbon Silicon (EFG) Wafers and Solar Cells. Research and Applications10:129–140.

[135] Kao I, Yang F (2001) Free Abrasive Machining in Slicing Brittle Materialswith Wire. Journal of Electronic Packaging ASME Transactions 123:254–259.

[136] Kikuchi M (2010) Advances in Manufacturing Equipment for Thin Film SiSolar Cells. Proceedings of JSPE Ultra-precision Positioning Technical Committee,41–72.

[137] Kim DS, Nakayashiki K, Rohatgi A, Rounsaville B, Yelundur V (2004) High-Efficiency Solar Cells on Edge-Defined Film-Fed Grown (18.2%) and StringRibbon (17.8%) Silicon by Rapid Thermal Processing. Applied Physics Letters84:145–148.

[138] King R, Mitchell K, Reiss J (1996) Characterization of Diffusion LengthDegradation in Czochraski Silicon Solar Cells. Applied Physics Letters 68:3302–3304.

[139] Kirk AP (2010) Proposal for a High Voltage AlGaAs/AlGaAs/GaAs Triple JunctionPhotovoltaic Cell. Solar Energy Materials and Solar Cells 94(12):2442–2445.

[140] Kittler M, Morgenstern G, Seifert W (1993) Influence of Dislocation Densityon Recombination at Grain Boundaries in Multicrystalline Silicon. Semicon-ductor Science and Engineering 8:1687–1691.

[141] Knier G (2002) How Do Photovoltaics Work?NASA Science: Science News.[142] Konaga M, Yamaguchi M, Kondo M (2010) Fundamentals and Applications of

Solar Cells, Baifukan, Tokyo.[143] Konagai M (2011) Present Status and Future Prospects of Silicon Thin-Film

Solar Cells. Japanese Journal of Applied Physics 50:030001.[144] Krebs FC (2009) Roll-to-Roll Fabrication of Monolithic Large-Area Polymer

Solar Cells Free from Indium-Tin-Oxide. Solar Energy Materials and Solar Cells93(9):1636–1641.

[145] Kumar Dhungel S, Yoo J, Kim K, Jung S, Gosh S, Yi J (2006) Double-Layer Anti-Reflection Coating of MgF2 and SiNx for Crystalline Solar Cells. Journal of theKorean Physical Society 49:885–889.

[146] Kumar D, Kumar R, Sharma M, Singh PK, Srivastava SK (2012) Silver CatalyzedNano-Texturing of Silicon Surfaces for Solar Cell Applications. Solar EnergyMaterials and Solar Cells 100:33–38. (in press).

[147] Kuna M, Liedke T (2011) A Macroscopic Mechanical Model of the Wire SawingProcess. International Journal of Machine Tool & Manufacture 51:61–64.

[148] Kutsukake K, Morishita K, Nakajima K, Stokkan G, Takahashi I, Usami N(2010) Generation Mechanism of Dislocations During Directional Solidifica-tion of Multicrystalline Silicon Using Artificially Designed Seed. Journal ofCrystal Growth 312:897–901.

[149] Kymakis E, Spyropoulos GD, Stylianakis M, Stratakis E (2011) PlasmonicOrganic Photovoltaics Doped with Metal Nanoparticles. Photonics and Nanos-tructures – Fundamentals and Applications 9(2):184–189.

[150] Latour M, Masson G, Papoutsi M, Rekinger M, Theologitis IT (2012) GlobalMarket Outlook for Photovoltaics 2013–2017. in Winneker C, (Ed.) EuropeanPhotovoltaic Industry Association (EPIA), .

[151] Lewis NS (2007) Toward Cost-Effective Solar Energy Use. Science315(5813):798–801.

[152] Li X, Wang P, Yang D, Yu X (2012) Thin Czochralski Silicon Solar Cells Based onDiamond Wire Sawing Technology. Solar Energy Materials & Solar Cells98:337–342.

[153] Lorenzo E (1994) Solar Electricity: Engineering of Photovoltaic SystemWinneker C, (Ed.) European Photovoltaic Industry Association (EPIA), ProgSevilla, Spain59–692013–2017.

[154] Martin de Nicholas S (2012) a-Si:H/c-Si Heterojunction Solar Cells: BackAssessment and Improvement, South University, Doctorate, Paris.

[155] McMahon TJ (2004) Accelerated Testing and Failure of Thin-Film PV ModProgress in Photovoltaics: Research and Applications 12:235–248.

[156] McMahon T, Quintana M, Tillerson J, Ton D, Zweibel K (2007) AcceleAging Tests in Photovoltaics, Solar Energy Technologies Program, U.S. Dement of Energy.

[157] Melkote SN, Wu H (2012) Effect of Crystallographic Orientation on DuScribing of Crystalline Silicon: Role of Phase Transformation and Slip. Mrials Science and Engineering A 549:200–205.

[158] Melkote SN, Wu H (2012) Modeling and Analysis of Ductile-to-BTransition in Diamond Scribing of Silicon: Application to Wire Saof Silicon Wafers. Journal of Engineering Materials and Technology ATransaction 134:041011.

[159] Morgan J (2010) Comparing Energy Costs of Nuclear, Coal, Gas, Wind, and SNuclear Fissionary.

[160] Narayanan S, Zulper J, Yun F, Wenham SR, Sproul SB, Chong CA, Green(1990) 18% Efficiency Polycrystalline Silicon Solar Cells. Proc. 21st

Photovoltaic Specialists Conference, 678–680.[161] Rupnowski P, Sopori B (2009) Strength of Silicon Wafers: Fracture Mecha

Approach. International Journal of Fracture 155:67–74.[162] Sandru O (2009) Quantasol Breaks Solar Efficiency Records with Loca

Optimized Cells, The Green Optimistic.[163] Sato K (2011) Chapter 2 in Fundamentals of Solar Cells, SoftBank Creative, To[164] Schumann M, Perez O, Riepe S (2011) The Solar Cell Wafering Pro

Manufacturing the Solar Future: The 2011 Production Annual86–93.[165] Seale E (2003) Photovoltaic (Effect). Encyclobeamia.[166] Sopori Bea (2007) National Solar Technology Roadmap: Wafer-Silicon pv,

Energy Technologies Program, U.S. Department of Energy.[167] Sunlight Electric, L. Photovoltaic History. Sunlight Electric.[168] SunPower (2014) Most Efficient Solar, SunPower.[169] Tee TY (2010) Methodology for Integrated Vapor Pressure, Hygroswelling

Thermo-Mechanical Stress Modeling of IC Packages, Moisture SensitiviPlastic Packages of IC Devices, Springer, New York221–243.

[170] Van Duyne RP, Willets KA (2007) Localized Surface Plasmon ResonSpectroscopy and Sensing. Annual Review of Physical Chemistry 58:2297.

[171] Van Nieuwenhuysen K (2010) Epitaxially Grown Emitters for Thin

Silicon Solar Cell Result in 16% Efficiency. Thin Solid Films 518:S80–S8[172] Verlinden P, YingBin Z, ZhiQiang F (2013) Cost Analysis of Curren

Production and Strategy for Future Silicon PV Modules, 28th EU PVSEC, PFrance.

[173] Wesoff E (2013) CIGS Solar Update: New Efficiency World Record; MiaSoHanergy; Solar Frontier Ships, Greentechsolar.

[174] Wild-Scholten MJde (2014) Life Cycle Assessment of Photovoltaics Status 2Part 1 Data Collection, SmartGreenScans.

[175] Zook JD (1980) Effects of Grain Boundaries in Polycrystalline Solar CApplied Physics Letters 37:223–226.












































































































role of surfaces and interfaces in solar cell manufacturing · of surfaces and interfaces in solar...

Documents