00

The Australian Research Council (ARC) Centre of Excellence for Advanced Silicon Photovoltaics and Photonics offi cially came into being at the University of New South Wales (UNSW) on 13th June, 2003. The Centre‘s mission is to advance silicon photovoltaic research on three separate fronts, as well as applying these advances to the related fi eld of silicon photonics. The educational activities of the former Key Centre for Photovoltaic Engineering have also been successfully integrated into the new Centre.

Photovoltaics is the process by which light, usually sunlight, is converted directly into electricity when absorbed in devices known as solar cells. Silicon is the most common material used to make these cells, as well as being the material most widely used in microelectronics.

Photovoltaics is currently the world's most rapidly growing energy source, with markets increasing at a compounded rate of 35%/year over the last 8 years and a massive 67% during 2004. Most product sold is based on “fi rst-generation” solar cells that use silicon wafers, similar to the wafers used in microelectronics. The Centre leads the world in this area, holding the international records for the highest-performing silicon cells in most major categories, including that for the outright highest-performing device. First-generation Centre research addresses the dual challenges of reducing cost and improving effi ciency. The Centre made further progress during the year in applying the Centre’s “buried-contact” approach to silicon wafers doped with phosphorus, rather than the standard boron dopant. The problem with boron is that, under sunlight, it interacts with trace quantities of oxygen in the wafers, producing defects that reduce cell output. Phosphorus does not have this problem, but it is more challenging to make high-performance cells from phosphorus-doped wafers. Good progress was made in achieving the Centre’s aims in this area, with energy conversion effi ciency above 22% measured for phosphorus-doped cells made during the year.

Silicon is quite a brittle material so silicon wafers have to be reasonably thick, about one-third of a millimetre, to be suffi ciently rugged for processing into solar cells. However, without this mechanical constraint, the silicon in principle could retain acceptable performance even if very thin, over 100 times thinner. Centre researchers have pioneered an approach where such very thin silicon layers are deposited directly onto a sheet of glass with the glass providing the required mechanical strength. Such a “second-generation” approach gives an enormous cost saving since the expensive processes involved in making wafers are no longer required, there is an enormous saving in silicon material, and the cells can be made over the entire area of large glass sheets.

The Centre is at the forefront of international research with this “second-generation”, “silicon on glass” approach. A partner in the Centre is CSG Solar, an originally Sydney-based company formed to commercialise this approach. With some assistance provided by Centre researchers, CSG Solar improved the energy-conversion effi ciency of this approach to 9.1% during 2004. Even though CSG Solar’s silicon layer is about 200 times thinner than a wafer, this effi ciency is close to the 10-15% range demonstrated by commercial wafer-based solar cell modules. Complementary work wholly within

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

DIRECTORS' REPORT

1

00

the Centre demonstrated improvement in output voltage of its EVA, ALICIA and ALICE thin-fi lm solar cell concepts. A signifi cant factor behind these improvements was the acquisition of a major new processing facility that became available during 2004. This acquisition would not have been possible without the existence of the Centre. As a result of a combined ARC Centre / UNSW initiative, the Centre was able to procure a fully-operational clean room, equipped with operational state-of-the-art equipment for research, at a fraction of the equipment’s replacement cost. This facility was surplus to the needs of Pacifi c Solar, the forerunner to CSG Solar, with eff orts now focusing on technology commercialisation.

The silicon thin-fi lm approach has a large cost advantage over the wafer-based approach, due mainly to reduced material costs. However, in large enough production volumes, such material costs will eventually dominate the costs of even the thin-fi lm approach. This led to the Centre’s interest in advanced “third-generation” thin-fi lm solar cells, targeting signifi cant increases in energy-conversion effi ciency. Higher conversion effi ciency means more power from a given amount of material, reducing costs.

The Centre’s experimental third-generation program is concentrating on “all-silicon” tandem solar cells, where high energy-threshold cells are stacked on top of lower-threshold devices. The silicon bandgap threshold is controlled by quantum-confi nement of carriers in small silicon quantum-dots dispersed in an amorphous matrix of silicon oxide or nitride, or eventually carbide. Other approaches being explored are based on up-conversion or down-conversion of the energy of photons in the solar spectrum, to manipulate this spectrum so it is more suitable for conversion by the cell. Cells based on “hot” carriers are also being investigated since operationally they would be very simple devices, although their implementation poses almost overwhelming challenges.

The fi nal Centre research strand involves silicon photonics where the emphasis is upon using our experience with solar cells, using light to produce electricity, to the reverse problem of engineering silicon devices that use electricity to produce light. Devices like this would be of great interest in microelectronics since they would allow optical processes to be integrated with the normal electronic functions on the same silicon microchip. The Centre now holds the international record for the light emission performance from bulk silicon, in both electroluminescent and photoluminescent categories. Emphasis is now upon incorporating these improvements into silicon microchip-compatible light emitting devices using silicon-on-insulator wafers.

In addition to these four research strands, the activities of the former Key Centre for Photovoltaic Engineering have been integrated into the ARC Centre of Excellence. The Centre has successfully achieved all of the performance targets in the teaching area originally agreed with the Australian Research Council. The fi rst students from the Bachelor of Engineering (Photovoltaics and Renewable Energy) program graduated during the year. This program has been enormously successful, attracting some of the best students entering the University. The Centre’s second undergraduate program, leading to a Bachelor of Engineering (Renewable Energy) completed its second year of operation.

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

2

DIRECTORS' REPORT

After a somewhat later commencement date than originally anticipated, the Centre is now pushing forward at full speed on its 5-year plan. We thank all those who contributed to its success during 2004, with special thanks to the Australian Research Council, the NSW Department of State and Regional Development and the University of New South Wales for their on-going support. The present continues to be a most exciting time for photovoltaics with the industry showing another year of very rapid growth. The year also saw heavy demand pushing oil and coal prices to recent highs and atmospheric Greenhouse gas concentrations continuing to escalate alarmingly. A focus on more sustainable ways of generating future global energy needs is clearly needed, with the Centre playing an important role in contributing to such a focus, both nationally and internationally.

Professor Stuart R. Wenham, Director

Professor Martin A. Green,Executive Research Director

00

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

DIRECTORS' REPORT

3

Technology Commercialisation

2004 was an exciting year for the photovoltaic industry with global production increasing by a massive 67% to above 1GW/year. It was also an important year for the commercialisation of the Centre’s technology. The Centre’s fi rst generation “buried contact” technology benefi ted from licensee BP Solar’s move to a larger facility at Tres Cantos in Spain, with buried-contact production exceeding $100 million per year, for the fi rst time. Accumulated sales are now likely to exceed $1 billion by 2008. The commercialisation of the Centre’s second generation crystalline silicon on glass (CSG) thin-fi lm technology also made an important leap forward with the formation of a new company, CSG Solar, to take the technology into production in 2006. In a 40 million Euro ($70million) initiative, construction of a 9,000 m2 manufacturing facility began in February, 2005 in Thalheim, Germany, to take advantage of Germany’s booming photovoltaic market. The company plans to sell 1.4 m2 modules initially of 7.5-8.5% effi ciency from early in 2006.

Education

The Centre’s Bachelor of Engineering in Photovoltaics and Solar Energy is the fi rst program of its kind in the world. Its importance has been recognised through the receipt of the Education and Awareness Award at the 2004 Energy and Water Green Globe Awards held by the Department of Energy Utilities and Sustainability (DEUS) on the 23rd November 2004.The Energy and Water Green Globe Awards encourage greenhouse reduction and sustainable use of resources by recognising excellence in the implementation and promotion of energy and water effi ciency initiatives by businesses, individuals, government departments, and members of the renewable energy industry and retail sector. The Photovoltaics and Solar Energy program at UNSW received the Education and Awareness award for achievement in education and awareness about sustainable energy or water supply. It was received on behalf of the Centre by the Head of School, Dr Richard Corkish.

Thin-Film Silicon

Good progress was achieved with all three Centre thin-fi lm cell technologies, EVA, ALICIA and ALICE, as demonstrated by large improvements of 90-150 mV in open-circuit voltages over the year. All three technologies now have voltages in the 400-460 mV range and voltages of 500-550 mV seem well within the reach. A signifi cant fraction of these improvements was due to additional equipment (in particular, a rapid thermal annealer and a PECVD cluster tool) that became available during 2004 at the Centre’s Bay Street research facility. This facility includes a state-of-the-art cleanroom acquired in January from the local photovoltaic company CSG Solar (formerly Pacifi c Solar). Thin-fi lm cell metallisation and interconnection also progressed strongly, leading to the group’s fi rst reasonably effi cient metallised cells (4 % using CSG Solar material). In the characterisation area, a powerful new method was developed for measuring the base doping concentration and the junction built-in potential of non-ideal devices. The thin-fi lm group has again been active in generating intellectual property, with the fi ling of a provisional Australian patent on thin-fi lm cell interconnection methods and an International PCT patent application on a glass texturing method.

00

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

4

HIGHLIGHTS

2004 World Technology Award For Energy

On Friday, 8th October at a gala dinner in the rotunda of the historic San Francisco City Hall, the World Technology Network presented its annual awards to “outstanding innovators within the technology arena”. The overall winner in the Energy category was the Centre’s Executive Research Director, Professor Martin Green in recognition of his work boosting commercial prospects for solar power generation using silicon solar cells. The awards event and the associated World Technology Summit was held in association with major players on the technology front including NASDAQ and Microsoft as well as Time, Fortune and Science magazines. In accepting the award, Professor Green expressed his appreciation for the recognition of his work by such an esteemed group and thanked the many students and colleagues who had contributed. He said his work had been motivated by the clear need for a low-cost method of generating solar electricity and believed the recent work with which he had been involved on silicon thin-fi lm cells had fi nally made this possible. The World Technology Network is composed of over 700 scientists, entrepreneurs, fi nanciers, policy makers and other leading players in the technology fi eld. Nominations for the World Technology Awards are accepted from the Network membership, with members then ranking the fi nal list of nominees. Professor Green’s work has been recognised by several other major international awards including the Australia Prize in 1999 and, more recently, the 2002 Right Livelihood Award, also known as the Alternative Nobel Prize.

Nanostructured Silicon Cells

The work on an innovative “all-silicon” nanostructured, stacked solar cell took an important step forwards in 2004. In addition to the fi rst electrical measurements on Si quantum dots (QDs) in an oxide matrix, Si QDs in a Si3N4 matrix were successfully grown by both PECVD and sputtering. Thin Si rich layers in nitride were demonstrated by SIMS and QDs were observed in high resolution TEM, with lattice planes clearly visible in many of the dots. This indicates successful transfer of the Si QD technology from oxide to nitride and is signifi cant in that electrical transport in the nitride will be much faster than in the oxide because of the lower barrier heights. This will reduce the tightness of specifi cation for QD density and size distribution required for the engineered wide-bandgap material to be used in an all-Si tandem cell. (See Si Nanostructures in section 4.5.) This followed on from the use of these quantum dots as selective energy contacts for the Hot Carrier Cells project. Negative diff erential resistance (NDR) was measured in a double barrier resonant tunnelling structure with 4nm Si quantum dots (QDs) as the resonant centres in an oxide matrix, using 5nm SiO2 barriers.

Wafer-based cells

As mentioned world-wide production of wafer-based silicon cells increased by a staggering 67% during 2004, with the Centre maintaining a position at the forefront of the associated international research and development. During the year, the Centre moved a step closer to a new world record on phosphorus-doped, n-type silicon, by shifting the boron-diff used emitter from the front to the rear surface of the cells. These rear emitter, n-PERT cells demonstrated 22.1% effi ciency, measured at UNSW relative to a Sandia-calibrated reference cell. Exceptionally high open-circuit voltage of 706 mV from these n-PERT cells shows the low levels of carrier recombination, and the potential for even higher performance. The Centre also continued its work on high-effi ciency buried contact solar cells on n-type silicon wafers, leading to cell effi ciencies as high as 19.2% on textured phosphorus-doped wafers and 16.8% on untextured, oxide-coated Czochralski-grown n-type wafers. Work continued with Centre partners on getting these and related improvements into commercial production. 00

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

HIGHLIGHTS

5

Director Stuart Ross Wenham, BE BSc PhD UNSW, FTS, SMIEEE, FIEAust(Scientia Professor)

Executive Research DirectorMartin Andrew Green, BE MEngSc Qld., PhD McMaster, FAA, FTS, FIEEE(Scientia Professor and Federation Fellow)

Deputy DirectorsArmin G. Aberle, BSc MSc PhD Freiburg, Dr habil Hannover, SMIEEE, MDPGGavin Conibeer, BSc MSc London, PhD Southampton Christiana B. Honsberg, BEEE MSc PhD DelawareJianhua Zhao, BE ME Nanjing I.T., PhD UNSW

Business & Technology ManagerMark D. Silver, BE UNSW, GMQ AGSM

Undergraduate Academic AdvisorJeff rey E. Cotter, BEE MSc PhD Delaware

Postgraduate Program Co-ordinatorAlistair B. Sproul, BSc PhD UNSW

Associate ProfessorsArmin G. Aberle, BSc MSc PhD Freiburg, Dr habil Hannover, SMIEEE, MDPGChristiana B. Honsberg, BEE MSc PhD DelawareThorsten Trupke, PhD Karlsruhe Jianhua Zhao, BE ME Nanjing I.T., PhD UNSW

Senior LecturersJeff rey E. Cotter, BEE MSc PhD DelawareAlistair B. Sproul, BSc Sydney, PhD UNSWGeoff rey J. Stapleton, BE UNSWMuriel Watt, BSc N.E., PhD Murdoch

Research FellowsPatrick Campbell, BSc BE PhD UNSWRichard Corkish, BEng (Com) RMIT, PhD UNSWXiming Dai, BSc Zhejiang, PhD UNSWAihua Wang, BE Nanjing IT, PhD UNSW

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

6

STAFF LIST

Annual R

eport 20

04

Staff List

Post-Doctoral FellowsRobert Bardos , BSc (Hons) MelbourneAndrew Brown, BE W’gong, PhD UNSW Eun-Chel Cho, BE ME Kwangwoon, PhD UNSW Gavin Conibeer, BSc MSc London, PhD SouthamptonTammy Humphrey, BSc UNSW (part-time from June 2004)Hamid Mervaarz, BE (Iran), PhD UNSWTom Puzzer, BSc PhD UNSWBryce Richards, BSc VUW, MEngSc PhD UNSWPer Widenborg , MSc Stockholm, PhD UNSW

Adjunct FellowsKylie Catchpole, BSc PhD ANUDidier Debuf, BE ME PhD UNSW

Visiting Professors/FellowsDavid Cahen (Weizmann Inst. Of Science, Israel)Rulong Chen (Suntec, China)Silke Krawietz, DipL. Arch. PhD TU Darmstadt Hua Li (Suntec, China)Keith McIntosh, BSc Syd., PhD UNSWTim Ohno, BScPhys. (Hons) Alberta, PhD MarylandStefan Reber, BSc PhD Freiburg Peter Würfel, Dipl. Physics PhD Karlsruhe

Research AssociatesYidan Huang, BSc AnhuiDaniel Inns, BE UNSW (until January 2004)Bernard Vogl, BE Regensburg (until June 2004)Guangchun Zhang, BE ME Shandong

Visiting ResearchersJens BirkholzHolger HabenichtStephan Trunk

Technical Support StaffGordon Bates, BA(IndDes) UTSBruce Beilby, BSc Monash, PhD UNSWWan-Lam Florence Chen, BE UNSWMatthew Edwards, BE UNSWKate Fisher, BE UNSWJamie Green, BE UNSW Mark Griffi n, BE UNSW

00

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

7

Annual R

eport 20

04

Staff List

Robert Largent, AS USATom Puzzer, BSc PhD UNSWTimothy Seary, InstrTradeAdvCertIndElectrElecTrade TAFENicholas Shaw, BE UNSW , PhD UNSWBelinda ThorneJules Z.S. Yang, BSc Eastern ChinaAlan Yee, BE UTS

Casual Teaching StaffAnna Bruce, BE UNSWMarlon Kobacker, BE BA UNSWBryce Richards, BSc VUW, MEngSc PhD UNSWRobert Passey, BSc UNSW, BAppSc Murdoch, MSc Murdoch, PhD UNSW

Casual StaffNino BorojevicDamian CamorealeKian ChinZoe JenningsNancy Sharopeam

Administrative Office ManagerTrichelle Burns, BCom UNSWLisa Cahill

Financial OfficerJulie Kwan

Computer System OfficerLawrence Soria, AssocDipCompAppl W’gong

Administrative StaffJenny HansenJill LewisAna Naumoska

PhD StudentsMalcolm Abbott, BE UNSWWan-Lam Florence Chen, BE UNSWYoung Hyun Cho, BSc Hanyang U, MEngSc UNSWPeter Cousins, BE UNSWMatthew Edwards, BE UNSW (since September, 2003)

00

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

8

STAFF LIST

Annual R

eport 20

04

Staff List

Thipwan Fangsuwannarak, BEng V’kul, MEng C’kornJiun-Hua “Allen” Guo, BSc MSc, Nat.Taiwan U.Anita Ho, BE UNSWYidan Huang, BSc AnhuiDaniel Inns, BE UNSWChu-Wei “Scott” Jiang, BSEE Nat.Taiwan U, MEngSc Qld.Kuo-Lung “Albert” Lin, BE Tatung, MSc Liverpool (October, 2004 graduation)Ly Mai, BE UNSWSupriya Pillai, BTech M.G.UGiuseppe Scardera, BSc McGill , BASc OttawaAvi Shalav, BSc MSc MasseyDengyuan Song, BSc MSc HebeiAxel Straub, BE ME UlmMason Terry, BSEET Oregon IT, MSEE WashingtonAttachai Ueranantasun, BE KMUTT, T’land, MEngSc UNSWTimothy Walsh, BSc Melbourne, (Hons) ANUPer Widenborg, BSc MSc Stockholm, PhD UNSW (April, 2004 graduation)Johnny Wu, BE BSc Qld.Guoxiao “James” Yao, BE UT Zhegiang, MS Sweden

Masters StudentsNatapol Chuangsuwanich, BEng Chulalongkorn (until March 2004)Jirka Stradal Shahril Sulaiman Bernard Vogl, BE Regensburg (from June 2004)

Visiting Students (Practicum)Violaine Barroux

Undergraduate Research StudentsDaniel BallNino BorojevicSin Von ChanHoward ChengJesse CopperJoel CourtneyDavid DiBonne Eggleston Raphael GebsBrett HallamHolger HabenichtJackson HeSuzie Hunter

00

Tarek Kassab Daniel KongAndrew LiFred Martin-BruneJohanna MayShervin MotaharIvano PolaEdwin PinkDaniel ParkerJames RuddAdaline SugiantoBelinda Thorne

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

9

4.1 INTRODUCTION TO RESEARCH

Ann

ual R

epor

t 2

00

4

R

esea

rch

Photovoltaics, the direct conversion of sunlight to electricity using solar cells, is recognised as one of the most promising energy options for a sustainable future. The Centre commenced in mid-2003, combining previous disparate strands of work supported under a variety of programs into a coherent whole, addressing the key challenges facing the field of photovoltaics, as well as “spin-off” applications in microelectronics and optoelectronics.

The photovoltaics research is divided into three interlinked strands addressing near-term, medium-term and long-term needs, respectively. The present photovoltaic market is dominated by “first-generation” product based on silicon wafers, either single-crystalline as in microelectronics or a lower-grade multicrystalline wafer (Fig. 4.1). This market dominance is likely to continue for at least the next decade. Production of first-generation product is growing rapidly, with the technological emphasis upon streamlining manufacturing processing to reduce costs while, at the same time, improving energy conversion efficiency. Other key issues involve reducing the manufacturing spread on multicrystalline wafer lines caused by variability in wafer quality (typically giving 20% spread in cell output) and elimination of the effects of boron-oxygen defects in both types of wafers. Under illumination, these defects become active and reduce the performance of most commercial modules by about 3%. They also constrain the specification of the starting silicon wafer, thereby restricting cell design possibilities. Also important is the reduction of the thickness of the starting silicon wafer without losing performance, to save material costs, and the development of low-cost techniques for reducing reflection from multicrystalline cells.

The Centre’s first-generation research is focussed on these key issues. Major emphasis is upon the “buried-contact” solar cell, originally developed by Centre researchers, the first of the modern high-efficiency cell technologies to be successfully commercialised (Fig. 4.1). Centre research seeks improvements to these devices to increase efficiency, particularly for devices fabricated on thin wafers. Also being investigated is the development of buried-contact sequences for substrates doped with phosphorus, rather than boron, to avoid the boron-oxygen defect problem previously noted. With support from the University, the Centre also completed installation during 2004 of a remote-plasma silicon nitride deposition system that will allow the issue of production spread with multicrystalline wafers to be investigated. The nitride coatings from the new system will also complement texturing approaches to reducing reflection losses in such wafers.

10

RESEARCH

Figure 4.1: “First-generation” wafer-based technology (BP Solar Saturn Module, the photovoltaic product manufactured in the highest volume by this company in Europe, using UNSW buried-contact technology).

Ann

ual R

epor

t 2

00

4

R

esea

rch

11

Since wafers are expensive and need quite elaborate and expensive encapsulation, because they are brittle and also thermally mismatched to the glass coversheet, first-generation technology is inherently material-intensive. To avoid the associated cost penalties, several companies worldwide are attempting to commercialise “second-generation” thin-film cell technology based on depositing thin layers of the photoactive material onto supporting substrates or superstrates, usually sheets of glass (Fig. 4.2). Although materials other than silicon have been suggested for these films, silicon avoids problems that can arise with these more complex compounds due to stability, manufacturability, moisture sensitivity, toxicity and resource availability issues. CSG Solar, a partner in the Centre, is commercialising an original approach by Centre researchers that is unique in that it is based on the use of the same high quality silicon used for first-generation production, but deposited as a thin layer onto glass.

Centre support for CSG Solar is provided through characterisation services and consultancies with Centre staff. The Centre's level of collaboration is steadily increasing as this technology becomes commercially available and the need for confidential research decreases. The Centre currently maintains a largely independent program addressing alternative solutions to those adopted by CSG Solar for producing high-performance “silicon-on-glass” solar cells.

At the present time, the second-generation thin-films are just starting to enter the commercial market. Past failures with alternative thin-film technologies have made the market somewhat cautious. However, the “crystalline-silicon-on-glass” (CSG) approach developed by CSG Solar appears to avoid past deficiencies and to have some additional strengths, compared to wafer-based product, apart from lower cost. The successful commercialisation of such a product planned for 2006 is expected to lead to a very different manufacturing cost structure. However, costs again are expected increasingly to become dominated by material cost as production increases, for example, by the cost of the glass sheet on which the cells are deposited.

More power from a given amount of material is possible by increasing energy-conversion efficiency. This leads to the concept of a third-generation of solar cell distinguished by the fact that it is both high-efficiency and thin-film. To illustrate the cost leverage provided by efficiency, Fig. 4.3 shows the relative cost structures of the three generations being studied by the Centre. This figure plots efficiency against manufacturing cost, expressed in US$/square metre. First-generation technology has relatively high production cost per unit area and moderate likely efficiencies (10-20%). The dotted lines in Fig. 4.3 show the corresponding cost/watt, the market metric. Values as low as US$1/watt may be feasible by increasing the efficiency while reducing manufacturing cost, but this is the likely limit of this approach (present manufacturing costs for large manufacturers are as low as US$2-3/watt).

Figure 4.2 Example of “second-generation” thin-film technology (module fabricated on CSG Solar’s pilot-line in Sydney, based on thin-films of polycrystalline silicon deposited onto glass, again UNSW-developed technology).

Second-generation thin-film technology has a different cost structure as evident from this figure. Production costs per unit area are a lot lower, since silicon wafers are a lot more expensive than glass sheet. However, likely energy-conversion efficiencies are lower (5-15%). Overall, this trade-off produces costs/watt estimated as about 2 to 3 times lower than those of the wafer product, in large production volumes.

The third-generation is imagined as a thin-film technology, which therefore has manufacturing costs similar to second-generation, but is based on operating principles that do not constrain efficiency to the same limits as conventional cells (31% for non-concentrated sunlight for these). Unconstrained thermodynamic limits for solar conversion are much higher (74% for non-concentrated light), giving an indication of the potential for improvement. If a reasonable fraction of this potential can be realised, Fig. 4.3 suggests that third-generation costs could be lower than second-generation by another factor of 2 to 3.

Of the third-generation options surveyed by Centre researchers, “all-silicon” tandem cells based on bandgap-engineering using nanostructures was selected as the most promising for implementation in the Centre’s timescale (Fig. 4.4). This involves the engineering of a new class of mixed-phase semiconductor material based on partly-ordered silicon quantum-dots in an insulating amorphous matrix. Photon up- and down-conversion as a way of “supercharging” the performance of relatively standard cells forms a second line of research. A third is the investigation of schemes for implementing hot-carrier cells. Although prospects for demonstrating these are deemed small in the time-scales involved, this phase is being continued due to an overlap with the engineered quantum-dot semiconductor work mentioned above, together with the appealing operational simplicity that may be possible with this approach.

100100

8080

6060

2020

00 100100 200200 300300 400400 500500

4040

Effic

ienc

y, %

US$0.10/W US$0.20/W

US$1.00/W

US$3.50/W

Present limit

ThermodynamicLimit

US$0.50/W

Cost, US$/m2

Ann

ual R

epor

t 2

00

4

R

esea

rch

12

Figure 4.3: Efficiency and cost projections for first-, second- and third-generation photovoltaic technology (wafers, thin-films and advanced thin-films, respectively).

Ann

ual R

epor

t 2

00

4

R

esea

rch

13

The fourth “spin-off” Centre strand of silicon photonics draws upon elements of all three of the photovoltaic strands. The aim is to develop silicon light-emitting-diodes (LEDs) that can be integrated into silicon integrated circuits, with their output capable of modulation at gigahertz frequencies (Fig. 4.5). This is closely related to the high-efficiency cell work in our first-generation strand of research, since the associated Centre researchers have demonstrated the highest-performing LEDs based on bulk silicon by a very clear margin, internationally. The work also draws upon the expertise in the second-generation strand since thin films of silicon are integral to the silicon-on-insulator (SOI) wafers used in this work. “Light-trapping” is also very relevant to both strands. The common interest in quantum-confinement in silicon also provides strong links to the third-generation strand. Another objective being pursued in the silicon photonic strand is the demonstration of the first silicon laser. This is the “holy grail” of the silicon photonics field.

Figure 4.4: Conceptual design of an all-silicon tandem cell based on Si-SiO2 ( or Si-Si3N4 or Si-SiC) quantum dot superlattices. Two solar cells of different bandgap controlled by quantum dot size are stacked on top of a third cell made from bulk silicon.

Cell 3 - Si QD, dia. < 2nm, Eg = 2.0 eV - less than 200nm thick - Voc = 1180mV, Jsc = 10m A.cm2

Cell 2 - Si QD, dia. < 4nm, Eg = 1.5 eV - less than 200nm thick - Voc = 880mV, Jsc = 10m A.cm2

Cell 1 - Bulk Si, Eg = 1.1 eV - less than 1000nm thick - Voc = 560mV, Jsc = 10m A.cm2

- Si QD, dia. < 4nm, Eg = 1.5 & 2.0eV - less than 40nm thick - Specific R < 5ohm. cm2

Junctions

- Bulk Si, Eg = 1.1 eV - Sheet rho < 5000 ohm/square

Rear BSF

Emitter - Si QD, dia. < 2nm, Eg>2.0 eV(or bulk Si) - less than 40nm thick - Sheet rho < 2500 ohm/square

Ann

ual R

epor

t 2

00

4

R

esea

rch

14

Figure 4.5: One of the concepts for an integrated silicon light-emitter/modulator presently being investigated experimentally. Modulation is sought using the quantum-confined Stark effect in thin silicon-on-insulator (SOI) layers.

Ann

ual R

epor

t 2

00

4

R

esea

rch

RESEARCH

15

4.2 Facilities and Infrastructure

The ARC Centre of Excellence for Advanced Silicon Photovoltaics and Photonics is located at the Kensington campus of the University of New South Wales, about 6 km from the heart of Sydney and close to its world famous beaches including Bondi, Coogee and Maroubra (Fig. 4.6).

Organisationally, the Centre of Excellence is located within the Centre for Photovoltaic Engineering (CPVE) within the Faculty of Engineering. The Centre of Excellence has a large range of laboratory facilities (Fig. 4.7); these include the Photovoltaics Research Laboratory, the Device Characterisation Laboratory, the Optoelectronic Research area, the Thin-Film Cell Laboratory and the Industry Collaborative Laboratory. Off campus the Centre has a Thin-Film Cleanroom facility at Botany, 5km south-west of the main campus. During 2004 the development and acquisition of laboratory equipment and infrastructure continued with specifi c details about signifi cant additions found under the laboratory headings which follow. Systems work is also facilitated through the equipment at the Little Bay Research Facility (near Maroubra, see Fig. 4.6) which towards the end of 2004 was being prepared for relocation to a new site on the Kensington campus. Another important resource is the Semiconductor Nanofabrication Facility jointly operated by the Faculty of Science and the Faculty of Engineering. In addition a school offi ce area for the CPVE has been established on the Lower Ground fl oor Most of these facilities are shown in Fig. 4.7 and further details on each are given below.

Additional equipment commonly used for solar cell work is available on the University campus. Included in this category are electron and focussed ion beam microscopes, X-ray diff raction, Raman spectroscopy, surface analysis and photoluminescence equipment. TEM, elipsometry and high-precision I-V equipment is also regularly accessed at Sydney University.

Sydney CBD

Kensington

BotanySydney Airpor t

University ofUniversity ofNew South WalesNew South Wales

Botany LaboratoryBotany Laboratory

CSG Solar Pty LtdCSG Solar Pty Ltd

Bondi

Coogee

Maroubra

Figure 4.6: Centre of Excellence location in Sydney.

Ann

ual R

epor

t 2

00

4

R

esea

rch

The Centre of Excellence has two computer networks. One is used for research and general administrative purposes, and the other by students enrolled in the Undergraduate Degree Courses in Photovoltaics and Renewable Energy. The Research and Administrative network consists of 3 fi le-servers, two Intranet servers, two internet web-servers and over 70 workstations. An additional 14 computers are dedicated to the computer control of laboratory and other equipment. The computer resources are used for general administrative purposes, document control, laboratory support; modelling/simulations; Internet access, equipment control; and maintaining the Centre’s presence on the Internet. The Centre also has a computer controlled SCADA system and PLC network for monitoring the research laboratories and related infrastructure. The Student Computer network comprises of two computer laboratories of total 83 m2 with 4 servers, a UNSW custom CSE computer based router and 26 workstations. Students enrolled in the Undergraduate Degree Courses use these computers for computer-related coursework and Internet access.

There is also an Internet capable web-server that gathers and displays data collected from solar arrays on the roof of the building. These data can be viewed using a web browser and can be made available for Internet access. The Centre of Excellence also shares a networked UNIX machine with the School of Computer Science and Engineering. In 2004 the Centre allocated each postgraduate research student a dedicated personal computer in addition to access to existing shared computers.

The Centre laboratory facilities are developed and maintained by the Laboratory Development and Operations Team. During 2004, the team, under the leadership of Mark Silver, comprised an additional 5 full-time and 12 casual employees, including: electrical and industrial design engineers, a computer/network manager, electronic and computer technicians, and administrative staff .

16

1

2

3

4

G

LG

High Efficiency Call Lab

Buried Contact Cell Lab

Office Photovoltaics Special Research Centre

Office Centre of Excellence for AdvancedSilicon Photovoltaics and Photonics

Industry Collaboation Lab.

Thin-Film Cell Lab

Device Characterisation and Laser Lab

Accommodation

Student Computer Lab

Workshop

Centre for Photovoltaic Engineering Office

Student Project/Seminar Room

Figure 4.7: ARC Centre of Excellence laboratories and other facilities within the Electrical Engineering building.

Ann

ual R

epor

t 2

00

4

R

esea

rch

17

Photovoltaics Research Laboratories

The Centre houses is the largest and most sophisticated bulk silicon solar cell research facility in Australia, incorporating both the High Effi ciency/LED and Buried Contact Cell Laboratories. Total laboratory processing space of over 500 m2 is located in the Electrical Engineering Building and is serviced with fi ltered and conditioned air, appropriate cooling water, processing gas, de-ionized water supply, chemical fume cupboards and exhausts. There is an additional 540 m2 area immediately adjacent to the laboratories for the accommodation of staff , research students, school offi ce and laboratory support facilities. Off site, areas totalling 200 m2 are used for the storage of chemicals and equipment spare parts.

The laboratories are furnished with a range of processing and characterisation equipment including 26 diff usion furnaces, 6 vacuum evaporation deposition systems, 2 laser-scribing machines, rapid thermal annealer, four-point sheet-resistivity probe, quartz tube washer, silver, nickel and copper plating units, infrared and visible wavelength microscopes, 3 wafer mask aligners, spin-on diff usion system, automated photoresist dual-track coater, photoresist spinner, electron beam deposition system, TiO2 spray deposition, belt furnace, manual screen printer and a laboratory system control and data acquisition monitoring system.

The laser scribe tool, shown in Figure 4.8, has a 20 watt Nd:YAG laser for infrared operation (1064 nm) and an optional frequency doubler for green operation (532 nm). The work stage is CNC controlled allowing 1 micron positional accuracy and table speeds approaching 25 cm/second across an area of 15 cm by 15 cm. The tool is used primarily for Buried Contact solar cell fabrication, cutting 35-micron wide laser grooves as deep as 100 microns into silicon wafers. It can also be used to cut other suitable materials, such as stainless steel.

Figure 4.8: CNC Laser Scribe Tool

Ann

ual R

epor

t 2

00

4

R

esea

rch

Device Characterisation Laboratory

This laboratory is located on the lower ground fl oor of the Electrical Engineering Building. Associated with it is the Optoelectronic Research Lab, reception area, seminar room, offi ces for Centre staff interacting with the public and industry, including the Business & Technology Manager, External Relations Manager and Design Assistance Division Manager and computer workstations for the device modelling activities of the Centre.

The Device Characterization Laboratory houses characterisation equipment including “Dark Star”, the Centre’s station for temperature controlled dark current-voltage measurements, the Centre’s Fourier-transform infrared spectroscopy system (FTIR), admittance spectroscopy system, microwave carrier lifetime system, ellipsometer, photoconductance decay equipment, wafer probing station for SOI LED work, open circuit voltage decay measurement system (Suns-Voc)infrared microscope and equipment for spectral response and related optical measurements.

Optoelectronic Research Area

This facility was established in 2002 with several visible and near-infrared semiconductor diode lasers and optical and electrical instrumentation. This was enhanced with a second optical bench and several further pieces of equipment in 2003. The facility is used for photoluminescence and electroluminescence measurements in the visible and infrared spectral range up to wavelengths of 2500nm; photoluminescence excitation spectroscopy; luminescence experiments with simultaneous two-colour illumination and Sinton lifetime testing with the conventional fl ash-light replaced by a high-power light emitting diode array. An area separate from the Device Characterisation Area was necessary in order to meet stringent standards for avoidance of laser eye and skin exposure for users and others. It shares cryogenic cooling equipment with the Device Characterisation Area.

Thin-Film Cell Laboratory

This 40 m2 laboratory is equipped with a range of equipment for thin-fi lm deposition and patterning, including a plasma-enhanced chemical vapour deposition (PECVD) system, a sputtering system, a reactive ion etcher, a resistively heated vacuum evaporator, and an optical microscope with digital image acquisition system. Also used by the Laboratory is an electron-beam vacuum evaporator for silicon physically located within the

18

Figure 4.9: Optical characterisation bench

Photovoltaics Research Laboratory. This Si evaporator is also equipped with an ionizer unit and a sample heater, enabling fast-rate Si homoepitaxy at temperatures of about 500-600 °C by means of ion-assisted deposition (IAD). Other equipment of use in thin-fi lm projects is located within the Semiconductor Nanofabrication Facility.

The PECVD system, shown in Fig. 4.10, has a 40x20 cm2 process platen and can handle large-area silicon wafers as well as smaller pieces. Two types of plasma excitation (remote microwave and direct RF) are available. The machine is used for the low-temperature deposition of thin dielectric fi lms (silicon nitride, silicon dioxide) and of amorphous silicon. The dual-cylinder, remote microwave plasma source produces excellent-quality silicon nitride and silicon dioxide fi lms, with precise control over the stoichiometry at temperatures up to 500°C. Amorphous and microcrystalline silicon fi lms can also be deposited in the system.

In 2004 the reactive ion etcher (RIE) was upgraded with methane and hydrogen gas feeds along with a vacuum pump upgrade to provide capability for gallium arsenide substrates. The vacuum pumping system improvements delivered lower base pressure, faster pump down, and higher processing throughput. A dry pump helium leak detector was also commissioned which permitted contamination-free leak testing of vacuum

systems, rapid vacuum leak fault fi nding, and the rigorous safety testing of hazardous gas delivery systems.

Semiconductor Nanofabrication Facility

The Centre also owns equipment within, and has access to, the Semiconductor Nanofabrication Facility (SNF) at the University. This is a joint facility shared by the Faculties of Science and Engineering and houses a microelectronics laboratory and a nanofabrication laboratory for e beam lithography. In 2004 a collaborative project commenced with the SNF to e-beam pattern a photonic crystal on the surface of the Centre’s SOI LED structures for enhanced optical emission. The SNF provides an Australian capability for the fabrication of advanced nanoscale semiconductor devices and their integration with microelectronics. SNF research projects form an integrated eff ort to fabricate innovative semiconductor nanostructures using the latest techniques

of electron beam patterning and scanning probe manipulation. A major applied objective of the facility is the development of a prototype silicon nuclear spin quantum computer.

00

Ann

ual R

epor

t 2

00

4

R

esea

rch

19

Figure 4.10: Remote plasma PECVD machine.

Ann

ual R

epor

t 2

00

4

R

esea

rch

Industry Collaborative Laboratory

This 120 m2 laboratory houses equipment needed for many of the industry-collaborative research activities in the Buried-Contact Solar Cell group. The laboratory was refurbished in 1999 and several new pieces of infrastructure have been acquired or constructed since, including: a belt furnace; a state of the art laser micromachining tool; a new PECVD deposition system (located in the adjacent thin fi lm solar cell laboratory); and a TiO2 spray deposition station. In 2004 a high temperature semiconductor muffl e furnace was added along with the commissioning of a manual screen printer whilst the group awaits delivery of their production screen printing unit.

Cleanroom Facility in Bay Street, Botany

During 2004 the Centre added a 120 m2 cleanroom facility in Bay Street, Botany to its infrastructure, greatly improving its experimental capabilities in the area of thin-fi lms. This cleanroom is equipped with several fume cupboards, two tube furnaces, an electron-beam vacuum evaporator, a thermal vacuum evaporator, a glass washing machine, a rapid thermal processing (RTP) machine, and a 5-chamber cluster tool. The cluster tool presently features four plasma-enhanced chemical vapour deposition (PECVD) chambers and one lamp-heated vacuum annealing chamber. The PECVD chambers enable the low-temperature deposition of dielectric fi lms (silicon oxide, silicon nitride, etc) and amorphous silicon fi lms (either n- or p-doped or undoped). Furthermore, samples can be hydrogenated by PECVD using a hydrogen plasma at substrate temperatures of up to 480°C. During 2004 the Centre purchased a low-pressure chemical vapour deposition (LPCVD) system, an infrared NdYAG laser scriber and a box furnace for sample annealing. The LPCVD system is capable of depositing doped crystalline silicon on glass and with its additional remote plasma source is currently engaged in hydrogenation work. The NdYAG laser is used for scribing silicon fi lms and other suitable metal and dielectric materials.

20

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

RESEARCH

21

4.3 FIRST GENERATION: WAFER-BASED PROJECTS

4.3.1 BURIED CONTACT SOLAR CELL GROUP

University Staff:Dr. Jeff rey Cotter (group leader)A/Prof. Christiana HonsbergProf. Stuart Wenham

Project Scientists and Technicians:Dr. Ximing DaiJamie GreenKate FisherBudi Tjahjono Nicholas Shaw

Research Associates:Dr. Robert BardosDr. Keith McIntoshDr. Tom PuzzerDr. Alistair SproulTed Szpitalak

Postgraduate Students:Malcolm AbbottFlorence ChenPeter Cousins (until 12/04)Matthew EdwardsKate FisherJiun-Hua “Allen” Guo (until 01/05)Anita HoLy MaiGuoxiao “James” YaoAttachai “Tao” Ueranantasun

Undergraduate Thesis Students (4th year):Jackson HeNino Borojevic

Undergraduate Summer Scholars (1st Year):David DiBrett HallamAdaline SugiantoBelinda Thorne

Visiting Students:Ivano Pola (until 03/04)

The Buried-Contact Solar Cell Group aims to develop new solar cell structures and novel process technologies specifi cally for commercially relevant silicon wafers. The group has a broad spectrum of research and development activities that address the evolving nature of commercial silicon solar cells. The Group’s main activities are focused on developing high-effi ciency, thin-wafer Buried Contact solar cells, developing of low-cost processing technologies and related device designs, and transferring BC technology to industrial and technical collaborators.

4.3.1.1 HIGH-EFFICIENCY BURIED CONTACT SILICON SOLAR CELLS

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

A. LOSS ANALYSIS AND DESIGN ISSUES FOR THIN SILICON SOLAR CELLS MANUFACTURED ON FZ(B), MCZ(B), CZ(Ga) AND CZ(B) WAFERS.

There is a push within the silicon solar cell industry to reduce manufacturing costs and to improve cell performance by reducing wafer thickness using higher lifetime wafers, for example FZ(B), MCZ(B) and CZ(Ga). In 2004, we investigated the manufacturing and design issues critical to the success of this approach through the analysis of double-sided buried contact solar cells fabricated on thin and thick commercial silicon wafers. These devices were fabricated on fi ve commercially available wafer types — FZ(B), MCZ(B), CZ(Ga) and two diff erent CZ(B) wafers — at two wafer thicknesses, 150 mm and 245 µm. Analysis of the results highlight the importance of both design related issues (e.g. light trapping), and manufacturing related issues (e.g. process-induced defects) to the successful manufacture of commercial high-effi ciency silicon solar cells.

The double-sided buried contact (DSBC) solar cell (see Figure 4.11) was proposed in 1990 as a high effi ciency solution to the performance limitations of using full area rear aluminium alloyed contacts. It uses existing laser groove technology to form localised boron grooves to contact the base and a fl oating-junction to passivate the rear surface. The performance of DSBC solar cells on commercially available boron-doped Czochralski (CZ) material is limited by bulk recombination. Therefore, the performance of the DSBC solar cell can potentially be improved either through a reduction in the wafer thickness, or through the use of higher lifetime materials.

The use of thin wafers in high effi ciency solar cells, like the DSBC solar cell, has the potential to provide both a performance improvement and a cost reduction. The reduction in the size of the bulk region when using thinner wafers, reduces the sensitivity of cell performance to bulk lifetime. This is expected to result in a performance improvement in thin DSBC solar cells on CZ(B) wafers. A further advantage of using thinner wafers is an increase in silicon yield resulting from ingot slicing. This has been predicted to result in a reduction in material costs. The use of higher lifetime materials in high-effi ciency solar cells is also expected to provide an improvement in performance. The recent discovery of light-induced degradation in CZ(B) has prompted the consideration of new wafer materials for commercial applications. These new materials — FZ(B), MCZ(B) and CZ(Ga) — off er possible performance improvements through higher stabilised lifetimes.

22

Figure 4.11. Schematic of a double-sided buried contact (DSBC) solar cell.

Ann

ual R

epor

t 2

00

4

R

esea

rch

23

Experiment

DSBC solar cells were manufactured on fi ve commercially available wafer types - FZ(B), MCZ(B), CZ(Ga) and two diff erent CZ(B) wafers - at two diff erent thicknesses - 150 µm and 245 µm. The characteristics of these wafers are shown in Table 4.1.

This solar cell processing sequence consists of the following steps:

1. Emitter Windowing. An additional step prior to the standard DSBC solar cell process was included to minimise the losses of edge recombination on these small-area devices. A thermal silicon dioxide masking layer (3000 Å) was grown on the front surface, before photolithographically opening an 8 cm2 square window.

2. Texturing. Random pyramids were etched into the exposed silicon on the front and rear surfaces using a 2% sodium hydroxide based texturing solution at 90°C.

3. Emitter Diff usion and Oxidation. Phosphorus collecting and fl oating junctions were diff used into the exposed silicon on the front and rear surfaces before the thermal growth of the fi nal surface silicon dioxide layer (3500 Å).

4. Contact formation. A laser and selective phosphorus diff usion were used to form the front laser groove contacts. Similarly, the rear contacts were formed using a laser and selective boron diff usion.

5. Metallisation. Electroless palladium, nickel and copper layers were used to provide electrical contact to the front and rear grooves, and to form the conductive metal fi ngers.

The manufacturing process was monitored using photoconductance lifetime. The implied open-circuit voltage, and inverse eff ective-lifetime versus minority-carrier-density curves were measured after each of the three high-temperature processing sequences. The fi nished devices were characterised in terms of the recombination sources using I-V, Suns-Voc and m-V curves, and carrier generation using EQE, IQE and refl ection. The cells were stabilised under 1-sun illumination before repeating the device characterisation.

Material Resistivityb [Oi] (Ω.cm) (ppma)FZ(B) 0.8−1.2 0.0a

MCZ(B) 0.5−1.5 4.7CZ(Ga) 0.6−1.6 14.4CZ(B) 0.6−1.6 12.0CZ(B) 0.8−1.2 15.1

a Below measurement limitb Manufacture specifi cation

Table 4.1: Characteristics of each wafer material prior to cell manufacture.

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

Cell Results

A summary of the 1-sun parameters for typical cells of each material and thickness are shown in Table 4.2. A comparison of the 245 µm and 150 µm CZ(B) thick cells demonstrates the expected improvement in the VOC due to a reduced sensitivity in device performance to bulk lifetime. This improvement is not observed on the higher lifetime materials, demonstrating that the performance of these devices is not material lifetime limited. A signifi cant drop in the JSC was observed on all of the thinner devices due to a combination of the reduced absorption thickness and the poor light trapping within these bifacial cells. This eff ect overshadows the improvements in VOC resulting in a reduction in effi ciency for all thin devices. The manufacture of DSBC solar cells on the higher lifetime materials (FZ(B), MCZ(B) and CZ(Ga)) provided an improvement in performance. This performance improvement was a result of an improvement in the VOC, which was signifi cantly higher on both the 245 µm and 150 µm thick devices. This dramatic improvement in VOC for these materials over CZ(B) on 150 µm wafers demonstrates the importance of bulk lifetime to performance, even on thin wafers.

24

Table 4.2. DSBC solar cells manufactured on CZ(B) material after stabilisation. Rear illumination parameters are shown in brackets.

+ Cells had high series resistance

CZ(B) A 245 640 36.2 77.1 17.8

(633) (26.2) (78.1) (13.0)

CZ(B) B 245 642 36.3 77.4 18.1

(638) (27.8) (77.8) (13.8)

CZ(B) A 150 646 35.5 74.4 17.1

(644) (28.4) (78.2) (14.3)

CZ(B) B 150 645 35.6 76.7 17.6

(642) (28.8) (77.8) (14.4)

FZ(B) 245 662 36.5 76.8 18.6

(657) (28.9) (78.9) (15.0)

MCZ(B) 245 656 36.4 72.3+ 17.3

(654) (29.4) (75.1) (14.5)

CZ(Ga) 245 663 36.6 76.8 18.6

(660) (31.4) (78.0) (16.2)

FZ(B) 150 661 35.9 72.1+ 17.1

(657) (31.1) (74.3) (15.2)

MCZ(B) 150 663 36.2 75.6 18.3

(661) (31.4) (78.6) (16.3)

CZ(Ga) 150 657 35.7 75.4 17.7

(656) (29.4) (79.1) (15.3)

W VOC JSC FF ηMaterial (µm) (mV) (mA/cm2) (%) (%)

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

25

Although the use of thin wafers and higher lifetime wafers provides a performance improvement to the DSBC solar cell, the fi nal performance of these cells is sub-optimal. Analysis of the electrical and optical characteristics, shown in Figure 4.12 and Figure 4.13, as well as in-process photoconductance lifetime measurements reveal the performance of these devices are not limited by wafer material, classic junction-shunting, series resistance or even fl oating-junction shunting. Instead these devices are limited by device design and manufacturing issues; (i) bulk asymmetric Shockley-Read-Hall (SRH) recombination induced during boron diff usion, (ii) bulk lifetime degradation during the thermal oxidation of the textured surfaces, (iii) diff usion saturation currents (J0), and (iv) optical losses. The following section outlines these losses and their impact on cell performance.

Figure 4.12. Electrical characteristics of a 150-µm thick DSBC solar cell manufactured on a CZ(B) wafer; (a) 1 sun front illuminated J-V (•) and Suns-VOC (∆) curves; (b) the Suns-VOC (∆), dark-IV (•) and m-V curves. The dashed and solid lines are the theoretical model.

Figure 4.13. EQE, IQE and refl ection for a 150-µm thick DSBC solar cell manufactured on a CZ(B) wafer.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

Diffusion Induced Defects

The asymmetric Shockley-Read-Hall recombination observed in the Suns-VOC and m-V curves (see Figure 4.12b), has been attributed to boron and phosphorus misfi t dislocations generated during the groove diff usion process (see later). Asymmetric-SRH recombination is SRH recombination where the majority carrier lifetime is signifi cantly larger than the minority carrier lifetime (i.e for p-type τn << τp).

The presence of asymmetric-SRH recombination centres in the bulk region of the DSBC solar cell has an adverse eff ect on the front and rear illuminated short circuit currents. Asymmetric-SRH recombination results in high recombination for low injection (i.e. low bulk lifetime at short-circuit). This has the eff ect of reducing the collection probability of carriers generated away from the front junction, and results in a poor red response (see Figure 4.13) and low JSC values particularly under rear illumination (see Table 4.2). This reduction in rear illuminated J- is smaller for thinner devices, see Table 4.2, as the collecting junction is closer to the rear surface.

The presence of asymmetric-SRH recombination centers in the bulk region of the DSBC solar cell also lowers the fi ll factor resulting in a soft lighted I-V curve, see Figure 4.12a. This eff ect is discussed in more detail in a subsequent section (Section 4.3.1.4).

Oxidation Induced Stress

The bulk lifetimes extracted from the photoconductance lifetime measurements after emitter-diff usion and oxidation (see Figure 4.14) for the textured samples compared to a planar reference sample demonstrate a signifi cant degradation in bulk-lifetime. This eff ect is not related to the wafer material (see Figure 4.14), but instead to the formation of defects during the thermal oxidation of upright random textures. These defects are generated to relieve the large stresses resulting from oxidising sharp features.

26

Figure 4.14. The photoconductance lifetime curves after the emitter diff usion and oxidation process for a 1 Ω.cm FZ(B) wafer; 280-µm thick planar surfaces (), and 245-µm thick textured surface (∆).

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

27

The presence of bulk SRH recombination associated with oxidation induced stress in the DSBC solar cell reduces the VOC. A comparison of the VOC of the diff erent wafer materials (see Table 4.2) does not demonstrate the expected spread for the diff erent material types. Instead, these VOC values refl ect the high-injection bulk lifetime measurements after the emitter diff usion and oxidation process (see Figure 4.15). The planar samples are unaff ected by this mechanism, hence refl ect the expected bulk lifetimes of the diff erent materials. Furthermore, the thinner samples are more sensitive to the adverse eff ects of oxidation induced stress, a phenomenon that is still under investigation.

Saturation Current

Analysis of the Suns-VOC curve (see Figure 4.12b) demonstrates that saturation current is a dominant recombination source at VOC. The saturation currents measured after each high temperature process are shown in Table 4.3. It can be concluded that further optimisation of all three diff used regions is required.

Table 4.3. Average saturation currents contribution for each diff usion process for a textured wafer.

Diff usion Process J0 (fA.cm-2)Emitter and Floating-Junction 60Phosphorus Contact 50Boron Contact 35Total 145

Figure 4.15. The high-injection bulk lifetime extracted from the photoconductance lifetime curves after the emitter diff usion and oxidation process for each sample compared to the identically processed planar samples.

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

Optical Design Losses

The optical losses within the DSBC solar cell consist of front surface refl ection (i.e. fi ngers, texture), escape losses, and internal parasitic losses. The front surface refl ection losses of the DSBC solar cell are a design limitation. Refl ection measurements of the upright random texture demonstrate ideal behaviour, and the broadband refl ection due to fi ngers is typical only 2−4%. The rudimentary silicon dioxide anti-refl ection coating is suboptimal in this case, see Figure 4.13. The bifacial nature of the DSBC solar cell results in large escape losses that increase as the device thickness is reduced. Finally, optical measurements of transmission and refl ection at long wavelengths have identifi ed a large internal parasitic optical loss, a phenomenon that is also still under investigation.

Design and manufacturing issues are critical to achieving effi ciency improvements on thin or higher lifetime silicon wafers. Oxidations and diff usions have been demonstrated as critical manufacturing issues; and optical design, saturation currents and surface coating selection as critical design issues. These issues need to be overcome to enable the DSBC solar cell, and other high effi ciency structures, to take advantage of reduced manufacturing costs, and improved performance available from the use of thinner and/or higher lifetime wafers.

B. 19.2% EFFICIENCY N-TYPE LASER-GROOVED SILICON SOLAR CELLS

N-type silicon wafers have been found to have higher bulk lifetimes compared to those of boron-doped p-type silicon wafers with the same resistivity, and proved to have no light-induced degradation associated with the boron-oxygen complex. In 2004, we demonstrated the use of laser grooved BC technology to fabricate Interdigitated Backside Buried Contact (IBBC) solar cells, as well as the more conventional DSBC solar cell. Three obstacles that have hindered the performance of n-type Interdigitated Backside Buried Contact (IBBC) solar cells—parasitic shunt resistance, metallization issues, and optimization of the diff used regions—are discussed. An effi ciency of 19.2%, achieved on the n-type IBBC solar cell, demonstrates the high-effi ciency potential of buried contact silicon solar cells and of n-type silicon wafers. Moreover, all BC solar cells made on phosphorus-doped n-type silicon wafers, regardless of the crystalline growth techniques, show no light-induced degradation.

Design and Fabrication of N-Type IBBC and DSBC Solar Cells

Table 4.4 lists the processes of the n-type IBBC solar cell (see Figure 4.16), which are similar to those of the DSBC solar cells, except the location and the layout of the contact grooves. Solar cells are 8 cm2 and are fabricated on (100) phosphorus-doped 1 Ω.cm FZ, CZ, and multicrystalline wafers respectively. The n-type CZ wafers were supplied by Shin-Estu Handotai Co. Ltd (SEH) with oxygen and carbon concentrations of less than 14.8 ppma and 0.24 ppma respectively. The n-type multi-crystalline wafers were supplied by Eurosolare SpA. These multi-crystalline wafers did not receive gettering prior to the fabrication.

28

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

29

Figure 4.16. (a) Schematic cross section of the n-type IBBC solar cell and (b) DSBC solar cells(not to scale)

Cu

p++ groove

p+ emitter

Illumination

n-type Si n++ groove

SiO2

SiO2

n++p+ overlap

n+ FSF

n-type Si

Illumination

Cu

SiO2

p+ emitter

p++ groove

SiO2

n+ BSF

n++ groove

Process Steps

FSF formation 1. Phosphorus solid source diff usion

2. TCA oxidation forming etching mask

3. Spin-on photoresist on FSF side

4. NaOH etching on the backside n+ diff usion

p+ emitter 1. Boron liquid source deposition and drive-in

diff usion 2. HF dip Oxidation

Oxidation 1. Dry-wet-dry oxidation

2. Nitrogen anneal

n++ groove 1. Laser groove scribing and groove etching

formation 2. Phosphorus solid source diff usion and drive-in

3. Dry oxidation to form diff usion mask against the following

p++ groove diff usion

p++ groove 1. Laser groove scribing and groove etching

formation 2. Boron liquid source deposition and drive-in

Metallization 1. Groove deglazing

2. Palladium chloride activation

3. Electroless nickel plating and sintering

4. Electroless copper plating

5. Oxide anti-refl ection fi lm thinning

Table 4.4. Process for n-type IBBC/DSBC solar cells

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

Cell DevelopmentParasitic Shunt Resistance

A parasitic shunting eff ect dominated early IBBC devices, not only degrading solar cell performance, but also making optimization of IBBC solar cells almost impossible. The main shunting routes in n-type IBBC solar cells are concluded as follows: (1) the emitter contact metal touching the n-type substrate, which is either due to non-uniform boron deposition, or diff usion-induced misfi t dislocations that lead to excessive shunting leakage by means of recombination via the dislocation core state; (2) the base contact metal touching the p+ emitter, attributed to either the phosphorus groove diff usion being unable to compensate for the boron emitter diff usion, or the junction depth located in the diff usion overlap regions being not deep enough to prevent nickel from spiking through the groove diff usion, as illustrated in Figure 4.17. Both of these shunting paths can be removed by well-controlled boron diff usions. The shunt resistance of the IBBC cells increased by more than two orders of magnitude after eliminating the shunt mechanisms, leading to an improvement in FF from 0.71—0.73 to 0.74—0.76, and an increase of average absolute effi ciency of more than 0.65%.

Metallization

Laser-grooved solar cell designs with more sophisticated contacting schemes, for example the DSBC and the IBBC solar cells, require more care in the metallization process because both phosphorus and boron doped grooves are used. Adding to the problem is the fact that the etching rates and the thicknesses of the borosilicate glass (BSG) and borophosphosilicate glass (BPSG) fi lms inside these grooves and on the wafer surface are diff erent. Furthermore, the simultaneous nucleation of nickel on heavily phosphorus- and boron-doped grooves is diffi cult due to the preferential deposition of nickel on n++ diff used silicon regions. An improved electroless metallization process with two aspects—improved deglazing selectivity and improved nucleation and deposition of nickel and copper using immersion palladium activation—has been developed. In terms of etching selectivity of BSG/SiO2 and BPSG/SiO2, 1%HF, 1% NH4F dilute BHF solution is recommended for deglazing DSBC and IBBC cells. This dilute BHF is easy to use and avoids the risk of etching the underlying silicon.

30

Figure 4.17. Suprem-4 simulation of the n++-p+ diff usion overlap region, showing that the shunting can occur (a) at the localized, uncompensated diff usion region, or (b) at the shallow junction.

When palladium activation is incorporated in metallization, heavily doped p-type grooves are plated uniformly and the specifi c contact resistance is signifi cantly improved. The improvement in n-type grooves is not so obvious and over-plating is observed, although this over-plating is greatly reduced if palladium sintering is carried out after activation. N-type DSBC solar cells metallized by the improved process show signifi cantly smaller series resistance and more narrowly distributed FF compared to those without activation. The improvement on the absolute effi ciency is more than 3%.

Optimization of Diffusion Regions

Controlling and reducing recombination in the diff used regions has long been recognized as an essential path to achieving high effi ciency solar cells. Therefore, recombination in the front n+ FSF, the rear p+ emitter, as well as in the contact groove regions of the IBBC solar cells has to be minimized. Boron diff usion is the most critical process in the fabrication of the n-type solar cells, because of (1) its high potential to introduce diff usion-induced misfi t defects; (2) diffi culties in achieving uniform diff usion; and (3) higher eff ective surface recombination velocity of oxide passivated boron-doped silicon surfaces.

Photoconductance lifetime measurements are utilized extensively as the process monitoring and optimization tools at UNSW. Parameters, such as the eff ective lifetime, emitter saturation current density of the diff used layer, and the implied open-circuit voltage, obtained from the photoconductance lifetime measurement are used to evaluate the diff used regions of the solar cells. We found that the optimal sheet resistances of FSF for planar and textured surfaces are 120 Ω/sq. and 105 Ω/sq., respectively. Optimal sheet resistance for the boron emitter is around 65 Ω/sq. right after the diff usion and becomes approximately 100 Ω/sq. after the diff usion-mask oxidation. Moreover, sheet resistance as heavy as 10-20 Ω/sq. for the boron groove diff usion and 5-10 Ω/sq. for the phosphorus groove diff usion, without introducing the diff usion-induced dislocations, have been achieved. These optimization results led to the absolute effi ciency improvement on the planar n-type IBBC solar cell of more than 0.6%.

N-Type Solar Cell Results and Analysis

Based on the above-mentioned improvements on the parasitic resistance and the diff usion region optimization, excellent n-type IBBC and DSBC solar cell results have been achieved. Table 4.5 summarizes the output parameters of the n-type BC solar cells. PC-1D modelling was used to extract the eff ective front and rear surface recombination velocities by the curve fi tting of the quantum effi ciency curves. The eff ective front surface recombination velocity of the IBBC device is approximately 350 cm/s and 600 cm/s for the planar and textured surface respectively, and the rear surface recombination velocity ranged from 4800 cm/s to 7500 cm/s, depending on the process of the p+ boron emitter. The eff ective front and rear surface recombination velocities of the DSBC device are 4200 cm/s and 2500 cm/s. These values are excellent considering no hydrogenation treatment, such as alneal and forming gas anneal, is applied to further improve the surface passivation.

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

31

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

The bulk lifetime extracted from cell IBBC-D8 was more than 1.5 ms, showing that the laser-grooved processing was able to maintain the high post-processing bulk lifetime required for the back collecting-junction device. The bulk lifetime extracted from cell DSBC-B4-8 was approximately 550 µs, suggesting that the optimal wafer thickness is 150 µm when IBBC device is optimised. The extracted post-processing bulk lifetime of the multicrystalline based cells is about 10—20 µs, indicating that the low-quality substrates are better suited to the DSBC design. Signifi cant effi ciency improvement on n-type mc- Si solar cell can be expected after the implementation of gettering, light trapping, and optimizing cell processing.

Excellent results on the n-type IBBC solar cells have been achieved - 19.2% front illumination effi ciency on textured fl oat-zoned silicon wafers (single-layer SiO2 anti-refl ection coating). It is concluded that the IBBC design is superior to the DSBC design in terms of both VOC and JSC on high-quality silicon wafers, due to smaller eff ective front surface recombination velocity and no grid shading loss. However, a choice between IBBC and DSBC designs would be largely governed by wafer bulk lifetime and the quality of the front surface passivation in a commercial production environment. The laser-groove processing sequence is proved to have the high post-processing bulk lifetime required for the IBC solar cell design. It is suggested that the laser-grooved IBBC process with some modifi cations is a potentially commercial and cost eff ective technology for the fabrication of the commercial backside contact solar cells. The BC solar cells fabricated on the phosphorus-doped, n-type silicon show no light-induced degradation. All phosphorus-doped, n-type silicon substrates are superior to the boron-doped, p-type Czochralski silicon substrates in terms of voltage stability.

Light-Induced Degradation Test

The light-induced degradation evaluation was carried out on the sunlight simulator at UNSW. The light source consists of four 300 W ELH halogen lamps to simulate one-sun AM1.5G sunlight, which is calibrated by a secondary reference solar cell. The sample stage incorporates a cooling system, which enables the temperature of the measured

32

Table 4.5. Output parameters of n-type BC solar cells measured at UNSW under one-sun, 25°C.

*Fabricated cells are 8 cm2, and are cleaved from their host wafers.

Cell ID Wafer Thickness Surface Voc Jsc FF η Voc Jsc FF η type (µm) front/rear (mV) (mA/cm2) (%) (mV) (mA/cm2) (%)

IBBC-D8 FZ 175 planar/planar 676.8 32.8 0.766 17.0 676.0 30.5 0.762 15.7DSBC-D12 FZ 175 planar/planar 668.6 32.2 0.770 16.6 667.2 31.5 0.772 16.2IBBC-T4 FZ 190 textured/planar 664.0 37.9 0.765 19.2 658.8 28.4 0.775 14.5IBBC-B1-4 CZ 200 planar/planar 663.5 32.1 0.764 16.3 661.7 29.0 0.757 14.5DSBC-B4-8 CZ 215 planar/planar 665.0 32.6 0.776 16.8 663.4 30.6 0.778 15.8IBBC-I6 mc-Si 190 planar/planar 545.7 12.7 0.738 5.1 570.8 27.1 0.745 11.5DSBC-I4 mc-Si 190 planar/planar 586.2 30.0 0.755 13.3 571.7 19.9 0.758 8.6

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

RESEARCH

33

cell to be kept constant at 25°C. The light soaking duration is around 10 hours. Figure 4.18 shows the voltage stability of IBBC and DSBC solar cells fabricated on the various phosphorus- and boron-doped crystalline silicon wafers. Light-induced degradation was only found in CZ(B) DSBC solar cell, for which the VOC decreased from 651 mV to 635 mV. All phosphorus-doped silicon BC solar cells, regardless of the crystalline growth techniques, show no degradation.

4.3.1.2 STENCIL PRINT SILICON SOLAR CELLS

In 2004, we demonstrated the utility of laser machining stainless steel stencils for printed silicon solar cells. Laser-formed stencils are potentially cheaper than foils and screens formed by lithographic processes and may have some advantage in terms of durability, and they may also have some advantage for alignment between stencil printed features and other laser processed features on the silicon wafer. The standard laser scribing tool used for buried contact solar cell fabrication at UNSW is capable of cutting features of less than 35 µm in stencil foils 100 µm thick. By serendipity, the scribing process converts the slag by-product to an iron compound that can be selectively etched, leaving clean, well-formed stencil features. The foils have proved to be printable, with reasonably well formed features in printed and fi red pastes.

Figure 4.18. Voltage stability under ELH halogen light illumination of IBBC and DSBC solar cells fabricated on phosphorus- and boron-doped crystalline silicon wafers.

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

Stencil Fabrication Process

Metal stencils are made up of metal foils with openings where the metal paste is to be printed onto the solar cell surface. Stencils have several potential advantages over the conventional stainless steel wire mesh screen with a patterned emulsion layer, including:.

Stencils are fabricated at UNSW using a custom tool constructed by 181 Engineering Co. that comprises a 15-watt Nd:YAG laser beam incident upon a CNC stage that is controlled by a G-code MMI interface. The stage has a positional accuracy of less than 1 µm and can be moved along any linear or curved path at speeds of up to 10 inches per second.

The stage is setup to provide good vacuum adhesion and thermal contact of the foil to the stage. This keeps the foil fl at during the cutting process, ensuring that the laser is well-focused over the entire cutting area. More importantly, good thermal contact between the foil and the stage prevents heat build up and local thermal expansion in the foil. This helps prevent wrinkling and deformation of the foil during cutting. Careful control of the cutting parameters, primarily the laser power, Q-switch frequency and table speed, is also important in avoiding deformation of the foil during cutting. Stainless steel foils between 50 and 100 µm thick can be cut with beam powers of less than 2 watts, provided the beam is well focused (for example, to a spot size of less than 40 µm). Slots as narrow as 35 µm can be cut into 80-µm thick foils. The resulting cut has a slightly tapered profi le that is 5—10 µm thicker on the cutting side.

The laser ablates the stainless steel during the cutting process, and some of the molten material re-deposits and solidifi es onto the stainless steel foil, leaving rough and poorly formed stencil features (see Figure 4.19a). Fortuitously, the ablation process alters the properties of the steel slag suffi ciently to allow it to be removed by using a selective chemical etch process that attacks the slag at a much higher rate than the un-ablated stainless steel foil. A number of etches work well, as stainless steel is fairly resistant to chemical etching. We have had good success etching with HF: NH4F:H2NO3 : DI H2O at a ratio of 3 : 5 : 10 : 82 at 50ºC for about 30 minutes. Figure 4.19b shows SEM photomicrographs of typical 50-µm wide fi ngers formed in a 50-µm thick stainless steel foil before and after etching.

34

RESEARCH

• the ability to print fi ne lines with no breaks in the pattern • the ability to obtain higher paste printed height via varied thickness of metal foils; • no obstruction of printed paste transfer due to almost 100% open print area; • high printed line defi nition due to slower wearing and less deformation of the metal foil; • reduction of the downtime of the printer due to easy cleaning, less wear and long life span of the stencils

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

35

Grid Pattern Formation

One of the challenges specifi c to using a stencil to form a standard solar cell print pattern is maintaining dropouts (for example, the middle circle of a printed letter ‘O’) and tines (for example the two right-pointing arms of the printed letter ‘E’) in good form. One approach is to use short “bridges” where fi ngers meet bus bars and along the length of the bus bar to provide continuity in the foil. Figure 4.20a shows a portion of a grid pattern formed in this way, illustrating the short bridge separating the fi nger and the bus bar.

After printing, during the paste levelling step, the paste fl ows enough to allow the fi ngers to connect together and the bus bar to connect to the fi ngers, as shown in Figure 4.21a. This approach works well, but requires careful control over paste rheology and the bridge dimensions. It also presents a trade-off between ensuring that all fi ngers are suffi ciently joined to the busbar (to avoid high series resistance in the grid pattern) and minimizing the width of the printed fi nger. For example, the pattern shown in Figure 4.21a was printed through a 45µm wide fi nger opening and its fi nal width after fi ring was approximately 65 µm.

Figure 4.19. (a) Scanning electron micrograph of a laser-formed fi nger after cutting and before slag etching. (b) After etching. The resulting fi nger pattern is 50 µm wide. The stencil is 50 µm thick.

Figure 4.20. (a) Micrograph of a stencil showing busbar, fi nger and bridge features. (b) Plan view of a fi nished stencil. The patterned fi ngers are ~60 µm wide and the stencil is ~95 µm thick.

Ann

ual R

epor

t 2

00

4

Fir

st G

ener

atio

n

An alternative approach to the dropout and tines problem takes advantage of the good control of the cut depth of the laser. By controlling the laser power, it is possible to cut partially through a foil in some regions of a stencil and fully through a foil in other regions of a stencil. In this way, it is possible to cut multiple level stencils that have high strength and good print defi nition. Figure 4.22a shows a prototype two level stencil, which has a 15-µm deep “print level” and a 90-µm deep “mesh level” cut in the same laser session on the same foil.

In the two-level stencil, the second mesh level provides the strength required to maintain the continuity of the foil, while the print level defi nes the print area. Importantly, both levels can be cut in the same laser scribing session, making the process simple and fast. Figure 4.22b shows one stencil design concept that uses individual pulses of the laser to form individual holes in the mesh level, although many interesting and diff erent stencil designs possible. Further work is needed to develop the cutting process and to design the stencils for optimum stencil strength and fi lm printing.

36

Figure 4.21. (a) After printing, the paste fl ows to join fi nger and busbar. (b) Dektak profi le of the printed fi nger. The fi nger is 68 µm wide by 25 µm high.

Figure 4.22. (a) SEM image of a prototype two-level stencil. The print level was cut fi rst; it is 25 µm wide and 15 µm deep. The mesh level was cut second; it is 25 µm wide and 90 µm deep. (b) Initial design of a two-level stencil, showing arrangement of the print and mesh levels to improve the strength of the stencil.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

37

Fine Line Printing

One of the prospects for stencil printing is fi ne line printing with high aspect ratios. The laser stencil cutting process has demonstrated stencil fi nger widths as narrow as 35 µm in an 80-µm thick stainless steel foil. Using a print paste of suitable rheology, continuous fi ngers with good aspect ratios can be printed through these narrow openings. Figure 4.23a shows initial print tests on a variety of stencil opening geometries using Ferro 33384 silver print paste, including the stencil dimensions (in Figure 4.23b) and the fi nal dimensions of the printed, dried and fi red fi ngers (in Figure 4.23a). Initial aspect ratios of around 1/4 (H/W) on fi nger widths of about 50 µm has been achieved. A maximum print thickness of around 35 µm for very wide fi ngers (>130 µm) has also been achieved. Further improvement is expected with the optimization of the paste properties, the stencil design and the printing conditions and there is good potential for forming fi ne lines with higher aspect ratios.

Hybrid Buried Contact/Stencil Printed Solar Cells

One potential feature of using laser formed stencils arises from the fact that the laser tool used for stencil fabrication can be used for other types of laser processing in the solar cell fabrication sequence. This gives a small advantage in terms of aligning stencil-printed metal features to laser-processed regions of the silicon wafer. For example, stencil patterns can be made that nearly exactly correspond to a laser scribed groove pattern such as the type used in the well-established selectively diff used emitter of the buried contact solar cell process. This is possible due to the high precision and repeatability of the laser cutting process. The good pattern match can be expected to facilitate alignment

Figure 4.23. (a) Photomicrograph of grid fi ngers printed through various stencil openings. (b) Side on diagram of the stencil dimensions used for the print tests. The foil is about 80 µm thick.

Figure 4.24. Hybrid buried contact/stencil printed (BCSP) silicon solar cell design.

n+

n++SiO2

Al

Stencil-printed Ag paste into laser-formed grooves

p++P-type

n+

n++SiO2

Al

Stencil-printed Ag paste into laser-formed grooves

p++P-type

Ann

ual R

epor

t 2

00

4

Fir

st G

ener

atio

n

in the aim of incorporating selectively diff used emitters in printed silicon solar cells.In 2004, we produced the initial set of hybrid solar cells that combine stencil printed metallisation with laser-grooved selectively diff used emitters. The structure for such a solar cell is shown in Figure 4.24.

This structure has a lightly doped emitter on the wafer surface (>100 Ω/sq. sheet resistance) and diff used grooves with heavily doping (<10 Ω/sq. sheet resistance) inside the wafer, capturing the benefi t of good current and voltage typical of well-passivated selectively diff used emitters. The front metal contact is buried in the groove as well, which makes it possible for the metal contact to have a large height-to-width aspect ratio (high-to-width) and to make good ohmic contact between the Ag paste and the silicon material. The large aspect ratio metal contact, in turn, allows a large number of closely spaced fi ne metal fi ngers on the front, resulting lower shading and lateral series resistance losses and thus further enhancing the cell short-circuit current and fi ll factor. Hybrid BCSP solar cells were fabricated on 1 Ω.cm, p-type, 300-µm Cz silicon wafers with a cell area of 4 cm2. The hybrid solar cells were fi rst textured with anisotropic sodium hydroxide/isopropyl alcohol etch. Next, the emitter was phosphorus diff used (to 100 Ω/sq.) followed by a thermal oxidation for front surface passivation and to form a rudimentary anti-refl ection coating. Shallow and wide grooves into the front surface, followed by a weak isotropic sodium hydroxide etch to remove slag and laser damage. The grooves were then selectively phosphorus diff used (to 10 Ω/sq.) followed by a phosphorosilicate glass removal step. Subsequently, the rear Al metal electrode was printed on the rear surface and fi red, and fi nally, the metal electrodes were stencil printed and fi red in the grooves.

Figure 4.25 shows the initial print results of printing the fi ngers and bus bar of the hybrid BCSP solar cells. The printing was performed with a simple hand-operated printer with simple mechanical alignment between the print features and the grooves. The results of the preliminary cells are shown in Table 4.6. Signifi cant improvement is expected when the print process is transferred to the new Ekra screen printer due to arrive in early 2005 and after the process is further optimized.

38

Figure 4.25. (a) SEM image of a laser-scribed fi nger with stencil printed paste. The printed fi nger is about 60 µm wide and 20 µm high inside a 70 µm by 25 µm groove. (b) SEM image of a laser-scribed busbar with stencil printed paste. The printed busbar is about 230 µm wide and 35 µm high.

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

39

4.3.1.3 REAR LOCALISED-CONTACT BURIED CONTACT SILICON SOLAR CELLS

In 2004, the novel localised rear contacting scheme has been further developed. The aim of the work is to form small area low resistance contacts on the rear of photovoltaic devices without requiring photolithography based techniques thereby allowing the remainder of the rear surface to be well passivated by thermally grown silicon dioxide layer.

This simple and low cost technique utilises aluminium spiking to randomly reduce localized regions of a surface passivating silicon dioxide layer at elevated temperature (but below the eutectic temperature of aluminium and silicon, i.e., 577ºC) to form conductive paths to the underlying substrate, see Figure 4.26. The aluminium can be further utilised to induce solid phase epitaxial growth of p+ silicon in these localized regions via subsequent amorphous silicon deposition and low temperature sinter, see Figure 4.27. This p+ silicon can be used to isolate the high recombination velocity Al/Si interface from the active regions of the device.

Table 4.6: The electrical output of the fi rst generation stencil-printed SP/BC hybrid solar cell on textured, 1 Ω-cm Cz wafer.

JSC VOC (mV) FF EFF (%) Rsh (Ω.cm2) Rs (Ω.cm2)

(mA/cm2)

AVERAGE 31.6 594 0.74 13.9 1416 0.63

BEST CELL 33.2 597 0.76 15.0 1260 0.50

SiO2

Si substrate

Al doped p+ Si

Displaced Al

Ni / Cu

Ni / Cu

SiO2

Si substrate

Al doped p+ Si

Displaced Al

Ni / Cu

Ni / Cu

Figure 4.26. Schematic diagram of a buried contact solar cell with rear aluminium localised contacts

Figure 4.27. Schematic diagram of a buried contact solar cell with epitaxially grown p+ rear silicon localised contacts.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

This technique is potentially suitable for commercial photovoltaic devices after further fabrication optimisation and process simplifi cation.

One highlight of the work is the identifi cation of the cause of VOC and Tbulk degradation in some devices by using the photoconductance lifetime measurements (see Table 4.7). Results show that the long heat treatment required for the formation of localised contacts is responsible for (1) damaging the surface passivation quality of the oxide even in regions where contacts have not yet been formed and (2) introducing contaminants into the substrate thereby degrading the fi nal minority carrier bulk lifetimes τb. Hence, even if there has been an improvement at the p+Si/Si interface in terms of reduced Seff , the improvement has been masked by the domination of the high Seff associated with the oxidised regions that limits the open circuit voltages to relatively low values. The degradation of τb is also found to be directly linked to the duration of the heat treatments (see Figure 4.28) although in an industrial setting, the latter should be relatively easy to solve.

40

Table 4.7: PCD lifetime measurement results of test devices with Al and p+ Si rear localized contacts. Aluminium has been removed prior measurements

Figure 4.28: Bulk minority carrier lifetimes plotted against the total duration of heat treatments received by test devices that had diff erent rear designs

Sample ID Contacts Heat treatment Heat treatment τb Seff

for spiking at for p+ Si epitaxial growth at (µs) (cm/s)

520ºC (hrs) 530ºC (hrs)

LTT11 Al 0.5 N/A 2100 11

LTT12 Al 5 N/A 240 128

LTT14 Al 15 N/A 43 339

LTT13 p+ Si 5 4 99 813

PL-1-I p+ Si 9 4 26 300

LTT15 p+ Si 15 4 20 82

Ann

ual R

epor

t 2

00

4

Fir

st G

ener

atio

n

RESEARCH

41

Another highlight of the work is the development of several variations of the rear localised contacting scheme. The most successful of these demonstrates signifi cant performance enhancement relative to the conventional BCSC. It involves the use of a shadow mask during Al deposition prior to heat treatment, see Figure 4.29. Devices that utilise the improved techniques achieve improvements in VOC of 20-25 mV and in JSC of 5-10%. After the inclusion of p+ Si, these devices reach a conversion effi ciency of 17.4% under one sun AM1.5G illumination at 25ºC without surface texturing or front anti-refl ection optimisation, see Table 4.8 This is a signifi cant improvement in performance over the standard BCSC with rear alloyed aluminium back surface fi eld.

4.3.1.4 SILICON SOLAR CELL PROCESS ENGINEERING

A. BORON DIFFUSION-INDUCED DEFECTS IN CRYSTALLINE SILICON SOLAR CELLS

Introduction

It is important to the performance of high effi ciency mono-crystalline silicon solar cells that boron and phosphorus diff usions processes do not degrade the quality of the silicon wafer, especially by introducing chemical impurities or lattice defects. The diff usion of phosphorus or boron causes stress on the silicon lattice as a result of atomic radius mismatch. The covalent radii of boron and phosphorus are 0.88 Å and 1.06 Å respectively, compared to 1.18 Å for silicon. This mismatch results in a localised contraction of the silicon lattice around the substitutional dopant atoms. For low concentrations of dopant atoms, the lattice accommodates this stress elastically. However, when the total concentration

Figure 4.29: Schematic of improved rear contacting scheme using a combination of shadow mask and heat treatments to form p+ silicon localised contacts

a-Si

Si substrate

Al

Si substrate

Displaced Al

p+ Si

Excess a-Si

Table 4.8: Results of a 1 Ω.cm p-type buried contact solar cell with rear localised p+ silicon contacts in shadow mask defi ned regions

VOC (mV) 635

JSC (mA/cm2) 33.6

η (%) 17.4

FF 0.813

RS (Ω.cm2) 2.10

RSHUNT (Ω.cm2) 1.05x106

Ann

ual R

epor

t 2

00

4

Fir

st G

ener

atio

n

of dopants exceeds a critical concentration the lattice can no longer accommodate elastically, and misfi t dislocations are generated. It is well known that diff usion-induced misfi t dislocations are half-loop dislocations nucleated at the diff usion surface in the <110> directions. These half-loop dislocations consist of two screw dislocations perpendicular to the diff usion surface and an edge dislocation line parallel to the diff usion surface in the <110> direction. At diff usion temperatures, the two screw dislocations glide out of the wafers edge within seconds, and the edge dislocation lines glide into the substrate in response to the stress gradient. The collision of several edge dislocations forms a cross hatch dislocation network characteristic of diff usion-induced misfi t dislocation (see Figure 4.30). Although the formation and movement of misfi t dislocations have been extensively studied, the impact on recombination statistics and the performance of a silicon solar cell has not been previously established.

Sample Preparation

Samples for boron and phosphorus diff usion were prepared on (100) 1 Ω.cm p-type fl oat-zoned (FZ) wafers as follows:

Sample A: Boron misfi t dislocations were intentionally generated by diff using each side with a heavy planar boron diff usion (10 Ω/sq.). This was achieved using thestandard 45 minute deposition cycle at 900°C followed by a 270 minute drive-in at 1060°C.

Sample B: A lighter boron diff usion (80 Ω/sq.) was conducted to avoid the generation of boron misfi t dislocations. This was achieved using a standard 45 minute deposition cycle at 900°C followed by a 30 minute drive-in at 1060°C.

Sample C: Phosphorus misfi t dislocations were intentionally generated by diff using each side with a heavy planar phosphorus diff usion (10 Ω/sq.). This was achieved using a 75 minute solid source deposition at 940°C followed by a 120 minute drive-in at 1000°C.

Sample D: A lighter phosphorus diff usion was conducted to avoid the generation of phosphorus misfi t dislocations. This was achieved using a 15 minute solid source deposition at 850°C followed by a 140 minute oxidation cycle.

After processing, the injection-dependent eff ective lifetime of each sample was measured using the photoconductance lifetime technique. Finally, the dislocations within each sample were characterised using the Yang defect etch (CrO3:HF). The samples were then deglazed in HF before cleaving along the 110 plane using a diamond tipped pen to obtain a cross section of the bulk region. The cleaved wafers were Yang etched for 10 minutes before examining both the surface and cross section under an optical microscope to establish a 3D picture of the resulting dislocation network.

Boron Misfit Dislocations

Examination by optical microscopy of the two boron diff used samples after Yang etching demonstrated misfi t dislocations were generated on Sample A (see Figure 4.30), and few misfi t dislocations were generated on Sample B (not shown). The photomicrograph of Sample A after Yang etching (see Figure 4.30a) exhibits the cross-hatch pattern characteristic of misfi t dislocations. This cross hatch pattern is the result of etching the edge dislocation lines, which are parallel to the top surface and in the <110> directions. The misfi t dislocation network was observed to penetrate over 100 µm into the bulk region (see Figure 4.30b). 42

Ann

ual R

epor

t 2

00

4

F

irst

Gen

erat

ion

43

The inverse eff ective-lifetime versus minority carrier density curve for Sample A (see Figure 4.31) directly after diff usion exhibited asymmetric Shockley-Read-Hall (SRH) recombination. We defi ne asymmetric-SRH recombination as SRH recombination where the majority carrier lifetime is signifi cantly larger than the minority carrier lifetime (i.e. for p-type silicon τn << τp). Since, in this case, the high J0 associated with the heavy surface diff usions dominated the photoconductance lifetime, a CP-etch was used to etch back the diff usions to ~100 Ω/sq. before remeasuring. This reduced the J0 and enabled a clearer observation of the asymmetric-SRH recombination (see Figure 4.31).

Figure 4.30. Photo of the boron misfi t dislocation network observed on Sample A after Yang etch (a) the planar (100) surface, and (b) (110) surface cross-section.

Figure 4.31. (a) PCD curve demonstrating asymmetric-SRH resulting from the boron misfi t dislocation network (Sample A); directly after boron diff usion () and after etching the diff usion back (∆). The theoretical model shown as a dashed-line contains; J0=44 fA.cm-2, Bulk-SRH τn=τp=115 µs, and asymmetric-SRH τn=83 µs, τp/τn=80. (b) PCD curve for the boron diff used control sample (Sample B) demonstrating reduced asymmetric-SRH (). The theoretical model shown as a dashed-line contains; J0=160 fA.cm-2, Bulk-SRH τn=τp=210 µs, and asymmetric- SRH τn=625 µs, τn/τp=80.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

The surface of Sample B was virtually dislocation free, with only a few small areas of boron misfi t dislocations observed around the edge of the wafer. The corresponding inverse eff ective-lifetime versus minority carrier density curve (see Figure 4.31a) exhibited asymmetric - SRH, however the extracted lifetime parameters are signifi cantly better than the than those of Sample A. It can be concluded, therefore, that the formation of a boron misfi t dislocation network during boron diff usion generates bulk asymmetric-SRH recombination centers.

Phosphorus Misfit Dislocations

Examination by optical microscopy of the two phosphorus diff used samples after Yang etching demonstrated misfi t dislocations were generated on Sample C (see Figure 4.32), and no misfi t dislocations were generated on Sample D.

The inverse eff ective-lifetime versus minority carrier density curve for Sample C (see Figure 4.33a) directly after diff usion also exhibited asymmetric Shockley-Read-Hall (SRH) recombination. The surface of Sample D was dislocation free, and the corresponding inverse eff ective-lifetime versus minority carrier density curve (see Figure 4.33b) exhibited no asymmetric-SRH. It can be concluded that the presence of a phosphorus misfi t dislocation network within the bulk causes bulk asymmetric-SRH recombination.

44

Figure 4.33. (a) PCD curve demonstrating asymmetric-SRH resulting from the phosphorus misfi t dislocation network (Sample C). The theoretical model shown as a dashed-line contains; J0=1250 fA.cm-2, Bulk-SRH τn=τp=1000 µs, asymmetric-SRH τn=18 µs, τn/τp=92 and the depletion region modulation (DRM) eff ect. (b) PCD curve for the phosphorus diff used control sample (Sample D) demonstrating no asymmetric-SRH. The theoretical model shown as a dashed-line contains; J0=48 fA.cm-2, Bulk-SRH τn=τp=600 µs, and the depletion-region modulation eff ect.

Figure 4.32. Photo of the phosphorus misfi t dislocation network observed on the (100) surface of Sample C after defect delineation using a Yang etch.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

45

Effect of Misfit Dislocation on Cell Performance

The eff ect of these boron- and phosphorus-misfi t dislocations on the performance of high-effi ciency silicon solar cells was investigated by manufacturing a double sided buried contact solar cell using the same phosphorus diff usion process as Sample C for the front contact diff usion, and the same boron diff usion process as Sample A for the boron contact diff usion.

The resulting light I-V characteristics of the cell were typical of a planar DSBC solar cell; VOC = 666 mV, JSC=30 mA/cm2, FF=76% and Eff =15.2%. Similarly, the Suns-VOC dark I-V and local ideality (m-V) characteristics were typical (see Figure 4.34). A theoretical model including; edge recombination (m=2), emitter recombination (m=1), Auger recombination (m=0.775) and asymmetric-SRH recombination was developed to match both these measured curves.

This model fi t clearly demonstrates that the recombination within the device changes from high bulk recombination dominated by bulk asymmetric-SRH at low operating voltages, to low bulk recombination dominated by emitter recombination at high operating voltages (see Figure 4.34). This results in low short-circuit current and poor fi ll factor, however the open-circuit voltage is less aff ected because the recombination rate is reduced at higher operating voltages. This change in the relative recombination (i.e. eff ective lifetime) with device voltage (i.e. minority carrier density) aligns well with the previous PCD measurement of samples containing misfi t dislocation networks, therefore, it is likely that the poor current and fi ll factor of the DSBC is caused by process-induced dislocations introduced during the boron and/or phosphorus diff usion processes.

Figure 4.34. Suns-VOC and mV-curve for a DSBC solar cell with misfi t dislocations; experimental data (+, ), and modelling (dashed and solid lines). The eff ect of this asymmetric-SRH on light-IV parameters is a reduction in both the short-circuit current and fi ll factor.

Ann

ual R

epor

t 2

00

4

D

irec

tors

' Rep

ort

00

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

A. LASER-INDUCED DEFECTS IN CRYSTALLINE SILICON SOLAR CELLS

The impact of crystal damage generated during laser processing on solar cell performance is uncertain. In 2004, we investigated laser-induced defects that result from a laser ablation process similar to that used to form grooves in the buried contact solar cell. The fi rst part of the work focused on the formation of the defects and used the Yang etching technique to investigate the impact of chemical etching and thermal cycles on their propagation. The second part of the work took a close look at the electrical properties of laser-induced defects, experimentally investigated the recombination of minority-carriers at laser-induced defects, and the potential of these defects to shunt pn-junctions.

Formation and Propagation of Laser-induced Defects

The fi rst part of the work verifi ed the presence and type of defects caused by laser ablation. The eff ect of pre-scribe etching and thermal cycles on defect formation was investigated.

The dominant Yang etch feature on the surface and the cross section was characteristic of an edge dislocation. After laser scribing a thin strained layer exists close to the surface, if this defected region is not removed during etching, defects propagate into the bulk region during subsequent thermal cycling, forming a network of dislocations. With correct etching or no subsequent thermal cycles, defect formation does not occur.

In the second part of the work, various amounts of laser-induced damage were intentionally generated within the bulk region and electrically characterized with respect to bulk recombination and cell performance.

46

Figure 4.35. (a) Laser scribe cross section, showing laser induced defects. (b) Top view.

Bulk Recombination

Wafers containing diff erent amounts of laser damage were characterized by photoconductance lifetime measurements.

It was found that an increase in laser-induced defects results in a reduction in eff ective lifetime. Furthermore, the shape of the curves in Figure 4.36 at low injection levels is characteristic of asymmetric Shockley-Read-Hall recombination with low electron lifetime and high hole lifetime. We conclude that laser-induced defects lower the eff ective lifetime of a sample through the mechanism of asymmetric Shockley-Read-Hall recombination. This technique allows us to use photoconductance lifetime measurements to detect the presence of laser damage in other laser based processes, for example laser texturing.

Junction Shunting

Standard Buried Contact cells containing diff erent amounts of laser damage were electrically characterized to investigate the potential impact of laser damage on the structure. The laser-induced defects did not cause shunting of the junction but did increase the J02 component and reduce the open circuit voltage. Reduced quantum effi ciency in the laser damaged samples at the blue end of the spectrum indicates enhanced recombination at the front surface and in the emitter.

00

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

47

Figure 4.36. Inverse eff ective lifetime versus excess minority carrier density for samples etched for diff erent times, 1-sun implied open circuit voltage in parentheses.

Figure 4.37. (a) Dark I-V curves of buried contact solar cells () with and (∆) without laser damage. (b) Spectral response of buried contact solar cells () with and (∆) without laser damage.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

Buried contact solar cells, fabricated with laser-induced defects, were not shunted and only displayed a small decrease in electrical performance. It is concluded that for these structures, laser damage does not have a large impact on the electrical properties of the device and that laser defects were not able to cause junction shunting.

In the case of laser ablation, dislocation formation can be avoided by a post-laser process etch in NaOH or by avoiding post-laser thermal cycling. The high implied open-circuit voltages observed on well etched samples supports the use of laser ablation with high-effi ciency silicon solar cell processing.

C. ANISOTROPIC TEXTURE ETCHING WITH KOH/IPA/SI SOLUTIONS

A number of suggestions have been put forward to explain how anisotropic etching of silicon wafers in weak alkali solutions produces upright random pyramids on the surface. Low concentration alkali solutions are primarily used in conjunction with photolithography for shaping small structures in silicon. Such solutions achieve this due to their anisotropic nature (ie: etching occurs at diff erent rates in diff erent crystallographic directions) and the masking eff ect of a patterned oxide. It has been reported that adding isopropyl alcohol (IPA) to etching solutions makes the etched surfaces generally become smoother but unwanted pyramid structures can form on the etched structures. In most applications this is a nuisance, but in solar cells this pyramid formation can be used to advantage. For wafers surfaces covered in pyramids, refl ection of light from the front surface can be reduced from about 30% to as low as 8%. The problem of creating pyramid structures evenly over the whole wafer surface is as diffi cult as avoiding them in other applications.

While the mechanisms of Si etching in alkali solutions are discussed at length in the literature, little has been published about the mechanisms of pyramid formation. One study used a Monte Carlo simulation and known activation energies and etch rates to model the etching of a silicon wafer which had localised stabilisation points across the surface. The study showed that pyramids formed at the stabilised points and concluded that impurities in the alkali solutions must form a mask on the surface. Anecdotally, this seems to be the most commonly held view but there is little experimental evidence for this as it is diffi cult to prove. Therefore, understanding how texturing solutions behave empirically can provide useful insight and help to achieve better textured surfaces for solar cell applications.

Experimental

Four diff erent, low concentration, KOH based solutions were investigated (Table 4.9). Each of the diff erent solutions was made from a base solution of 3% KOH in DI H2O. In two of the solutions, 3 g of Si wafer (a mixture of p- and n-type, FZ and CZ) per litre of solution was dissolved into the base solution prior to texturing. Each of these types of solution (with and without Si) mixed either with or without IPA. Texturing in each solution was carried out in a beaker with a lid to minimise evaporation. In all cases the solution was stirred with a tefl on coated magnetic stirrer and heated to 90°C before adding the IPA and the wafers. Wafers were textured at 90°C. Eight 1 Ωcm, (100), n-type wafers were texture-etched in each of the four solutions, and two wafers were removed every 15 minutes.

48

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

49

RESULTS

Figure 4.38 to Figure 4.41 show how the texturing process proceeds in the diff erent solutions. The wafers textured with Solution A start with a sparse coverage of pyramids which results in the fi nal pyramid coverage and size distribution being very uneven. Adding IPA to the solution (Solution B) results in a much denser coverage of pyramids and the fi nal pyramids are larger. Dissolving Si into the base solution (Solution C) also results in a much denser pyramid coverage (compared to Solution A), but in this case, the pyramids are extremely small and appear to be more rounded at the apex and at the bases compared to the pyramids formed in Solution B. Adding both Si and IPA (Solution D) results in medium sized pyramids with sharp features.

3% KOH 60 ml/l IPA 3 g/l Si

Solution A

Solution B

Solution C

Solution D

Figure 4.38: Textured with Solution A

Figure 4.39: Textured with Solution B

Figure 4.40: Textured with Solution C

Figure 4.41: Textured with Solution D

Table 4.9: Texturing solution components.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

Reflectance Measurements

Refl ectance curves (measured on a Cary500 spectrophotometer) for the four diff erent textured surfaces are shown in Figure 4.38a and Figure 4.38b. These graphs show that, given enough time, all the solutions will eventually reduce the refl ection to around 8% minimum. It would seem that the diff erent size and shape of the pyramids (shown in Figure 4.38 to Figure 4.41) has little impact on the fi nal refl ection value, as long as the surface is completely covered with pyramids. Solution C achieves this minimum refl ection value in the fi rst 15 minutes while the other solutions take quite a bit longer.

D. SILICON NITRIDE COATINGS FOR BURIED CONTACT SOLAR CELLS

PECVD silicon nitride is an excellent anti-refl ection coating material and it has also fi rmly established itself as an excellent material to passivate the surface of crystalline silicon wafers. Previous studies conducted by various research groups indicate that the passivation quality provided by PECVD silicon nitride is comparable to that of thermally grown oxide. One of the advantages of PECVD SiN over thermally grown oxide is that good AR coating or surface passivation can be satisfactorily achieved at the relatively low deposition temperature, typically in the vicinity of 400°C.

Initial evaluation on the surface passivation quality provided by the Centre’s new Roth and Rau laboratory-type remote microwave PECVD system was conducted. This report focuses on the passivation quality on various surfaces on n-type crystalline silicon wafers. The surfaces studied include bare and textured planar surfaces, phosphorus diff usion on planar and textured surfaces, and boron diff usion on planar surfaces. For the cases of surface passivation on diff used layer, thermally grown oxide was used as a baseline to compare the quality of the surface passivation.

Non-diffused Silicon Surfaces

Several batches of fl oat zoned, 260-µm thick, 1 Ω.cm, (100), n-type silicon quarter wafers were prepared in the following sequence. First, the wafers underwent standard saw damage removal etch, followed by a standard RCA clean, and an HF dip prior to

50

Figure 4.42: Refl ectance curves for each solution after 15 minutes texturing

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

51

PECVD deposition. Some of the wafers were textured using a standard anisotropic sodium hydroxide etch process. Silicon nitride fi lms were deposited on the bare wafers with three diff erent refractive indices (~1.95, 2.3 and 2.75) by varying the gas fl ow ratio of silane and ammonia. A clean polished wafer was included in each deposition run to monitor the thickness and refractive index of the fi lms by ellipsometry. The deposited fi lms were typically around 65-85 nm thick. The eff ective minority carrier lifetime was measured versus injection level for all samples using the generalised quasi-steady state photoconductance lifetime measurements (see Figure 4.43).

The eff ective LLI minority carrier lifetime of three diff erent refractive index silicon nitride layers on non-diff used planar surfaces in as-deposited condition are summarised in Table 4.10. The results show that higher the refractive index results in higher eff ective minority carrier lifetime for planar-surface cases. The textured-surface cases, however, do not follow this trend. Instead, the silicon nitride fi lm with a refractive index of 2.35 achieved the best lifetime amongst the three diff erent refractive indices. The fact that the eff ective LLI lifetime is signifi cantly lower on textured surfaces compare to planar surfaces is important for device applications because of the importance of texturing in reducing front-surface refl ection. Further investigation and optimisation on textured surfaces is planned for 2005.

Figure 4.43. Photoconductance lifetime curves for various silicon nitride fi lms deposited directly on n-type silicon wafers. (a) planar surface (b) textured surface.

Table 4.10. A summary of the eff ective minority carrier lifetime achieved by various silicon nitride fi lms on bare silicon surfaces.

Surface Type Eff ective Minority Carrier Lifetime (µs)

n = 1.95 n = 2.35 n = 2.75

Planar 380 770 1300

Textured 6.5 250 100

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

Diffused Silicon Surfaces

For the cases of diff used surfaces, a standard phosphorus or boron diff usion was carried out, followed by a dry-wet-dry oxidation process to result in a fi nal sheet resistance for the diff used layers of 100-200 ohm/sq. The eff ective minority carrier lifetime was measured versus injection level for these samples using the photoconductance lifetime measurement technique. Next, the oxide was stripped off one side of the wafers, followed by a second standard RCA clean. The three silicon nitride fi lms described above were deposited on the oxide-stripped side and then they were measured again using the photoconductance lifetime technique (see Figure 4.44).

On phosphorus diff used, planar surfaces, the eff ective lifetime is as good as a thermally oxidized diff used silicon surface, highlighting the excellent quality of the deposition equipment and process. More work is needed to optimise the deposition conditions for textured surfaces, and especially for boron diff used surfaces.

The bulk lifetime τbulk, emitter saturation current density Joe, and implied-Voc of thermally grown silicon dioxide and deposited silicon nitride fi lms on diff used planar surfaces extracted by curve fi tting the photoconductance lifetime data is shown in Table 4.11. In particular, the surface passivation quality of low refractive index silicon nitride fi lms (n=1.95) is signifi cantly worse than that of silicon dioxide. On the other hand, the passivation quality provided by high refractive index silicon nitride fi lms (n=2.77) is comparable to that of the silicon dioxide.

52

Figure 4.44. Photoconductance lifetime curves for various silicon nitride fi lms deposited on diff used n-type silicon wafers (a) phosphorus diff used, planar surface (b) phosphorus diff used, textured surface (c) boron diff used, planar surface. The solid symbols show the photoconductance lifetime curves for oxidized phosphorus-diff used silicon surfaces.

Table 4.11. A comparison of Joe,τbulk, and implied-Voc of silicon dioxide and silicon nitride passiv-ation layers on phosphorus diff used emitters with a planar surfaces.

Silicon dioxide SiN, SiN, SiN,

(baseline) n=1.95 n=2.35 n=2.75

Joe (A/cm2) 3.6 x 10-14 2.0 x 10-13 5.2 x 10-14 4.3 x 10-14

τbulk (µs) 3700 330 1800 2800

Impiled Voc (mV) 708 650 694 703

Ann

ual R

epor

t 2

00

4

Fir

st G

ener

atio

n

53

Initial work indicates that the prospects are good for the use of remote microwave PECVD silicon nitride in high-effi ciency silicon solar cells, and work in 2005 will focus on device applications.

Overall, the initial fi lms deposited in the Roth & Rau remote microwave PECVD demonstrate excellent passivating quality, especially considering that “turn-key” deposition parameters were used and little optimisation of the deposition parameters has been carried out. As expected, the fi lm composition, as indicated by the refractive index, plays a signifi cant role in determining the passivation quality of the fi lm. On planar surfaces, the best LLI eff ective lifetime is equivalent to surface recombination velocities of lower than 10 cm/sec.

4.3.1.5 RAYSIM 6 - A FREE GEOMETRICAL RAYTRACING PROGRAM FOR SILICON SOLAR CELLS

In 2004, the Centre released its freeware geometrical ray tracing program, called RaySim. RaySim is a graphical user interface (GUI) software program for geometrical ray tracing, primarily for silicon solar cell design and development, although it has been used for other applications such as design of static concentrators and simulation of conformally textured thin silicon fi lms. The fi rst version of RaySim was written in 1997 for the investigation of light trapping in thin silicon solar cells, and more recently to support modelling and simulation research at the University of New South Wales’ Centre of Excellence in Silicon Photovoltaic and Photonic Devices. The RaySim program is written in Visual C++ with Microsoft Windows MFC support, using the MS Visual IDE.

The simulator has a number of features that make it fast, fl exible and easy to use. The simulator has a pseudo-3D display mode that allows users to visualize models and traces. It also uses dialog-based GUI for building optical models and specifying simulation parameters, making model building and testing fl exible and straightforward. The simulator is also multi-chromatic in that it can co-trace an arbitrary spectrum of wavelengths with dispersive optical materials and surfaces. Perhaps its best feature is that it’s free, available for download from http://www.pv.unsw.edu.au//links/products.asp.

Some features of the program include: - Geometrical ray tracing of linear, time-invariant materials - Fast, threaded trace algorithms - Multi-chromatic tracing- Visual GUI model building and pseudo-3D visualization - A variety of optical models for optical materials and interfaces- Optical path-length calculation

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

RaySim Guide

RaySim uses the familiar MS Windows graphical user interface for model building and tracing, including the traditional menu bars, tool bars and active graphical elements. A key feature is the pseudo-3D display of the model and of partial traces. The user can rotate and zoom a stick-fi gure representation of the optical model, both with and without trace data, with simple mouse movements. In combination with the GUI interface, the pseudo-3D display makes visualization easy and model construction fast.

Ray tracing simulation requires a great deal of data to describe everything from the details of the spectrum/distribution of incident bundle of rays, to the geometrical extents and optical properties of the various parts of the optical model, to the results of a trace simulation. RaySim uses a collapsible tree control, similar to the Windows Explorer style representation of folders, sub-folders and fi les, to allow easy access to the large quantity of various data needed for a trace simulation, for example, as shown in Figure 4.45.

A typical optical model is comprised of a number of layers that represent diff erent optical materials. Users can select from constant or wavelength-dependent optical models (specifi ed in an external fi le) for each layer. The physical extent of each layer is defi ned any number of planes that form a “light-tight” surface around the layer. A variety of diff erent optical models are available, ranging from specular and diff use refl ectors to simple interfaces and single layer anti-refl ection coated interfaces. A variety of auxiliary plane types are also available to aid the simulation, for example, by randomizing or translating rays as they pass through the model. RaySim can model the shape of any surface that can be reasonably well approximated by a collection of parallelepiped and/or triangular facets. The optical model, the set of parameters that controls a trace simulation, and the simulation results are organized into a RaySim batch. The simulation parameters specify the input bundle of rays (which can be spatially and/or angularly distributed and/or multichromatic) and parameters that control how RaySim spawns sub-rays during simulation. Results collected during simulation are also stored in the batch, and batches are further organized into Documents, which can be saved and opened later for further editing and simulation.

Users can quickly and easily edit any of the various optical model data or simulation control data using the collapsible tree control. Double clicking an entry for a layer, plane or batch on the tree calls up a dialog box that allows the properties of that object to be entered or edited. New batches, planes or layers can also be easily inserted, deleted, copied and pasted using the collapsible tree control. This, combined with the pseudo-3D visualization, allows simulations to be constructed, visualized and tested quickly and easily, with no prior knowledge of the C++ language and without requiring cumbersome input fi les to describe optical models.

54

Figure 4.45. RaySim organization of data

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

55

RaySim Output

RaySim tracks every ray in the incident bundle of input rays, accumulating absorbed light in every layer and at every plane. This allows users to examine the response of the model in detail as a function of incident position (x, y and z), incident angle (azimuth and declination) and starting wavelength. For example, Figure 4.46a shows a line-scan response of a tapered, conformal thin silicon fi lm (in air) on a textured glass substrate. The input bundle of rays comprises a line of several normally incident, monochromatic rays perpendicular to the crest of the texture. The simulation results (see Figure 4.46b) predict the spatially varying absorbance, refl ectance and transmittance of the thin fi lm structure.

Multi-dimensional analysis is also possible, for example to generate contour plots of the spatial or angular response of an optical model. Figure 4.47 shows an example x-y analysis of a wedge-shaped static concentrator module. Figure 4.47a shows the model with a few trace rays originating at the thin edge of the wedge. Figure 4.47b shows the static concentrator’s response as a function of the position of the input ray. RaySim could be similarly used to plot the concentrator’s response as a function of input azimuth and declination angle.

Multi-chromatic analysis is also possible, for example to generate absorbance, refl ectance or transmission spectra of an optical model. Figure 4.48a shows a typical simulation trace of a 200-µm thick silicon wafer with upright random pyramids on both surfaces. The simulation uses dispersive bulk optical properties for the silicon layer and an input bundle of rays that spans 410–1290 nm. Figure 4.48b shows the plotted results of the simulation, with and without a rear detached refl ector.

Figure 4.46. (a) Raytrace simulation of a tapered, conformal thin-fi lm crystalline silicon fi lm on a textured glass substrate. (b) Spatially varying absorption, refl ection, and transmission through the thin fi lm.

Figure 4.47. (a) Ray trace simulation of a wedge-shaped static concentrator, showing a few traces originating from the thin end of the wedge. (b) Contour plot of the absorption at the solar cell as a function of the position of the incident ray.

solar cell

static concentratorlense

Thin Film Silicon on Textured Glass

0

0.2

0.4

0.6

0.8

1

-20 -10 0 10 20 Position (um)

A R T

incident raysair

silicon

glass

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

Two additional analysis features are spectrum weighting of the input bundle of rays and path length calculation. Users can specify an external fi le that contains a wavelength-dependent weighting for the input bundle of rays. For example, using a wavelength-banded AM1.5G spectrum in the units of mA/cm2 (per wavelength band) allows the subsequent calculation of the maximum available current density (MACD) for an optical model. The MACD is a good way to compare various optical models.

RaySim also allows users to specify path-length analysis for any layers in the model. In this case, the absorption is arbitrarily set to zero and RaySim accumulates the intensity-weighted length of all rays that pass through that layer. The result is the eff ective optical path length of the layer.

Fast, Accurate, Threaded Tracing

Speed and accuracy are traded off in ray tracing simulations, generally, due to the large number of trace operations required to arrive at an accurate result. For example, simulation of a textured silicon wafer with an idealized rear diff use refl ector over a spectrum 43 of wavelengths (spanning 410–1250 nm) might require upwards of 20 million trace operations to reach an error margin of less than about 1%.

RaySim uses a number of techniques to minimize trace time, including effi cient computation algorithms, fast storage of trace data and results data, and threaded operation of the trace kernel. One unique feature of RaySim is the ability of the trace algorithm to switch between ray spawning and Monty-Carlo propagation. Ray spawning, whereby a ray is split into two or more sub-rays whenever it intersects a partially transmitting interface or diff use refl ecting surface, is employed initially to quickly distribute the light within the model.

Ray spawning, while it is very accurate, creates a very large number of rays, which can be very time consuming to trace to completion. RaySim changes to a Monte-Carlo method during a trace simulation once a ray has met the trace parameters set by the user. Monte-Carlo tracing is relatively fast, as no additional rays are spawned within the model. The accuracy of the Monte-Carlo simulation is enhanced by fi rst distributing light within the model by spawning. RaySim can perform almost 8 million trace operations per minute on a 1.1 GHz, Windows XP computer, depending on the optical model.

56

Figure 4.48. (a) Ray trace simulation of a silicon wafer with random upright pyramids on both surfaces. (b) Absorbance, Refl ection and Transmission versus wavelength with a detached randomizing rear refl ector (open symbols) and without a rear refl ector (fi lled symbols).

silicon

air

EVA

Upright Random Pyramids in Silicon

0

0.2

0.4

0.6

0.8

1

400 600 800 1000 1200 Wavelength (nm)

A,R

,T A R T A

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

57

Optical Models

RaySim incorporates a growing number of bulk and interface optical models. At present, bulk media can be described by constant or dispersive refractive index, absorption coeffi cient and scattering coeffi cient (for scattering media such as pigmented diff use refl ectors). Interfaces can be modeled by constant or dispersive simple refl ective interfaces (specular refl ectors and diff use (Lambertian), idealized refractive interfaces (unity transmission, except for total internal refl ection) and non-dispersive, non-absorbing single layer antirefl ection coatings. There are also a number of utility interfaces, such as random ray shifting planes and ray spawning planes to aid simulation and for special applications.

Limitations and Future Features

RaySim uses geometrical ray tracing techniques and ignores coherence eff ects. Because of this, small features, like optical gratings, or thin layers, like antirefl ection coatings are not simulated accurately. RaySim can presently simulate interference eff ects of single-layer antirefl ection coatings, however (by calculating the transmission coeffi cient and not by ray tracing techniques) and optical models for multilayer anti-refl ection coatings is planned for the future. RaySim also simulates linear and time-invariant optical materials.

A RaySim document can have up to 10 batches, 10 layers per batch and 25 planes per layer, and the input bundle of rays can contain any number of rays. Larger optical models with a larger number of input rays tend to require a disproportionately longer time for simulation, however, because of the overhead associated with memory management.

New bulk and interface optical models can be easily included in the source code, provided a there is a computable stochastic model that describes the optical eff ect. Presently, there is a signifi cant eff ort at UNSW to develop RaySim to simulate other interesting optical conversion processes, including downshifting, down conversion and upconversion. The most straightforward of the three optical eff ects, downshifting, is presently in beta testing, with some interesting results from luminescent concentrator simulations, as shown in Figure 4.49.

Obtaining Free RaySim

A compiled executable fi le for Windows 2000 and Windows NT (and later versions), with some auxiliary data fi les, is available, free, with some restrictions, from http://www.pv.unsw.edu.au//links/products.asp. Help fi les (usually under construction) are available online at this location, as well.

Figure 4.49. (a) Ray trace simulation of a hexagonal shaped luminescent solar concentrator (LSC), showing propagation of wavelength-downshifted light through the luminescent layer (the middle layer in the fi gure). (b) response, along the axes shown at left, of solar cells arranged on the six edge faces.

Fraction of Light Reaching LSC Edges

0

0.02 0.04 0.06 0.08 0.1

0.12 0.14

-1000 -500 0 500 1000 Position (a.u.)

Res

pons

e Absorbed Side1 Side2 Side3 Side4 Side5 Side6

incident ray

1

2

3

4

56

4.3.2 HIGH EFFICIENCY SILICON CELL GROUP

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

University Staff : A/Prof. Jianhua Zhao (group leader)Prof. Martin Green Prof. Stuart Wenham

Research Fellows: Dr. Aihua Wang

Research Assistant: Guangchun Zhang

Technical Staff : Jules Yang

4.3.2.1 REAR BORON EMITTER CELLS ON N-TYPE SUBSTRATES

In the previous two years, to avoid problems with boron and oxygen related carrier lifetime degradation [J. Schmidt et al, “Investigation of Carrier Lifetime Instabilities in CZ-Grown silicon”, 26th IEEE PVSC, pp.13, 1997], the high effi ciency cell group at the Centre has been focusing on fabricating solar cells on n-type CZ silicon substrates, avoiding boron dopant in the device bulk. PERT (passivated emitter, rear totally-diff used) cells on CZ and FZ n-type silicon substrates with a traditional front boron diff used emitter have demonstrated high effi ciencies of 21.1% and 21.9%, respectively [J. Zhao et al, 29th IEEE Photovoltaic Specialist Conference, New Orleans, p.218, (2002)]. However, it was later found that most of the cells suff ered large losses in their Voc up to 100 mV after 2 years in storage. A further degradation also was observed when these cells were then illuminated under one-sun intensity from ELH lamps.

In 2004, a new rear boron emitter cell structure has been investigated using n-type substrates. Figure 4.50 shows this structure. Moving the boron doped emitter to the rear surface is expected to reduce the boron-oxygen degradation problem, since this eff ect is enhanced by illumination. Initially, 0.9 Ω-cm, n-type phosphorus doped FZ substrates of 400 µm thickness were used for these cells.

Since the rear emitter boron diff usion was over the entire rear surface, the rear emitter cells have to be scribed from the wafer before measurement (in contrast to the standard research PERL cells which are mostly measured while embedded in the silicon wafer without scribing). Hence, the performance of these scribed rear emitter cells suff ers by about 4% relatively when compared to un-scribed small-area research cells, due to the extra edge recombination loss. As a reference point, the best scribed PERL cell on p-type substrates of the same 22 cm2 area as the present n-type cells had demonstrated 23.7% effi ciency [J. Zhao et al, “20 000 PERL Silicon Cells for the ‘1996 World Solar Challenge’ Solar Car Race”, Progress in Photovoltaics, Vol. 5, pp. 269, 1997], while the best small 4cm2 cell tested while embedded in a wafer had a signifi cantly higher effi ciency of 24.7%.

58

RESEARCH

double layerantireflection coating

finger “inverted” pyramids

oxiderear contact

n-silicon

Fig. 4.50: Rear boron emitter PERT cell structure on n-type silicon substrate.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

59

Small 4 cm2 rear emitter n-type cells were initially investigated. However, due to the relatively larger edge-scribing damage, these small cells had relatively low Voc around 680mV. Some earlier batches of the 4 cm2 small rear emitter n-type cells had even lower Voc of only 650 mV, which gave a false conclusion concerning the potential of the rear emitter structure which delayed the project for some time. In the present research phase, we fabricated large area rear emitter cells of 22 cm2 area in an eff ort to reduce the edge recombination loss, which resulted in high Voc of over 700 mV.

4.3.2.2 PERFORMANCE OF REAR BORON EMITTER CELLS ON N-TYPE SUBSTRATES

Table 4.12 lists the performance of the large-area rear emitter n-PERT cells, which were measured at Sandia National Laboratories under 100mW/cm2, AM1.5 global spectrum at 25ºC. Voc is over 10 mV higher than the best smaller rear emitter cells. This confi rms that these n-type cells have no fundamental problems for very high performance. Most of the loss mechanisms can be further reduced to achieve even closer Voc values to their p-type counterparts.

As shown in Table 1, these 22 cm2 large-area rear emitter cells have relatively low Jsc of around 39.0 mA/cm2. This is over 2 mA/cm2 lower than the equivalent PERL cells on p-type FZ substrates. Due to the rear emitter structure, the light generated minority carriers have to diff use through the entire wafer thickness of about 400 µm to reach the rear emitter to be collected. This increases the recombination loss. Furthermore, the high surface recombination velocity at the scribed cell edge can cause much higher recombination loss than for front emitter cells, since most of the minority carriers are generated close to the front surface. However, the edge recombination loss should be signifi cantly reduced if the cell area can be increased, which is the case for commercial production cells. Reducing the substrate thickness can also reduce the edge recombination loss.

The fi ll factors of these cells are slightly lower than for 22 cm2 p-type PERL cells. It was found that silver plating to the metal grid of the rear emitter n-type PERT cells was much slower than for the p-type PERL cells. Hence, these cells had less than optimum metal grid thickness. However, this can be easily adjusted for the future silver plating processes to produce the same metal thickness. Hence, the rear emitter cells ultimately are expected to have similar FF to the p-type PERL cells.

Table 4.12. Performance of scribed 22 cm2 large-area rear emitter n-PERT cells measured at Sandia National Laboratories under the 100mW/cm2, AM1.5 global spectrum at 25ºC

Cell ID Jsc Voc FF Effi ciency

(mA/cm2) (mV) (%)

Wnp01-1-1 39.05 696 0.774 21.02

Wnp01-1-2 39.05 697 0.767 20.88

Wnp01-2-2 38.89 694 0.768 20.73

Ann

ual R

epor

t 2

00

4

Fir

st G

ener

atio

n

The highest effi ciency of these rear emitter cells was 21% measured at Sandia National Laboratories under the 100 mW/cm2 AM1.5 global spectrum, at 25ºC. This is a signifi cantly high performance, considering the cells are large-area application-ready cells. This performance can be improved by further optimisations such as reducing substrate thickness, increasing cell area, optimising substrate resistivity and further optimising the rear boron emitter diff usion profi les.

The spectral response of these rear emitter cells was measured at Sandia National Laboratories, as shown in Figure 4.51. These cells have clearly non-ideal internal quantum effi ciencies of just above 90% across nearly the whole absorbable spectrum. This reduces short-circuit current density for these rear emitter cells. The 400 µm thick substrates and the low resistivity of 0.9 Ω-cm have resulted in an increased carrier loss during transportation toward the rear junction. Thinner substrates and higher substrate resistivity are expected to improve the spectral response of these cells.

Figure 4.52 shows the calculated relation between cell effi ciency and substrate thickness, as simulated by PC-1D. Low surface recombination velocities of 20 cm/s, high carrier lifetime of 1 ms, and 0.9 Ω-cm substrate resistivity were used in the calculation. When the substrate thickness is reduced from 400 µm to 200 µm, both Voc and Jsc are expected to increase, due to the higher minority carrier density at the rear emitter and an increased spectral response. However, when the substrate thickness is reduced to less than 200 µm, the silicon is not suffi ciently thick to absorb all the incident light, and hence current density reduces. The effi ciency reaches a plateau, even if the calculated Voc keeps increasing for thinner substrates. Experimentally, the cell Voc would not increase as quickly as the calculation suggests with the reduced substrate thickness, since the surface passivation will become more and more diffi cult for extremely thin substrates.

60

Fig. 4.51: Spectral response of a rear emitter n-type PERT cell.

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

61

4.3.2.3 CELL STABILITY

One of the major problems with our previous front emitter n-type PERT cells was their instability during storage in a nitrogen box, as well as under one-sun illumination. However, it is found that the rear emitter n-type PERT cells have very stable and even improved performance under one-sun illumination by ELH lamp light source for many hours. Figure 4.53 shows these results.

As shown in Figure 4.53, all the parameters of the rear boron emitter n-type PERT cells had been improved during one-sun ELH lamp illumination. The cell effi ciencies were improved by 0.2% to 0.4% absolute. Most of these gains came from the Jsc improvements, which is a sign of reduced total recombination loss. All the other parameters have also improved.

Most possibly the high recombination at the scribed edges has been reduced during illumination. The scribed bare silicon surface might be oxidised slightly during the exposure, which can marginally reduce the edge surface recombination. It is also possible that electrical charge can be deposited on the edge silicon surface, which can also change the edge recombination velocity.

Figure 4.52. PC-1D simulated relations of (a) effi ciency; (b) Voc; and (c) Jsc as a function of the substrate thickness. Reducing the substrate thickness from 400 µm to 200 µm is expected to signifi cantly increase all three parametres.

Figure 4.53. The perfor-mance of rear boron emit-ter n-type PERT cells dur-ing 1 – 2 days one-sun illumination by ELH lamp. All parameters of these cells improved with illumination time. Most possibly, the edge recombination may have been reduced by illumi-nation process.

Ann

ual R

epor

t 2

00

4

Fir

st G

ener

atio

n

It is clear that moving the boron diff used emitter from the front to the rear surface has signifi cantly improved the cell stability.

4.3.2.4 RECENT THIN CELL PERFORMANCE

The latest batch of rear emitter n-PERT cell has just been processed in the Centre. Higher resistivity 1.5 -cm substrates of 270 µm thickness were used for these cells. The performance of these cells was measured relative to a Sandia calibrated cell. Table 4.13 list these results.

Although still to be confi rmed by Sandia, these latest cells have shown excellent Voc of 706 mV, which is just 3 mV below the best p-type PERL cells. This clearly demonstrates the potential of these rear emitter n-PERT cells. The current density is also 1 mA/cm2 higher than the previous thicker rear emitter cells, possibly due to an improved spectral response. High effi ciencies of about 22% with their relatively simple processing requirements could make these cells highly attractive for commercial use.

Unfortunately, these rear emitter cells still have relatively low fi ll factors. The cause of this will be further investigated in the future.

Rear emitter cells on n-type CZ substrates will also be investigated in the near future. Since high carrier lifetimes of over 5 ms have been demonstrated from n-type SEH CZ substrates, the n-type CZ rear emitter cells are expected to give similar performances to the n-type FZ cells.

62

Table 4.13. Performance of scribed 22 cm2 large-area rear emitter n-PERT cells measured relative to cells calibrated at Sandia National Laboratories under the 100mW/cm2, AM1.5 global spectrum at 25ºC

Cell ID Jsc Voc FF Effi ciency

(mA/cm2) (mV) (%)

Wnrj4-3b 40.1 706 0.782 22.10

Wnrj4-3a 40.0 705 0.780 22.02

Wnrj4-1b 40.0 706 0.766 21.61

Ann

ual R

epor

t 2

00

4

Fi

rst

Gen

erat

ion

63

4.4 SECOND GENERATION: THIN-FILMS

Ann

ual R

epor

t 2

00

4

Sec

ond

Gen

erat

ion:

Thi

n-Fi

lms

University Staff:A/Prof. Armin Aberle (group leader)Prof. Martin GreenDr. Alistair SproulProf. Stuart Wenham

Technical Staff:Dr. Bruce Beilby (part-time)Mark Griffi nJürgen Weber (part-time)

Research Fellows/Associates:Dr. Patrick Campbell (part-time)Yidan Huang (part-time)Dr. Per Widenborg

Research Students:Natapol Chuangsuwanich (Masters)Peter Harris (PhD)Daniel Inns (PhD)Oliver Kunz (Masters)Dengyuan Song (PhD)Jirka Stradal (Masters)Axel Straub (PhD)Mason Terry (PhD)Timothy Walsh (PhD)

Undergraduate Thesis Students:Violaine Barroux

Research Assistants:Daniel BallSin Von ChanRaphael GebsHolger HabenichtSuzie HunterDaniel KongFred Martin-BruneShervin Motahar

64

RESEARCH

Ann

ual R

epor

t 2

00

4

Sec

ond

Gen

erat

ion:

Thi

n-Fi

lms

65

4.4.1 SUMMARY

The primary aim of the Centre’s Second-Generation group (or “Thin-Film Cell group”) is to develop polycrystalline silicon (poly-Si or, equivalently, pc-Si) thin-fi lm solar cells on glass, an approach that is widely recognised as being a pathway towards substantially lowering the cost of photovoltaic (PV) solar electricity. Previous work by the group has led to the invention of three poly-Si on glass thin-fi lm solar cells (EVA, ALICIA, ALICE) and a glass texturing method (AIT). Based on these earlier results, the group’s main areas of research in 2004 have been:• Optimisation of the open-circuit voltage Voc of EVA cells on planar and textured glass; • Optimisation of the Voc of ALICIA cells on planar glass; • Optimisation of the Voc of ALICE cells on planar glass, using PECVD for silicon deposition; • Optimisation of the AIT glass texturing method; • Fabrication of EVA cells on AIT-textured glass;• Development of novel metallisation and interconnection methods for poly-Si solar cells on glass;• Development of novel characterisation methods for poly-Si solar cells on glass.

The group has also been busy with taking over a semiconductor cleanroom facility from CSG Solar (formerly Pacifi c Solar) at 82 Bay Street in Botany, about 6 km from UNSW’s main campus. Support for purchasing the equipment was provided by a UNSW Capital Grant. Acquired equipment items include a glass washing machine, two fume cupboards for chemical processing, a spinner for photoresist deposition, two tube furnaces, two metal evaporators, a silicon deposition system (LPCVD), a rapid thermal anneal system, and a multi-chamber cluster tool for PECVD deposition of dielectrics (SiN etc) and doped amorphous silicon. The cleanroom equipment is set up for square glass plates with size 15 x 15 cm2. The cleanroom is connected to a gas delivery system (providing a variety of semiconductor-grade gases such as silane, nitrogen, hydrogen, phosphine, diborane, etc), as well as appropriate gas monitoring systems, a deionised water system, a gas exhaust system, and an air conditioning system. The systems servicing the cleanroom continue to be owned, operated and maintained by CSG Solar. Also available in an adjacent room is a 1064-nm Nd:YAG laser for solar cell scribing work. Furthermore, offi ce space and a characterisation area (featuring a solar cell I-V tester and a Suns-Voc tester) are available in the neighbouring (also UNSW-owned) building at 78 Bay Street.

Progress with the open-circuit voltages Voc of the three investigated cells (EVA, ALICIA, ALICE) during 2004 has been very good, as shown in Fig. 4.54. It becomes increasingly clear that all three PV technologies have the potential for achieving voltages of above 500 mV and hence are promising candidates for effi cient poly-Si thin-fi lm solar cells on glass. Since solar cells need both voltage and current, our emphasis will now increasingly shift to the short-circuit current density Jsc of the devices. Substantial Jsc improvements are possible by implementing a light trapping scheme. This will be realised by texturing the Si-facing surface of the glass with our AIT method and by adding a back surface refl ector.

In 2004, the Thin-Film Cell group has again been active in generating intellectual property, with the fi ling of provisional Australian patent 2004 903028 on thin-fi lm cell interconnection methods and the submission of a draft to Unisearch Ltd. for a provisional Australian patent application on thin-fi lm cell metallisation. Furthermore, the group has been very active with generating research papers, as evidenced by 5 journal papers, 19 conference papers, and the submission of 16 further papers (including 8 journal manuscripts) that are presently in the review process.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

4.4.2 BACKGROUND INFORMATION ON THIN-FILM PV

It is widely agreed that thin-fi lm photovoltaics has the potential to generate solar electricity at much lower cost ($/kWh) than is possible with a Si wafer based technology. Silicon thin fi lms on foreign supporting materials are particularly interesting due to their non-toxic nature, the abundance of Si, the huge technological experience with Si in the microelectronics industry, and Si’s well-matched bandgap energy with respect to photovoltaic conversion of sunlight. Various high-temperature as well as low-temperature approaches are currently being investigated to explore how suitable Si thin-fi lms can be fabricated on foreign supporting materials [1-5] (references are listed at the end of section 4.4) or can be transferred onto such materials [6-8]. The Si materials presently being investigated internationally for PV applications range from amorphous Si (a-Si:H) over nanocrystalline Si, microcrystalline Si (µc-Si:H; grain size g < 1 µm) and polycrystalline Si (poly-Si or, equivalently, pc-Si; 1 µm < g < 1000 µm) to multi- and singlecrystalline Si. Glass is a particularly promising supporting material as it is cheap, transparent (enabling superstrate and/or bifacial modules), long-term stable, and readily available in large quantities. The drawback with glass sheets is their limited thermal stability, excluding the use of lengthy high-temperature (> 650 °C) processing steps during solar cell fabrication. Our work focuses on pc-Si because we believe that, on a life-cycle basis, PV modules using this material will, in the not-too-distant future, produce the cheapest PV electricity. This assessment is based on the assumption that pc-Si thin-fi lm PV modules will have the same excellent long-term stability as today’s standard Si wafer-based PV modules. One of the advantages of pc-Si over a-Si:H and nc/µc-Si:H is the much higher (orders of magnitude) lateral conductance of (heavily) doped thin fi lms. Therefore, no transparent conductive oxide (TCO) is required on the illuminated solar cell surface for cell metallisation, enabling potentially very inexpensive metallisation and interconnection methods. The most effi cient pc-Si solar cells made as yet at low temperature on a foreign supporting material have an effi ciency of 9.2 % and were made by solid phase crystallization (SPC) of PECVD-deposited a-Si on a metal substrate [9]. The most advanced pc-Si solar cell on glass is probably the CSG technology of UNSW spin-off company CSG Solar where the entire solar cell structure is deposited onto textured glass as amorphous material (using PECVD) and then crystallised by SPC [4,10]. The pilot line CSG modules have effi ciencies of up to 8.2 % and open-circuit voltages of close to 500 mV/cell. The CSG devices have excellent light trapping properties, enabling 25 mA/cm2 of short-circuit current with 1.5 µm thin Si fi lms [4]. This shows that pc-Si thin-fi lm solar cells on glass merely require a minority carrier diff usion length of a few microns in the absorber region to achieve > 8 % effi ciency. The CSG technology is deemed ready for mass production and the fi rst factory is presently being built in Germany.

66

Fig. 4.54: Evolution of the Voc of the three poly-Si thin-fi lm solar cells on glass under development in the Cen-tre’s Thin-Film Cell group (EVA, ALICIA, ALICE). Also shown, for comparison, is the Voc of CSG Solar’s poly-Si on glass technology.

Jan990

100

200

300

400

500

600

Jan00 Jan01 Jan02 Jan03 Jan04 Jan05

CSG

4.0 %7.25 %

8.2 %

8.0 %

ALICIAALICE

EVA

Year

Voc

mV

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

67

4.4.3 MOTIVATION BEHIND GROUP’S WORK

The Thin-Film Cell group’s research aims at the realisation of novel pc-Si thin-fi lm solar cells on glass. One aim is to establish potentially cheaper Si deposition methods than PECVD (~30 €/m2 for 1.5 µm thick SPC pc-Si fi lms [10]) for the realisation of such cells. Another aim is to realise pc-Si fi lms with better electronic quality (larger grain size, longer diff usion length, etc) compared to the existing technologies, enabling higher cell effi ciency. Cell thickness is generally about 2 µm. In the following we describe progress obtained during 2004 with EVA, ALICIA and ALICE cells. The fi rst two cells are made by vacuum evaporation, whereas ALICE cells can be made by either vacuum evaporation or PECVD. Evaporation has the advantage of being a fast and inexpensive Si deposition method. Because our silicon evaporator is not equipped with a vacuum annealer, no evaporated ALICE cells have been made in 2004 and instead PECVD was used for ALICE work. Our R&D work during 2004 focused on improving the open-circuit voltage Voc of these three solar cells. As can be seen in Fig. 4.54, the rate of progress was very good and the Voc of the best cells now exceeds 450 mV. Given this momentum, we are optimistic that we will soon be able to improve the Voc of one (or even all) of these thin-fi lm PV technologies to over 500 mV. Such a Voc is an important prerequisite for the realisation of > 10 % effi cient pc-Si thin-fi lm solar cells on glass.

4.4.4 CHALLENGES WITH C-SI THIN-FILM CELLS

4.4.5 HOW WE TACKLE THESE CHALLENGES

While tempered soda lime glass is the cheapest PV-compatible glass (~10 €/m2), we use a more expensive fl oat glass (Borofl oat33 from Schott AG, Germany, ~25 €/m2 [11]) because this glass has a better thermal stability. As a consequence, cell processing can occur at higher temperatures and/or with higher thermal budget, providing scope for signifi cantly improved Si material quality. To achieve light trapping, we texture the Si-facing surface of the glass plate with a novel method called AIT (aluminium-induced texture). Low-temperature pc-Si formation on glass is realised using three diff erent methods developed by us. The resulting solar cells are called EVA, ALICIA and ALICE. The fi rst two cells are made by vacuum evaporation of Si in a non-ultra-high vacuum (non-UHV) environment,

The main challenges associated with c-Si thin-fi lm solar cells on glass are:• Glass is the ideal supporting material (cheap, long-term stable, transparent), but doesn’t permit lengthy high-temperature steps. Thus, a low-thermal-budget Si formation method is required that is cheap but nevertheless gives good-quality c-Si.• Glass is amorphous and hence c-Si on glass is polycrystalline. Thus, a cheap and effi cient method for passivating grain boundaries (and bulk defects) is required.• c-Si is a rather poor absorber of near-infrared light. To maintain a high current in thin cells, an excellent light trapping scheme is required.• The fabricated thin-fi lm material needs to be scribed into individual cells and these need to be metallised and interconnected. Thus, an inexpensive cell metallisation and interconnection method is required.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

whereas ALICE cells can be made by either vacuum evaporation or PECVD. Evaporation in a non-UHV environment is interesting for PV as it is a particularly fast and inexpensive silicon deposition method. It also has an excellent Si source material usage and avoids the use of toxic gases. Rapid thermal annealing is used to reduce the density of point defects in our cells and to increase the fraction of electrically active dopants. For grain boundary and bulk defect passivation we use a standard parallel-plate PECVD machine and strike a hydrogen plasma. A range of cell metallisation and interconnection methods is being explored by us, the common features of which are the use of a laser to separate cells from each other, and a completely self-aligning metallisation process, SAMPL (for self-aligning maskless photolithography).

4.4.6 OPTIMISATION OF EVAPORATED CELLS

A EVA solar cells

EVA stands for “solid phase crystallisation of evaporated Si”. The cell structure (see Fig. 4.55) is glass/SiN/n+/n-(or p-)/p+. The a-Si is evaporated at low temperature (~200 °C) and then crystallised by atmospheric-pressure SPC in a tube furnace (N2, 600 °C, 48 h). Good pc-Si material can be made on Borofl oat33 glass by this approach, as shown by the cross-sectional transmission electron microscope (XTEM) image in Fig. 4.56. The Si grains are up to 1.5 µm wide and are preferentially (111)-oriented. The good crystalline material quality is also confi rmed by Raman and UV refl ectance measurements [12].

68

Glass

SiNn+

p-p+

Fig. 4.55: EVA pc-Si solar cell structure.

Fig. 4.56: XTEM image of EVA material on planar glass.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

69

During 2004, an investigation into the impact of rapid thermal annealing (RTA) on planar EVA solar cells was performed [13]. Rapid thermal annealing is used to perform a high-temperature (> 700°C) process step primarily for dopant diff usion, point defect annealing and dopant activation on thin-fi lm transistors, solar cells, and other devices. The eff ect of a RTA process on the Suns-Voc curve (i.e., pseudo I-V curve) of a single sample is illustrated in Fig. 4.57. The pseudo I-V curve is analysed with a 2-diode model with fi xed ideality factors of 1 and 2 and a shunt resistance. The fi t parameters V1 and V2 are the fi tted 1-Sun Voc of the n = 1 and n = 2 diodes, respectively. RTA plateau time was 9 min at 900 °C. The Voc is dominated by n = 2 recombination in both cases, although slightly more so after RTA. An increase in Voc from 138 mV to 171 mV is seen.

The eff ect of hydrogen passivation on the Suns-Voc curve of the same sample is illustrated in Fig. 4.58. Using a 9-min, 900 °C RTA and 480 °C RF PECVD hydrogen passivation, the as-crystallized 1-Sun Voc increased by a factor of 3.2 (from 138 to 443 mV), V1 by a factor of 2.3 (from 201 to 457 mV), and V2 by a factor of 3.2 (from 152 to 493 mV). The eff ective ideality factor of the fi tted Suns-Voc curve is calculated to be about 1.4. In this case the Voc shifts from being n = 2 recombination dominated (after RTA) to n = 1 recombination dominated (after hydrogenation) [13].

In Fig. 4.59, a comparison of measured Voc after RTA (lower curve) and after hydrogenation (upper curve) as a function of the RTA plateau time is shown for the 900 °C RTA process. The eff ect of the RTA plateau time on the measured Voc is dramatic, especially after hydrogen passivation. The optimal 900°C RTA plateau time for the Voc of this solar cell structure is around 540 s. For this time, maximum point defect removal and dopant activation is thought to be achieved while the negative impact on electrical device performance from smearing of the emitter and the resulting shunt resistance is still negligible. The best Voc obtained in the course of this study was 454 mV. With optimization in RTA profi le and hydrogenation, a Voc of 500 mV is clearly within the reach of the EVA technology.

Fig. 4.57: Comparison of Voc, V1 and V2 before RTA (left group) and after RTA (right group).

Fig. 4.58: Comparison of Voc, V1 and V2 after RTA (left group) and after hydrogen passivation (right group).

69

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

In parallel to the above RTA and hydrogenation studies, Mesa-type EVA cells were made from hydrogenated fi lms (~2 µm thick). The best effi ciency achieved so far is 1.4 % (Voc = 345 mV, Jsc = 6.0 mA/cm2, FF = 0.64) [12]. These results were obtained on planar glass. EVA cells were also made on AIT-textured glass. The surface topography of such a cell (measured by an atomic force microscope) is shown in Fig. 4.60, whereas Fig. 4.61 shows a cross-sectional transmission electron microscope (XTEM) image of such a cell. The best voltage and effi ciency on textured glass so far are 260 mV and 1.1 %, respectively [14]. These values are somewhat lower than on planar glass due to shunting and/or crystallisation problems.

70

Fig. 4.59: Voc after RTA (lower curve) and after hydrogen passivation (upper curve) as a function of the RTA plateau time. RTA plateau temperature = 900 °C.

Fig. 4.60: AFM image of the surface of an EVA poly-Si fi lm made on SiN-coated textured glass (image area = 10x10 µm2).

Fig. 4.61: Cross-sectional TEM image of an EVA cell made on SiN-coated AIT-textured glass.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

B. ALICIA solar cells

ALICIA stands for aluminium-induced crystallisation ion-assisted deposition [15]. The idea behind ALICIA cells is to directly (i.e., epitaxially) grow the crystalline absorber layer on a hydrogen-terminated seed layer made by AIC (aluminium-induced crystallisation) on glass. The AIC precursor structure is glass/SiN (100 nm)/Al (~200 nm, evaporated)/a-Si (~300 nm, sputtered). This structure is annealed for 12 h at 425 °C in a tube furnace at atmospheric pressure and then the Al and the excess Si are removed. The resulting average grain size is about 10 µm, see Fig. 4.62. The high crystal quality of AIC seed layers is confi rmed by UV refl ectance, Raman, and TEM measurements [16].

For low-temperature Si epitaxy we use IAD (ion-assisted deposition) because this method is capable of high-rate Si growth at low (i.e., borosilicate glass compatible) temperatures of about 600 °C. Figure 4.63 schematically shows an ALICIA solar cell. The two IAD-grown pc-Si layers have a combined thickness of about 2 µm and are deposited in less than 30 minutes (this time includes sample heating, Si deposition, cooling and unloading of the sample). We have developed an IAD process that is capable of achieving good-quality epitaxial Si in a non UHV environment [17,18]. This makes IAD Si epitaxy potentially suitable for the PV industry.

Fig. 4.62: FIB micrograph (top view) of a 300 nm thick AIC pc-Si seed layer on planar glass.

Fig. 4.63: Schematic of an ALICIA pc-Si thin-fi lm solar cell on glass. Note that the drawing is not to scale with respect to grain size and layer thicknesses.

71

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

UV refl ectance, Raman and TEM (see Fig. 4.64) all show that ALICIA material is of high structural quality. Using SIMS (secondary ion mass spectroscopy), we analysed the concentration levels of various elements (N, O, C, Ga, P, B, Al, Fe, Cu, Mo, Na), both before and after a RTA treatment (4 min at 900 °C). The SIMS results verifi ed that high-rate Si growth by IAD keeps the concentrations of the investigated contaminants below critical levels, despite the non-UHV environment. As an example, Fig. 4.65(a) shows the concentrations of N, O and C throughout a 1300 nm thick ALICIA cell before RTA treatment. As can be seen, the density of these contaminants is < 1018 cm-3 in the base region of the cell and hence at an uncritical level. (Remark: The SIMS data at a depth > 800 nm should be treated with care because they are very likely aff ected by an artefact (enhanced sputter rate along the grain boundaries, see inset in Fig. 4.65(a)), causing the impurity densities to increase much earlier than they should due to localised sputtering into the AIC fi lm). Figure 4.65(b) shows the dopant concentration profi le (P, Al, Ga) in an ALICIA solar cell before (solid lines) and after the rapid thermal anneal (dotted lines). As can be seen, the Ga concentration is below 1x1016 cm-3 throughout the entire solar cell and hence Ga plays no role with respect to the device’s active doping profi le. The Al concentration in the range 0-700 nm appears to be unaff ected by the above artefact. A straight-line behaviour is apparent, both before and after RTA, showing that the Al concentration falls off exponentially with increasing distance from the AIC seed layer, as expected from standard diff usion theory. The RTA treatment increases the Al concentration in the base region by about a factor of 10. After RTA, linear extrapolation of the straight line towards the AIC seed layer shows that the Al concentration is dangerously high (> 1x1017 cm-3) at depths in the range 850-1130 nm. This indicates that the duration of our 900 °C RTA process (4 min) should be reduced to minimise excessive outdiff usion of Al from the seed layer into the base region of the solar cell. The phosphorus concentration in the base region, before RTA, is uniform (~5x1016 cm-3) and the n+-doped back-surface-fi eld (BSF) region, which can be considered a “dead layer” for solar cell work, is narrow (~100 nm) and heavily doped (P concentration increases sharply from 5x1016 cm-3 in the base to 6x1018 cm-3 in the BSF region). Such a doping profi le appears well suited for solar cells. The RTA step, while benefi cial in terms of reducing the density of point-like defects and dopant activation, causes a major redistribution (“smearing”) of the phosphorus profi le due to diff usion. As a result, the BSF region has widened to over 600 nm and its peak concentration has reduced by a factor of two. Given that the heavy doping reduces the diff usion length in the base (and hence the short-circuit current of the solar cell), it seems that a shorter RTA time should be used to maximise the solar cell effi ciency.

72

Fig. 4.64: Bright-fi eld XTEM image of an ALICIA pc-Si solar cell on planar glass.

Ann

ual R

epor

t 2

00

4

Sec

ond

Gen

erat

ion:

Thi

n-Fi

lms

During 2004, a detailed investigation and optimisation of the thermal profi le of the IAD silicon deposition process of ALICIA cells was performed. Prior to loading of the samples into the IAD machine, they are chemically cleaned to remove contaminants from the seed layer surface and then dipped in hydrofl uoric acid (HF) to remove the surface oxide which forms during cleaning. This sequence creates a hydrogen-terminated Si surface, temporarily preventing the re-oxidation of the Si surface. Samples need to be heated from room temperature to 450-600 °C for epitaxial growth, and during this heating the hydrogen desorbs and epitaxy is started. The hydrogen layer on the Si surface can thus be considered as a sacrifi cial protective layer [17]. Hence the heating phase has to be a trade-off between a stable hydrogen termination and a high enough temperature for good epitaxial growth. Initially, samples were slowly pre-heated for 8 minutes to approximately 450 °C then rapidly heated to growth temperatures with a lamp power of 1 kW for 6 minutes. Growth was begun at various times during the fi nal heating step. A broad but clear peak can be seen in Fig. 4.66 at around 300 seconds for both the as-grown and the RTA-treated samples. The superiority of intermediate heating times can be explained by the fact that for excessively long heating times the hydrogen termination of the growth surface deteriorates (resulting in poorer quality growth), whereas for short heating times the glass substrate has not yet reached the minimum temperature necessary for good-quality epitaxial growth.

Fig. 4.65: (a) SIMS results for the O, C and N contamination in an ALICIA cell before RTA treatment. (b) SIMS results for the dopant atom concentration (P, Al, Ga) before (solid lines) and after (dotted lines) RTA treatment. The vertical dashed line in both graphs marks the interface between the AIC seed layer and the IAD-grown fi lm.

Fig. 4.66: Average open-circuit voltage versus the time to start of epitaxy. The open squares represent the values obtained directly after epitaxy (as-grown) while the solid squares represent the values after rapid thermal anneal treatment.

73

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

Figure 4.67 shows the dependence of the average Voc on the fi nal heater temperature Tfi nal. The fundamental trend in the data, both before and after RTA, is that the Voc (i.e., the material quality) improves with increasing fi nal heater temperature. This is a common eff ect in Si epitaxy. However, superimposed on this fundamental trend, both before and after RTA, is a severe reduction of the Voc at fi nal heater temperatures of about 620-640 °C. This eff ect is very likely due to the fact that the thermal expansion coeffi cient of the glass substrate (Borofl oat33) is signifi cantly larger than that of c-Si in this temperature range [11], causing severe stress in the Si fi lms epitaxially grown at these temperatures.

Using planar glass, ALICIA cells have achieved voltages of up to 420 mV in 2004, currents of up to 11.4 mA/cm2, and effi ciencies of up to 2.2 % [19,20].

4.4.7 OPTIMISATION OF PECVD ALICE CELLS

ALICE stands for aluminium-induced crystallisation solid-phase epitaxy. The idea behind ALICE cells is to deposit the absorber onto an H-terminated AIC seed layer at very low temperature (~200 °C) as amorphous material and then to crystallise the amorphous material in a thermal anneal at elevated temperature (570-600 °C). This method is related to solid-phase crystallisation (SPC), however, because of the presence of a crystalline seed layer, it is actually a solid-phase epitaxy (SPE) process. The key feature in the ALICE process is a crystallographic transferral of information during a thermal anneal from the seed layer into the crystallising a-Si overlayer [21]. Figure 4.68 shows the ALICE precursor structure. The precursor formation is followed by a thermal anneal, whereby the a-Si crystallises via SPE. To obtain high-quality poly-Si via SPE of evaporated a-Si, it is important that the thermal anneal is performed in-situ [21], i.e. without breaking the vacuum after a Si deposition. Because our PECVD cluster tool does not yet feature a vacuum anneal chamber (this work is still in progress), the SPE anneal of PECVD a Si:H was performed ex-situ in a conventional tube furnace. Nevertheless, UV refl ectance, optical transmission microscope and FIB all show that the ex-situ annealed PECVD-fabricated ALICE material is of a similarly high structural quality as in-situ annealed evaporated ALICE material [22]. Figure 4.69 shows a FIB micrograph of PECVD-made ALICE material. It can be seen that the large grains of the AIC poly-Si seed layer have been transferred into the SPE material. The best Voc of PECVD ALICE cells achieved in 2004 is 423 mV, see Fig. 4.70. These results clearly show that ALICE is a promising novel pc-Si on glass thin-fi lm PV technology.

74

Fig. 4.67: Average open-circuit voltage versus the fi nal heater temperature Tfi nal. The open squares represent the values obtained directly after epitaxy (as-grown) while the solid squares represent the values after a rapid thermal anneal treatment.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

75

4.4.8 GLASS TEXTURING

The AIT (aluminium-induced texture) glass texturing process consists of 3 steps: (i) Deposition of a thin Al fi lm onto glass; (ii) annealing of the sample at 500-650 °C for ~6 h to allow Al to reduce SiO2 at the interface; (iii) wet-chemical removal of the Al and the interfacial reaction products using two etches (fi rst H3PO4, then a HF:HNO3 mix). A key feature of the AIT process is the experimentally observed fact that the reaction between

Fig. 4.68: Schematic of the SPE precursor structure of ALICE solar cells.

Fig. 4.69: FIB micrograph (45° tilt) of a 1500 nm thick, PECVD-fabricated ALICE pc-Si solar cell on planar glass.

Fig. 4.70: Suns-Voc data (symbols) and 2-diode model analysis (lines) of the 423-mV PECVD-fabricated ALICE cell.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

Al and SiO2 is spatially non-uniform along the surface of the glass pane. The AIT process is very versatile and, by varying the HF:HNO3 ratio, enables the realisation of a variety of glass textures, including textures with sub-micron feature sizes. Both mildly textured and heavily textured glass surfaces can be realised. A focused ion beam (FIB) microscope image of an EVA pc-Si fi lm on an AIT-textured glass substrate with texture feature sizes of about 1 µm is shown in Fig. 4.71 [23].

To quantify light trapping, we measure the wavelength-dependent internal absorption effi ciency (IAE) of the samples, i.e. the ratio of absorbed photons to photons entering the fi lm [24]:

R (refl ectance) and T (transmission) are measured quantities and Rf is the front refl ectance (i.e. the fraction of incident photons not coupled into the fi lm). Rf of each Si fi lm is determined using a nonlinear extrapolation method we have developed [25]. Figure 4.72 shows the IAE of 3-µm EVA pc-Si fi lms on planar glass and on two diff erent AIT-textured glasses [23]. These measurements were performed in the substrate confi guration and did not use a back surface refl ector (BSR) on the air side glass surface. Also shown (curve labelled “randomizer”) is the theoretically expected IAE for a perfectly randomizing texture. It can be seen that the 1:20 AIT sample comes quite close to the randomizer curve, a clear proof that the 1:20 AIT texture provides very good light trapping.

76

Fig. 4.71: FIB microscope image of an EVA pc-Si fi lm made on AIT-textured glass (HF:HNO3 ratio = 1:20).

Fig. 4.72: IAE of 3-µm EVA Si fi lms on 3 diff erent glass substrates (substrate confi guration, no BSR). Also shown is the expected result for an ideally randomizing sample (squares).

(1)

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

4.4.9 CELL METALLISATION AND INTERCONNECTION

In 2004, a range of possible thin-fi lm cell metallisation and interconnection schemes have been explored. One such interconnection scheme is based on laser scribing of Si fi lms to divide a large-area cell into many smaller cells of equal size. During this laser scribing process, the Si fi lms are covered with doped spin-on glass fi lms, aiming at the realisation of grooves with heavily n- and p-doped sidewalls. A schematic cross section of a device based on this scheme is shown in Fig. 4.73. After the cells have been divided, and the sidewalls appropriately doped, adjacent cells are interconnected using SAMPL, a novel self-aligning maskless photolithography method [26].

One such heavily doped laser-formed groove sidewall, and the metal deposited using the SAMPL method, are shown in Fig. 4.74. The metal (Al) makes ohmic contact with the heavily doped p-type sidewall of one cell, and the heavily doped n-type sidewall of the adjacent cell, thus forming a string of series connected cells. Suns-Voc measurements have confi rmed the series interconnection of the individual cells in devices made by this method [26]. Patent protection has been sought for this cell interconnection method in June 2004 with the fi ling of provisional Australian patent 2004903028.

Location of p-n junction

Location of 2ndlaser scribe

Location of 1stlaser scribe

Alp+ Si

p Si

n+ Si

SiNglass substrate

Fig. 4.73: Schematic cross section of a device based on asymmetric sidewall doping and SAMPL metallisation. Shown is one thin-fi lm cell, and the series interconnections to the adjacent cells.

77

Fig. 4.74: FIB microscope image of a heavily doped Si cell sidewall formed by laser scribing. A trench is milled into the sample to reveal the cross section of the contact region

Ann

ual R

epor

t 2

00

4

Sec

ond

Gen

erat

ion:

Thi

n-Fi

lms

4.4.10 NOVEL CHARACTERISATION METHODS FOR POLY-SI SOLAR CELLS ON GLASS

C-V measurements on non-ideal diodes by impedance analysis

For new solar cell technologies (e.g. pc-Si on glass), high-quality devices (“ideal” devices) are not readily available. However, for such technologies it is important to understand the activation and incorporation of dopants in the base layer to further optimise and improve the device. Typically, the base doping concentration is measured using Hall measurements, secondary ion mass spectroscopy (SIMS), or capacitance-voltage (C-V) measurements [27]. C-V measurements, being non-destructive, are the obvious choice for PV applications since they can be performed on the solar cell itself and give the active doping density. Commonly the diff erential depletion region capacitance C is measured with a capacitance meter at a fi xed frequency for various reverse biases. For such measurements, low series resistance Rs, high shunt resistance Rsh, and a low dark saturation current are essential to obtain reliable results. Hence impedance spectroscopy was previously used to interpret the data of non-ideal devices [28].

In 2004 we introduced a novel method for obtaining the C-V characteristics of non-ideal p-n junction solar cells and demonstrated that impedance spectroscopy can be directly used to determine the capacitance on non-ideal devices [29]. The method is based on the analysis of the measured frequency dependence of the sample’s impedance Z (“Z analysis method”). It enables the accurate determination of the base doping density and the junction built-in potential from the resulting C-V characteristics. The method was tested on ideal Si wafer devices and then transferred to non-ideal devices (ALICIA pc-Si thin-fi lm solar cells on glass). Using the impedance analysis method, the infl uence of rapid thermal annealing (RTA) and hydrogenation on the active base doping density of ALICIA solar cells was investigated. Temperature dependent Z analysis is used to further investigate the infl uence of defects and impurities on the active base doping density.

Principle of the method

The magnitude and phase of the impedance is measured using a commercial instrument (Hewlett Packard, impedance/gain-phase analyser, model 4194A) at 402 diff erent frequencies in the range 100 Hz to 40 MHz, using a small (10 mV) ac measuring signal superimposed on a constant dc reverse bias (see Fig. 4.75). The obtained experimental data is then fi tted using the model shown in Fig. 4.75 (inset), with C, Rsh, Rs and L as fi t parameters. The sample measured in Fig. 4.75 is a polycrystalline Si p-n junction diode on glass with a uniformly doped one-sided abrupt junction (ALICIA cell). The sample has a high Rs and a low Rsh. The sample is contacted with metallic needles and, for shielding purposes (illumination, electric noise), enclosed by a metal box. If not stated otherwise the samples are measured at 295 K. As can be seen, good agreement between experiment and theory is obtained. Then, by repeating the procedure for a number of fi xed reverse bias voltages, the C-V curve across a suffi ciently large reverse bias range is obtained. Finally, the classic 1/C2 vs. V representation is used to determine the sample’s doping concentration on the lightly doped side of the p-n junction and its built-in potential. The main advantages of this approach are (i) its insensitivity to non-ideal device properties like a large diode leakage current under reverse bias and (ii) its simplicity of interpretation since no appropriate measurement regime (i.e., appropriate ac frequency) for the C-V measurement has to be found.

78

Ann

ual R

epor

t 2

00

4

Sec

ond

Gen

erat

ion:

Thi

n-Fi

lms

79

Verification of the method using Si wafer solar cells

Results on wafer-based Si solar cells were used to verify the method and compared to a conventional C-V measurement. Excellent agreement was achieved for wafer-based Si solar cells between the two methods. Moreover the reproducibility of the method was found to be extremely good. To investigate the infl uence of the parasitic series and shunt resistances, an ‘ideal’ device (Si wafer solar cell) was modifi ed by adding various combinations of known external shunt and series resistances. This allows mimicking a non-ideal device under test. It is emphasised that the ability to measure these resistances accurately is an essential requirement for the analysis of non-ideal devices, as will be shown later. Good agreement between the external series resistance and the measured series resistance was obtained. The deviation was in the range of 2-3 %, regardless of the used combination of series and shunt resistance. However, large deviations in the capacitance were observed. The lower curve in Fig. 4.76(a) shows a typical example with a series resistance of 470 Ω and a shunt resistance of 470 Ω. For comparison the result obtained on the original sample (i.e., without external resistances) is shown as well (upper curve). The strong deviation in the slope of the two straight lines is caused by the dc voltage drop across the series resistance, which is not negligible if Rs is not negligibly small compared to Rsh. Thus, the 1/C2 data has to be plotted versus the dc voltage across the shunt resistance. This voltage is given by

with Vdc.appl being the applied dc bias voltage (see also inset in Fig. 4.75). Since the small-signal (ac) series and shunt resistances should, in fi rst-order approximation, be the same as the respective large-signal (dc) resistances, the voltage drop across the series resistance can easily be taken into account. Using Eq. 2 and the resistance values measured with the Z analysis method, good agreement between the 1/C2 data of the original device and those of the externally modifi ed devices is obtained. As an example, Fig. 4.76(b) shows the corrected plot where 1/C2 is plotted versus the dc voltage across the shunt resistance. This clearly shows that the capacitance can be correctly determined even in the presence of a low shunt resistance and/or a high series resistance. Hence the need for an ideal test device is not given anymore. It is noted that, regardless of the measurement method used, a correct analysis is only possible if the shunt and series resistances are known at the corresponding operating point, i.e. at each reverse bias voltage used. This is a clear advantage of the presented method since it allows the determination of all four circuit components (C, Rs, Rsh, L) at each reverse bias voltage used.

Fig. 4.75: Measured frequency dependence of magnitude and phase of the impedance Z (solid lines) of an ALICIA solar cell for a fi xed applied dc reverse bias of -0.5 V. Also shown (dotted lines) are the two fi tted theoretical curves from which the numerical values of the circuit components are obtained. Inset: Small-signal equivalent circuit model of a non-ideal p-n junction diode in reverse bias. A small ac signal with variable frequency is superimposed on a fi xed applied dc voltage.

(2)

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

Application of the method to ALICIA solar cells

The investigated ALICIA pc-Si thin-fi lm solar cells on glass have the following structure: 3 mm glass / 70 nm SiN / 200 nm p+ AIC / 50-100 nm p / ~2000 nm n- / 100 nm n+. The samples were rapid thermal annealed (RTA) at 900 °C for 240 s and then hydrogenated at a substrate temperature of ~480 °C in a RF hydrogen plasma. After the hydrogenation one sample was cut into two pieces and these were annealed in a nitrogen-purged furnace for 30 min at either 360 °C or 420 °C.

To give an example for the earlier statement that non-ideal p-n junction diodes cannot be reliably analysed using the standard (i.e., fi xed-frequency) C-V method, Fig. 4.77 shows a comparison of the capacitance-voltage data obtained from a pc-Si solar cell with the Z analysis method (fi lled squares) and the fi xed-frequency C-V method at 1 kHz (open triangles) and 100 kHz (fi lled triangles). The Z analysis method gives a doping density of (5.1 ± 1.2)x1016 cm-3 and a built in potential of 0.58 ± 0.12 V. As can be seen in Fig. 4.77, the scatter in the experimental 1-kHz fi xed-frequency data is very large and hence neither N nor Vbi can be determined reliably. The scatter is due to the fact that the measured signal is dominated by the sample’s shunt resistance. Despite a low scatter in the experimental 100-kHz fi xed-frequency data, this data must also be rejected because it produces a far-too-high junction built-in potential.

80

Fig. 4.76: Plot of 1/C2 versus voltage (a) before and (b) after correction for the dc voltage drop across the series resistance. The fi lled squares are for the original device (i.e. low Rs and high Rsh). The open squares represent the externally modifi ed device (Rs = 470 Ω, Rsh = 470 Ω). The corrected data was obtained from the measured shunt and series resistance using Eq. 2. The dashed lines are linear fi ts to the experimental data, producing the doping level N on the lightly doped side of the junction and the junction’s built-in potential, Vbi.

Fig. 4.77: Capacitance-voltage data obtained on a pc-Si solar cell with the Z analysis method (squares) and the fi xed-frequency C-V method at 1 and 100 kHz (open and fi lled triangles). The dc voltage shown is the applied dc voltage for the fi xed-frequency method and the dc voltage across the shunt resistance for the Z analysis method. The solid lines are linear fi ts to the experimental data, producing the doping level N on the lightly doped side of the junction and the junction’s built-in potential Vbi.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

Figure 4.78 shows the measured active base doping density (i.e. the free charge carrier density n in the base region) for the as-grown sample, after RTA, and after hydrogenation. The RTA step increases the base doping from 1.6x1016 to 3.4x1016 cm-3. Correspondingly, in the highly doped layers (AIC seed layer, IAD top layer), a reduction in the sheet resistance is observed. This can be attributed to a better incorporation of the dopants in the crystal and hence a higher fraction of electrically active dopants. Furthermore, SIMS measurements (see Fig. 4.65(b)) indicate that an additional increase in the base doping is caused by diff usion (“smearing”) of phosphorus from the highly doped n+ back surface fi eld region into the base layer. The hydrogenation step increases the doping density in the lowly doped base region to 1.2x1017 cm-3, which agrees very well with the doping density measured by SIMS. This increase can be explained by a trapping of the free carriers due to a high defect density prior to hydrogenation. Defects are then passivated during hydrogenation, causing an increase in n. While this eff ect is dominant in the lightly doped material (defect density similar to doping density), dopant passivation (i.e., neutralisation) is dominant in the heavily doped layers, as refl ected in the increase of the sheet resistance in Fig. 4.78.

To further improve the understanding of the behaviour in the lowly doped material, the hydrogenated samples were annealed at two diff erent temperatures and then characterised with temperature-dependent impedance measurements (T range = 23-150 ºC). While for the sample annealed at 360 ºC (fi lled squares in Fig. 4.79) almost no temperature dependence is observed, a strong temperature dependence is found for the sample annealed at 420 ºC (open squares).

81

Fig. 4.78: Base doping concentration (solid squares) measured with the impedance analysis method after each manufacturing step and the sheet resistance of the AIC seed layer (fi lled circles) and the IAD top layer (open circles).

Fig. 4.79: Free carrier densities n vs. 1000/T for the two anneal temperatures. The symbols represent the experimental data. The lines are the fi ts obtained from the model.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

To explain this behaviour, the data was modelled using a simple model consisting of 3 energy levels:• ND represents the phosphorus doping concentration at the energy level ED = 1.075 eV above the valence band edge. • NO represents unintentional n-type doping from donor-like defects or impurities (e.g. oxygen) at the energy level EO.• NT represents unintentional p-type doping from acceptor-like defects or impurities at the energy level ET .

The bandgap Eg was kept constant at 1.12 eV. Using these energy levels, charge neutrality requires

Inserting Boltzmann statistics for non-degenerate semiconductors into Eq. 3, the following neutrality condition is obtained:

Solving Eq. 4 numerically for EF, the free carrier concentration n can be calculated for each temperature [27]. Under the condition n >> p we obtain

and hence n can be directly measured using impedance analysis as long as frequency dependences of the defects are negligible.

To fi t the measured data, the parameters ND, NO, NT, and EO were varied while ET was set to 0.5 eV. Please note that other values will result in diff erent defect densities NT. The infl uence, however, is small for energy levels close to 0.56 eV (midgap) since the energy level is fully occupied with electrons and therefore always negatively charged. Using EO = 0.92 eV, NO = 3x1016 cm-3, ND = 1.1x1017 cm-3 and a trap density NT = 8x1015 cm-

3, good agreement between the experimental data and the modelled data is obtained for the sample annealed at 360 °C. To model the sample annealed at 420 °C, EO = 0.92 eV, NO = 5x1016 cm-3, ND = 1.1x1017 cm-3 and a trap density NT = 5x1016 cm-3 are used. This indicates that an increased defect density mostly causes the diff erence in the temperature dependence. This behaviour can be easily understood by the larger loss of hydrogen at higher anneal temperatures. This was also observed in a lower open-circuit voltage measured on the sample annealed at 420 °C (290 mV, compared to 350 mV after the 360 °C anneal).

82

(5)

(3)

(4)

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

83

These results show that impedance analysis yields easily interpretable results and reliable values for the doping density and the junction built-in potential, for both ideal and non-ideal p-n junction diodes. Measurements on ALICIA pc-Si thin-fi lm solar cells on glass show that Fermi-level pinning and contaminants have a strong impact on the free carrier density in the lowly doped base region. The insight gained from these measurements into the base doping density, a crucial parameter for solar cell performance, makes the impedance analysis method an invaluable tool for further device optimisation.

References for Section 4.4

1 H. Keppner, J. Meier, P. Torres, D. Fischer, and A. Shah, Appl. Phys. A 69, 169-177 (1999).2 T. Kieliba, S. Bau, D. Osswald, and A. Eyer, Proc. 17th European Photovoltaic Solar Energy Conference, Munich, 2001, p. 1604.3 K. Yamamoto, A. Nakajima, M. Yoshimi, T. Sawada, S. Fukuda, K. Hayashi, T. Suezaki, M. Ichikawa, Y. Koi, M. Goto, H. Takata and Y. Tawada, Proc. 29th IEEE Photovoltaic Specialists Conference, New Orleans, 2002, p. 1110.4 P.A. Basore, Proc. 3rd World Conference on Photovoltaic Solar Energy Conversion, Osaka, 2003, p. 935.5 A.G. Aberle, P.I. Widenborg, A. Straub, and N. P. Harder, Proc. 3rd World Conference on Photovoltaic Energy Conversion, Osaka, 2003, p. 1194.6 C. Berge, R.B. Bergmann, T.J. Rinke, and J.H. Werner, Proc. 17th European Photovoltaic Solar Energy Conference, Munich, 2001, p. 1277.7 K. Feldrapp, R. Horbelt, R. Auer and R. Brendel, Progress in Photovoltaics 11, 105-112 (2003).8 A. Blakers, K. Weber, V. Everett, S. Deenapanray, J. Babaei, and M. Stocks, Proc. 19th European Photovoltaic Solar Energy Conference, Paris, 2004, p. 431.9 T. Matsuyama, N. Terada, T. Baba, T. Sawada, S. Tsuge, K. Wakisaka and S. Tsuda, J. Non- Cryst. Solids 198-200, 940 (1996)10 P.A. Basore, Proc. 19th European Photovoltaic Solar Energy Conference, Paris, 2004, p.455. 11 http://www.schott.com/german/12 D. Song, A. Straub, P. Widenborg, B. Vogl, P. Campbell, Y. Huang, and A.G. Aberle, Proc. 19th European Photovoltaic Solar Energy Conference, Paris, 2004, p. 1193.13 M.L. Terry, A. Straub, D. Inns, D. Song, and A.G. Aberle, Proc. 31st IEEE Photovoltaic Specialists Conference, Orlando, USA, Jan. 2005 (in press).14 D. Song, P.I. Widenborg, A. Straub, P. Campbell, N. Chuangsuwanich, Y. Huang and A.G. Aberle, Proc. 31st IEEE Photovoltaic Specialists Conference, Orlando, USA, Jan. 2005 (in press).15 A.G. Aberle, A. Straub, P.I. Widenborg, A.B. Sproul, Y. Huang, and P. Campbell, Progress in Photovoltaics 13, 37-47 (2005).16 P.I. Widenborg, T. Puzzer, J. Stradal, D.H. Neuhaus, D. Inns, A. Straub, and A.G. Aberle, Proc. 31st IEEE Photovoltaic Specialists Conference, Orlando, USA, Jan. 2005 (in press).17 A. Straub, N.-P. Harder, Y. Huang, and A.G. Aberle, J. Crystal Growth 268, 41-51 (2004).18 A. Straub, D. Inns, M.L. Terry, Y. Huang, P.I. Widenborg, and A.G. Aberle, submitted to J. Crystal Growth, Dec. 2004.

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

83

Ann

ual R

epor

t 2

00

4

Sec

ond

Gen

erat

ion:

Thi

n-Fi

lms

19 A. Straub, D. Inns, O. Kunz, M.L. Terry, P.I. Widenborg, A.B. Sproul, and A.G. Aberle, Proc. 31st IEEE Photovoltaic Specialists Conference, Orlando, USA, Jan. 2005 (in press).20 D. Inns, A. Straub, M. Terry, Y. Huang, and A.G. Aberle, submitted to 20th European Photovoltaic Solar Energy Conference, Barcelona, June 2005.21 P.I. Widenborg, A. Straub, and A.G. Aberle, J. Crystal Growth (in press).22 P.I. Widenborg, J. Weber, M.L. Terry, and A.G. Aberle, submitted to 20th European Photovoltaic Solar Energy Conference, Barcelona, June 2005.23 N. Chuangsuwanich, P. Campbell, P.I. Widenborg, A. Straub, and A.G. Aberle, Proc. 31st IEEE Photovoltaic Specialists Conference, Orlando, USA, Jan. 2005 (in press).24 P. Campbell, Glass Technology 43, 107-111 (2002).25 N. Chuangsuwanich, P.I. Widenborg, P. Campbell, and A.G. Aberle, Tech. Digest 14th International Photovoltaic Science and Engineering Conference, Bangkok, 2004, p. 325.26 T.M. Walsh, S.R. Wenham, and A.G. Aberle, Proc. 31st IEEE Photovoltaic Specialists Conference, Orlando, USA, Jan. 2005 (in press).27 P. Blood and J.W. Orton, Techniques of Physics, The electrical characterization of semiconductors: Majority carries and electron states (Academic Press, 1992).28 M. Meier, S. Karg, and W. Riess, J. Applied Physics 82, 1961 (1997).29 A. Straub, R. Gebs, H. Habenicht, S. Trunk, R.A. Bardos, A.B. Sproul, and A.G. Aberle, J. Applied Physics (in press).

84

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

85

Ann

ual R

epor

t 2

00

4

S

econ

d G

ener

atio

n: T

hin-

Film

s

85

4.5 THIRD GENERATION: ADVANCED CONCEPTS

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

University Staff:Dr. Gavin Conibeer (group leader)Dr. Richard Corkish Prof. Martin Green

Research and Postdoctoral Fellows:Dr. Eun-Chel Cho (to July, 2004)Dr. Ximing DaiDr. Tammy Humphrey (part time from June, 2004)Dr. Kuo-Lung Lin (from July, 2004)Dr. Tom Puzzer (part time)Dr. Bryce Richards Dr. Thorsten Trupke

Research Associates:Yidan HuangDr. Didier Debuf (Adjunct Fellow)

Higher Degree Students:Young Hyun ChoThipwan Fangsuwannarak Chu-Wei “Scott” JiangKuo-Lung “Albert” Lin Supriya PillaiGiuseppe ScarderaAvi Shalav

Visiting Researchers:Prof. David Cahen, Weizmann Inst, Tel Aviv (Jan to Feb, 2004) Prof. Kazuo Nagajima, Tohoku University, Katahira, Aoba-ku, Sendai, Japan (Feb, 2004)Prof. Yushi Xue, Dalian University, Dalian, China (Oct to Dec, 2004)

Visiting Students:Howard Cheng (“Taste of Research” scholar, January, 2004)Daniel Parker (3rd year U/grad thesis student, 2004)Edwin Pink (3rd year U/grad thesis student, 2004) James Rudd (“Taste of Research” scholar, December, 2004)

The work of the Third Generation Strand through 2004 has extended the work of 2003 and of the earlier ARC Third Generation Special Research Centre. The principal objective is to significantly, rather than merely incrementally, improve photovoltaic cell performance beyond that of present devices. In 2004, proofs of concept have been achieved in the three principal project areas: Si nanostructures, up-conversion and hot carrier cell contacts. Also in two of the subsidiary project areas there have been significant advances: with plasmon enhancement of electroluminescence achieved from Si LEDs and an approach to within 1% of photoluminescent cooling in GaAs devices. At the end of 2003 the third subsidiary project area on quantum antennae was brought to the conclusion that it is not an attractive route even in theory.

86

RESEARCH

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

87

During 2004 a decision was made to concentrate on the Si nanostructure approach, for two reasons. Firstly as this work is the most developed experimentally, it is thought likely that implementation of the “all-Si” tandem cell based on Si nanostructures is the most promising route for significant improvements in efficiency. Secondly, Si nanostructures are useful for all three of the major project areas: “all-Si” tandem; up-conversion; and hot carrier cell.

In addition, through 2004, work in the Third Generation Strand has enabled leverage of significant additional funding, with a project on high efficiency thermoelectrics, funded by Toyota Central Research Laboratories and Toyota Future Project Division, worth $180,000 over two years. This project started in Dec 2004.

Third Generation

The term Third Generation derives from the progression of photovoltaic technologies. First generation refers to high quality and hence low defect single crystal photovoltaic devices these have high efficiencies and are approaching the limiting efficiencies for single band gap devices (see below). However, such devices are labour and energy intensive and are not likely to achieve lower costs than US$1/W (see Fig. 4.80). Second generation technology involves low cost and low energy intensity growth techniques such as vapour deposition and electroplating. Such processes can bring costs down to a little under US$0.50/W (see Fig. 4.80) but because of the defects inherent in the lower quality processing methods, have much reduced efficiencies compared to First Generation.

Third Generation aims to decrease costs to below US$0.50/W, potentially to as low as US$0.20/W, by dramatically increasing efficiencies but maintaining the economic and environmental cost advantages of thin film deposition techniques (see Fig. 4.80). This is possible because such devices aim to circumvent the Shockley-Queisser limit for single band gap devices that limits efficiencies to the “Present limit” indicated in Fig. 4.80 of either 31% or 41% (depending on concentration ratio). This can be done by tackling the two very significant

loss mechanisms of non-absorption of below band gap photons and the fact that any photon above the band gap, of whatever energy, only produces one electron hole pair at the band gap energy, the rest of the photon energy being lost by thermalisation (see Fig. 4.81). These two mechanisms alone amount to the loss of about half of the incident solar energy in solar cell conversion to electricity. Such approaches do not in fact disprove

100100

8080

6060

2020

00 100100 200200 300300 400400 500500

4040

Effic

ienc

y, %

US$0.10/W US$0.20/W

US$1.00/W

US$3.50/W

Present limit

ThermodynamicLimit

US$0.50/W

Cost, US$/m2

Figure 4.80: Efficiency and cost projections for first-, second- and third-generation photovoltaic technology (wafers, thin-films, and advanced thin-films, respectively).

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

the validity of Shockley-Queisser limit, rather they avoid it by the exploitation of more than one energy level - in some form – for which this limit does not apply. The limit which does apply is the thermodynamic limit shown in Fig. 4.80, of 67% or 8.7% (again depending on concentration).

All Third Generation approaches are based on tackling one or both of the “below band gap” or “thermalisation” loss mechanisms mentioned above, as detailed below.

Third Generation Approaches to Tackling the Major Losses in PV Cells

Third Generation concepts are based on devices than can exceed the theoretical solar conversion efficiency limit for a single energy threshold material. This limit was calculated in 1961 by Shockley and Queisser as 31% under 1 sun illumination and 41% under maximal concentration of sunlight (46,200 suns, which makes the latter limit more difficult to approach than the former).

Shockley and Queisser [J App Phys, 32 (1961) 510-519] developed a detailed balance formalism which equated the number of electron hole pairs collected by the external circuit to a solar cell (i.e. the current divided by electronic charge) with the difference between the number of photons absorbed and the number emitted. This approach simplified the problem of accounting for all the absorptions and emissions of photons that occur in a cell. Shockley and Queisser used a number of idealising assumptions which, with a few modifications, give the maximum possible efficiencies from a solar cell. These assumptions included modelling the sun as a black body at 6000K emitting radiation described by the Planck equation; and the cell as a luminescent body at 300K emitting light above the band gap described by a modified Planck equation with emission exponentially enhanced by the bias on the cell. They also assumed zero contact resistance and infinite mobility which resulted in an external voltage for the cell equal to the split in quasi-Fermi levels for electrons and holes. Thus the losses in their cell were only the below band gap, thermalisation and junction losses ((1), (2) & (3) in Fig. 4.81). The band gap of their idealised cell was optimised to give maximum efficiency.

88

Figure 4.81: Loss processes in a standard solar cell: (1) non-absorption of below band gap photons; (2) lattice thermalisation loss; (3) and (4) junction and contact voltage losses; (5) recombination loss.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

89

Hence the routes to exceeding the Shockley-Queisser limit address the below band gap and the thermalisation loss mechanisms in Fig 4.81. Means to tackling these have been variously quoted as falling into three generic categories, namely: multiple energy threshold devices; modification of the incident spectrum; and use of excess thermal generation to enhance voltages or carrier collection. Specific directions for experimental research on each of these were set for the Third Generation SRC in 2002 and continued to be investigated in 2003. These continue to be the most promising routes to achieving the objectives of the Third Generation Strand, except there has been a shift in emphasis in 2004 to concentrate to a greater extent on the first of these using Si nanostructures, as these are useful for the other approaches as well. The specific technologies chosen address the generic approaches mentioned above and are respectively:

• Band gap engineering of silicon based nanostructures for silicon based tandems: This project seeks to exploit the quantum confined properties of Si quantum dots embedded in a SiO2 matrix (or matrix of another Si compound). The aim is to engineer a wider band gap material through formation of a quantum dot superlattice. The layers are fabricated by growth (sputtering or PECVD) of alternating layers of stoichiometric SiO2 and silicon rich oxide (SRO). On annealing the excess Si precipitates to form Si nano-crystals. The layers are 10nm down to 1.5nm and below 4nm the nanocrystals formed are uniform in size, spherical and exhibit good crystallinity, as shown by TEM data. PL data shows increased energy level optical transitions, with much greater luminescent intensity – indicating true quantum dot (QD) formation.

In 2004 characterisation of the electrical properties of Si QDs in oxide has begun, with promising, although still very high, resistivities measured. Also TEM characterisation has been extended to include a plan view technique rather than only cross-sectional. It has also been identified that the resolution limit for HRTEM imaging is being exceeded in imaging the Si QDs, but not the information limit. Therefore work has begun on quantifying TEM images more precisely with through focus series and by reconstructing lattice images from TEM diffractograms.

Furthermore in 2004, the work has been extended to include silicon nitride as a matrix for the Si quantum dots. Si3N4 has a lower barrier height than the oxide and hence a true superlattice and mini-band will form with a lower quantum dot density. Si QDs in nitride have been grown by both sputtering and PECVD with good HRTEM evidence of Si nanocrystal formation.

• Up- or down-conversion of the incident spectrum: Previous work in the Centre has shown that modification of the solar spectrum incident on a cell is one way to boost efficiency, either by down-converting one high energy photon to two photons just above the band gap or by up-converting two or more below band gap photons to a photon above the band gap. This effectively narrows the band width of the radiation incident on the cell and hence boosts efficiency. The important property for such devices is high quantum efficiency meaning that they must be very radiatively efficient. Experimental work is concentrating on up converters as conversion of any low energy photons that are normally wasted can boost current and increase efficiency.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

Theoretical work on the limiting efficiencies of various up/down conversion devices has continued as has experimental work on up-conversion using erbium doped phosphors. A quantum efficiency of 3.4% has been achieved for such phosphors over a very narrow wavelength range below the band gap and an increase in current demonstrated, if a very small one, for an up-converter mounted beneath a bifacial Si PV cell. This is a significant increase over the quantum efficiency announced last year and represents an important proof of principle of the device. Work continues to increase the quantum efficiency and to extend the wavelength range of absorption.

• Hot carrier solar cells: In a conventional cell photogenerated carriers from high energy photons thermalise with the lattice in a few picoseconds. Thus their excess energy is wasted. Hot carrier cells aim to slow this cooling such that the carriers can be collected before they thermalise. Collection must be through contacts that only conduct over a very narrow range of carrier energy so that the cold external contacts do not cool the hot carriers in the absorber.

An important proof of concept has been achieved in 2004, through the demonstration of negative differential resistance (NDR) for a sandwich structure of Si quantum dots between two thin oxide layers. The NDR is evidence for a resonant conduction across the two oxide layers – most likely by quantum mechanical tunnelling – which gives a larger probability of tunnelling conduction for a narrow range of carrier energies. NDR is a necessary pre-requisite for selective energy contacts.

Theoretical work has continued on the properties required of an absorber which slows carrier cooling. Two avenues have been identified both of which seek to restrict carrier cooling by restricting the range of phonon energies that can exist in a given material. This limits the possibility of optical phonons – heated by scattering with hot carriers– from losing their energy to acoustic phonons – and is known as the “hot phonon” effect. Enhancement of this effect can be achieved either by use of certain III-V compounds, such as InN, which have a large gap between the energies of optical and acoustic modes; or by the use of quantum well or quantum dot superlattices, in which the range of phonon energies and momenta is limited by the confinement caused by the folding of the mini-Brillouin zone in reciprocal space as the dimensions decrease.

Additionally, two other project areas are being pursued within the Strand:

• Thermophotonics: which aims to enhance the radiative efficiency of specific III-V based devices such that radiative cooling occurs under illumination. Such cooling is only a few percentage points from being achieved. The interest in this is not currently for thermophotonic devices but rather to explore the techniques required for very precise temperature measurement and to gain insight into highly radiatively efficient devices. These properties are useful for the Hot Carrier and the luminescent Up/Down Converter projects. Texturing of the back surface of a GaAs based device has been pursued. It is modelled that this - with the ZnSe dome already employed - should boost PL efficiency to 97.5% - as compared to the 96% achieved previously – even closer to the 98% required for radiative cooling. In addition similar experiments have been carried out for Si showing an 18.5% IQE and 3.1% EQE, independently confirming previous Centre measurement of very high PL EQE for Si.

90

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

91

• Surface Plasmons: these are the coherent oscillations of electrons near the surface of a conductor or semiconductor. They can be used to couple radiation into the surface modes of the material and hence can be used for effective light trapping in structures that are very thin. Reciprocally they can also enhance the electro-luminescence from a LED. In 2004 enhanced coupling of electro-luminescence has been demonstrated with Ag particles evaporated onto the surface of SOI LEDs fabricated in the Photonics group in the Centre.

Thus the Third Generation Strand has active programs in the most promising routes to devices that can demonstrate increase in solar cell efficiencies above the limit imposed by a single band gap. The approaches outlined above, are detailed in the sections that follow, preceded by a section on modelling of efficiency limits. The further significant progress, over and above that of 2003, that has been achieved in each of these areas is reported.

Efficiency Limits

Researchers: Gavin Conibeer, James Rudd, Thorsten Trupke, Martin Green, Howard Cheng, Richard Corkish, Chu Wei Jiang

Further work on the programs for calculating limiting efficiencies of various solar cell types has continued at a low level through 2004. The program models the limiting efficiencies for a given set of solar cell parameters, using the detailed balance formalism described in connection with Fig. 4.81. The simulation models the input from the sun as a black body at 6000K or from tabulated air mass data. It also models black-body emission from the cell at 300K modified by its band gap and exponentially enhanced by the bias on the cell. This modification is required because the forward bias caused by photogeneration, enhances emission by promoting recombination of electron-hole pairs. The model optimises the input variables to give maximum efficiency with a numerical integration of the modified Planck equation for the black body terms. Application of this model to multiple energy threshold cells has been reported previously with application to two terminal tandem, impurity photovoltaic (IPV) and intermediate band solar cells (IBSC); the difference between these types being in the detailed energy transitions allowed for a given number of energy levels. Also included was the spectral response of these cell types.

Figure: 4.82: Solar cell (cross hatched) with down converter on front of cell and up-converter behind cell. The dark arrows indicate photons absorbed from the sun; the white arrows indicate photons exchanged between the cell and the converter; the dark red arrows are photons emitted.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

In 2004 the work was extended to model the combination of both an up-converter and a down-converter on solar cell efficiency, see Fig. 4.82. It was found that the combination boosts the efficiency of a Si cell slightly to 42.3%, as compared to results with just an up-converter (36.3%), or just a down-converter (36.6%). Other work showed that a combination of two up-converters boosts limiting efficiency for a cell with floating band gap to 50.1%. As would be expected this is higher than for one up-converting level at 47.6%, but requires a band gap of 2.6eV. Work in this area will continue with multiple up and down conversion steps, although a further small modification to this program is required to achieve true limiting efficiencies at high band gaps. [1]

Multiple Junction Solar Cell Modeller: a GUI for the limiting efficiency programs

At the beginning of 2004, work by Howard Cheng, a Taste of Research student, continued the development of the Graphical User Interface (GUI) for these programs reported last year. At the end of 2004 James Rudd, a new Taste of Research student, continued this work and allowed implementation of this GUI for tandem cells of any number of levels up to 40, see Fig. 4.83.

This is already proving a useful resource for Centre members for straightforward calculation of limiting efficiencies. The database of calculated solutions will grow with use and allow an increasing chance of choosing suitable initial estimates for the input parameters that give convergent solutions.

92

Figure 4.83: Graphical User Interface for limiting efficiency programs.

[1] G. Conibeer, A.S. Brown, M.A. Green, MA, Proc. 19th European Photovoltaic Solar Energy Conference (Paris, June 2004) 274-277: “Efficiency limits for an ideal solar cell with combined up and down conversion”

Set up Screen Running Screen

s

f

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

93

Si-Based Nanostructures

Researchers: Eun-Chel Cho, Young Cho, Martin Green, Thorsten Trupke, Chu Wei Jiang, Gavin Conibeer, Yidan Huang, Tom Puzzer, Kuo-lung Lin, Thipwan Fangsuwannarak, Daniel Parker, Edwin Pink, Giueseppe Scardera

Collaborators: Dao Lapp, Swinburn University, MelbourneAndrei Nikulin, Monash University, Melbourne

The concept is to engineer the band structure of silicon based materials, with the aim of creating larger effective band gaps in materials which are based on abundant non-toxic silicon and its oxides, nitrides and carbides. The engineering takes the form of fabrication of silicon nanocrystals embedded in either a SiO2 , Si3N4 or SiC matrix. Nanocrystals less than 4nm in diameter behave as quantum dots exhibiting quantum confinement. With such Si quantum dots spaced sufficiently close together, overlap of the wavefuctions of quantum confined carriers in adjacent dots can allow the formation of a true superlattice with the confined states smearing out to form a mini-band. For sufficiently broad mini-bands this effectively results in a new semiconductor material with a wider band gap than Si. This in turn can be used to fabricate wide band gap p-n or p-i-n junction solar cells that can be used as tandem cell elements on top of normal Si cells, thus creating an “all-Si” tandem cell with voltage boosted from that of a normal Si cell.

As reported in last year’s annual report and elsewhere [2,3], such Si quantum dots have been fabricated by deposition of silicon rich layers interspersed with stoichiometric layers of SiO2 (an application of the method of M. Zacharias et. al. Phys Rev B 62 (2000) 8391 to photovoltaics). On annealing at 1100°C, silicon precipitates from the supersaturated solid solution to form nanocrystals. For layers of thickness less than about 4nm, the precipitation enters a regime of 2D diffusion in which the dot size is accurately controlled by the layer thickness.

Work in 2004 has continued to develop this technology with direct physical evidence of the crystalinity of these nanocrystals obtained from high resolution TEM, (see Fig. 4.84 b). During 2004 this technique has been refined to include plan view TEM images which indicate the spatial position of nanocrystals. This is shown in Fig. 4.84 c, which indicates the close spacing of these nanocrystals. Although the resolution is not yet sufficient in this case to ‘see’ crystal planes in the dots, it can be seen that the dots in this sample are about 2nm in diameter and have an average spacing of about two diameters.

Figure 4.84: a.) & b.) Cross-sectional TEM & HRTEM images of Si QDs in oxide; c.) Plan view TEM image of a single layer of Si QDs giving additional information on lateral spacing.

[2] Cho, EC; Cho, YH; Trupke, T; Corkish, R; Conibeer, G; Green, MA, Proc. 19th European Photovoltaic Solar Energy Conference (Paris, June 2004) 235-238. [3] Cho, EC; Cho, YH; Corkish, R; Xia, J; Green, MA; Moon, DS, Asia-Pacific Nanotechnology forum, (Cairns, 2003).

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

Optical MeasurementsYoung Cho, Eun-Chel Cho, Thorsten Trupke, Kuo-lung Lin

Optical results indicate quantum confined properties as evidenced by the increase in the photo-luminescent energy in PL experiments, see Fig 4.85. The data in Fig 4.85a (as presented in last year’s report) shows the increase in PL

energy with reducing dot size, which is attributed to increasing confined energy levels. It also shows a dramatic increase in PL intensity on going from 4.7nm to 3.5nm dot size. This correlates well with a greatly increased uniformity in Si quantum dot size as the deposited layer goes from 4.7nm to 3.5nm and is attributed to a transition from 3D to 2D diffusion of Si atoms to form nanocrystals. The large increase in PL intensity is due to the much greater signal at a given energy with good dot size uniformity. The fact that the intensity decreases again for smaller dot sizes is thought to be due to the twin effects of reducing overall volume of crystalline Si, and hence of PL absorption and output, and to the increasing surface area to volume ratio as the dots shrink, which increases the chance of recombination of photogenerated pairs.

The data in Fig 4.85b show an attempt to normalise for the volume and surface area to volume ratio that distorts the PL intensity data in Fig. 4.85a.). Shown here curves normalised to the Si volume for 2nm dots to the power of 1, 2 & 3. The first of these is for the volume effect, the second for the surface area to volume ratio effect (this will be squared because either electrons or holes can become trapped by defects) and the third for the combination of the two.

The result of this normalisation is a flatter curve but not one that is constant for data below 3.5nm dot size, as would have been expected. This could be due to two additional effects:

1. Nomalisation to surface area to volume ratio is only an approximation. It is actually the interfacial volume (i.e. that containing the defects) to total QD volume ratio which is relevant. As the dot size decreases this approximation becomes less valid (assuming constant interface width). Hence the ratio will increase more rapidly than linearly and again this effect will be squared.

94

Photo-luminescence at 15Kand normalised data

2 2.5 3 3.5 4 4.5 5 5.5 6QD dia (nm)

PL intensity

PL intensityNorm. to volume of 2nm QDs

Norm. to (SA:V ratio) 2 of 2nm QDsNorm. to Vx(SA:V ratio) 2 of 2nm QDs

Figure 4.85: a.) Previously presented, quantum confined energy in silicon quantum dots (15K), showing PL energy and integrated intensity; b.) PL intensity data normalised for decreasing volume and increasing surface area to volume ratios.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

95

2. The assumption that all excess Si precipitates to nanocrystals, may become increasingly less valid as dot size decreases. The annealing time required for complete 2D precipitation is likely to increase as thickness decreases and it is probable that this is not yet long enough. Hence decrease in volume with thickness could also be super-linear.

More samples are being fabricated and further PL data measured to investigate this further.

In 2004 further PL data has been obtained confirming that the quantum confinement effect in Si nanocrystals is apparent at room temperature. Fig. 4.86 shows new 300K data with further evidence for increased PL signal at 850nm (or 1.45eV) for 3.5nm nanocrystals, in agreement with Fig. 4.85a and in keeping with reduced effective mass calculation of the first confined energy level. Other higher energy levels are also apparent although not clearly defined. Also evident is a small peak shift to shorter wavelengths with increasing incident power, indicative of a slowing of thermalisation from higher confined energies to the first confined level.

The greatly increased strength of optical processes, as shown in Fig. 4.85 & 4.86, is fortuitous as it means that only very thin layers of these dots are required for strong absorption. For unconfined Si, our work with Si thin-films shows that less than 1,000nm of Si is required for high solar absorption, with an appropriate light-trapping scheme. Quantum dot devices of only 100 nm thickness seem feasible given the increased optical strength of the confined processes.

Electrical measurements

Edwin Pink, Thipwan Fangsuwannarak, Daniel Parker, Tom Puzzer, Yushi Xue

The small thicknesses also relax the requirements on carrier-mobility within the films. Carriers transfer between the dots by quantum-mechanical tunnelling, enhanced by resonance effects that are strongest when dots have periodic spacing. A systematic study has commenced in 2004, of the transport properties within these nanostructured-semiconductor materials using temperature-dependent conductivity measurements, see Fig. 4.87.

Wavelength (nm)

500 600 700 800 900

2.0 mW 1.5 mW 1.0 mW 0.5 mW 0.3 mW 0.1 mW 0.05 mw

PL intensity

300K

[4] L. Dao, X. Wen, M. Do, P. Hannaford, E-C Cho, Y. Cho, Y. Huang, M.A. Green, J Appl Phys 97 (2005) 013501: “Time integrated and time resolved PL studies of state filling & quantum confinement of Si QDs”

Figure 4.86: New PL data at 300K for 3.5nm Si QDs, showing shoulder at 850nm for first confined energy level and several unresolved higher energy levels [4].

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

Fig. 4.87 shows some early measurements of resistivity with temperature. The data shown are for a hydrogenated sample. Before hydrogenation resistivity at room temperature is about 105 Ω.cm; after hydrogenation it decreases dramatically to about 200Ω.cm. This is most likely due to passivation of defects acting as recombination centres. These defects could be on the surface of QDs or in the oxide, but more data are required to ascertain this. The activation energy, extracted from the slope of the plot, is approximately 17meV between room temperature and 120K and 10meV down to 90K before rising again around 80K. This is indicative of a change of mechanism at lower temperatures. However these are low activation energies and it is too soon to be sure of the mechanisms involved. Further studies, using this technique, of material prepared in different ways is providing information on conduction mechanisms and their dependence on preparation conditions.

Alternative matrixes for Si quantum dots: Martin Green, Eun-Chel Cho, Young Cho, Gavin Conibeer

Transport properties are expected to depend on the matrix in which the silicon quantum dots are embedded. As shown in Fig.4.88, different matrices produce different transport barriers between the Si dot and the matrix, with tunnelling probability heavily dependent on the height of this barrier. Si3N4 and SiC give lower barriers than SiO2 allowing larger dot spacing for a given tunnelling current.

96

Figure 4.87: Electrical resistivity and corresponding thermal activation energy of silicon QD superlattice layers in the lateral direction. The data shown is for a hydrogenated sample. Hydrogenation reducing resistivity by more than two orders of magnitude.

c-Si c-Si c-Si1.1 eV

0.5 eV

0.9 eV

1.1 eV

1.9 eV

2.3 eV

1.1 eV

3.2 eV

4.7 eV

Figure 4.88: Bulk band alignments between crystalline silicon and its carbide, nitride and oxide.

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

97

The important parameter in determining the degree of interaction between quantum dots is m*∆Ed2, where m* is the bulk effective mass in the respective band of the matrix, ∆E is the energy difference between this bulk band and the band formed by quantum dot interaction and d is the spacing between dots. At first sight, it would appear that the spacing of dots would have to be closest in the oxide, nitride and carbide, in that order. Similar deposition and quantum dot precipitation approaches should work for all.

Si quantum dots in a nitride matrixYoung Cho, Eun-Chel Cho, Giueseppe Scardera

For the above reason we have explored application of similar techniques to the growth of Si nanocrystals in silicon nitride by both sputtering and PECVD [5]. For sputtering, growth parameters are very similar to the oxide but with growth from a Si and Si3N4 target rather than from an oxide one. Results from TEM are very promising, see Fig 4.89, which shows crystalline nanocrystals in the nitride matrix.

In addition PECVD has been used for the growth of Si nanocrystals in nitride with a similar regime of growing alternate Si rich and stoichiometric nitride layers. Annealing is carried out again at 1100°C but with a pre-anneal at 500°C to drive off hydrogen incorporated from the PECVD process. Again HRTEM images showing even clearer nanocrystals have been obtained, see Fig. 4.90.

Figure 4.89: Sputtered Si nanocrystals in silicon nitride: a.) TEM showing multilayer structure; b.) HRTEM showing crystal structure in nano-crystals.

[5] Y. Cho et al: submitted to the 20th European Photovoltaic Solar Energy conf (Barcelona, June 2005): “Silicon Quantum Dots in SiNx matrix for Third Generation Photovoltaics”.

Figure 4.90: PECVD Si nanocrystals in silicon nitride: a.) TEM showing multilayer structure; b.) HRTEM showing crystal structure in nano-crystals.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

Further optical and electrical characterisation is underway on these nitride materials. In addition, growth of the carbide analogue is planned.

Analysis of HRTEMTom Puzzer, Gavin Conibeer, Yidan Huang, Giueseppe Scardera

High-resolution TEM promises to be the most powerful analytical technique for the study of the structural and crystallographic properties at the atomic scale of the Si quantum dots. In practice, the structural details we are interested in are beyond the point resolution of most TEMs. Under these conditions, conventional HRTEM is subject to a number of artefacts that make direct interpretation of the structure of our quantum dots problematic [Kohno et al.,“Misleading fringes in TEM images and diffraction patterns of Si nanocrystallites”, Cryst.Res.Technol. 38, 1082, 2003].

Fig. 4.91 shows an example of the contrast reversal that can occur when changing focus settings. A relatively stable image contrast with a range of defocus settings gives way to a rapidly changing contrast at certain defocus settings and even an apparent change in lattice spacing. This makes a naïve interpretation of HRTEM images unreliable for quantitative information. Comparison of such simulated data with actual through-focus images can determine the “real” crystallographic data. We are currently employing conventional image processing and Fourier techniques for diffractogram analysis. In addition, we are developing software models of the crystal structure of quantum dots which will be used to perform image simulations for arbitrarily oriented quantum dots. We are also investigating the application of exit-wave reconstruction techniques to the analysis and interpretation of HRTEM images [J.Smith,“The realization of atomic resolution with the electron microscope”, Rep. Prog. Phys. 60, 1513 ,1997].

Theoretical analysisChu-Wei Jiang, Martin Green, Gavin Conibeer, Giuseppe Scardera

Tunnelling and resonant tunnelling calculations, already being carried out, will be extended to take account of miniband formation and the fading of resonant behaviour as carriers undergo phonon interactions. Departures from periodicity are also being investigated with this approach. ‘Effective mass approximation’ modelling for large dot and potentially ‘ab-inito’ modeling for small dots is planned together with an extension of the effective mass approach to 1-D chains of dots. Additional work will explore the likely effect of overlapping bands upon spectral responsivity.

98

Figure 4.91. Through-focus series of images for defocus values, ∆f = -400nm to -580nm for (111) oriented silicon showing complete contrast reversal with variations in objective lens focus. Images simulated using the NCEMSS package (Roar Kilaas, MCEM/MSD Lawrence Berkeley Laboratory).

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

99

Up/Down-Conversion

Researchers: Bryce Richards, Avi Shalav, Thorsten Trupke, Martin A. Green

Collaborators:S. Glunz, ISE, Freiburg, Germany H.U. Güdel, K. Krämer, University of Bern, Switzerland

Due to the discrete band structure of semiconductors, only photons with energy equal to or greater than the bandgap will be absorbed and may contribute to the electrical output of a photovoltaic (PV) device. Photons of lower energy are transmitted through the solar cell and do not contribute to the electrical output. Such losses are known as sub-bandgap losses and are one of the main loss mechanisms that limit the efficiency of conventional silicon (Si) solar cells.

Recently, we have developed the first prototype of a bifacial silicon solar cell with erbium-doped sodium yttrium fluoride (NaY1-xF4:20% Erx

3+) up-conversion (UC) phosphors adhered to the rear surface, as shown in Fig. 4.92. Trivalent erbium (Er3+) is an ideal candidate for single wavelength near infrared (NIR) UC due to its ladder of nearly equally spaced energy levels that are multiples of the 4I15/2 4I13/2 (~1540 nm) transition, as shown in Fig. 4.93. This device gave a measured external quantum efficiency (EQE) of 2.5% under 5.1 mW laser excitation at a wavelength of λ = 1523 nm, corresponding to an internal quantum efficiency (IQE) of 3.8% [6].

[6] A. Shalav, B.S. Richards, T. Trupke, K.W. Krämer and H.U. Güdel, Appl. Phys. Lett. 86, 2005, 013505.

Incident light

bifacial solar cell

up-converter

reflector

bifacial solar cell

up-converter

reflector

Fig: 4.92: Schematic diagram of a bifacial solar cell with an up-converter placed at the rear to capture sub-bandgap light.

Fig. 4.93: 2- and 3-step UC processes between two Er3+ ions. Energy relaxation from one Er3+ ion (the sensitizer) can result in energy transfer to a neighbouring Er3+ ion (activator) giving rise to higher energy photons. Solid lines represent photon absorption (up) and emission (down), dotted lines represent energy transfer, wavy lines are phonon (radiationless) emission. [F. Auzel, Chem. Rev. 104, 2004, pp139-173]

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

Fig. 4.94 shows early results of an EQE measurement on a bifacial cell with and without an Er doped UC phosphor on the back surface. This clearly shows an increase in the spectral response, albeit very small, over the wavelength range 1480 to 1580nm.

The EQE is the ratio of the number of charge carriers or electron hole (e-h) pairs (φe-h) collected by the solar cell to the number of photons (φp) incident on the cell, the latter being reduced by optical transmission and reflection losses.

A linear relationship exists between generated e-h pairs and number of incident photons (i.e.φe-h ∝ φp.) and, as a result, the EQE is not dependent on the incident power intensities. However, UC processes within a rare-earth ion are dependent on the incident light intensity, and hence φe-h ∝ (φp)n for some real number n within a rare-earth UC device. Therefore, the EQE of our UC device can now be written as:

For the case of two low energy photons being absorbed and a single high energy photon being emitted, the probability of this two-step UC event occurring is proportional to the square of the number of incident photons (n = 2). In this case, a linear increase in the EQE is therefore expected with increasing incident light intensities. In general, an exponential dependence of EQE on φpn-1 would imply a n-photon step UC process. However, it has been shown that for increasing incident power, the value of n decreases [M. Pollnau, D.R. Gamelin, S.R. Lüthi, and H.U. Güdel, Phys. Rev. B, 61(5), 2000, pp. 3337 -3346.]. This decrease in n can be modelled utilizing the rate equations of each energy level.

The earlier results were improved by, firstly, using a host material that was more transparent in the NIR (white oil, mixed with a rubberiser) and also easier to work with [8]. Secondly, a greater fraction of NIR light was absorbed by using a higher phosphor

100[7] T. Trupke, A. Shalav et al, Proc PVSEC (Bangkok, 2004).[8] A. Shalav, B.S. Richards, K.W. Krämer and H.U. Güdel, Proc. of 31st IEEE Photovoltaic Specialists Conference, Orlando, Florida, Jan 2005, (in press).

(1)

(2)

Figure 4.94: EQE of a bi-facial solar cell with and without NaYF4:20% Er3+ phosphors on the back surface [7].

to host weight ratio of 4:1 and applying a thicker layer of 1.6mm. Finally, evaporated silver was used as a rear reflector instead of white paint. Progress has also been made in investigating a wide range of Er3+ concentrations, ranging from x = 0.02 to 1.0. The improved EQE is shown in Fig. 4.95 a and b as a function of wavelength and incident laser power, respectively. The peak EQE at a wavelength of 1523 nm is now 3.4% [8]. The internal quantum efficiency of the luminescence process is estimated to be 5.7%, taking into account optical losses in the host polymer and solar cell [6].

It has been determined that the 20% Er3+-doped phosphor has a higher luminescence compared with the 2%, 50% and 100% doped samples. The measured EQE is increased by 0.9% absolute over previous work by making only minor changes in device structure, while there is still plenty of room for further improvement.

Luminescent solar concentrators

Researchers: Bryce Richards, Avi Shalav, Richard Corkish

In down-conversion, more than one low energy photon is generated from each high energy photon, hence an external quantum efficiency (EQE) greater than 100% is required. In contrast, the luminescent solar concentrator (LSC) described here relies on a fundamentally simpler single-photon wavelength changing process, referred to as luminescence down-shifting. While the EQE of this process is less than unity, and thus conversion efficiencies are reduced, successive internal reflections within the host medium carry the luminescent photons to the edge of the plate where they can be collected by solar cells whose areas are small compared to that of the sheet surface. Thus, the LSC is a static (non-tracking) concentrator that is also able to work effectively under diffuse radiation conditions.

Luminescent solar concentrators (LSC) were the focus of much research in the 1970-1980s due to their promise of generating 20% more electrical energy per day than a PV device of the same peak watt (Wp) rating and at a lower cost per watt peak ($/Wp). Fig. 4.96 shows a cross-sectional diagram of a traditional LSC and how light, absorbed and subsequently re-emitted isotropically by a luminescent centre, is transported to solar

Fig. 4.95: a.) Graphs showing: a.) enhanced spectral response of a silicon device in the NIR (2mW laser power); and b.) variation of UC EQE process with incident laser power (λ= 1523nm).

00.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

1 2 3 4 5 7

Incident Laser Power (mW)

EQE

(%)

(b)

EQE

(%)

1480 1520 15600

1

2

Wavelength, λ (nW)

(a)

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

101

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

cells at the edges of the sheet by total internal reflection (TIR). While the down-shifting step inherently results in the loss of energy as heat within the luminescent centres, and hence a reduction in the LSC energy conversion efficiency (ηLSC - light to electrical power), the total power (P) generated by the LSC system can be much greater than that of the PV cell alone because of the concentration of light at the edge of the module by TIR. There are many advantages to a LSC over a standard PV module, including:

• At realistic concentrations of up 20 suns, the LSC technology is compatible with high-efficiency silicon solar cells, such as the buried-contact or Sliver solar cell.• The concentrator is static, thus no tracking mechanism is required to follow the sun, and good performance is achieved even under cloudy conditions. • As the LSC solar cells see only above-bandgap light, the PV cells operate cooler and more efficiently than a standard PV module, which, at 60°C, exhibits a 15% decrease in performance.• The narrow band of luminescence is close to the maximum spectral response of the solar cells.• Finally, it should be noted that the performance of the LSC system can be further enhanced by optimising the solar cell to better match the LSC luminescence spectrum and output intensity, e.g. adjusting the anti-reflection coating thickness, using heavier emitters and moving the collecting junction farther away from the front surface, and widening the front contact spacing.

One of the most apparent and, until now unaddressed, loss mechanisms is the escape cone loss. With a typical transparent sheet of material with refractive index n = 1.5, the critical angle for TIR to occur is χ= sin-1 (1/n) = 42°. Therefore, luminescence emitted at angles >42° will be transported by TIR to the edge of the sheet, resulting in ηtrap = 74.5%, while the remaining (25.5%) luminescence emitted at angles less than χ departs via the escape cone through the front or rear of the LSC. In order to combat the 25% escape cone losses, we have developed the low critical-angle-loss (LowCAL) LSC. This concept requires, firstly, multiple luminescent species being present within a host material. When organic dyes are mixed, it has been shown that energy transfer processes between the dyes results in 95% of the luminescence being emitted from the highest wavelength luminescence centre, and that 99% of the light emitted by the first dye is collected by a second before it leaves the LSC. The LowCAL LSC takes advantage of both of these facts, and retains the light that would have left the LSC through the escape cone by placing a “hot-mirror” (HM) at the front surface. This HM is wavelength-selective, permitting the transmission of shorter wavelength sunlight through the front surface of the device, while reflecting the longer wavelength light, as shown in Fig. 4.97.

102

PV

Cel

l

t

L

Incident Light

Escaped Light

Escape Cone

Surface ScatteringReflection

Self-Absorbtion

Absorbtion

TIR

Fig. 4.96: Cross-sectional diagram of a traditional LSC with luminescence centres (•) and PV cells attached to the edge. Light is transported to the edge via TIR, incurring some losses along the way, e.g. self-absorption, scattering and transmission within the escape cone angle (χ).

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

103

Thus, higher energy light is allowed to pass through to the LowCAL LSC by the HM and is then down-shifted to lower energy light, which is prevented from leaving the sheet due to the HM, thus providing up to a 25% gain in ηLSC. The HM also relaxes the constraints on self-absorption, as re-emitted light can no longer leave the LSC via the escape cone, and also on the surface flatness, as TIR is no longer the sole mechanism for retaining luminescence within the LSC. However, all existing HMs are optimised for transmission of visible wavelengths only (see Fig. 4.97) as they are typically used to reduce the amount of heat transmitted in powerful lamps (hence the name). The graph in Fig.4.98 compares the luminescence being emitted via the escape cone of samples with and without a HM present. With no HM, emission from the rare-earth luminescence centres occurs in both the visible and NIR, whereas when a HM is placed on the top surface of the LSC stack all NIR luminescence is retained within the stack.

It is yet to be determined whether or not the performance of HMs at angles other than their design angle will result in increased performance over the day.

Hot Carrier Cells

The concept underlying the hot carrier solar cell is to slow the rate of photoexcited carrier cooling, caused by phonon interaction in the lattice, to allow time for the carriers to be collected whilst they are still “hot” thus enhancing the voltage of a cell. Thus it tackles the major PV loss mechanism of thermalisation of carriers (loss (2) in Fig. 4.81). To be effective such carriers must be collected over a very small energy range with selective energy contacts so as to prevent cold carriers in the contacts cooling the hot carriers to be extracted. In thermodynamic terms the carriers are thus collected with a very small increase in entropy. Ideally this collection would be isoentropic using mono-energetic contacts, see Fig. 4.99.

Fig. 4.97: Transmission and reflection curves of a standard “hot mirror”, with a reflectance cut-on (λcut-on) of about 720 nm.

Fig. 4.98: NIR luminescence is retained within the LSC by the HM, while visible light can still enter/exit the device.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

The challenges for such a hot carrier cell fall into two categories [P. Würfel, “Solar Energy Conversion with Hot Electrons from Impact Ionisation”, Sol. Energy Mats. and Sol. Cells. 46, 43 (1997). The first concerns absorption under such difficult constraints and retardation of the thermalisation mechanisms relative to radiative recombination rates. Progress on the theory of phononic engineering to achieve slowing of carrier cooling is reported below. Significant experimental progress on the second category – for selective energy contacts – is also reported below.

Selective Energy Contacts

Researchers:Chu-Wei Jiang, Eun-Chel Cho, Gavin Conibeer, Martin Green, Tammy Humphrey, David Cahen

In 2004 a significant proof of concept has been achieved with negative differential resistance observed at room temperature for a double barrier quantum dot structure. The device structure is based on our Si quantum dot technology discussed above in the Si nanostructures section and is shown in Fig. 4.100.

104

Figure 4.99: Schematic and band diagram of an ideal hot carrier cell. The absorber has a hot carrier distribution at temp TH. Carriers cool isoentropically in the mono-energetic contacts to TA. The difference of the Fermi levels of these two contacts manifests as a difference in chemical potential of the carriers at each contact and hence an external voltage, V.

Figure 4.100: a.) Device structure with single layer of Si quantum dots sandwiched by SiO2 ; b.) HRTEM of the structure showing the Si quantum dot layer.

Gate electrode

Poly Si

Si dotsSubstrate electrode

WaferWafer

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

105

The device uses a degenerately doped wafer and highly doped capping layer to provide the conducting contacts either side of the resonant structure. The layer of Si quantum dots and oxide layers are grown by the reactive sputtering technique discussed above in the Si nanostructures section. The band structure and intended I-V behaviour are shown in Fig. 4.101.

The actual I-V results at room temperature are shown in Fig. 4.102.

Fig.4.102 shows definite evidence for NDR in this structure at room temperature – a very encouraging result. The data are as yet preliminary, only show limited repeatability and are not particularly well peaked at the resonant energy. Also although NDR is a necessary pre-requisite for total energy filtering it is not direct proof of a total energy filter. Nonetheless together with the evidence for quantum dot formation in the structure shown in Fig. 4.100b, this represents a very encouraging proof of concept of energy filtering.

Ec

I

V

Gate n-type degenerate wafer

EfE

Ef

I

V

I

V

Figure 4.101: Intended band structure and NDR I-V behaviour.

Figure 4.102: Actual I-V behaviour at 300K, showing NDR for two different devices on one wafer.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

Further work is now underway on similar structures, refining the technique to investigate local I-V curves on the scale of a few Si quantum dots, with the intention of improving the quality factor of the resonant peak.

In addition, modelling of the 1D resonant tunnelling has been carried out with a model developed for perturbations in the position of a defect in an oxide barrier [9]. The theory is being extended to a 2D model.

Hot Carrier Absorbers

Researchers: Gavin Conibeer, Martin Green, Tammy Humphrey

In last year’s report equivalences between several approaches to modifying phononic band structures were noted and that these might be useful in reducing carrier cooling. These approaches are the application of superlattices to thermoelectrics and the consequent phonon Bragg reflections at mini-zone boundaries; the slowing of carrier cooling in superlattices due to a “phonon bottleneck” effect; and the observation of large gaps between acoustic and optical phonon energies in many III-V bulk materials.

In 2004 work on the theory of this hot “phonon bottleneck” or “hot phonon” effect has been tentatively applied to explaining slowing of carrier cooling in superlattices. In parallel the applicability of “phononic band gap” of some bulk III-Vs (particularly InN) to slowing of cooling has also been pursued [10].

The “hot phonon effect” refers to the build up of a population of high energy optical phonons as a result of scattering from hot electrons (or holes), for instance from photogeneration. The size of this population depends on both the temperature and number of hot carriers and the routes for decay of the hot phonons after they have scattered with carriers. The “bottleneck” comes about if the routes for decay of hot optical phonons are limited and when this occurs further cooling of carriers is impeded, thus slowing carrier cooling overall.

Optical phonons are the oscillations of planes of atoms in such a way that there is no net change of momentum. i.e. they can only occur when there are at least two atoms in the unit cell and require these two atoms to vibrate in opposite directions at any given time, such that there is no net change of momentum. Optical phonons couple very effectively to hot electrons (and holes). Acoustic phonons differ in that all atoms in the plane oscillate in the same direction at once. Hence they transfer momentum, heat or sound through a crystal. They only couple very weakly to hot carriers. (For the purposes of this discussion longitudinal and transverse optical phonons behave in qualitatively the same way, as do transverse and longitudinal acoustic phonons.)

As identified by Klemens [P.G. Klemens, Phys. Rev. 148 (1966) 845], there is a selection rule that operates in the scattering of optical phonons, as illustrated schematically in Fig. 4.103.

106 (10) G. Conibeer, M.A. Green, Proc. 19th European Photovoltaic Solar Energy Conference (Paris, June 2004) 270-2: “Phononic band gap engineering for Hot Carrier Cell absorbers”. [9] C-W. Jiang, M.A. Green, E-C. Cho, G. Conibeer, J Appl Phys 96 (2004) 5006: “Resonant tunnelling in defects through an insulator” G. Conibeer, M.A. Green, Proc. 19th European Photovoltaic Solar Energy Conference (Paris, June 2004) 270-2: “Phononic band gap engineering for Hot Carrier Cell absorbers”.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

107

Decay of an optical phonon at zone centre can only occur via creation of two equal and opposite longitudinal acoustic phonons. For the ideal case this is the only possibility. Work in the Centre proposes that the observation of a hot phonon bottleneck effect slowing carrier cooling arrises because of the restriction in optical phonon decay caused by this mechanism.

Such cooling has been observed by several authors, usually in III-V materials and only at very high injection levels. However the effect is seen to be enhanced in multiple quantum well and superlattice structures, see Fig. 4.104. This is attributed to the reduced dimensionality restricting phonon modes further and hence enhancing the hot phonon effect.

Work in the Centre has led to the suggestion of a possible mechanism for this superlattice enhancement as being due to the restriction in allowed acoustic phonon modes caused by the mini-Brillouin zone edge reflections in a superlattice (or “mini-zone folding”). Fig. 4.105 shows this reduced zone and the fact that optical phonons are forced to decay by the emission of many low wavenumber acoustic phonons.

Figure 4.103: An optical phonon can only decay into two longitudinal acoustic phonons [Klemens, 1966]

-100

100

300

500

700

900

0 50 100 150time (ps)

Car

rier t

empe

ratu

re (K

) No hot phononsHot phonon modelExpt InGaAsExpt 15nm MQW

Figure 4.104: Example of time resolved PL carrier cooling data at 4K, for a laser excited carrier density ~5x1019 cm3 , showing a good fit for a hot phonon model and also reduced carrier cooling rate for MQW. [D.J. Westland et. al., Solid Sate Eelectronics, 31(1988)431; P.A. Snow, et.al., Superlattices and Microstructures, 5 (1989) 595]

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

In Fig. 4.105 folded acoustic branches are labelled with the value m. For m>0 the modes are optical-like in that they have non-zero energy at zone centre. This in turn means that they cannot take part as pure acoustic phonons in the operation of the Klemen’s mechanism and force a multiple emission of folded acoustic phonons – a less likely event.

The imaginary solutions shown correspond to phonon frequencies for which there is a complete Bragg reflection at the mini-zone boundary. This restricts the transmission of phonons within the structure and is particularly marked effect for the hypothetical material shown in Fig. 4.105 b.

It is further suggested that the extra restriction caused by a quantum dot (as opposed to quantum well) superlattice structure would enhance this effect and hence the hot phonon effect even more. However there are qualitative differences in the phonon modes – particularly interface modes - allowed in such structures, that make a comparison more complex. The further inter-relation of these factors and their effect on carrier cooling in superlattices, including the Si quantum dot structures discussed elsewhere in this report, is being investigated.

Bulk III-Vs exhibit large phononic band gaps. The largest is shown by InN due to the large difference in masses between its constituent atoms, see Fig. 4.106.

108

0

1

2

3

4

5

6

7

0 0.2 0.4 0.6 0.8 1Wavevector, q (R & I) (PI/d)

Freq

uenc

y, O

meg

a (2

v(1)

/d)

q(real)q(imag)

0

1

2

3

4

5

6

7

0 0.2 0.4 0.6 0.8 1Wavevector, q (R & I) (PI/d)

Freq

uenc

y, O

meg

a (2

v(1)

/d)

q(real)q(imag)

Figure 4.105: Mini-Brillouin zone of a superlattice showing folded acoustic modes using the calculation of [Santos and Ley, Superlattices and Microstructures, 5 (1989) 43].: a.) for a calculation based on a Si/Ge superlattice, as in the reference and b.) a hypothetical material with enhanced acoustic impedance, F.

Figure 4.106: Phonon dispersion for InN [V. Davydov, et. al., Appl Phys Lett, 75 (1999) 3297].

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

109

Note that the gap between acoustic and optical modes is larger than the highest acoustic energy or frequency. This means that the Klemens mechanism of decay of an optical phonon into just two LA phonons is not possible. This should slow the carrier cooling significantly. There is some evidence for this [F. Chen, A.N. Cartwright, Appl Phys Lett, 83 (2003) 4984] but the data are not directly comparable to other carrier cooling data.

There is some preliminary evidence of a slowing of carrier cooling in our Si nanostructures from time resolved PL, probably due to relaxation via the multiple confined energy levels in the Si QDs [11]. However this is yet to be corroborated. Work is continuing on identifying the effect of these and other mechanisms on carrier cooling and their potential link to the low dimensional phonon restrictions discussed above.

Thermoelectric cells

Researchers:Tammy Humphrey, Gavin Conibeer, Martin Green

As reported in last year’s report a spin-off of theoretical work in 2003 has included the application of the concept of energy-selective electron transport, used in hot carrier solar cells, to thermoelectrics and thermionics. During 2004, establishment of a separate programme, outside the Centre of Excellence, to further such non-photovoltaic applications, has been pursued. This has resulted in leverage of significant additional funding of A$180,000 from Toyota Central R & D Labs, Nagoya and Toyota Future Projects Division, Higashi-Fuji, Japan, for a project that commenced in December 2004.

Work in the Centre has shown that properly designed thermoelectric nanomaterials such as quantum dot superlattices with narrow peaks in their electronic density of states, operate with an efficiency approaching the theoretical (Carnot) limit in the absence of phonon heat leaks. Expressions have been derived for the theoretical limiting power and efficiency of thermionic power generators and refrigerators, demonstrating that, in principle, increases in both power and efficiency can be gained via energy filtering of electrons in these devices.

This theory shows that Carnot efficiency can be obtained in thermoelectric devices using the same thermodynamic mechanism used to obtain Carnot efficiency and Landsberg efficiency in thermionic and hot carrier solar cell devices respectively. In all three cases an ideal energy filter is needed to restrict the transport of electrons to those for which the opposing effects of the temperature and electrochemical potential gradient exactly cancel, where the occupation of states for electrons is equal in the hot and cold reservoirs.

While in solar cells maximizing the efficiency is identical to maximizing the power, this is not the case in thermionic and thermoelectric heat engines in which heat is supplied by non-solar sources. Due to non-idealities such as phonon mediated heat leaks, it is usually better in practice to maximize the power rather than the efficiency of these devices.

[11] L. Dao, X. Wen, M. Do, P. Hannaford, E-C Cho, Y. Cho, Y. Huang, M.A. Green, J Appl Phys 97 (2004) 013501: “Time integrated and time resolved PL studies of state filling & quantum confinement of Si QDs”

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

In 2004, we have extended our previous work to consider filters of finite width, which produce finite power, but with an efficiency less than the Carnot limit. In the case of a thermoelectric where electrons move diffusively from hot to cold, it was found via self-consistent numerical calculations that the power generated can be increased by up to 60% if the bandstructure is graded (or doping is inhomogeneous) so that the energy gap between the band-edge and the electrochemical potential is proportional to the temperature as a function of position in the material.

An interesting result arising out of this work has been the observation that a common thermodynamic model can be developed for thermoelectric, thermionic and conventional solar cell devices, along with other heat engines (of primarily theoretical interest) such as the thermally pumped laser [13]. This group of heat engines does not require isothermal heat transfer to achieve Carnot efficiency, unlike other ‘conventional’ heat engines such as Carnot, Otto or Brayton cycles. Instead, they all transfer heat directly between hot and cold reservoirs via particle exchange, and all require an ideal energy filter to achieve Carnot efficiency. From the point of view of fundamental thermodynamics, the interesting aspect of this result is its implication that two systems can be in equilibrium even if they have different temperatures, provided the interaction between them is limited to the exchange of particles with a single energy.

Supporting project areas

Two projects have continued into 2004, because of their role in supporting and of giving further insight into the work in the three main project areas. (A third project, on quantum antennae, has not been continued.)

Surface texturing for Thermophotonic photoluminescent cooling

Researchers: Kylie Catchpole, Kuo-Lung Lin, Martin Green, Patrick Campbell

Collaborators: Dr Fay Stanley, Centre for Quantum Computing, UNSW

110

Figure 4.107: “Loop” plots of the efficiency versus the power of a thermoelectric nanomaterial, in which the voltage is varied from zero to open-circuit as the loop is traversed anticlockwise. The three cases (a), (b), and (c) consider different values of the phonon thermal conductivity in the thermoelectric, the red/solid curves are for devices with a graded bandstructure and the blue/dotted curves for those with an ungraded bandstructure. The four curves of each colour indicate the results for different width energy filters (arrows indicate decreasing width, from 250meV to 10meV), under the assumption that the number of states available for electrons remains constant as the filter width is changed. It can be seen that the power and the efficiency is increased by grading the bandstructure of thermoelectric nanomaterials.

[12] T. E. Humphrey and H. Linke “Inhomogeneously doped thermoelectric nanomaterials” Proceedings of the International Thermoelectrics conference, Adelaide (2004). Condmat/0407506. T. E. Humphrey and H. Linke “Quantum, cyclic and particle exchange heat engines” Proceedings of Frontiers of Quantum and Mesoscopic Thermodynamics, Prague (2004). Condmat/0407508.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

111

As detailed in last year’s report, this project seeks to achieve photoluminescent cooling of a highly radiatively efficient double heterostructure GaAs device. The original motivation was to fabricate devices suitable for thermophotonic cells, in which the broadband emitter of a thermophotovoltaic system is replaced with a heated light emitting diode. This in turn illuminates a second room temperature photovoltaic diode of slightly narrower band gap. The rate of energy transfer for a given emitter temperature increases with this arrangement and emission is concentrated in an energy range more suited for conversion by the receiver. Such a device requires a high proportion of the incident heat energy to be re-emitted by the heated diode. Thus in order to achieve net conversion of heat to electricity, a very high external quantum efficiency (EQE) LED is required and hence a very high radiative recombination efficiency, so that the LED cools when a voltage is applied. As a first step towards this goal, we have been aiming to achieve cooling of an optically pumped GaAs double heterostructure. In this scheme laser light at the bandgap energy Eg is used to excite electron-hole pairs, which then thermalise with the lattice and recombine, emitting light with energy Eg + kT. Thus the required EQE is Eg/(Eg+kT)~ 98% for GaAs with Eg = 1.4 eV.

Thin heterostructure devices were fabricated by epitaxial lift off technique (ELO) to improve coupling of light from the device. Extraction of light can be further enhanced by coupling to a high refractive index hemispherical dome. ZnSe has been used for this dome, which although having a slightly low index for coupling to GaAs, is highly transparent over the wavelength range of interest, see Fig. 4.108.

In 2003 photoluminescent measurements and a combination of thermal and photoluminescent measurements indicated EQEs of 90% and 92 ± 3% respectively, (the difference being due to slightly different injection levels). For the photoluminescent measurement these results were boosted to 96% when the device was coupled to the ZnSe dome – this being tantalizingly close to the 98% required for cooling.

In 2004 in addition to further confirmation of the thermal photoluminescence results, numerical modelling of photon recycling and recombination inside devices has indicated that the layer quality of the GaAs heterostructures was excellent and suggested even higher EQEs are achievable for undoped samples. The EQE could be further increased

Figure 4.108: An ELO film mounted on a sapphire substrate. Coupling to a high refractive index hemisphere was used to increase the EQE in our measurements.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

with better light extraction by texturing one of the capping layers together with coupling to the ZnSe dome on the other side, see Fig. 4.109. This increase in EQE should be about 1.5% giving a predicted EQE of about 97.5% - even closer to the cooling threshold.

Preliminary work carried out on such surface texturing on AlGaAs/GaAs ELO films using nano-scale texturing by wet etching. A nanoscale dot array pattern was created with electron beam lithography (EBL), with variation of the e-beam dose and gap between dots. Samples were then etched in citric acid for various times. The aim was to achieve a close-packed density of circular holes with a depth around 250nm. Characterisation was carried out by SEM and AFM, with the results shown in Fig. 4.110.

Thus the holes achieved are of the size and shape required but rather too shallow. The next stage is to measure the photoluminescence from these textured layers in combination with the ZnSe dome as well as possible improvement of the nanoscale texturing using reactive ion etching, which has recently been upgraded. Furthermore, in 2004 the photoluminescence measurement method has been extended to Si samples, with internal quantum efficiencies of 18.5% indicated (3.1% EQE). These measurements provide an independent confirmation of the high photoluminescence EQEs measured previously for Si at the Centre [14].

Surface plasmons

Researchers: Kylie Catchpole, Supriya Pillai, Thorsten Trupke, Martin Green

112

Textured Surface

GaAs active layer

AlGaAs capped layer

Figure 4.110: SEM and AFM images of surface textured with EBL and citric acid etching. Indicating approximately circular holes of diameter about 350nm, minimum spacing 250nm and depth 50nm.

Figure 4.109 Schematic diagram of textured back surface for optimal light extraction from double heterostructure GaAs thin films.

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion:

Adv

ance

d C

once

pts

113

As reported last year, surface plasmons are being investigated as a means to enhance coupling of light into (or out of) Si PV cells (or LEDs).

As indicated in Fig. 4.111, a coherent oscillation of electrons in a conductor (or surface plasmons) can be induced in metallic nano-particles on the surface of a semiconductor by incident light. For certain wavelengths near the metal plasmon frequency these plasmons can resonate with particular guided modes in the thin semiconductor layer. The conditions required depend on the thickness and refractive index of the layer and the size and spacing of the nano-particles. [In fact any material that can support a dipole, e.g. polar dye molecules, can be used to support palsmons. However we are presently most interested in metals because of their ease of deposition.]

Such enhancement of absorption will be reciprocal and hence should also manifest as enhancement of electro-luminescence from an LED. We are attempting to achieve such luminescence enhancement in the SOI LEDs discussed in section 4.6 on Silicon Photonics.

In our experiments, silver nano-particles were used. Silver was evaporated onto the LED surface which was then annealed such as to form a random array of small particles of the order of 100nm in size [see Fig. 4.112a].

Figure 4.111: An illustration of how surface plasmon resonances on metal nano-particles scatter incident light into guided modes of a thin semiconductor layer.

Figure 4.112: a.) SEM image of Ag islands on LED surface, major axis 105nm, minor axis 74nm; b.) Enhancement of electroluminescence signal possibly as a result of plasmon coupling.

[14] K. R. Catchpole, “Silicon photoluminescence external quantum efficiency determined by combined thermal/photoluminescence measurements”, Semiconductor Science and Technology, Vol 19, pp1411 (2004).

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

The devices were tested in forward bias with the spectral luminescence results shown in Fig. 4.112 b. An enhancement of approx. 30% in luminescence was observed, with a greater effect at shorter wavelengths. This latter indicating that it is more than just a light trapping scattering effect. These are preliminary experiments with nano-particle size and shape and semiconductor layer thickness unoptimised. Significantly larger enhancement, as well as further insight into the exact mechanisms involved, are expected with further experiments.

Antenna Reception of Solar Energy

Researchers Richard Corkish, Martin Green, Tom Puzzer

No further work has been done on this topic following the conclusion last year that the approach cannot increase limiting efficiencies even in theory. However a watch is being kept on the literature in this area.

114

Ann

ual R

epor

t 2

00

4

Thi

rd G

ener

atio

n: A

dvan

ced

Con

cept

s

115

Ann

ual R

epor

t 2

00

4

Th

ird

Gen

erat

ion

00

RESEARCH

4.6 Silicon Photonics Research Group

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

University Staff A/Prof. Jianhua Zhao (group leader: integrated LEDs)Dr. Thorsten Trupke (group leader: silicon laser)Professor Martin GreenProf. Stuart Wenham

Research FellowDr. Aihua Wang

Adjunct FellowDr. K. Catchpole

Research StaffDr. Robert BardosDr. Tom Puzzer

Research AssistantGuangchun Zhang

Research StudentSupriya Pillai

Technical StaffJules Yang

CollaboratorsSusan AngusDr. Fay HudsonA/Prof. Andrew Dzurak, Centre of Excellence for Quantum Computing, UNSW Prof. Peter Würfel. University of Karlsruhe

4.6.1 SOI LED Devices

Since the demonstration of high electroluminescence and photoluminescencs efficiencies from the bulk silicon devices [M. Green et al, “Efficient Silicon Light Emitting Diodes”, Nature, Vol. 412, pp. 805, 2001; T. Trupke et al, “Very Efficient Light Emission from Bulk Crystalline Silicon”, Applied Physics Letters, Vol. 82, pp. 2996, 2003], one of the Centre’s main interests has been the fabrication of an efficient silicon light emission device (LED) on Silicon-on-Insulator (SOI) substrates. The primary goal for this work is to use these devices for light signal transmission both within devices on a SOI substrate and to the outside world. The first aim was to demonstrate electroluminescence, the second to demonstrate the quantum confinement effect in nano-scale layers of silicon, and the third to realise modulation of the electroluminescence of LED by applying a controlling gate signal.

A facility for SOI LED fabrication and measurement has been established, and is currently being upgraded to meet ever more demanding requirements. During 2004, the electroluminescent performance of LED has been improved. A strong electroluminescence signal was detected from devices of 300 µm by 24 µm, with the external quantum efficiency measured as 2x10-6 ± 50%. Electrical modulation of SOI LED current output has been demonstrated, with frequency response up to 5MHz. 116

RESEARCH

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

117

In this work, commercial silicon SOI wafers from Soitec (France) were used for device fabrication. The top silicon layer is p-type, 14-20 Ω-cm resistivity, (100) orientation, and 205 nm thick. The buried box oxide is 400 nm thick. Figure 4.113 shows a schematic diagram of the cross-sectional view of the light emitting diode design. During the device processing, the top silicon layer is first etched into a 300 µm by 24 µm rectangular island. Phosphorus and boron are then diffused into stripes at each side of this silicon island to form the p-i-n structure with a lightly doped base width of about 10 µm, as shown in Figure 4.113. The final thickness of the top silicon layer is estimated to be about 50 nm, with some uncertainty due to non-uniform final oxide thickness. An aluminium layer was thermally evaporated onto the top surface, with this layer subsequently patterned and annealed to form the metal contact to the p-i-n diode.

Figure 4.114 shows a section of the fabricated SOI device. It is seen that the boron and phosphorus diffused stripes at both sides of the silicon island appear as the darker coloured regions. This is due to reduced silicon layer thickness in the regions. Under the centre gate metal, the oxide was removed and regrown, to produce a thinner oxide layer for a stronger gate control. The silicon layer thickness in this region is also reduced, also producing a darker colour in Figure 4.114 (the oxide removal and regrowth consumes some underlying silicon). Such an oxidation method can be used repeatedly to reduce the centre silicon layer thickness to the desired thickness.

Figure 4.113: Cross sectional view of the SOI light emitting diode.

Figure 4.114: A top view photo of the fabricated SOI light emitting diode.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

The processing of a SOI LED includes six oxidation steps, eight chemical etching steps, two diffusions and six photolithography processes. One of the major aims during the SOI LED processing is to achieve the desired thicknesses of silicon and oxide layers at the each section of the device. In order to achieve the required device processing control, computer simulation was performed using commercial software (ATHENA). The simulation result is shown in Figure 4.115.

According to the simulation result, the silicon layer in gate region can be thinned to 10nm using chemical etching and oxidation. One finished device (without chemical etching for silicon thinning) was inspected using TEM. The thickness of gate region is about 90 nm, as shown in Figure 4.116. This sample was used as a reference in subsequent LED processing. Chemical etching was added to the LED processing sequence to further reduce the silicon layer thickness. On some samples, very thin silicon layer appear to have been obtained, possibly less than 20 nm according to the reference sample. On some other samples, the silicon layer in the device was completely removed due to over-etching.

118

RESEARCH

Figure 4.115. ATHENA simulation results of LED processing control.

Figure 4.116. Cross-sectional TEM picture on gate region of a finished LED.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

RESEARCH

119

Figure 4.117 shows the measured light emission spectra from this SOI silicon LED, as measured perpendicularly to the front surface of the device. All four measured spectra have the same peak wavelength of 1135 nm, which is a result of band edge emission. In Figure 5, published data for silicon’s absorption coefficient [M. J. Keevers and M. A. Green, Appl. Phys. Lett. 66, 174 (1995)] were used to calculate the SOI LED spectrum according to the generalized Planck equation (dashed line) [P. Würfel, J.Phys.C, 15, 3967 (1981)]. This gives a very close fit to the experimental spectrum.

The peak of the spectrum corresponds to sub-bandgap processes in which a photon and a phonon are emitted simultaneously (the bandgap of silicon corresponds to a wavelength of 1102 nm). In the spectral range 1190 nm to 1250 nm, two phonons are emitted with one photon. The inflection in the theoretical spectrum around 1060 nm represents the spectral range where a phonon is absorbed, when a photon is emitted. This spectral feature is more pronounced in the theoretical curve than in the experimental spectrum. The external quantum efficiency of the SOI LED was measured to be 2x10-6 ± 50%, not including the light emission from the edge of the device. Figure 4.118 compares the SOI LED results to previously published spectra from a planar and from a textured bulk crystalline silicon LED [M. Green et al, “Efficient Silicon Light Emitting Diodes”, Nature, Vol. 412, pp. 805, 2001]. The SOI LED has a similar peak wavelength and a similar long wavelength spectrum as the planar c-Si LED. However, in relative terms, it has a significantly enhanced short-wavelength emission in the spectral range <1100 nm. In bulk silicon LEDs, reabsorption of spontaneously emitted photons leads to a reduction of the emitted photon flux especially at short wavelength. Contrarily, in the silicon SOI LED all photons are generated in a very thin layer of silicon, which is located next to the device surface, and hence the photon reabsorption is minimal. The relative intensity in the emission spectrum at short wavelength is therefore higher. The textured bulk LED in Figure 4.119 has a significant light trapping performance, and has an effective device thickness equivalent to about 40 times the geometric thickness. As well as increasing the peak emission by over 10 times, the emission peak is also shifted to a longer wavelength of 1160 nm due to the enhanced short wavelength re-absorption, in relative terms, caused by the extended average path.

Figure 4.117: The forward bias light emission spectra of a SOI p-i-n LED at different currents.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

4.6.2 Modulation of LED Output

As shown in Figure 4.113 and Figure 4.114, a controlling gate is located on top of the thin oxide over the middle of the SOI LED. Figure 4.114 shows one of the tested I-V curves of the SOI LEDs. In the figure, the top curve is the standard I-V curve of a pn diode. However, when a controlling voltage signal is applied to the gate, the diode current can be reduced or increased, depending to the polarity of the gate voltage and the device processing conditions. This is because the gate voltage can induce either electrons or holes in the under lying silicon layer. Under some conditions, this cuts off the current channel in the silicon layer.

Applying such a controlling voltage signal to the gate is also expected to modulate the electroluminescence output of the LED. Unfortunately, the present infrared sensor used in the light detection system has a frequency limit only in kHz range, so that the electroluminescence modulation of LED output could not be measured in the Centre yet. To improve the speed of the infrared sensor unit, some rather complicated modifications are required, which will be made in the near future. Therefore, we measured only the electrical frequency response of the LED presently. The measurement setup is shown in Figure 4.120.

The forward current of the LED was modulated by a square-wave signal applied to the gate. A typical result is shown in Figure 4.121. The pink output signal clearly responds to the blue input signal with a delay time, as apparent in Figure 4.121. A response time constant of about 0.1 µs is demonstrated. This corresponds to a switching frequency limit of about 5 MHz. We believe the modulation effect can be further improved.

120

Figure 4.118: Comparison of the EL spectra from the SOI LED and from bulk Si LEDs with planar and textured front surfaces.

Figure 4.119. The I-V curve of the SOI LED. The top curve is the standard I-V curve of a pn diode. However, when a controlling voltage signal is applied to the gate, the diode current can be reduced or increased, depending to the polarity of the gate voltage and the device processing conditions.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

121

The emphasis in future research will be upon establishing a dry etching facility to improve device processing, especially for the metallization, the reduction of the thickness of the silicon layer to realise quantum confinement effects, and the upgrading of the measurement system for LED performance

Signal Generator

DC Power Supply

OscilloscopeCH1 CH2

1k

Al

Figure 4.120. Electrical frequency response measurement setup.

Figure 4.121. Electrical frequency response of LED008-1-C8.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

4.6.3 Design of a SOI Silicon Waveguide

The next research goal is to demonstrate light propagation within an all-silicon circuit by using a SOI silicon waveguide. The design of such a SOI waveguide is shown in Figure 4.122. The purple T shaped structure in Figure 4.122 is the silicon waveguide with 6 m silicon width. At each end of the T waveguide, there is a SOI LED diode, with the same structure as the ones shown in Figures 4.113 and 4.114.

One LED is used to emit light into the silicon waveguide, while the other 2 diodes are used to detect the light received through the waveguide. Other controlling structures are also incorporated along the SOI waveguide, such as control gates and pn-junction isolation rings. These devices are designed to cut-off electrical signals to make sure that any signal detected is optical. These silicon waveguide devices will be fabricated in 2005.

4.6.4 SOI LED Project / Photonic Crystals / Surface Plasmons

One important aspect of the SOI LED project is the enhancement of the external luminescence quantum efficiency, which is given by the number of photons emitted per electron driven through the device. Due to the small thickness of the active silicon layer in over SOI LEDs of typically < 100 nm, the surface recombination is the dominant recombination channel even with good quality thermal oxide passivated surfaces, resulting in relatively low internal luminescence quantum efficiencies (IQE). Because a dramatic improvement of the surface recombination velocity seems unrealistic, our efforts focus on a modification of the spontaneous radiative recombination to achieve higher IQE. The latter is targeted firstly by two dimensional photonic crystal structures within the active silicon area (i.e. a modification of the photonic density of states) and secondly by thinning

122

1

2

3

4

5

6

7

8

9

Silicon Island

Boron Diffision

Phosphorus Diffusion

Silicon Thinning

Contact Holes

Metalization

Metalization

Silicon formation

Phosphorus isolation lines

4010

Figure 4.122. SOI silicon waveguide design. The purple line is the silicon layer. One of the LED is used to emit light into silicon waveguide, while the other 2 diodes are used to detect the light received. The dimensions are in microns.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

123

the active layer to a thickness of < 4 nm, where quantum size effects become relevant (i.e. by a modification of the electronic band structure of the material). In another related project we aim to enhance the EQE for a given IQE by coupling of internally generated photons to the outside world by means of surface plasmon effects.

4.6.4.1 Photonic Crystals

Photonic crystals can be considered the optical analogues to electronic semiconductors. A continuous density of states exists for photons in a vacuum and also in bulk material. A band structure in the photonic density of states however results from a geometric structure in which the refractive index varies spatially on a length scale of the wavelength of light, in much the same way as electronic bands result from the periodic variation of the electrostatic potential in a crystal. An example for a two dimensional photonic crystal in silicon is shown in Figure 4.123 [Birner, A., et al., Silicon-based photonic crystals. Advanced Materials (Weinheim, Germany), 2001. 13(6): p. 377].

Antireflection coatings or Bragg mirrors are examples of simple one dimensional photonic crystals. Most photonic crystal based research is currently directed towards devices in which radiative recombination is impaired by the existence of a partial or, in the case of three dimensional structures, a complete bandgap in the photonic density of states. On the other hand, photonic crystals also allow radiative recombination to be enhanced if the photonic density of states can be enhanced in the spectral range around the bandgap, where spontaneous emission occurs. In most related applications, radiative centres such as dye molecules or quantum dots are introduced into the transparent periodic environment of a photonic crystal host to modify the emission.

We aim to process 2D photonic crystal structures directly into the active light emitting material itself and benefit from the fact that silicon is transparent to a large extent to its own luminescence. Figure 4.124 shows a theoretical photonic band structure for a two dimensional triangular array of air holes in silicon with spacing a. [Zelsmann, M., et al., Seventy-fold enhancement of light extraction from a defectless photonic crystal made on silicon-on-insulator. Applied Physics Letters, 2003. 83(13): p. 2542]. The left axis represents the so called reduced frequency, which is which is the spacing, a, divided by the wavelength of interest. The upper grey band in Figure 4.124 represents the spectral region, where spontaneous emission occurs in bulk silicon (for a lattice parameter of 640nm). A relatively flat photonic band with a correspondingly large density of states falls into that grey area.

Figure 4.123. A 2D photonic crystal structure (after Birner et al.).

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

This enhanced photonic density of states (compare to bulk Si) is expected to result in an enhanced spontaneous emission. Seventy fold increase of the emission rate due to such 2D photonic crystal structures was reported for photoluminescence experiments on SOI samples at low temperatures (80K) [Zelsmann, M., et al., Seventy-fold enhancement of light extraction from a defectless photonic crystal made on silicon-on-insulator. Applied Physics Letters, 2003. 83(13): p. 2542]. We are aiming to achieve similar or even higher performance improvements in our SOI LEDs at room temperature. Test devices are made in collaboration with the Centre for Quantum Computing, UNSW, where a sophisticated electron beam lithography system and reactive ion etching (RIE) equipment are used to transfer the desired two dimensional patterns into the SOI layer. The remaining device processing and characterisation of the devices is carried out at our Centre. In our initial studies, the process parameters for pattering of the SOI layers (electron beam and RIE) are being investigated and optimised. Figure 4.125 shows a SEM picture of a test structure that has been written into a PMMA mask on top of an oxidised silicon wafer. In subsequent steps the PMMA will be used as a mask for RIE processes, by which this pattern is transferred first to the top oxide of the devices and then into the active silicon layer. A subsequent cleaning and oxidation will remove the etching damage and passivate the exposed surfaces. In the early stages of this project, we mainly aim to demonstrate a significant performance improvement of SOI LEDs at room temperature by the photonic crystal structures. Photonic crystal structures, however offer many other possibilities such as guiding light in specific directions through a SOI layer, which may be beneficial, for example, for optical coupling between emitters and receivers located on the same chip. Such investigations will also shed light on the question whether the absorption properties of thin film solar cells can be improved by similar photonic crystal structures.

124

Figure 4.124. A theoretical photonic band structure for a two dimensional triangular array of air holes in silicon with spacing a (after Zelsmann et al.)

Figure 4.125. SEM picture of a test structure that has been written into a PMMA mask on top of an oxidised silicon wafer.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

125

In the early stages of this project, we mainly aim to demonstrate a significant performance improvement of SOI LEDs at room temperature by the photonic crystal structures. Photonic crystal structures, however offer many other possibilities such as guiding light in specific directions through a SOI layer, which may be beneficial, for example, for optical coupling between emitters and receivers located on the same chip. Such investigations will also shed light on the question whether the absorption properties of thin film solar cells can be improved by similar photonic crystal structures.

4.6.4.2 Surface Plasmons

A surface plasmon is a combination of an electromagnetic wave and a coherent oscillation of the electrons at the interface between a metal and a dielectric. Surface plasmon effects can be utilised, for example, to enhance the absorption in very thin silicon films (e.g. SOI samples covered by a random array of metal islands). For instance a twelve fold absorption enhancement at ~800 nm due to these effects has been reported for SOI samples in [Stuart, H.R. and D.G. Hall, Absorption enhancement in silicon-on-insulator waveguides using metal island films. Applied Physics Letters, 1996. 69(16): p. 2327]. According to Kirchhoff’s law, an enhanced absorptance also corresponds to an enhanced emissivity, which is why these effects should also be suitable to enhance the emission efficiency of SOI LEDs.

The advantage here is that the absorptance / emissivity only needs to be enhanced over a narrow wavelength range (~1000-1200nm). In contrast, in photovoltaic applications of plasmon effects, e.g. for efficient light trapping schemes, the effects would have to be optimised for a much wider spectral range. Particular care would have to be taken that performance gains in a certain spectral regime are not outweighed by stronger performance losses in other spectral regions.

In initial proof of concept studies, the EL spectrum of SOI LEDs was measured before and after deposition of silver islands onto the top oxide of existing SOI LED devices. Figure 4.126 shows an SEM picture of the SOI surface after the deposition of the islands, showing that the island size varies from tens to hundred nanometres. Figure

4.127 shows that a ~30% enhancement of the EL intensity is observed due to the deposition of the silver islands. Whether this effect results from surface plasmon effects or from a geometric light scattering effect is not clear yet. The enhancement is stronger at short wavelengths leading to a small blue shift of the spectrum, which indicates that the enhancement is not a simple light trapping effect. It must be noted that, in these initial

Figure 4.126. A SEM photo of the deposited nanometre silver islands for surface plasmon effect.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

studies, none of the system components were optimised. We expect to observe much larger effects by using a thinner top oxide (~30 nm compared to ~100 nm in present devices) and a slightly thicker active silicon layer (200-220 nm compared to currently

~160 nm) which will both enhance the coupling between the metal particles and the guided optical modes within the silicon layer. A stronger enhancement will also allow a more conclusive identification of the surface plasmon effect.

In contrast to the photonic crystal approach, the application of surface plasmon effects is not suitable to overcome the problem of the surface dominated recombination within the devices. It is rather a suitable technique to couple internally generated photons out of the device more efficiently, i.e. to maximise the external luminescence quantum efficiency for a given, in our case relatively low, internal quantum efficiency.

126

Figure 4.127. The measured enhanced EL emission from the SOI LED before and after silver dots deposition.

Ann

ual R

epor

t 2

00

4

S

ilico

n P

hoto

nics

127

4.7 Collaborative Research

Ann

ual R

epor

t 2

00

4

C

olla

bora

tive

Res

earc

h

Monash University, Melbourne, Australia

Continuation of a collaboration with Dr. Andrei Nikulin of the School of Physics and Materials Engineering, Monash, on synchrotron radiation characterisation of Si nanostructures. The collaboration resulted in a synchrotron experiment at the Australian National Beamline Facility on the Photon Factory synchrotron facility at Tsukuba, Japan, in November 2004. Funding for this experiment was the result of a successful proposal by Centre members from the ANBF. Dr. Nikulin was involved in helping to run the experiment with Centre members and in analysis of results. This continued a collaboration started in 2003 on a similar experiment on a Si quantum well structure.

Swinburne University, Melbourne, Australia

Dr. Lapp Dao and colleagues of the BSEE, Swinburne, carried out time resolved photoluminescence characterisation of the Centre Si nanostructures. Evidence for quantum confinement in the Si QDs was obtained at room tempertature and tentative evidence for carrier cooling restricted to the allowed energy levels in Si QDs. (See section on Si nanostructures in section 4.5 and ref. Dao et.al.)

Fraunhofer-Institute for Solar Energy Systems, Freiburg, Germany University of Bern, Switzerland

These collaborations were on the Up/Down conversion project, see Up/Down-conversion in section 4.5. H.U. Güdel, K. Krämer from Bern provided the Er doped phosphors and gave advice on results of absorption and PL measurments. S. Glunz from ISE, provided some of the bifacial cells used in the up-conversion device experiments.

Centre for Quantum Computing, UNSW

Dr. Fay Stanley collaborated in the Electron Beam Lithography used for texturing the surface of the GaAs based devices used in the photoluminescent cooling experiments, see Supporting Projects in section 4.5.

Australian Nuclear Science and Technology Organisation, Lucas Heights, NSW

The Centre was again successful in an application to AINSE for time on Secondary Ion Mass Spectrometry at ANSTO in 2004. In addition this time the grant included time on Rutherford Backscattering at ANSTO. Characterisation primarily involved Si nanostructures with high levels of depth resolution achieved for SIMS, better than 5nm with 2nm layers identified (this being independently measured by TEM). RBS was used for calibration of these data and in addition RBS channelling was used for further structural characterisation of Si nanostructure samples used in the synchrotron experiments.

128

RESEARCH

Ann

ual R

epor

t 2

00

4

Col

labo

rati

ve R

esea

rch

129

Weizmann Institute, Tel Aviv, IsraelTohoku University, Katahira, Aoba-ku, Sendai, Japan Dalian University, Dalian, China

Collaboration with the above was by way of prolonged visits by researchers from these institutions.

Prof. David Cahen, from the Weizmann Inst, visited the Centre for two months from January to February, 2004. He was principally involved in work on Hot Carrier cells but also provided useful contributions to other areas of Third Generation Strand activities.

Prof. Kazuo Nagajima, from Tohoko University, visited the Centre in February. He was involved in disseminating his work on Si:Ge microstructures.

Prof. Yushi Xue, from Dalian University, visited the Centre for two months from October to December, 2004. She worked on improving the techniques for electrical measurement of the Si nanostructures.

Utrecht University, The Netherlands

During 2004 the group has continued its collaboration with Prof. Ruud Schropp and co workers at the Debye Institute at Utrecht University, The Netherlands, in the area of amorphous and crystalline silicon depositions by hot-wire chemical vapor deposition (HWCVD). HWCVD is interesting for thin-film PV because it is an inexpensive silicon deposition method that is easily upscalable to large-area (~1 m2) solar modules. During 2004, joint Masters student Jirka Stradal has finished his thesis and has published several papers on his work. One of the papers has won him the “Marie Curie Early Stage Researcher Poster Award” at the 3rd International Conference on Hot-Wire CVD (Cat-CVD) Processes. The joint work investigated HWCVD epitaxial thickening of poly-Si seed layers on glass made at UNSW by aluminium-induced crystallisation (AIC) of amorphous silicon. Another project was the deposition of amorphous solar cell precursor structures at Utrecht and subsequent processing of these films into crystalline solar cells at UNSW.

CSG Solar, Botany

During 2004, the group has been closely collaborating with UNSW spin-off company CSG Solar (formerly Pacific Solar) at Bay Street, Botany. As outlined earlier, the group has taken over an operational cleanroom research facility from CSG Solar, whereby the cleanroom services (gases, exhaust, toxic gas monitoring system, etc) are provided by CSG Solar. Based on a user-pays system, CSG Solar continued to use several UNSW-owned equipment items in the cleanroom facility. The company also provided invaluable assistance with respect to maintaining an OH&S system for the cleanroom facility that meets Australian standards and legislation. Another area of collaboration was the provision of CSG thin-film material to the group for solar cell metallisation and interconnection work. Furthermore, Centre directors Stuart Wenham and Martin Green have been consultants for CSG Solar during 2004.

Ann

ual R

epor

t 2

00

4

C

olla

bora

tive

Res

earc

h

Toyota Central Research Laboratories, Nagoya, Japan and Toyota Future Project Division, Higashi-Fuji, Japan

During 2004 a proposal submitted to Toyota on high efficiency thermoelectrics, was pursued. This resulted in a visit by Mr. Nagashima, Dr. Motohiro and Dr. Takeda from Toyota in February, presentations and continued negotiations on the form of the project. The project was finally funded by Toyota Central Research Laboratories and Toyota Future Project Division, and is worth $180,000 over two years. The project started in December 2004 and will employ a Postdoctoral Fellow investigating aspects of high efficiency thermoelectric devices. An extended visit by Drs. Motohiro and Takeda in December 2004 initiated the work on this project.

UNSW Centre for Energy and Environmental Markets (CEEM)

The PV Centre is one of the founding members of a new UNSW Centre for Energy and Environmental Markets. CEEM involves collaboration between the Faculty of Engineering, the Faculty of Commerce and Economics, and the Australian Graduate School of Management. It undertakes interdisciplinary research in the design, analysis and performance of energy and environmental markets and their associated policy frameworks. It was established in 2004, largely as a response to government initiatives to reorganise infrastructure industries (such as electricity, gas, water and telecommunications) and to rely increasingly on markets in tradable environmental instruments as a form of environmental regulation.

CEEM brings together researchers in UNSW and partner organisations, contributing to UNSW’s ability to provide world-class research, advice and education. Dr Muriel Watt is involved with the research strand focussing on “Policy framework and policy instruments in energy and the environment”. Other strands are Energy market design, Derivative markets, Experimental markets, Applications of artificial intelligence and Economic valuation methodologies and their application to energy and environmental issues.

Australian Business Council for Sustainable Energy (BCSE)

PV Centre staff liaise regularly with the BCSE on PV industry development and policy issues. Dr Muriel Watt and Dr Richard Corkish regularly attend the BCSE PV Directorate meetings while in 2004 the BCSE funded a research project undertaken by the PV Centre on the use of PV for peak load reductions. A number of consultancies have also been funded and longer term collaborative arrangements are now under discussion.

130

Ann

ual R

epor

t 2

00

4

Col

labo

rati

ve R

esea

rch

131

Suntech-Power Company, China.

Research TeamProf Stuart Wenham (UNSW – Team Leader)Anita Ho (Postdoc Fellow – UNSW)Ly Mai (PhD student – UNSW)Budi (PhD student – UNSW)Dr Z. Shi (Suntech)Dr. Jingjia Ji (Suntech)A Zhu (Suntech)Li Hua (Suntech)Victor Chen (Suntech)

Project: Innovative Emitter Design and Metal Contact for Screen-printed Silicon Solar Cells

Aim:The broad aim of this work is to develop the next generation of screen-printed solar cell for implementation on the Suntech-Power production line. In particular, the fundamental limitations of the conventional screen-printed solar cell that have limited its performance for the last 30years have been identified, and innovative approaches to redesigning the emitter and front metal contact have been devised and will be developed and analysed in this work.

Background:

Screen-printed solar cell technology dominates commercial photovoltaic manufacturing, with well over 50% share of international markets. Despite the dominance of this technology, this solar cell design shown in Fig 4.128, has significance performance limitations that limit the cell efficiencies to well below those achievable in research laboratories around the world. In particular, the front surface screen-printed metallisation necessitates a heavily diffused emitter to achieve low contact resistance and also to achieve adequate lateral conductivity in the emitter since the metal lines need to be widely spaced compared to laboratory cells to avoid excessive shading losses. Such cells therefore typically have emitters with sheet resistivities in the range of 40-50 ohms per square, which inevitably give significantly degraded response to short wavelength light. To raise this sheet resistivity to above 100 ohms per square as required for near unity internal quantum efficiencies for short wavelength light, serious resistive losses are introduced, both in the emitter and the contact resistance at the metal to n-type silicon interface.

patterned metal contact3mm

phosphorus

bulk of wafer

rear metal contact

150-200

n++

p+

p-type

metal

Figure 4.128: Conventional screen-printed solar cell with heavily diffused emitter and large metal/silicon interface area for the front surface metallisation.

Ann

ual R

epor

t 2

00

4

C

olla

bora

tive

Res

earc

h

Furthermore, the conventional design of fig 4.128 has quite poor surface passivation in both the metallised and non-metallised regions. Even if good ohmic contacts could be made to more lightly doped emitters, the large metal/silicon interface area would significantly limit the voltages achievable due to the high levels of recombination in these regions and hence contribution to the device dark saturation current. These voltage limitations are not of major significance at the moment due to the limitations imposed by the substrates. However, In the future as wafer thicknesses are reduced to improve the device economics, the cells will have the potential for improved open circuit voltages, but only provided the surfaces, including under the metal, are well passivated.

The aim in this work is to develop and demonstrate an innovative emitter design for the screen-printed solar cell that overcomes the current and voltage limitations imposed by the design shown in Fig 4.128, while retaining compatibility with existing equipment and infrastructure currently used for the manufacture screen-printed solar cells.

Innovative Emitter Design:

The new emitter design is shown in Figure 4.129. The top surface is diffused to 100 ohms per square, while the heavily diffused grooves act as semiconductor fingers to carry the current to the screen printed silver fingers that run perpendicular to the grooves as shown in Figure 4.130. Also not shown in Fig 4.129 is the surface passivating dielectric that not only passivates the lightly diffused surface so as to give near unity internal quantum efficiencies for short wavelength light, but it also isolates the metal from these same regions to minimise the device dark saturation current. The metal only contacts the silicon within the heavily diffused grooves, therefore giving a low area contact while still achieving low contact resistance. Fill factors above 77% have been demonstrated with this structure on large area devices of approximately 150cm2, verifying the effectiveness of this contacting scheme.

132

30 microns

p-type silicon

45 microns

5 / n++

100 / n-type

n++

rear surface of solar cell

Figure 4.129: Cross-section of the proposed innovative emitter design using semiconductor fingers, developed to address the fundamental limitations of screen-printed metal contacts with their inability to produce fine lines and make ohmic contact to lightly diffused emitters.

Figure 4.130: Screen-printed fingers running perpendicular to the heavily diffused grooves where electrical contact is made. A dielectric/AR coating passivates the top surface and isolates the metal from the lightly diffused top surface.

Ann

ual R

epor

t 2

00

4

Col

labo

rati

ve R

esea

rch

133

The grooves are typically spaced less than a millimetre apart so as to minimise resistive losses within the lightly diffused emitter, while the screen-printed metal lines can be spaced significantly further apart than in normal screen-printed cells of Fig 4.128 due to the comparatively excellent lateral conductivity of the emitter achieved by the very heavy doping within the grooves. This concept of semiconductor fingers does not appear to have ever been used in commercial solar cells, and has considerable appeal as it facilitates good conductivity within the emitter, but without the normal trade-off found in screen printed cells where such regions of good emitter conduction are located at the top surface and therefore degrade the cell spectral response and current generating capability due to the corresponding extremely short minority carrier diffusion lengths in such regions.

Fig 4.131 shows a photo of a screen-printed metal line crossing one of the heavily phosphorus diffused laser grooves. The silicon is only exposed within the grooves, with the screen-printed metal having been shown to make excellent ohmic contact to the heavily phosphorus diffused silicon in these regions. Both thick oxides and silicon nitride layers, when used with appropriate pastes, appear to provide adequate protection to the lightly diffused surface regions, preventing the screen-printed metal from contacting the silicon.

A simplification of the proposed emitter design is to apply a phosphorus doped passivating dielectric after lightly diffusing the top surface. The laser scribing conditions for groove formation are then modified so as to melt the silicon rather than ablate it, thereby allowing large amounts of phosphorus to penetrate into the molten silicon, producing heavily doped channels rather then grooves, as shown in Fig 4.132. This avoids the need for etching the grooves and subsequently diffusing the groove walls. The performance of these devices however does not currently match that of the devices produced using the emitter design of Fig 4.129.

Fig. 4.131: Photo of a screen-printed metal line making excellent electrical contact to one of the heavily phosphorus diffused laser grooves running perpendicular to the metal line. The dark regions are textured and lightly diffused to 100 Ω/sq, and well passivated by a thick thermally grown oxide

Ann

ual R

epor

t 2

00

4

C

olla

bora

tive

Res

earc

h

Financial Support

Suntech-Power has been particularly generous, not only funding all the collaborative research conducted in China, but also fully funding two research positions at UNSW for a period of 12 months commencing in March 2004. The latter alone is equivalent to a cash contribution of approximately $100k, while the inkind contribution through the former has been valued in excess of $200k. In addition, Suntech funded all the living, travel and accommodation expenses for the 12 staff and students who worked at Suntech on the collaborative research. Also during 2004, Suntech-Power awarded a Postdoctoral Fellowship to Dr Anita Ho which commenced in January 2005.

Staff Exchange

Professor Stuart Wenham made 4 trips to Suntech-Power during 2004 in relation to the described collaborative research projects. These trips ranged from periods of 1 week to 2 weeks each, with the combined period being 6 weeks. Professor Martin Green also made one visit to Suntech-Power several days. In addition, one PhD student and 9 undergraduate students were involved in the research work at Suntech-Power for varying lengths of time during 2004. Staff from Suntech-Power also regularly visit UNSW for training, information exchange and to participate in the collaborative Research.

134

Dielectric

Dopant source100 / n-type

n++

p-type silicon

rear surface of solar cell

Figure 4.132: Cross-section of the simplified version of the proposed innovative emitter design using semiconductor fingers, developed to address the fundamental limitations of screen-printed metal contacts with their inability to produce fine lines and make ohmic contact to lightly diffused emitters. In this design, the heavily diffused grooves of Figure 4.129 are replaced by heavily doped channels produced by laser doping, that run perpendicularly to the metal fingers.

Ann

ual R

epor

t 2

00

4

Col

labo

rati

ve R

esea

rch

135

Ann

ual R

epor

t 2

00

4

Edu

cati

on

5.1 Summary

The ARC Centre for Advanced Silicon Photovoltaics and Photonics incorporates the activities of the former ARC Key Centre for Photovoltaic Engineering. The former Key Centre started in 1999, after the award of special funding from the Australian Government to promote teaching and industry collaborative research in the area of photovoltaics. One of the primary initiatives of this Key Centre was the establishment of the world’s first undergraduate Engineering degree in Photovoltaics and Solar Energy. This was one of only eight such Key Centres awarded Australia-wide across all disciplines, demonstrating the importance the government placed on this new and exciting field of photovoltaics. In 2003 the Key Centre and its activities were incorporated into the new ARC Centre of Excellence, awarded to the same team at the University of New South Wales.

The Centre offers undergraduate, postgraduate and research programs encompassing a range of aspects relating to the photovoltaic and renewable energy industries. These programs have been developed in consultation with representatives from industry to ensure graduates are appropriately qualified to enter the field upon completion of their studies. The first group of students graduated from the inaugural Photovoltaics and Solar Energy program in April 2004, and most of those students have since gained employment in the industry, or spent 2004 commencing research programs or undertaking postgraduate studies in Commerce. The Centre has seen students graduate from all of its educational programs except for the newest undergraduate program in Renewable Energy Engineering which commenced in 2003, and the recently introduced Master of Philosophy.

Throughout 2004 the Centre has continued to participate in promotional activities which, in addition to various scholarship programs, help to attract very bright students to the Centre’s educational programs. Students are also attracted by the variety of projects they become involved with throughout their studies, and in 2004 the Centre’s undergraduate program in Photovoltaics and Solar Energy was recognised by being awarded a Green Globe Award from the Department of Energy, Utilities and Sustainability. Receipt of this award is testament to the hard work of Centre academic and general staff who are involved with the Centre’s educational activities and continue to refine and develop courses taught as part of the Centre’s undergraduate and postgraduate programs.

136

5. EDUCATION

Ann

ual R

epor

t 2

00

4

Edu

cati

on

137

5.2 Undergraduate Educational Programs

The development of the Centre’s undergraduate programs was driven by predictions of massive job creation and anticipated growth in the photovoltaic and renewable energy industries of over 30% per annum. This growth rate has been achieved and it is expected that there will be a growing lack of appropriately qualified engineers to fill the positions created by this continued growth in the future. As a result, the development and implementation of the world’s first degree in Photovoltaics and Solar Energy in 2000 was soon followed by the introduction of a broader undergraduate engineering degree in Renewable Energy Engineering in 2003.

At the end of 2004, there were a total of 125 students enrolled in undergraduate programs offered by the Centre. Approximately 30% of these students were female which is unusually high for engineering programs which typically have a participation rate of 10-15% by women. Students who enrol in these programs are generally concerned and enthusiastic about making a personal contribution to society and the environment. They are also attracted by the opportunity of working with cutting edge technology.

5.2.1 Bachelor of Engineering in Photovoltaics and Solar Energy

This four year program focuses on photovoltaic devices, and includes training in technology development, manufacturing, quality control, reliability and life cycle applications, system design, maintenance and fault diagnosis, marketing, policy development and the use of other renewable energy technologies. Considerable emphasis is placed on gaining hands-on experience of working with photovoltaic devices, modules and systems particularly by involving students in a project in their second year, a unique feature when compared to other Australian engineering programs. This program can also be combined with a Bachelor of Science or a Bachelor of Arts, and more information, including program outlines, can be found at www.handbook.unsw.edu.au.

Another unique feature of this program is the “strand”, a second area of specialisation which complements a students study of photovoltaics. Students can take strands covering a variety of areas including computing, electronics, mathematics, physics, mechanical engineering, civil engineering and architecture. The aim of the strand is to provide students with broader engineering backgrounds important for the cross-disciplinary nature of photovoltaic applications.

The first student graduated from this degree program in October 2003 and another 22 students graduated during 2004. One of these students was awarded a University Medal for outstanding academic achievement throughout his studies. A few of these students are undertaking research with the Centre or at the Australian National University, while others are completing a Master of Commerce degree program, or have gained employment in the industry. Graduates from this program have even gained employment in the booming photovoltaic market in Germany. It is expected that a further 11 students will graduate in April 2005 after completing their studies in December 2004.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

5.2.2 Bachelor of Engineering in Renewable Energy Engineering

The Centre’s Renewable Energy Engineering program commenced in 2003, and is a much broader program compared to the Photovoltaics and Solar Energy degree. In addition to photovoltaic devices, students study solar architectural technologies, wind energy, biomass, solar thermal, and renewable energy policy as part of their core curriculum. Ten students commenced this program in 2003 and by the end of 2004 there were a total of 23 students enrolled. A detailed program outline is available from www.handbook.unsw.edu.au.

In 2004 the Centre obtained approval to offer this program in combination with a Bachelor of Science. This new combined degree will appeal to students with broader interests, and enables students to study areas from Science which complement their study of renewable energy engineering.

5.2.3 New Combined Bachelor of Engineering/Bachelor of Commerce Undergraduate program

During 2004 the Faculty of Engineering in association with the Faculty of Commerce, developed a 5.5 year combined Bachelor of Engineering/Bachelor of Commerce program. Staff from the Centre were involved with this process and as of 2005 students with an appropriate academic background will be able to combine either of the Centre’s two undergraduate programs with a Bachelor of Commerce.

The broad aims of this new combined degree are to provide students with a strong science background, in-depth competency in one engineering discipline, and further skills in technical management, accounting, commerce, economics and marketing. Engineers who graduate with the Bachelor of Commerce will be better able to obtain venture capital for commercialisation of products, to establish new companies, to manage projects, to strategically manage manufacturing and marketing, to accurately track and report financial status, to globalise the business internationally and to accurately predict economic trends.

It is expected that this new program combination will be popular among students, particularly those who enjoy engineering and would like to progress to entrepreneurial and managerial roles.

138

Ann

ual R

epor

t 2

00

4

Edu

cati

on

139

5.3 Postgraduate Educational Programs

The Centre for Photovoltaic Engineering offers one postgraduate coursework program and three research programs. These degrees are intended to provide students with an exceptional basis in advanced concepts and research in the photovoltaic area.

5.3.1 Master of Engineering Science in Photovoltaics and Solar Energy

The Master of Engineering Science is a coursework program developed to build on the previous engineering education of engineers from other engineering disciplines who are currently being attracted to the photovoltaics and renewable energy industries. Students study courses chosen from the areas of photovoltaic devices, photovoltaic systems and applications, and renewable energy technologies. During 2004, 6 new students commenced this program which is the highest number of first year enrolments since its commencement in 2002. 4 students also graduated from the program during the year, one of whom graduated in a ceremony in Beijing.

5.3.2 Research Programs – Doctor of Philosophy (PhD), Master of Engineering (ME) The Centre is internationally recognised for it’s research activities and research topics are available for students covering the entire photovoltaic sector, but with greatest emphasis on device theory, device and module design, balance of system components, and photovoltaic systems and applications. Research students are involved across all the Centre’s research groups and play an important role in the Centre’s activities. More detailed information about these programs is available at www.handbook.unsw.edu.au.

At the end of 2004 there were 21 research students enrolled, six of whom commenced a PhD program during the year. During 2004 two of the Centre’s students graduated from a PhD, and one graduated from a Master of Engineering. 5.3.3 New Master of Philosophy Program

In 2004 the Faculty of Engineering introduced a new research degree program called a Master of Philosophy. This program is a 3-semester research degree with a significant coursework component. As part of the program students complete 18 units of credit of coursework and a 54 unit of credit research project. It is possible for students to complete this degree in one calendar year if they complete their research during the summer period. A unique feature of the program is an oral defence requirement as part of the examination process. This new program will be included as part of the Centre’s educational offerings.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

5.4 Course Development

Course development work was undertaken by several academic staff members throughout 2004. While most course development was undertaken within the first four years of the Photovoltaics and Solar Energy program, courses still needed to be developed for the relatively new Renewable Energy Engineering program, and to increase the selection of professional electives for both undergraduate and postgraduate coursework students.

5.4.1 Solar Architectural Technologies

The course Solar Architectural Technologies forms part of the core curriculum for Renewable Energy Engineering students, and is taught in cooperation with the Faculty of the Built Environment. Throughout the course the use and effectiveness of electricity generation from renewable energy sources in buildings and domestic housing are considered in relation to energy efficient housing and effective solar utilisation. The course is coordinated by an academic from the Centre and in addition to attending a class taught by the Centre’s academic, students have the opportunity to take classes with Architecture students on various environmental, structural and building related aspects of housing design. Mr Marlon Kobacker, the first graduate of the undergraduate photovoltaics program, was the Course Coordinator for this course, and he was able to draw on relevant industry experience and postgraduate studies in Architecture for his teaching.

5.4.2 Sustainable Energy in Developing Countries

Mr Geoff Stapleton developed and taught a course on Sustainable Energy in Developing Countries in session 2 2004 as a professional elective for undergraduate and postgraduate coursework students. Geoff has a large amount of industry experience associated with the introduction of sustainable energy solutions in developing countries and students appreciated Geoff’s extensive knowledge and experience he brought to the course. The course covers many of the technical and non-technical issues relating to the introduction of photovoltaics and renewable energy systems and technology in developing countries. It is closely aligned with current national or international programs in developing countries such as the IEA PVPS Task IX. Recommended Practice Guides developed by industry expert groups were used throughout the course and covered the areas of financing and investment mechanisms, capacity building, implementation models and quality assurance. According to students one of the best features of this course was Geoff's use of real world examples and case studies.

140

Installation of photovoltaics on rooftops in developing countries

Ann

ual R

epor

t 2

00

4

Edu

cati

on

141

Another feature of the course appreciated by students was the opportunity to attend a briefing on PV in the International Energy Agency’s PV Power Systems Program and developing countries. This briefing was held at Sydney University and was attended by international experts who made presentations on their country’s expertise and experiences establishing photovoltaic systems in developing countries with reference to the IEA PVPS Task IX.

5.4.3 Energy Efficiency

An elective course for postgraduate and undergraduate students on Energy Efficiency was developed and taught in 2004 by Dr Alistair Sproul. If Australia and the world are to reduce emissions of greenhouse gases, both renewable energy and more efficient use of that energy will be required. Energy efficiency is the cheapest, fastest, safest and simplest way to reduce emissions. This course covers the various methodologies, technologies and policies that can be used to reduce energy use, while still producing what that energy is needed for – heat, light and movement.

Topics covered in the course include current and predicted energy use and associated Green-house gas (GHG) emissions; residential and commercial passive solar design; energy management programs; building and management systems; heating, ventilation and air conditioning; and consumer products and office equipment. The impact of transport is also covered, together with opportunities to reduce transport energy requirements through more efficient engines, public transport, and urban design. Industrial systems examined include heat recovery; cogeneration; compressed air and steam distribution; and motor systems, pumps and fans. Efficient use of water, and increased efficiency of water supply can also significantly reduce energy use. Various government policy measures at the local, state, commonwealth and international level are covered in terms of their effectiveness and relevance in Australia. Finally, barriers to improved energy efficiency such as up-front cost, lack of information, and the low cost of energy in Australia are examined. Assessments focus on student’s personal experience of energy use and real examples of energy practices at UNSW.

Students were very satisfied with this course and felt it was comprehensive, relevant, interesting and involved real life applications. They also appreciated Dr Sproul’s obvious enthusiasm and detailed knowledge in this area.

5.5 Student Projects

In the second year of the Photovoltaics and Solar Energy program students have the opportunity of undertaking

a year long project in the photovoltaic or renewable energy areas. The main emphasis of the course

is hands-on project management and project engineering. The course has a lecture component covering project engineering skills and practice, and each project has a research component, a planning component, a hands-on component and a reporting component. This course helps to prepare students for their fourth year

Ann

ual R

epor

t 2

00

4

Edu

cati

on

thesis, which is generally undertaken by all students enrolled in both the Photovoltaics and Solar Energy and Renewable Energy Engineering undergraduate programs.

5.5.1 Biodiesel Trailer

The Biodiesel Project aims to construct a small-scale, portable chemical factory to convert used vegetable oil, from UNSW’s food outlets, into diesel fuel for UNSW’s fleet of diesel cars. The project has now been running for three years and is at a stage where completion is in sight. Four second-year students who undertook the Biodiesel Project in 2004 designed two key subsystems, the fluid handling and mixing subsystem, and the electrical system. One of the key challenges was handling potentially explosive vapours and flammable liquids, and this necessitated a completely pneumatic fluid mixing and pumping system instead of the more common electrical pumping and mixing systems. The students also completed several construction projects, including mounting of the solar hot water and solar photovoltaic panels on the trailer and installing the reaction vessels and the solar hot water into the trailer. The majority of the parts and fittings required to complete the trailer's systems have been obtained and are ready for installation in 2005, when a new group of second-year undergraduate students will aim to complete the trailer and make the first batches of biodiesel fuel. This project provides students with hands-on experience in the design and construction area of the biomass renewable energy sector.

5.5.2 Developing Countries

The Centre has been involved with projects in developing countries for the last four years and this project was again the most popular among students in 2004. The application of photovoltaics and other renewable energy technologies can make the greatest difference to people’s lives and living standards in developing countries. As part of this project students become involved with a variety of work including photovoltaic module construction, encapsulation and testing, as well as designing solar lighting systems, solar electric lanterns and solar cookers. Fundraising and project management are also an important part of this project. Throughout 2004, students travelled to Nicaragua and Nepal as a result of work completed in previous academic sessions as part of this project, and another group of students commenced work on a project in Vanuatu.

In January 2004 a group of eleven students and two academic staff members travelled to Nicaragua to work with a Nicaraguan non-governmental organisation, Grupo Fenix, who conduct practical research into appropriate energy technologies and train local people, including land mine survivors, to build, maintain and operate solar ovens, PV lighting systems and battery rechargers. Grupo Fenix is associated with the Universidad Nacional de Ingeneria in Managua and its work impacts directly on rural people. One of its outstanding successes has been the community ownership and support of solar cookers in Madriz province. Throughout the previous academic year, the students had worked on technical projects relevant to the target community. Through 2003, one group of students field tested solar lanterns, while others built and tested solar crop dryers and solar cookers. Techniques for assembling PV modules as a cottage industry were also investigated, as were the construction, checking and maintenance of small stand-alone

142

Ann

ual R

epor

t 2

00

4

Edu

cati

on

143

PV systems. Each group followed up their particular work in Nicaragua with technology transfer, training of, and learning from local people as well as the construction, installation and maintenance of solar thermal PV systems.

Similarly, in February 2004, a group of 11 fourth year students travelled to central Nepal, the culmination of work commenced in 2001 as part of their second year project. Their journey to Nepal was originally to have taken place in September 2001, but was postponed due to security fears. Funds raised by the students for equipment, travel and other costs were retained and disbursed as part of their trip in 2004, where they installed a PV lighting system in a medical clinic in the village of Shanke Bazaar. This health centre, with a single doctor, serves a widespread community of around 30, 000 people. The system comprised a PV module, battery, a controller, mounting structures and fluorescent lights. Prior to the installation of this system by the Centre’s students the only lighting available to the Health Post was provided inadequately by torches or kerosene lights. During their stay in Nepal, students and supervisors stayed in the village to gain an appreciation of local lifestyle and customs.

Following on from this trip to Nepal, the 2004 second year developing countries project group began the year investigating further renewable energy options for the clinic and village of Shanke Bazaar. The doctor providing medical services at the clinic had requested a vaccine refrigerator and the students intended to provide a PV- powered, World Health Organisation approved chest refrigerator system as well as investigating the provision of improved task lighting, exterior lighting, improved battery and controller containment and

Students and staff in Nicaragua

Students and staff in Nepal

Ann

ual R

epor

t 2

00

4

Edu

cati

on

the use of micro-hydro using a nearby river for the purification of water supply. However, security concerns arising from political instability in Nepal led to a change of focus to Vanuatu, utilising as much as possible the previous work initially done with Nepal in mind. Students now intend to install a large number of small photovoltaic power systems for lighting, vaccine refrigeration and two-way radio in remote clinics and dispensaries on various islands in association with Rotary International, the Vanuatu Ministry of Health and BP Solar. The group of students are aiming to travel to Vanuatu mid-year 2005 to participate in system installations, testing, maintenance, usage monitoring and user education.

Participation in this project provides an opportunity for students to directly experience real-world applications of renewable energy technologies and to see them satisfying energy needs in the target community that would otherwise go unmet. Students find the experience extremely satisfying and rewarding. After staying with local families in their adobe huts with dirt floors, no running water, and pit toilets, students appreciate and understand the significance of the clean electricity that they have helped to install.

5.5.3 Micro-hydro Project

Micro-hydro is an excellent low cost source of off grid power, if you are fortunate to have access to a suitable water resource. This project commenced with five students who had no previous experience with micro hydro or hydraulics but were keen to take on the engineering challenge of designing and setting up a micro hydro system.

The project vision is to install a commercial micro hydro system, with additional measurement instrumentation, on campus at UNSW. In the absence of a natural elevated water resource a large storage tank on top of our five storey building provides an equivalent source. This phase of the project commenced with the investigation of suitable sites around the building taking into consideration the various issues to be addressed which, in addition to basic hydro suitability, included aesthetic impact, safety, security and accessibility for visitors. Several sites were short listed and concept drawings and documentation developed for inclusion in applications for site approval.

This phase concluded with a fully operational “mobile” micro hydro system which has the benefit of being operated either outside or inside the building in any weather. The system comprises of a 300W Pelton wheel micro hydro generator, electronic controller, storage batteries, electrical load and buffer tank on a heavy duty trolley. Additional valves and an electrically driven pump enables configuration of the system for operation of the hydro generator without the location constraint of connection to a falling water resource. In this mode the system can operate in a class room or be used in a traveling demonstration. When connected to a falling water resource such as from the storage tank on the roof, the same pump is suitably sized to prevent water wastage by recirculating the water back to the roof top storage tank.

This project gave students an opportunity to explore and gain experience in many areas including site selection, liaison with tradespersons and building managers, water flow theory, micro hydro generator selection, pipe sizing, materials selection, electrical cable selection, power circuit construction, measurement instrumentation, occupational health and safety issues, the specification of equipment, components and their procurement, installation, mounting and testing. The students were invited by Dr Jeff Cotter to demonstrate their rig to his SOLA2053 Sustainable and Renewable Energy Technologies class.

144

Ann

ual R

epor

t 2

00

4

Edu

cati

on

145

5.5.4 Thin-Film Solar Processing Equipment

Thin-film solar cells have the potential to revolutionise not only the solar cell industry, but also electricity generation for this planet in the future. UNSW has had particular success in this area, developing a thin-film crystalline silicon on glass technology which has become the focus of a $100 million commercialisation program. UNSW continues to develop new thin-film solar cell technologies in its laboratories. This project aims to give students broad experience in establishing and maintaining such a world-class research laboratory. This is achieved through a two-stage process, firstly students develop a base level understanding of equipment and systems within the lab, which include vacuum systems, gas delivery systems, leak testing equipment and plasma processing. The second stage involves the students integrating what they have learnt by applying it to a specific project in the thin film laboratory. This year the students designed, built and fitted a lifting mechanism for a vacuum evaporator chamber, undertook leak testing work on the system to achieve a better base pressure and improved existing equipment safety and operation documentation.

5.5.5 Fourth Year Thesis

The thesis is generally carried out in the last two sessions of an engineering undergraduate student’s studies. Students undertake directed laboratory and research work on an approved subject under guidance of the Centre’s academic staff. Typically, the thesis involves the design and construction of experimental apparatus together with practical tests. Each student is required to present a seminar, submit a written report, and present a poster as part of an Open Day. It is a chance for students to demonstrate what they have learned throughout their studies with the Centre.

Students working on the micro-hydro project and a picture of the micro-hydro generator during trials.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

The standard of theses submitted by students during 2004 was excellent. Application-based photovoltaic and renewable energy topics were the most popular with students also completing policy and laboratory-based research, and industry-linked projects. Areas investigated by students included:- wind energy; solar powered water pumping; characterisation of silicon-based nanostructures; the installation of PV on the UNSW Quadrangle Building and in developing countries such as Sri Lanka and Western Sahara refugee camps; sustainability in project development homes; PV solar thermal systems; micro-hydro; and an examination of the Australian Solar Cities program.

The two most outstanding theses will both receive the 2004 Photovoltaics Thesis Prize. These were ‘Energy Use in Project Homes, A Sustainability Study’ by Jessie Copper, and ‘Kill Counting Solar Mosquito Destroyer’ by James Tan. As part of her thesis, Jessie investigated energy efficient measures in project home designs using case studies and a software program she designed. Each measures impact on energy use, CO2 emissions, capital costs and periodic savings were analysed. She recognised the importance of economic benefits for driving changes in project homes which will lead to improved energy efficiency and sustainability. James’ thesis focused on the development of a solar powered system for attracting and killing mosquitoes which are recognised as transmitting fatal diseases particularly in developing countries. The system built by James also included a digital counter which was intended to add to the marketability of the product. These two theses will be nominated for the 2004 Wal Read Memorial Prize to be presented by the Australia New Zealand Energy Society.

146

A picture of James Tan’s Kill Counting Solar Mosquito Destroyer.

A screen shot of the Energy Consumption Program (ECP ) developed by Jessie.

Screenshot of the results page of the ECP, displaying the results of a case analysis.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

147

5.6 Education-related Prizes

5.6.1 2004 Energy and Water Green Globe Awards

The Centre’s Bachelor of Engineering in Photovoltaics and Solar Energy received the Education and Awareness Award at the 2004 Energy and Water Green Globe Awards held by the Department of Energy Utilities and Sustainability (DEUS) on the 23rd November 2004.

The Energy and Water Green Globe Awards encourage greenhouse reduction and sustainable use of resources by recognising excellence in the implementation and promotion of energy and water efficiency initiatives by businesses, individuals, government departments, and members of the renewable energy industry and retail sector. The Photovoltaics and Solar Energy program at UNSW received the Education and Awareness award for achievement in education and awareness about sustainable energy or water supply. It was received on behalf of the Centre by the Head of School, Dr Richard Corkish.

5.6.2 2004 Wal Read Memorial Prize

On the 3rd of December 2004 at the Solar 2004 Conference held in Perth, Florence Chen, a recent graduate from the Centre for Photovoltaic Engineering was awarded the 2004 Wal Read Memorial Prize. This prize was jointly won by Florence and a student from the University of Adelaide.

The Wal Read Memorial Prize is awarded by the Australia and New Zealand Energy Society, ANZSES, for a thesis completed in a student’s

Dr Corkish accepting the Green Globe Award from the Hon Frank Sartor MP

Florence Chen

Ann

ual R

epor

t 2

00

4

Edu

cati

on

final year of their undergraduate studies. The thesis must cover some aspect of solar energy which may include other forms of renewable energy such as wind or biomass, and can either cover technical or non-technical aspects of the application of solar energy.

Florence Chen graduated from the Centre’s Bachelor of Engineering in Photovoltaics and Solar Energy with Honours Class 1 in October 2004, and the title of her final year thesis was ‘Evaluation of Boron Solid Source Diffusion for High Efficiency Silicon Solar Cells’, which was supervised by Dr Jeff Cotter. The aim of the thesis was to investigate the potential of using boron nitride as a replacement for liquid-source diffusion currently

used for p-type diffusions in the Centre’s high efficiency buried contact solar cells. Her findings were published in a paper presented at the 2003 ANZSES Conference, and her thesis was also awarded the 2003 Photovoltiacs Thesis Prize. Florence is currently studying a PhD in Photovoltaics with the Buried Contact Group at the Centre.

5.7 Scholarships

The undergraduate programs at the Centre attract very bright students from across Australia. Even though the UAI cut off for the programs has been approximately 80.00 over the last few years, the majority of students who enrol in the programs have a UAI above 90, and almost a quarter have UAIs above 97. The Co-Op Scholarship Program and the Faculty of Engineering’s Rural Scholarship Program help to attract these bright students, while the Taste of Research Summer Scholarship Program provides high-achieving students with experience in the Centre’s laboratories and encourages them to pursue research careers.

5.7.1 Co-Op Scholarship Program

The Co-Op Program is an industry-linked scholarship program where students obtain a year of work experience with an industry sponsor as part of their undergraduate studies. In addition to their outstanding academic achievements, students are selected based on their involvement in school and community activities, their demonstrated leadership skills and their ability to communicate. The scholarship is valued at approximately $13, 400 per annum.

Two organisations, BP Solar and CSG Solar, sponsored scholarships for the undergraduate Photovoltaics and Solar Energy program in 2004, while 4th year co-op scholars spent 2004 working for 6 months at each of these organisations. Participation in this program enables students to apply the knowledge they have gained during their studies, and is also beneficial to the industry sponsor.

148

Ann

ual R

epor

t 2

00

4

Edu

cati

on

149

5.7.2 Rural Scholarship Program

The Faculty of Engineering established the Rural Scholarship Program in 2001 to encourage high-achieving students living in rural and isolated areas to study engineering. The scholarships are valued at approximately $8,500 per annum for four years of full-time study which eases the financial hardship of relocating to and living in Sydney. The Centre offered one of these scholarships for the commencement of the 2004 year and there was strong competition for this place with more than 40 applications listing Photovoltaics and Solar Energy as one of their preferences.

5.7.3 Taste of Research Summer Scholarship Program

The Taste of Research Summer Scholarship Program is primarily for high achieving 3rd year students, and in exceptional cases 2nd year students may be considered. As part of the program, engineering schools offer 10 week projects for students to complete during their summer break. These projects provide students with paid experience working as part of a research team, often in world class laboratories such as the Centre’s. From the 1st December 2003 to the 19th February 2004, three students worked with the Buried Contact Group on a variety of projects including the creation of a GCode Simulator for a CNC Laser Tool, research on photoelectrochemistry as a texturing technique

for silicon solar cells, and investigating the effect speed has on the laser texturing process for solar cells. Participation in these projects helps students further develop their technical skills as well as their written and oral communication as students have to write a report and present a poster on the outcomes of their research.

5.8 PVSOC

One characteristic of the Centre which has aided the success of the educational programs is the friendly atmosphere that is engendered by being a small school with highly motivated academic, and general staff. Students appreciate being able to form friendships and support networks with fellow students, as well as feeling comfortable and familiar with academic and administrative staff. PVSOC is a social committee established by the students which fosters this atmosphere with organised social events and activities to encourage student interaction. In 2004 the committee organised barbeques, a day playing paintball, pubcrawls in the city and inner west, a karaoke night , a soccer match, student participation in Solar House Day (organised by ANZSES), and finally the PVSOC Annual Dinner. Almost 120 students and staff attended this dinner where final year students were farewelled, staff thanked, and the 2005 PVSOC committee was elected. This dinner epitomised the friendly atmosphere of the Centre.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

5.9 Educational Promotional Activities

The Centre regularly participates in promotional activities organised by the Faculty of Engineering and Student Recruitment at UNSW. These events are important for increasing awareness and interest in the Centre’s educational programs.

5.9.1 UNSW Information Day

Local undergraduate students must apply for admission to UNSW programs through UAC, and the 5th January 2004 was the last day students could change their preferences for university degrees. Therefore the university hosted an information day on the 5th January to assist students obtain information to finalise their preferences. School administrative and academic staff attended this event and talked to many prospective students who were unsure of their career direction. Students received information packs and had the opportunity to ask questions of staff.

5.9.2 UNSW Courses and Careers Day

UNSW Courses and Careers Day is the annual information day for prospective students to obtain information about programs and student life at the university. As part of this day academic and administrative staff from the Centre attended information desks in the Roundhouse to provide advice, and information packs to prospective students. During the day Dr Alistiar Sproul also presented two lectures about the Centre’s programs as part of the lecture series organised centrally by the university.

5.9.3 Faculty of Engineering, “Ten Wonders of the UNSW Engineering World.”

Associated with Courses and Careers Day the Faculty of Engineering organised a special display of the “Ten Wonders of the UNSW Engineering World”. This exhibition was inspired by the ABC series “Seven Wonders of the Modern Industrial World”. The impact of engineering to health, lifestyle and prosperity was emphasised through displays and interactive activities. To launch the exhibition industry representatives were invited to a special preview cocktail function. The Centre participated in the event and showcased the latest silicon solar cell technology using a microscope, and demonstrated a virtual photovoltaic production line computer program used as part of the Centre’s teaching activities.

150

Ann

ual R

epor

t 2

00

4

Edu

cati

on

151

5.9.4 Faculty of Engineering Information Day for High School Students

In October each year, the Faculty of Engineering organises an information day to give high school students an opportunity to learn about engineering and the programs offered at UNSW. As part of the day students visit three engineering schools of their choice and engage in interactive activities aimed at demonstrating the relevant engineering area. Approximately 40 students attended the Centre as part of this day and they had the choice of competing in a Virtual World Solar Car Challenge in the Centre’s computer laboratories, or to learn about solar powered water pumping in the schools teaching laboratory. At the conclusion of the activities each student was issued with an information pack containing a promotional CDROM and brochures and pamphlets providing information on the Centre’s educational programs.

5.9.5 Honeywell Engineering Summer School

The Honeywell Engineering Summer School is an event held in December and is conducted by Engineers Australia. As part of the summer school high school students from across NSW and the ACT about to enter their final year take part in a week of activities which involves industry visits and lectures / demonstrations at a number of universities. In 2004 approximately 30 students visited the Centre where they constructed a capacitor using PET bottles, salt water and aluminium foil under the supervision of Mr Rob Largent. At the conclusion of Rob’s demonstration of charging and discharging the student-built capacitors, the students were issued with information packs on the Centre’s educational programs.

5.9.6 Postgraduate Information Sessions

UNSW Postgraduate Expo is held during October and is aimed at providing information about the range of postgraduate coursework and research opportunities at UNSW. As part of this event, the Faculty of Engineering organised information sessions at each School for 3rd and 4th year students to learn more about postgraduate coursework and research opportunities as well as scholarship information. The Associate Dean (Academic) of the Faculty, A/Prof Hesketh talked to Centre students about the outcomes of postgraduate studies, graduate attributes, differentiation of programs and the unique strengths of postgraduate programs at UNSW. Dr Jeff Cotter and research student Florence Chen were also in attendance to speak more specifically about the Centre’s research activities.

5.9.7 Mudgee TAFE Field Day

Staff and students from the Centre participated in a Field Day on the 16th and 17th July 2004 at Mudgee TAFE to promote sustainable energy, and educational programs on photovoltaics. Approximately 25, 000 people attended the

event and our staff in the Sustainable Energy Tent were busy throughout the weekend. In addition to distributing information about the Centre’s educational activities, the UNSW solar car was also displayed. This event obtained good publicity by being featured in the ABC Landline and on the local news.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

5.10 Educational Collaboration

A range of collaborations have been established between the Centre and other educational institutions and organisations. These collaborations involve the development and implementation of educational programs and courses, the provision of support for student projects and theses, and the exchange of students and staff.

5.10.2 APACE

Students involved with the 2004 Developing Countries project were fortunate enough to meet Paul and Donella Bryce from APACE, an organisation with considerable experience in hydro electricity projects in developing countries. Paul and Donella shared their experiences working with micro-hydro systems in the Solomon Islands with these project students.

5.10.3 Australia – Western Sahara Association

During 2004 the Centre established a collaborative project with the Australia-Western Sahara Association aimed at assisting long-term Saharawi refugees exiled in four camps near Tindouf, at the remote Western end of Algeria. A final year student designed photovoltaic power systems for the hospital and for the Women’s School at the Smara camp. The hospital is currently relying on a tiny photovoltaic power system and inconsistent and unreliable generator supply, and the new design allows for efficient lighting and ceiling fans throughout. The textiles and information technology school also has an unreliable and insufficient power supply since a donated generator has fallen into disrepair. Power is required for lighting, a ceiling fan, computers, a clothes iron, TV and video player. In each case, a rooftop photovoltaic array, charge controller, battery and inverter have been specified to meet the loads reliably, given the sunshine intensity at the site and the extreme temperatures.

5.10.4 BP Solar, Rotary International, and the Vanuatu Ministry of Health

The Centre commenced discussions with BP Solar, Rotary International and the Vanuatu Ministry of Health in the second half of 2004 regarding a potential collaboration on a developing countries project in Vanuatu in 2005. BP Solar also donated a charge controller for the developing countries project for use in 2004.

152

Pictures of the refugee camp in Smara, and a photovoltaic installation.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

153

5.10.5 Educational Experience

Educational Experience is a company involved with the distribution of educational products and in 2004 this organisation donated a lego eLAB Renewable Energy Set to the Centre to use for promotional activities. Using the kit, a windmill was constructed to demonstrate wind energy, and this was displayed as part of Courses and Careers Day 2004.

5.10.16 Universidad Nacional de Ingeneria

The Centre’s projects in developing countries require collaboration with aid organisations or else local institutions to facilitate the involvement of students from UNSW. In Managua, Nicaragua, an ongoing teaching and training collaboration has been established with Universidad Nacional de Ingeneria to facilitate the development and use of renewable energy technologies for the benefit of those living in remote villages. Undergraduate students from UNSW are able to make valuable contributions through the design, installation and testing of such systems while simultaneously receiving excellent education and training experiences. Staff from the Universidad Nacional de Ingeneria, in addition to collaborating in these activities, are able to provide the essential knowledge and experience of local issues, customs and culture.

5.10.6 Grupo Fenix Grupo Fenix is a non-profit organisation which collects and uses reject solar cells from US manufacturers to make into solar panels. The Centre’s developing countries project aims to work with organisations that employ local people and have a long term commitment in the target country, and one of the unique things about Grupo Fenix is their focus on creating jobs, especially for landmine victims. This organisation is based in Nicaragua and staff and students involved with the developing countries project in Nicaragua worked in association with this group.

5.10.7 Himalayan Light Foundation (HLF)

The Centre first became involved with the aid organisation Himalayan Light Foundation in October 2000, to coordinate the involvement of students in a project for the design, implementation and testing of photovoltaic systems in remote villages in Nepal. Thirteen students were scheduled to visit Nepal in September 2001. However, due to the terrorist attacks in the US, the trip was postponed. Towards the end of 2003 preparations were made for the students to travel to Nepal at the beginning of 2004 to complete the work started two years previously. Most of the original team were part of the 11 member group which travelled to Nepal in February 2004 under the supervision of Dr Alistair Sproul. In conjunction with HLF the UNSW team spent 6 days installing a solar photovoltaic lighting system on a Health Post in Shankhe, Nepal. Prior to this, the only lighting available to the Health Post was torches or kerosene lights that were quite inadequate. Future visits by UNSW students are intended to install vaccine refrigeration and to explore the possibility of generating electricity using micro-hydro turbines.

Ann

ual R

epor

t 2

00

4

Edu

cati

on

5.10.8 Kathmandu University

In conjunction with undergraduate projects offered by the Centre in Nepal, a relationship has been established with Kathmandu University. This provides students and staff of the Centre with excellent opportunities to be involved with projects that are relevant to situations in developing countries, such as micro-hydro projects, solar home systems for health posts and community buildings, and vaccine refrigeration. The collaboration is enhanced by visits to Kathmandu University by Centre staff, and reciprocal visits to UNSW by staff from Kathmandu University. There also exists a possibility for students enrolled at either organisation to participate in an exchange program as part of their studies.

5.10.9 Insulation Solutions

Insulation Solutions manufactures and sells insulation products that assist in the reduction of energy consumption. This organisation kindly donated rolls of “Sisalation” for use in the developing countries project. The first application will be in a workshop in a rural solar cooker workshop in rural Nicaragua in early 2005.

5.10.10 Murdoch University Western Australia

Scientia Professor Wenham was awarded an adjunct appointment at Murdoch University to help the university implement a new undergraduate degree in Renewable Energy Engineering which commenced in 2001. Prof Wenham still holds this appointment and is also a member of the Murdoch University Renewable Energy Engineering Board. This collaboration is important for the development and exchange of educational material for both institutions. A couple of the courses offered by the Centre were developed at Murdoch University under the direction of Dr Martina Callais.

5.10.11 PV Solar Tiles

During 2004, a fourth year student undertook a thesis project in collaboration with Mr Peter Erling from Sydney company, PV Solar Tiles Pty Ltd. The student monitored and analysed the performance of a building integrated photovoltaic system installed on an occupied Sydney house. This solar power system is not mounted on the roof, but is itself part of the roof of a new extension of the house. Care has been taken to allow air to flow behind the photovoltaic modules and be ducted into the original house when required for space heating. The student was able to conduct measurements over several months to quantify the heating benefits, thanks to the helpful cooperation of the house occupants and of the company.

5.10.12 Ropatec and FNQ Solar Power Specialists

The Centre was fortunate enough to receive a wind turbine in 2004 thanks to Ropatec and FNQ Solar Power Specialists. This turbine is intended to be installed in 2005 and be used as part of student projects and as a teaching resource in future years.

154

Ann

ual R

epor

t 2

00

4

Edu

cati

on

155

5.10.13 Solahart

Solahart is involved with the manufacture and sale of solar hot water systems and in 2004 the organisation agreed to donate a 150 litre water heater for use as part of the Centre’s educational activities.

5.10.14 Suntech Power Corporation

Suntech, a partly Australian-owned company operating in China, has been actively involved in assisting UNSW with the development, testing and evaluation of educational material in the PV area such as in the development, testing and evaluation of a software package called the Virtual Production Line. Significant numbers of UNSW undergraduate students also participate in industrial training at Suntech who find their time in China an extremely valuable educational and cultural experience. This collaboration has also resulted in Suntech staff visiting the Centre for approximately 12 months to work and gain experience in the Centre’s laboratories.

5.10.15 – ACRE

After nearly 5 years of operation, the ACRE Energy Policy Group (AEPG), run by Dr Muriel Watt, finalised its projects and was wound up when ACRE itself ceased operation in mid 2004. Final AEPG reports covered energy efficiency, renewable energy employment and PV for summer peak loads. The UNSW – ACRE collaboration facilitated both industry and government liaison, as well as educational collaboration, especially between Murdoch University and the PV Centre. These linkages remain in place and continue to provide important input to the PV Centre’s policy research and education.

5.10.16 – TAFE Colleges

Mudgee TAFE college offered its first course: “Introduction to Solar Energy” in November 2004. Dr Muriel Watt assisted the TAFE college in course development, materials supply and funding. BP Solar donated a range of different modules and the NRMA provided funding for purchase of other equipment and course materials. The course will be offered again in 2005 and other courses are planned.

Ann

ual R

epor

t 2

00

4

O

rgan

usat

iona

l Str

uctu

re

The Centre’s organisational structure and the way it relates to the Centre for Photovoltaic Engineering and the rest of the Faculty of Engineering is shown below. This is as foreshadowed in the Fig.H.1 for the Centre of Excellence grant application, reproduced as Fig. 5.33 below, apart from several minor changes described later.

Summarising the organisational structure, the Centre Director is accountable within the context of the University system to the Dean of Engineering through the Head of School of the Centre for Photovoltaic Engineering, and responsible to the Australian Research Council through the UNSW Research Office in terms of reporting against stated aims, objectives and expected outcomes.

The Centre Director is advised by an Advisory Committee consisting of representatives of end-users and end-user groups, senior UNSW management, and researchers from other national and international research groups with relevant interests. A Management Committee consisting of academic staff members affiliated with the Centre and Area Managers has responsibility for advising the Centre Director on issues relevant to the day-to-day operation of the Centre. These activities are supported by an Administrative Support team headed by Jenny Hansen and a Business and Operations team headed by Mark Silver.

156

6. ORGANISATIONAL STRUCTURE

Figure 5.33: Centre organisational structure.

Dean of Engineering

Other Schools

Advisory Committee

Research Programs

Management Committee

School AdministrationTrichelle Burns

UndergraduateEducation

Centre for Photovoltaic EngineeringHead: Dr Richard Corkish

AdministrativeSupport

Jenny Hansen

Business andOperations Support

Mark Silver

Centre Excellence inAdvanced Silicon Photovoltaics

and Photonics

Postgraduate Education- Coursework and Research

Dr. Jeff Cotter & Dr. Alistair Sproul

Director: Prof. Stuart WenhamExecutive Res: Prof Martin Green

Photovoltaicsand Solar Energy

RenewableEnergy

Engineering

Wafer-Based ProjectsMgr: A/Prof. Christiana Honsberg

Asst Mgr: Dr. Jeff Cotter

Silicon Thin FilmsManager: A/Prof. Armin Aberle

Third GenerationManager: Dr. Gavin Conibeer

Silicon PhotonicsManager: A/Prof. Jianhua Zhao

Ann

ual R

epor

t 2

00

4

O

rgan

usat

iona

l Str

uctu

re

157

The research activities of the Centre are divided into four interactive strands, with activities within each strand co-ordinated by an Area Manager (Fig. 5.33), each also appointed as a Deputy Director of the Centre. The experience within the Centre’s research team is attested by the fact that each of the four originally nominated Area Managers was previously an ARC-approved Joint or Administrative Director of an ARC Special Research Centre. Interactions with collaborating organisations are handled within the most relevant strand. Given the completion of course development for the undergraduate education program during 2002, the Centre’s educational development activities are focussed primarily on the postgraduate program required. As academics within the Centre for Photovoltaic Engineering are affiliated with the Centre of Excellence, there is a natural path for input into the on-going development of undergraduate programs. The standard financial management system established at UNSW is used for recording the financial transactions of the Centre and for preparing Centre reports. Julie Kwan is Financial Officer within the Centre Administrative Support team with responsibilities for interfacing with this system. Mark Silver is Business, Technology and Operations Manager with responsibilities for interfacing with Unisearch for the exploitation of Centre-developed technology.

The changes from the original proposal are relatively minor. Firstly, in the absence overseas of the Manager for Wafer-Based Projects, Associate Professor Christiana Honsberg, Assistant Manager, Dr. Jeff Cotter has been taking responsibility for these projects. Dr. Jeff Cotter has accordingly, therefore, been included in the Centre’s Management Committee. Secondly, as mentioned in Section 7, point H.2, the Centre for Photovoltaic Engineering has been given all the same rights, powers and status as all the other Schools within Engineering but has not yet officially received the new name of School of Photovoltaics and Renewable Energy Engineering as originally proposed. The Head of School is Dr. Richard Corkish, as shown. Thirdly, Dr. Gavin Conibeer has been appointed manager for Third Generation Photovoltaics, replacing Dr. Richard Corkish. Fourthly, Trichelle Burns has been appointed to lead the School Administration and Lisa Cahill has taken on a part time role.

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

In its initial application, the Centre proposed 18 research milestones for its 5-year program, as well as 4 teaching milestones, 4 “linkage” milestones and 5 governance milestones. Progress towards these milestones is reported below.

Performance Measures(for end of 5-year program)

Research Milestones

A.1 Efficiency improvement of standard p-type silicon cell to 26% or n-type cell to 24%.

Good progress was made during the year with an unconfirmed efficiency above 22% measured for n-type cells, close to the international record of 22.7% for such cells.

A.2 Efficiency improvement of Si cell boosted by an up- or down- converter to 28%.

This milestone is linked to A.10. An increase in spectral response in the range 1480 to 1580nm (i.e. well below the band edge of Si) for a bifacial Si cell has been demonstrated (see Fig. 4.94) representing a very small increase in current, but nonetheless is a proof of concept. A patent application has been prepared on a new up-conversion approach.

A.3 Demonstration of large-area buried contact silicon solar cell of efficiency above 21% and/or multicrystalline buried contact solar cell above 18%.

Work in 2004 focused on furthering development and understanding of key solar cell processing techniques, design limitations and characterisation methods. One of the highlights in our process engineering efforts is the new understanding of the formation and role of process induced dislocations, especially related to their impact on solar cell performance. We have linked boron diffusion induced dislocations to the poor fill factor of double sided buried contact solar cells. This poor fill factor has been an unsolved mystery in double-sided buried contact solar cells for many years. More importantly to establishing this process-to-performance link, we have identified and/or established the set of tools that can be used for further research on many different types of solar cells. Of note here is the establishment of the photoconductance lifetime technique in combination with defect etching to identify the signature of the process induced defects.

The boron diffusion induced dislocation has been identified as one of the major efficiency limitations of double-sided buried contact solar cells, which reached efficiencies as high as 18.6% on FZ(B) and 18.6% on Cz(Ga) despite this limitation.

UNSW also continued work on high efficiency buried contact solar cells on n-type silicon wafers. The interdigitated back-sided buried contact cell design improves on previous designs by placing all electrical contact on the rear surface and avoiding grid shadowing losses. One of the highlights in this project is the identification and elimination of a parasitic loss mechanism that is troublesome for all similar solar cell designs. The loss is related to overlapping diffusion regions - a common occurrence in back-side solar cell designs. We have brought this loss under control by carefully controlling the diffusion processes with special consideration to the solid state solubility limits for phosphorus and boron diffusions, including the potential formation of phosphorus and boron silicides and impurity

158

7. PROGRESS AGAINST TARGETS

segregation in silicon dioxide layers. Our efforts have led to cell efficiencies as high as 19.2% on textured FZ(P) and 16.8% on untextured, SiO2 SLAR-coated Cz(P) wafers.

A.4 Demonstration of at least three new promising technologies for use in conjunction with commercial buried contact solar cells.

1. Stencil printed solar cells

The laser-formed stencil printing technology, demonstrated for the first time during 2004, is a novel, low cost technique for forming printing stencils for silicon solar cell metallisation patterns. It offers several key advantages of the current technique of using mesh screens for metal patter printing. One of the main advantages is fine line printing, which can reduce grid shadowing losses significantly. We have demonstrated continuous print line widths of less than 50 microns and further improvement in continuity and finger profile is expected as we develop the two-level stencil technique further. A second advantage stems from the ability to use the precision of the laser cutting tool to make both the stencil and the grooves of a selectively diffused emitter, which has potential to improve both current and voltage of printed silicon solar cells. In 2004, we demonstrated solar cell efficiencies of over 15% with the stencil print technology.

2. Hybrid buried contact/screen-printed solar cells

An innovative emitter structure for screen-printed solar cells has been developed in conjunction with Suntech-Power company and is described in the section on industry collaborative research. This new design addresses the fundamental limitations of screen-printed metallisation schemes that have limited the performance of screen-printed solar cells for more than 30 years. The approach captures the performance advantages of buried-contact solar cells with a selective emitter, low top surface shading losses, and a well passivated and lightly diffused top surface, while simultaneously retaining the simplicity and manufacturability of screen-printed solar cells including the use of all the same equipment and infrastructure.

3. Other Novel cell designs

In 2004, a new project on a simplified silicon solar cell design was initiated and is presently under development.

A.5 Demonstration of optical pathlength enhancement > 30 in a silicon thin-film.

Recent collaborative work with CSG Solar has resulted in optical pathlength enhancement above 20 in the thin crystalline layers deposited onto textured glass. Further improvement is expected in the near future. In parallel, the AIT glass texturing method (AIT = Aluminium-Induced Texture) is being developed and patented at UNSW.

00

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

159

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

A.6 Demonstration of series interconnection of silicon thin-film solar cells without the use of metal interconnects and improved durability and performance for such devices.

Cell interconnection completely without the use of metal has proved difficult to achieve so far. However, good experimental progress has been achieved for a method that uses a very small amount of metal per cell. A provisional Australian patent is presently being sought for this invention. The method is an extension of an earlier cell interconnection scheme that is protected by provisional Australian patent 2004903028 (filed on 4th June 2004).

Another area of progress worth mentioning is the development of a new metallisation scheme for a single thin-film cell. The method has been developed and tested using CSG Solar material and has so far given cell efficiencies of 4%

A.7 Demonstration of 11% efficient silicon thin-film cell deposited onto glass.

Work towards this goal progresses along two paths. One is the involvement of senior Centre staff as Consultants for UNSW’s spin-off company CSG Solar Pty Ltd (formerly Pacific Solar Pty Ltd), aiming at the improvement of the company’s CSG (crystalline silicon on glass) technology from 8.2 % efficiency in 2003 to 11 % in 2007. During 2004, CSG Solar has improved the efficiency of its CSG modules to over 9 %.

The other path is the independent research performed in the Centre’s thin-film group. As outlined in the group’s research report, three novel poly-Si thin-film solar cells on glass are being developed (EVA, ALICIA, ALICE). The focus in 2004 has been on improving the voltages of these cells. Good progress has been made with all three cells, with open-circuit voltages now in the range 400-460 mV. Simple solar cells (“Mesa cells”) with efficiencies of up to 2.2 % have also been made. Research during 2005 focuses on boosting the short-circuit current to above 20 mA/cm2 and on making functioning mini-modules using the Centre-developed cell interconnection method.

A.8 Demonstration of 10% photoluminescent efficiency in silicon and above 98% in GaAs, sufficient to produce device refrigeration.

As reported last year the 10% PL in Si has already been achieved. Progress towards the 98% for GaAs has been demonstrated in 2004, with development of the texturing technique required for the back surface. Modelling in 2004 has shown that this should bring the 96% already reported with a ZnSe dome to a 97.5% external quantum efficiency – remarkably close to the 98% figure calculated to be required for PL cooling.

A.9 Efficiency improvement for a silicon based tandem cell over a single cell baseline.

Further work has built on the success with Si quantum dots (QDs) in SiO2 reported last year, with the PL evidence for enhancement of the energy of radiative emission combined with the increased intensity due to localisation in QDs.

160

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

161

This work includes initial measurement of the electrical properties, with promising, although still high, resistivities. It has been demonstrated that these can be reduced to about 103 Ω.cm with hydrogen passivation, with further information on activation mechanisms obtained (see Fig. 4.87).

Furthermore, there has been success with the fabrication of the analogous structure of Si QDs in Si3N4 , with TEM evidence of nano-crystals by both sputtering and PECVD. The advantage of the nitride matrix is that the lower barrier height should improve conductivity for a given QD density, although optical and electrical properties are still being tested.

A.10 Demonstration of photon down-conversion quantum efficiency above 50% and/or up-conversion efficiency above 10% and demonstration of performance improvement in a baseline cell by either approach.

Up-conversion external quantum efficiencies (EQEs) have been increased to 3.5% for the erbium doped phosphors from the 1% reported last year – in conjunction with further demonstration of increased spectral response for a bifacial Si cell (see Fig. 4.94 & 4.95). Nonetheless, although this is very promising, it represents only a very tiny increase in cell efficiency as yet. Further improvements in EQE and in improving the spectral range of absorption are being pursued.

A.11 Demonstration of two key elements required for hot carrier cell implementation: energy selective contacts and enhancement of radiative recombination rates relative to relaxation rates of photogenerated carriers. Further improvement on selective energy contacts, to that reported lat year, has been achieved with the demonstration of negative differential resistance (NDR) in samples containing a single layer of quantum dots sandwiched between two oxide layers (see Fig. 5.02). This indicates resonance in the tunnelling current across the two oxide barriers – this being a necessary pre-requisite for selective energy contacts. Further work on improving the quality factor of this NDR response is in progress.

For the absorbers important development in the understanding of carrier cooling has been achieved. In particular the specific mechanisms which can slow cooling by enhancing the “phonon bottleneck effect” in QW and QD superlattices are being identified. Experimental evidence of such reduced cooling in QW superlattices has been identified in the literature. (See section “Hot Carrier Absorbers”.)

A.12 Demonstration of the quantum confined Stark effect in silicon as the basis of a high-speed silicon modulator).

Devices of an appropriate structure were fabricated during 2003, but silicon layers needed to be thinner to demonstrate this effect. Progress was made in 2004 in reducing the thickness of these layers, that are now close to the desired value.

A.13 Evaluation of the optical constants of quantum confined silicon SOI devices as a function of layer thickness.

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

Further work on quantifying the evidence for QDs is continuing. This includes analysis of HRTEM and high resolution SIMS. As reported last year, accurate measurements of confined dimensions are required to convert PL data to optical constants.

A.14 Improvement of light emission efficiency of silicon diodes above levels already demonstrated, to maintain Australian leadership and as a test bed for integrated devices.

Australia retains leadership in the bulk silicon luminescence field, although efforts during the year concentrated on integrated devices.

A.15 Demonstration of high light emission efficiency in silicon LEDs integrated into microelectronic chips.

Efficiencies of 2x10-6 have been demonstrated, although further improvements are expected as devices become thinner.

A.16 Demonstration of high speed modulation of output of silicon light emitting diodes.

Modulation has been demonstrated, but not yet at high speed.

A.17 Demonstration of operational microelectronic circuits communicating at high data rates using silicon light emission and detection.

This is a target for the final year of the program. To date, we have demonstrated the targeted communication between bulk silicon devices, probably for the first time, but data rates are slow, as expected. Test structures for demonstrating this at the chip level have been designed.

A.18 Investigate the feasibility of obtaining lasing action in both bulk and quantum-confined silicon and, if feasible, demonstrate the first laser in these materials.

Several new mechanisms for demonstrating lasing in bulk silicon continue to be investigated.

Teaching Milestones

C.1 At least 50 honours students completing their degrees by 2007

The Centre of Excellence has performed particularly well in the educational areas with 25 students having already completed their degrees with honours by the end of 2004. Eighteen of these graduated during 2004 while the other seven have completed their degrees and will graduate in May 2005. Based on the performance and the numbers of the students enrolled in the Centre’s degree programs, this milestone will be comfortably achieved by the target date at the end of 2007.

162

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

163

C.2 At least 20 new postgraduate students involved by 2007

Following the enrolment of 4 new postgraduate students in 2003, a further 6 new postgraduate research students have enrolled during 2004. This takes the total to 10 for the first two years of CoE operation making the Centre well on track for achieving this milestone.

C.3 The involvement of at least 10 postdoctoral research fellows by 2007

The Centre of Excellence has already achieved this 5-year target with the involvement of a further 4 postdoctoral research fellows during 2004, in addition to the 9 that became involved in the Centre’s research activities during 2003. The new postdoctoral research fellows during 2004 were Dr. Per Widenborg, Dr. Kuo-Lung Lin, Dr. Thorsten Trupke and Dr. Didier Debuf. In addition, Dr Peter Cousins became involved in the Centre’s research activities for several months following the completion of his PhD prior to subsequently taking up a position with the company Sunpower.

C.4 The development of at least 4 courses on photovoltaics suitable for internet-based delivery with at least 1 such internet-based course offered each year

Three courses have either been developed or are under development in preparation for internet delivery. One course, based on the postgraduate course Photovoltaics SOLA9001, has been completed and successfully run with course enrolments from many countries. Two other courses are under development, one based on the course Solar Cell Technology and Manufacturing SOLA9006 while the other is based on the undergraduate course Sustainable and Renewable Energy Engineering SOLA2053. The latter will be offered electronically via Web CT for local students during session 2, 2005, to demonstrate suitability for internet-based delivery.

“Linkage” Milestones

G.1 Total Average Cash and equipment Contribution of at least $600,000 p.a. from UNSW

During 2004, as documented in the Financial Section, UNSW has made cash contributions to the Centre of approximately $1 million, well above the target allocation of $600k. In addition, with the Centre receiving “school” status, EFTSU income has been awarded to the Centre with earnings for 2004 being close to an additional $1 million.

G.2 Allocation of at least 70m2 office space and 180m2 laboratory space by UNSW

The Centre’s space needs are being constantly reassessed according to need. With greater than expected growth in numbers of PhD students, post doctoral researchers and visiting Fellows/Academics associated with the Centre of Excellence, UNSW has already provided more than 80 m2 of office space, above the 70m2 originally anticipated. Regarding Laboratory space, an additional 200m2 of space has been made available for the Centre of Excellence, primarily to accommodate the new state-of-the-art facilities at Bay St, Botany.

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

G.3 Support for at least 4 academic staff affiliated with the centre from EFTSU or related income

UNSW has again generously exceeded its targeted support during 2004 through the provision of staff funded from EFTSU generated income or related sources. The same staff supported during 2003 and listed in the 2003 Annual Report have again been supported during 2004, and will continue to be supported during 2005.

G.4 Average of at least 3 weeks per year of interchange between collaborating organisations

This target has been comfortably exceeded again during 2004 as documented in the section on collaborative research.

Governance Milestones

H.1 Establishment of Advisory and Management Committees during 2003.

As was reported in the 2003 Annual Report, this milestone has been achieved. The Advisory Committee of the ARC Centre of Excellence met in June 2004, while the Management Committee meets fortnightly. The membership of the latter includes the Centre Director (S. Wenham), the Executive Research Director (M. Green), the 4 Deputy Directors (G. Conibeer, J. Zhao, J. Cotter and A. Aberle), the Head of Characterisation (A. Sproul), the Head of School for the Centre for Photovoltaic Engineering (R. Corkish) and the Centre’s Business and Technology Manager (M. Silver). The membership of the Advisory Committee is:

• Professor Mark Wainwright, Pro-Vice Chancellor (Research), UNSW; • Mr. David Hogg, Managing Director, Pacific Solar Pty. Ltd.; • Mr. David Jordan, Manager of New Technology, BP Solar International; • Dr. Zhengrong Shi, Managing Director, Suntech Power Corporation; • Mr. Mark Fogarty, Executive Director, Sustainable Energy Development Authority of NSW; • Dr. Andres Cuevas, Senior Lecturer, ANU; • Prof. Peter Würfel, Institut fur Angewandte Physik der Universität Karlsruhe;• Prof. Albert Polman, Head, Nanophysics Dept., and Group Leader, Optoelectronic Materials, FOM-Institute for Atomic and Molecular Physics, Amsterdam;• Prof. Stuart Wenham, Director, Centre of Excellence; • Prof. Martin Green, Executive Research Director, Centre of Excellence.

H.2 Formation of the School of Photovoltaic and Renewable Energy Engineering and the appointment of a Head of School during 2003.

This milestone was achieved during 2003 as was reported in the 2003 Annual Report.

164

Ann

ual R

epor

t 2

00

4

P

rogr

ess

Aga

inst

Tar

gets

165

H.3 Establishment of the Management Structure of Figure H.1 by June, 2003, with appropriate documentation of responsibilities of all staff and managers.

The Management structure of Figure H.1 was successfully established in early 2003 as was reported in the 2003 Annual Report. During 2004 minor changes have taken place with T. Burns taking responsibility for School Administration. The roles and responsibilities for most staff are defined, in most cases, in their official Position Descriptions, held by Human Resources Department. These are routinely reviewed and revised at times of appointments and reappointments. In select cases, such as for the Head of School who is responsible to the Dean of Engineering, special documentation exists: http://www.hr.unsw.edu.au/poldoc/hos.htm. Management, staff and student responsibilities for occupational health and safety are documented by the UNSW Risk Management Unit

H.4 Consolidation of New Centre Strategic Plan in New Document by June, 2003.

Following the achievement of this milestone in 2003, the strategic plan is regularly reviewed in consultation with the Head of School and the Dean of Engineering.

H.5 Document Management and Advisory Committee Meetings at least monthly and annually, respectively.

An Advisory Committee meeting was held in June 2004, while the Management meetings are scheduled fortnightly. At the latter, reports have been written to document progress in each of the research areas.

Ann

ual R

epor

t 2

00

4

Fi

nanc

ial S

umm

ary

Centre 2004 Income

The total income for 2004 for the Australian Research Council (ARC) Centre of Excellence in Advanced Silicon Photovoltaics and Photonics 2004 was $4.41 million. This does not include funding approaching $1 million from the Host Institution based on EFTSU and related income. This income has been generated by the Centre through its educational activities by virtue of the fact that it is now equivalent to a School within the Faculty of Engineering. This income has been used for the development of new courses and teaching materials and to fund the salaries of most of the academic staff associated with the Centre.

The largest component of the Centre’s income was from the ARC totalling $2.7 million (Centre grant of $2.6 million with the remaining $135k being support for an existing Linkage grant with BP Solar).

The second largest cash component of income was Host Institution support, with UNSW contributing almost $1

million excluding EFTSU and related income. This was well above the targeted $600k, and excludes the direct funding of seven full-time or part-time academic staff who have been actively involved in the Centre’s activities.

The next largest cash component of income was $304k from the State Government through the Department of State and Regional Development (DSRD). This funding is particularly important for the purchase of new equipment and the development of facilities, and has contributed significantly to the acquisition and development of the new state-of-the-art research facilities at Botany.

The 2004 income from industry included cash contributions totalling almost $0.5 million although much larger in-kind contributions for the collaborative research were provided. The largest overall contributor in 2004, as in 2003, was Suntech-Power Company who funded two full-time research positions in the Centre’s laboratories, while simultaneously supporting the collaborative research at their own facilities. Including the funding of collaborative research costs, scholarships, royalties, consulting and rental of facilities, the major cash contributors to the Centre’s 2004 income were Pacific Solar (now CSG Solar) with a contribution of approximately $202k and BP Solar with $201k. Other cash or in-kind contributions have been made by the Max Plank Institute, FOM Institute, Eurosolare, ANSTO, Ian Potter Foundation, the Australian CRC for Renewable Energy, the Australian Business Council for Solar Energy, the Fraunhofer Institute, SunPower, ENEA, the Paul Scherrer Institute and the Australian National University. These organisations contribute to the Centre through the expertise and experience they offer in collaborative research areas and the in-kind support they provide through access to equipment, facilities and personnel not available at UNSW.

166

8. FINANCIAL SUMMARY

Fig 8.1: Centre income by source of income.

Fig 8.2: Industry contributions.

Ann

ual R

epor

t 2

00

4

Fi

nanc

ial S

umm

ary

167

Other Centre contracts, consulting work and technology transfers are conducted through Unisearch, the commercial arm of the University. These are handled on behalf of the Centre of Excellence and its staff through the Unisearch accounts and are not included in this financial report.

The Centre also earns income through the sale of educational CDs, books and computer software, with a combined income for 2004 of $21k.

Centre 2004 Expenditure of ARC Grant

The total expenditure of ARC grant monies during 2004 was $2.52 million, close to the $2.55 million awarded. By far the largest component of this expenditure was on salaries, with the employment of 20 full-time staff and a further 50 part-time or casual staff, including student scholarships. The combined expenditure on salaries and scholarships during 2004 was $1,674k. This included 10 full-time researchers: P. Widenborg, A. Wang, G. Zhang, R. Bardos, E.C. Cho, G. Conibeer, X.M. Dai, T Humphry, K. Lin and B. Richards. To support the research activities, 8 technical support staff were employed by the Centre during 2004, namely M. Silver, M. Griffin, L. Soria, T. Seary, A. Yee, J. Green, J. Yang and R. Largent. Two full-time administrative staff were also employed by the Centre during 2004, J. Hansen and J. Kwan. The latter is the Financial Officer for the Centre of Excellence, while the former carries out secretarial, administrative and receptionist duties.

Equipment expenditure from the ARC Centre grant was kept relatively small for 2004 at $167k due to the support from the State Government through DSRD and the host institution UNSW. These provided significant levels of funding specifically targeting equipment and infrastructure development. The majority of equipment expenditure from these latter sources was for a sophisticated screen-printing machine and in equipping the new Botany facilities with a range of processing capabilities. This allowed the ARC Centre funds to be used primarily for small laboratory equipment items such as hotplates, temperature controllers, pumps, an ultrasonic cleaner, a small processing furnace, specialised computing facilities, test equipment and a range of spare parts.

The Centre has a very active program with visiting academics, fellows and researchers. Where appropriate, the Centre contributes to rental accommodation costs, with a total of $31k being spent in this way during 2004.

The total expenditure from the ARC grant on consumables and maintenance for 2004 was $543k. These costs are strongly dominated by laboratory consumables to support the device research, such as high purity gases, chemicals and general laboratory supplies.

The expenditure on travel during 2004 from the ARC grant was $103k. The primary purposes for this travel were for staff and students participating in collaborative research projects with overseas industry partners such as Eurosolare in Italy and Suntech-Power in China, and for the attendance at international conferences such as the European Photovoltaic Solar Energy Conference in Paris in June 2004 and the 8th China Photovoltaic Conference in Shenzen, China, in November 2004. Fig 8.3: Expenditure of ARC income by

expenditure category

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

BOOK CHAPTERS

R. Corkish, “Solar Cells”, in “Encyclopedia of Energy”, Elsevier Press, pp. 545-557, 2004 (invited).

R. Corkish, “Photovoltaic Materials”, in “Encyclopedia of Chemical Processing”, submitted for 2005 publication, Marcel Dekker Inc. (invited).

M.A. Green, “Photovoltaic Applications of Nanostructures”, in A.A. Baladin and K.L. Wang (eds.), “Handbook of Semiconductor Nanostructures and Devices”, 2005 (invited).

M.A. Green, “High Efficiency Silicon Solar Cell Concepts” in “Solar Cells: Materials, Manufacture and Operation”, T. Markvart and L. Castener (eds.), Elsevier, Oxford, 2005, pp. 190-214.

M.A. Green, “Eureka!” in “The Ideas Book”, L. Carroli (ed.), Brisbane, University of Queensland Press, 2005.

B.S. Richards and M.A. Green, “Photovoltaic Cells”, in “Encyclopaedia of Biomedical Engineering”, Wiley, 2005.

PATENTS, PATENT APPLICATIONS

A.G. Aberle, P.I. Widenborg and N. Chuangsuwanich, “Glass Texturing”, International PCT patent application PCT/.AU2004/000339 (filed on 19th March, 2004).

M.A. Green, “Artificial Amorphous Semiconductors and Application to Solar Cells”, Application No. 200402299 (filed April, 2004).

T. Walsh, A.G. Aberle and S.R. Wenham, “Thin Film Solar Cell Interconnection”, Australian patent provisional application 2004903028 (filed on 4th June, 2004).

PAPERS IN SCIENTIFIC JOURNALS

A.G. Aberle, A. Straub, P.I. Widenborg, A.B. Sproul, Y. Huang and P. Campbell, “Polycrystalline Silicon Thin-Film Solar Cells on Glass by Aluminium-Induced Crystallisation and Subsequent Ion Assisted Deposition (ALICIA), Progress in Photovoltaics, Vol. 13, pp. 37-47, 2005.

A.B. Aberle and D. Inns, “Sheet Resistance Profiling and p-n Junction Localisation in Poly-Si Thin-Film Solar Cells by Wet-Chemical Etching, Solar Energy Materials and Solar Cells (submitted).

K.R. Catchpole, K.L. Lin, M.A. Green, A.G. Aberle, R. Corkish, J. Zhao and A. Wang, “Thin Semiconducting Layers as Active and Passive Emitters for Thermophotonics and Thermophotovoltaics”, Solar Energy, Vol. 76, pp. 251-254, 2004.

168

9. PUBLICATIONS

Ann

ual R

epor

t 2

00

4

P

ublic

atio

ns

169

K.R. Catchpole, “Silicon Photoluminescence External Quantum Efficiency Determined by Combined Thermal/Photoluminescence Measurements”, Semiconductor Science and Technology, Vol. 19, pp. 1411, 2004.

Eun-Chel Cho, Peter Reece, Martin A. Green, James Xia, Richard Corkish, Mike Gal, “Clear Quantum Confined Luminescence from Crystalline Silicon/SiO2 Single Quantum Wells”, Applied Physics Letters, Vol. 84, p. 2286, 2004.

Eun-Chel Cho, Peter Reece, Martin A. Green, James Xia, Richard Corkish, Mike Gal, “Clear Quantum Confined Luminescence from Crystalline Silicon/SiO2 Single Quantum Wells”, Virtual Journal of Nanoscale Science and Technology April, 2004.

Eun-Chel Cho, M.A. Green, J. Xia, A. Nikulin and R. Corkish, “Atomistic Structure of SiO2 /Si/SiO2 Quantum Wells with an Apparently Crystalline Silicon Oxide, Journal of Applied Physics, Vol. 96, p. 2311, 2004.

N. Chuangsuwanich, P.I. Widenborg, P. Campbell ad A.G. Aberle, “Light Trapping Properties of Thin Silicon Films on AIT-Textured Glass”, Solar Energy Materials and Solar Cells (submitted).

L. Ferraioli, P. Maddalena, E. Massera, A. Parretta, M.A. Green, A. Wang and J. Zhao, “Evidence for Generalised Kirtchhoff’s Law from Angle-Resolved Electroluminescence of High Efficiency Silicon Solar Cells”, Applied Physics Letters, Vol. 85, pp. 2484-2486, 2004.

L. Ferraioli, P. Maddalena, A. Parretta, A. Wang and J. Zhao, “Current-Voltage Characteristics of High-Efficiency Silicon Solar Cells from Photoluminescence”, Applied Physics Letters, Vol. 85, pp. 4422-4424, 2004.

M.A. Green, K. Emery, D.L. King, S. Igari and W. Warta, “Solar Cell Efficiency Tables (Version 23)”, Progress in Photovoltaics, January, 2004, pp. 55-62.

M.A. Green, P.A. Basore, N. Chang, D. Clugston, R. Egan, R. Evans, J. Ho, D. Hogg, S. Jarnason, M. Keevers, P. Lasswell, J. O’Sullivan, U. Schubert, A. Turner, S. R. Wenham and T. Young, “Crystalline Silicon on Glass (CSG) Thin-Film Solar Cell Modules”, Solar Energy, Special Issue on Thin Film Photovoltaics, Vol. 77 pp. 857-863, 2004 (invited).

M.A. Green, “Recent Developments in Photovoltaics”, Solar Energy, Vol. 76, pp. 3-8, 2004 (invited).

M.A. Green, K. Emery, D.L. King, S. Igari and W. Warta, “Solar Cell Efficiency Tables (Version 24)”, Progress in Photovoltaics, 2004, pp. 465-372.

M.A. Green, “Price/Efficiency Correlations for 2004 Photovoltaic Modules”, Progress in Photovoltaics, Vol.13, pp. 85-87, 2005.

G. Kymaravelu, M.M. Alkaisi, A. Bittar, D. Macdonald and J. Zhao, “Damage Studies in Dry Etched Textured Silicon Surface”, Current Applied Physics, Vol., 4, pp. 108-110, 2004.

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

K.L. Lin, K.R. Catchpole, P. Campbell and M.A. Green, “High External Quantum Efficiency from Double Heterostructure InGaP/GaAs Layers as Selective Emitters for Thermophotonic Systems”, Semicond. Sci. Technol. Vol 19, pp. 1268-1272, 2004.

B.S. Richards, S.R. Richards, M.B. Boreland, D.N. Jamieson, “High Temperature Processing of TiO2 thin Films for Application in Silicon Solar Cells”, Journal of Vacuum Science and Technology A., Vol. 211, pp. 339-348, 2004.

B.S. Richards, "Comparison of TiO2 and Other Dielectric Coatings for Buried Contact Solar Cells: A Review," Progress in Photovoltaics, Vol., 12, pp. 253-281, 2004.

A. Shalav, B.S. Richards, K.W. Kramer and H.U. Gudel, “The Application of NaYF4:Er3+ Up-Converting Phosphorus for Enhanced Near Infrared Silicon Solar Cell Response”, Applied Physics Letters, Vol., 86, 2005.

D. Song, P. Widenborg, A. Straub, Y. Huang and A.G. Aberle, “Polycrystalline Silicon Films on Glass by Solid Phase Crystallization of Evaporated a-Si”, Solar Energy Materials and Solar Cells (submitted).

J. Stradal, G. Scholma, H. Li, C.H.M. van der Werf, J.K. Rath, P.I. Widenborg, P. Campbell, A.G. Aberle and R.E.I. Schropp, “Epitaxial Thickening by Hot Wire Chemical Vapor Deposition of Polycrystalline Silicon Seed Layers on Glass, Thin Solid Films (submitted).

A. Straub, D. Inns, M.L. Terry, Y. Huang, P.I. Widenborg and A.G. Aberle, “Optimisation of Low-Temperature Silicon Epitaxy on Seeded Glass Substrates by Ion-Assisted Deposition”, Journal of Crystal Growth (submitted).

A. Straub, N.-P. Harder, Y. Huang and A.G. Aberle, “High-Quality Homoepitaxial Silicon Growth in a Non-Ultra-High Vacuum Environment by Ion-Assisted Deposition”, Journal of Crystal Growth, Vol. 268, pp. 41-551, 2004.

A. Straub, R. Gebs, H. Habenicht, S. Trunk, R.A. Bardos, A.B. Sproul and A,G. Aberle, “Impedance Analysis – A Powerful Method for the Determination of the Doping Concentration and Build-In Potential of Non-Ideal Semiconductor p-n Diodes”, Journal of Applied Physics (in press).

A. Straub, P.I. Widenborg, N.-P. Harder, A.B. Sproul, Y. Huang and A.G. Aberle, “Present Status of ALICIA Solar Cells on Glass”, Solar Energy Materials and Solar Cells (submitted).

M.L. Terry, A. Straub, D. Inns, D. Song and A.G. Aberle, “Large Open-Circuit Voltage Improvement by Rapid Thermal Annealing of Evaporated Solid-Phase-Crystallized Thin-Film Silicon Solar Cells on Glass”, Applied Physics Letters (submitted).

T. Trupke, P. Wurfel and M.A. Green, “Comment on Three-Dimensional Phononic-Crystal Emitter for Thermal Photovoltaic Power Generation “ (Applied Physics Letters, Vol. 83, p. 380 2003), Applied Physics Letters, Vol. 84, pp. 1997-1998, 2004.

170

Ann

ual R

epor

t 2

00

4

P

ublic

atio

ns

171

P.I Widenborg, A. Straub and A.G. Aberle, “Epitaxial Thickening of AIC Poly-Si Seed Layers on Glass by Solid Phase Epitaxy”, Journal of Crystal Growth (in press).

J.E. Wu and A.B. Aberle, “Characterisation of Micrometer-Sized Inversion-Layer Emitters in Crystalline Silicon”, Solar Energy Materials and Solar Cells (submitted).

J. Zhao, “Recent Advances of High Efficiency Single Crystalline Silicon Solar Cells in Processing Technologies and Substrate Materials”, Solar Energy Materials and Solar Cells, Vol. 82, pp. 53-64, 2004.

J. Zhao, G. Zhang, T. Trupke, A. Wang, F. Hudert and M.A. Green, “Near Band-Edge Light Emission from Silicon Semiconductor on Insulator Diodes“, Applied Physics Letters, Vol. 85, pp. 2830-2032, 2004.

B.S.Richards and M.E.Watt (2004), “Permanently Dispelling a Myth of Photovoltaics via Adoption of a New Net Energy Indicator”, Renewable and Sustainable Energy Review. In press.

M. Watt and I. MacGill (2004) “Securing Australia’s Energy Future – A Lost Investment and Employment Opportunity?”, Solar Progress Vol 25, No. 3, October 2004.

P. J. Cousins, D. H. Neuhaus and J. E. Cotter, “Experimental verification of the effect of depletion-region modulation on photoconductance lifetime measurements”, Journal of Applied Physics, 95(4), pp1854-1858, 2004.

J. H. Guo and J. E. Cotter, “Laser-grooved Backside Contact Solar Cells with 680 mV Open-circuit Voltage,” IEEE Transactions on Electronic Devices, vol.51(12), pp.2186-2192, 2004.

J. H. Guo and J. E. Cotter, “Metallization Improvement on Fabrication of Interdigitated Backside and Double-sided Buried Contact Solar Cells,” Solar Energy Materials and Solar Cells, vol.86(4), pp.485-498, 2005.

Ho Anita Wing Yi, Wenham Stuart R. (2004), Buried Contact Solar Cells with Innovative Rear Localised Contacts, Prog. Photovoltaics. Res. Appl. 2004; 12:297-308

Richards, B. S.; Rowlands, S. F.; Ueranatasun, A.; Cotter, J. E.; Honsberg, C. B. “Potential cost reduction of buried-contact solar cells through the use of titanium dioxide thin films,” Solar Energy (2004), 76(1-3), 269-276.

P. J. Cousins, A. Guo and J. Cotter, “Influence of diffusion-induced misfit dislocations on high efficiency silicon solar cells”, IEEE Trans. Electron Devices. (submitted Jan, 2005).

P. J. Cousins and J. Cotter, “Minimising lifetime degradation associated with thermal oxidation of upright randomly textured silicon surfaces”, Sol. Energy Mat. and Sol. Cells (submitted Nov, 2004).

P. J. Cousins, N. Mason and J. Cotter, “Loss Analysis of double-sided buried contact solar cells on FZ(B), MCZ(B), CZ(Ga) and CZ(B) wafers”, IEEE Trans. Electron Devices. (submitted Jan, 2005).

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

J. H. Guo, P. J. Cousins, and J. E. Cotter, “Investigations of Parasitic Shunt Resistance in N-type Buried Contact Solar Cells,” Progress in Photovoltaics: Research and Applications, (submitted Dec, 2004).

J. H. Guo, B. S. Tjahjono, and J. E. Cotter, “Performance and Stability of Laser-grooved Buried Contact Solar Cells Made on N-type Commercial Silicon Wafers,” IEEE Transactions on Electronic Devices, (submitted Dec, 2004).

J. H. Guo and J. E. Cotter, “Optimizing the Diffused Regions of Interdigitated Backside Buried Contact Solar Cells,” Solar Energy Materials and Solar Cells, (submitted Dec, 2004).

M.D. Abbott and J.E. Cotter “ Optical and Electrical Properties of Laser Texturing for High Efficiency Solar Cells,” IEEE Transactions on Electron Devices, (submitted Jan, 2005).

K.R. McIntosh, J.E. Cotter and D. Swanson, “A Simple Ray Tracer to Compute the Concentration of a Photovoltaic Module”, Progress in PHotovoltaics (submitted, Dec, 2004)

D. Debuff, “General Theory of Carrier Lifetime in Semiconductors with Multiple Localized States”, Journal of Applied Physics, Vol. 96, No. 11, pp. 6454-6469, 2004.

B. S. Richards, “Photovoltaics Literature Survey (No. 28)”, Progress in Photovoltaics: Research and Applications, Vol. 12, pp. 63-65, 2004.

A. Straub, P.I. Widenborg, A. Sproul, Y. Huang, N.-P. Harder, A.G. Aberle, “Fast and Non-Destructive Assessment of Epitaxial Quality of Polycrystalline Silicon Films on Glass by Optical Measurements”, Journal of Crystal Growth, Vol. 265, pp. 168-173, 2004.

B.S. Richards, A. Lambertz, A.B. Sproul, “Determination of the Optical Properties of Non-Uniformly Thick Non-Hydrogenated Sputtered Silicon Thin Films on Glass”, Thin Solid Films, Vol. 460, pp. 247-255, 2004.

T. Trupke, P. Wurfel, “Improved Spectral Robustness of Triple Tandem Solar Cells by Combined Series/Parallel Interconnection”, Journal of Applied Physics, Vol. 96, No. 4, pp. 2347-2351, 2004.

C.-W. Jiang, M.A. Green, E.-C. Cho and G. Conibeer, “Resonant Tunneling through Defects in an Insulator: Modeling and Solar Cell Applications”, Journal of Applied Physics, Vol. 96, No. 9, pp. 5006-5012, 2004.

T. Trupke and R.A. Bardos, “Self-Consistent Determination of the Generation Rate from Photoconductance Measurements”, Applied Physics Letters, Vol. 85, No. 16, pp. 3611-3613, 2004.

172

Ann

ual R

epor

t 2

00

4

P

ublic

atio

ns

173

CONFERENCE PROCEEDINGS

A.G. Aberle and D. Inns, “Sheet Resistance Profiling and p-n Junction Localisation in Poly-Si Thin-Film Solar Cells by Wet Chemical Etching”, Technical Digest, PVSEC-14, Bangkok, Thailand, January, 2004, pp. 747-748.

A.G. Aberle, P.I. Widenborg, D. Song, A. Straub, M.L. Terry, T. Walsh, A. Sproul, P. Campbell D. Inns B. Beilby, M. Griffin, J. Weber, Y. Huang, O. Kunz, R. Gebs, F. Martin-Bruce, V. Barroux and S.R. Wenham, “Recent Advances in Polycrystalline Silicon Thin-Film Solar cells on Glass at UNSW”, Conf. Proceedings, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA, (in press).

A.G. Aberle, “Crystalline Silicon Thin-Film Solar Cells on Glass – Cheap Electricity from the Sun?”, Conf. Record, 16th Australian Institute of Physics Congress, Canberra, February, 2005.

A.G. Aberle, “Crystalline Silicon Thin-Film Solar Cells: Where are We? Where to Go from Here?”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

Eun-Chel Cho, Young Hyun Cho, James Xia, Richard Corkish, Gavin Conibeer, Yidan Huang, and Martin A. Green, “Solar Cell Emitters with Silicon Nanostructures”, Conf. Record, 14th PVSEC, Bangkok, Thailand, January, 2004, pp.249-250.

E.-C. Cho, Y.-H,. Cho, T. Trupke, R. Corkish, G. Conibeer and M.A. Green, “Silicon Nanostructures for All-Silicon Tandem Solar Cells”, 19th European Photovoltaic Solar Energy Conference, Paris, 2004, pp. 235-238.Eun-Chel Cho, Y.H. Cho, C.-W. Jiang, Y. Huang, D. Conibeer, R. Corkish and M.A. Green, “Silicon Nanostructures in Photovoltaics”, Korea Conference on Innovative Science and Technology (KCIST-2004), September, 2004.

E-C. Cho, Y.H. Cho, C-W. Jiang, Yidan Huang, G.J. Conibeer, R. Corkish, M.A. Green, “Silicon Nanostructures in Photovoltaics”, Korea Conference on Innovative Science and Technology (KCIST-2004): New Frontiers in Photovoltaics, Hotel Hyundai, Gyeongju, Korea, Sept. 1-4, 2004.

Y.H. Cho, E.C. Cho, Y. Huang, T. Trupke, G. Conibeer, Martin A. Green, “Silicon Quantum Dots in SiNx Matrix for Third Generation Photovoltaics”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June 2005.

N. Chuangsuwanich, P.I. Widenborg, P. Campbell and A.G. Aberle, “Light Trapping Properties of Thin Silicon Films on AIT-Textured Glass”, Technical Digest, PVSEC-14, Bangkok, Thailand, January, 2004, pp. 325-326.

N. Chuangsuwanich, P. Campbell, P.I. Widenborg, A. Straub and A.G. Aberle, “Light Trapping Properties of Evaporated Poly-Silicon Films on AIT-Textured Glass Substrates”, Conf. Proceedings, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA, (in press).

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

G. Conibeer and M.A. Green, “Phononic Engineering for Hot Carrier Solar Cell Absorbers”, 19th European Photovoltaic Solar Energy Conference, Paris, 2004, pp. 270-273.

G.J. Conibeer, A.S. Brown, T. Trupke, M.A. Green and A. Shalav, “Efficiency Limits for an Ideal Solar Cell with Combined Up and Down Conversion or for Two Level Up Conversion Using Blackbody Modelling and AM1.5G Data”, 19th European Photovoltaic Solar Energy Conference, Paris, June, 2004, pp.274-279.

G.J. Conibeer and B.S. Richards, “Comparison of PV and Photoelectrochemical Generators in a Hydrogen Storage SAPS”, International Conference on Materials for Hydrogen Energy, 2004, Sydney, Australia (in press).

G.J. Conibeer, M.A. Green, R. Corkish, Y. Cho, E-C. Cho, T. Fangsuwannarak, E. Pink, Y. Huang, T. Puzzer, B. Richards, A. Shalav, “Silicon Nanostructures for Third Generation Photovoltaic Solar Cells”, THINC-PV2 symposium European MRS conference, Strasbourg, May 2005 (invited).

G.J Conibeer, M.A. Green, “Phononic Band Gap Engineering for Hot Carrier Solar Cell Absorbers”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June 2005.

R. Corkish, T. Burns, S.R. Wenham, .B. Sproul, J. Cotter, C.B. Honsberg, A.G. Aberle, A. Bruce, T. Spooner, G. Stapleton, M. Watt, B.S. Richards, M.A. Green and L. Cahill, “Employment Outcomes for Australian Photovoltaic Engineering Graduates”, Conf. Record, 19th European Photovoltaic Solar Energy Conference, June, 2004, Paris, France, pp. 2373-2376.

R. Corkish, T. Burns, S.R. Wenham, A.B. Sproul, J. Cotter, C.B. Honsberg, A.G. Aberle, A. Bruce, T. Spooner, G. Stapleton, M. Watt, B.S. Richards, M.A. Green, L. Cahill, “Employment Outcomes for Australian Photovoltaic Engineering Graduates”, 19th European Photovoltaic Solar Energy Conference, Paris, 7-11 June 2004, pp. 3273 -3276.

R. Corkish, S.R. Wenham, A.B. Sproul, J. Cotter, C.B. Honsberg, M.A. Green, A.G. Aberle, G. Conibeer, J. Zhao, “Photovoltaics and Renewable Energy Education and Research at the University of New South Wales”, International Solar Energy Society Asia-Pacific 2004, Gwangju, South Korea, 17-20 October 2004, pp. 175-182 (ISBN: 89-954429-1-3).

R. Corkish, S.R. Wenham, A.B. Sproul, J. Cotter, C.B. Honsberg, A.G. Aberle, A. Bruce, T. Spooner, G. Stapleton, M. Watt, B.S. Richards, M. Kobacker, M.A. Green, “Australian Photovoltaic Engineering Graduates”, International Symposium on Renewable Energy Education, ISREE-10, Perth, 28 Nov. – 1 Dec. 2004.

R. Corkish, A.B. Sproul, A. Bruce, G. Stapleton and S.R. Wenham, “Developing Country Renewable Energy Projects at UNSW”, International Symposium on Renewable Energy Education, ISREE-10, Perth, 28 Nov. – 1 Dec. 2004.

R. Corkish and A. Blakers, “Renewable Fossil Fuels? Corruption of our Language”, Solar 2004, Perth, 1-3 Dec. 2004.

R. Corkish, A. Bruce, J. Bocking, M. Linney and A. Sproul, “Small Photovoltaic Systems in Rural Nicaragua”, Solar 2004, Perth, 1-3 Dec. 2004.

174

Ann

ual R

epor

t 2

00

4

P

ublic

atio

ns

175

R. Corkish, S.R. Wenham, A.B. Sproul, J. Cotter, C.B. Honsberg, M.A. Green, A.G. Aberle, G. Conibeer, "Education and Research at the Centre for Photovoltaic Engineering, University of New South Wales”, International Rio 5 World Climate and Energy Event, Rio de Janeiro, Brazil, 15-17 February 2005.

R. Corkish, S.R. Wenham, M.A. Green, A.B. Sproul, J. Cotter, A.G. Aberle, A. Bruce, “Materials Engineering Education in Two New Engineering Degree Programs at the Centre for Photovoltaic Engineering”, MRS Fall Meeting, Boston, Dec. 2005 (abstract submitted, invited).

J. Cotter, S.R. Wenham, G. Bates, A. Bruce, M.A. Green, C.B. Honsberg, R. Largent, M.D. Silver, A.B. Sproul, G. Stapleton and M. Watt, “Photovoltaics and Renewable Energy Engineering Degree Programs at UNSW – The First Four Years”, Technical Digest, PVSEC-14, Bangkok, Thailand, January, 2004, pp. 477-479.

M.A. Green, “Present and Future of Crystalline Silicon Solar Cells”, PVSEC-14, Thailand, January, 2004, pp.391-394 (invited plenary).

M.A. Green, Book Review, “Organic Photovoltaics: Concepts and Realization”, The Physicist, Vol. 41, January/February, 2004.

M.A. Green, “Update on Photovoltaic Solar Energy Conversion”, Engineers Australia, Sydney, 27th May, 2004.

M.A. Green, “Third Generation Photovoltaics: Theoretical and Experimental Progress”, 19th European Photovoltaic Solar Energy Conference, Paris, June, 2004, pp.3-8 (opening plenary).

M.A. Green, “Third Generation Solar: Future Photovoltaics”, Renewable Energy World, July/August, 2004, pp. 203-213 (invited).

M.A. Green, “Silicon-on-Insulator (SOI): A Path to High Density Silicon Optoelectronics”, 1st International Conference on Group IV Photonics, Hong Kong, 2004 (invited).

M.A. Green, “Nanomaterials for Photovoltaic Applications”, International Conference on Nanomaterials: Synthesis, Characterisation and Application, Kolkata, India, 2004 (invited).

M.A. Green, “ Renewable Energy and Its Potential”, IEEE Regional Meeting, Jaipur, India, 2004.

M.A. Green, “Crystalline Silicon Solar Cells: Status and Challenges”, Building Science Forum of Australia Seminar, “Solar Power and Buildings”, Sydney, October, 2004.

M.A. Green, “Solar Technologies & Global Potential of Solar Energy”, GCEP Solar Energy Workshop, Stanford University, October, 2004 (invited).

M.A. Green, “Third Generation Photovoltaics”, GCEP Solar Energy Workshop, Stanford University, October, 2004 (invited).

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

M.A. Green, “Photovoltaics – Electricity’s Future?”, Local Government Manager, Accent on Energy, December, 2004/January, 2005, p. 28.

M.A. Green, “Renewable Energy and Its Future”, International Rio 5 World Climate and Energy Event, Rio de Janeiro, Brazil, 15-17 February 2005.

M.A. Green, “State-of-the-Art of Solar Cell Development”, International Rio 5 World Climate and Energy Event, Rio de Janeiro, Brazil, 15-17 February 2005.

M.A. Green, “Status of Silicon on Glass, a New Production Technology", 2nd Photovoltaic Science Applications and technology Conference, Loughborough, U.K., April, 2005 (invited).

M. A. Green, E. C. Cho, Y. Cho, Y. Huang, E. Pink, T. Trupke, A. Lin, T. Fangsuwannarak, T. Puzzer, G. Conibeer, R. Corkish, “All-Silicon Tandem Cells based on “Artificial” Semiconductor Synthesised using Silicon Quantum Dots in a Dielectric Matrix”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June 2005.

T.E. Humphrey and H. Linke, “Inhomogeneously Doped Thermoelectric Nanomaterials” Proceedings, International Thermoelectrics Conference, Adelaide, 2004.

T.E. Humphrey and H. Linke”, Quantum, Cyclic and Particle Exchange Heat Engineers”, Proc., Frontiers of Quantum and Mesoscopic Thermodynamics, Prague, 2004.

D. Inns, A. Straub, M. Terry, Y. Huang and A.G. Aberle, “Improved Ion-Assisted Silicon Epitaxy Process on Seeded Glass by Optimised Sample Heating Procedure”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

C-W. Jiang, E-C. Cho, G.J. Conibeer, M.A. Green, “Silicon Quantum Dots: Application for Energy Selective Contacts to Hot Carrier Solar Cells”, Proc. 19th European Photovoltaic Solar Energy Conference, Paris, 2004, pp 80-83.

G. Kumaravelu, M.M. Alkaisi, D. Macdonald, J. Zhao, B. Rong, U. Traxlmayr and B. Vogl, “Minority Carrier Lifetime in Plasma Textured Silicon Wafers for Solar Cells”, Schell2004 Conference, Badajoz, Spain, May, 2004.

O. Kunz, A. Straub, A. Sproul and A.G. Aberle, “Determination of the Two-Diode Model Parameters of Non-Ideal Solar Cells”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

S. Pillai, K.R. Catchpole and A. Shalav, “Targeting Better Absorption at Longer Wavelengths Using Surface Plasmons”, Proc. 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA.

S. Pillai, K.R Catchpole, T. Trupke, M.A. Green “Application of Surface Plasmons for Increased Light Emission from Silicon Based LEDs”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

176

Ann

ual R

epor

t 2

00

4

P

ublic

atio

ns

177

N.R. Poespawati, A. Udhiarto, D. Hartanto and M.A. Green, “Si1-xGex/Si Solar Cell by Optically Thickness of Graded Si1-Gex Layer, Conf. Record, PVSEC-14, Thailand, January, 2004, pp.273-274.

B.S. Richards, A. Shalav and R.P. Corkish, “A Low Escape-Cone-Loss Luminescent Solar Concentrator”, 19th European Photovoltaic Solar Energy Conference, June, 2004, Paris, France, pp. 113-116, 2004.

B.S. Richards, C. Remy and A.I. Schafer, “Sustainable Drinking Water Production from Brackish Sources using Photovoltaic Power”, 19th European Photovoltaic Solar Energy Conference, Paris, France, pp. 3369-3372, 2004.

B.S. Richards and A. Shalav, “Luminescent Down-Converters in Polymer Hosts for Enhanced Solar Cell Performance”, presented at The International Conference in Synthetic Metals (ICSM): The Role and Impact of Nanoscience and Nanotechnologies, July, 2004, University of Wollongong (in press).

B.S. Richards and M.E. Watt, “Use of the Energy Yield Ratio as a Means of Dispelling One Myth of Photovoltaics”, Proceedings ANZSES Conference Solar 2004, December, 2004, Perth, Australia (in press).

A. Shalav, B.S. Richards, “Third Generation Photovoltaics: A Quest for Cheap Efficient Silicon Solar Cells”, Conf. Proceedings, The Environmental Engineering Research Event, pp. 45-53, 2004.A. Shalav, B.S. Richards, K.W. Cramer and H.U. Gudel, “Improvements of an Up-Conversion NaYF4:Er3+ Phosphor-Silicon Solar Cell Device for an Enhanced Response in the Near-Infrared”, Conf. Proceedings, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA.

D. Song, P. Widenborg, A. Straub, Y. Huang and A.G. Aberle, “Polycrystalline Silicon Films on Glass by Solid Phase Crystallization of Evaporated a-Si”, Technical Digest, PVSEC-14, Bangkok, Thailand, January, 2004, pp. 327-328.

D. Song, A. Staub, P. Widenborg, B. Vogl, P. Campbell, Y. Huang and A.G. Aberle, “Polycrystalline Silicon Thin-Film Solar Cells on Glass by Solid Phase Crystallization of In-Situ Doped Evaporated a-Si”, 19th European Photovoltaic Solar Energy Conference, June, 2004, Paris, France, pp. 1193-1196, 2004.

D. Song, P.I. Widenborg, A. Straub, P. Campbell, N. Chuangsuwanich, Y. Huang and A.G. Aberle, “EVA Polycrystalline Silicon Thin-Film Solar Cells on Textured Glass”, Conf. Proceedings, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA, (in press).

D. Song, P.I. Widenborg, A. Straub, D. Ins, M. Terry and A.G. Aberle, “Impact of Glass Substrate Topography on EVA Polycrystalline Silicon Thin-Film Solar Cells on Glass”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

J. Stradal, G. Scholma, H. Li, C.H.M. van der Werf, J.K. Rath, PI. Widenborg, P. Campbell, A.G. Aberle and R.EI. Schropp, “Epitaxial Thickening by Hot Wire CVD of Polycrystalline Silicon Seed Layers made by AIC on Glass”, Conf. Proceedings, 3rd International Conference on Hot-Wire CVD (Cat-CVD) Processes, Utrecht, Netherlands, August, 2004 (in press).

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

J. Stradal, G. Scholma, H. Li, C.H.M. van der Werf, J.K. Rath, P.I. Widenborg, P. Campbell, A.G. Aberle and R.EI. Schropp, “Hot Wire CVD for Epitaxial Thickening of Polycrystalline Silicon Seed Layers made by AIC on Glass”, Conf. Proceedings, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA, (in press).

A. Straub, R. Gebs, A.B. Sproul, R. Bardos, D. Inns, M.L. Terry and A.G. Aberle, “C-V Measurements by Impedance Analysis on ALICIA Polycrystalline Silicon Thin-Film Solar Cells on Glass”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

A. Straub, P.I. Widenborg, A. Sproul, Y. Huang, P. Campbell and A.G. Aberle, “Towards 300 mV ALICIA Polycrystalline Silicon Solar Cells on Glass”, 19th European Photovoltaic Solar Energy Conference, June, 2004, Paris, France, pp. 963-966, 2004

A. Straub, P.I. Widenborg, N.-P. Harder, A.B. Sproul, Y. Huang and A.G. Aberle, “Present Status of ALICIA Solar Cells on Glass”, Technical Digest, PVSEC-14, Bangkok, Thailand, January, 2004, pp. 29-30.

A. Straub, D. Inns, O. Kunz, M.L. Terry, P.I. Widenborg, A.B. Sproul and A.G. Aberle, “Towards 400 mV ALICIA Thin-Film Silicon Solar Cells on Glass”, (in press).

M.L. Terry, A. Straub, D. Inns, D. Song and A.G. Aberle, “Voc Improvement of Evaporated SPC Thin-Film Si Solar Cells on Glass by Rapid Thermal Annealing”, Conf. Proceedings, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA, (in press).

T. Trupke, A. Shalav, P. Wurfel and M.A. Green, “Efficiency Enhancement of Solar Cells by Luminescent Up-Conversion of Sunlight”, Conf. Record, International PVSEC-14, Bangkok, 2004, pp. 753-756 (invited).

T. Trupke, F. Hudert, P. Wurfel, R.A. Bardos, J. Zhao, A. Wang and M.A. Green, “Effective Excess Carrier Lifetimes Exceeding 100 ms in Float Zone Silicon Determined from Photoluminescence”, 19th European Photovoltaic Solar Energy Conference, Paris, 2004, pp.758-761.

T.M. Walsh, S.R. Wenham ad A.G. Aberle, “Novel Method for the Interconnection of Thin-Film Silicon Solar Cells on Glass”, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA, (in press).

P.I. Widenborg and A.G. Aberle, “Aluminium-Induced Crystallisation Bi-Layer Process of a-Si: Limiting Mechanism”, Technical Digest, PVSEC-14, Bangkok, Thailand, January, 2004, pp. 281-282.

P.I. Widenborg, A. Straub, Y. Huang and A.G. Aberle, “Solid Phase Epitaxy of a-Si on AIC Poly-Si Seed Layers on Glass Substrates”, 19th European Photovoltaic Solar Energy Conference, June, 2004, Paris, France, pp. 1205-1208, 2004.

P.I. Widenborg, T. Puzzer, J. Stradal, D.H. Neuhaus, D. Inns, A. Straub and A.G. Aberle, ”Structural Quality of Smooth AIC Poly-Si Films on Glass Substrates”, 31st IEEE Photovoltaic Specialists Conference, January, 2005, Florida, USA, (in press).

178

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

179

P.I. Widenborg, J. Weber, M.L. Terry and A.G. Aberle, “Solid Phase Epitaxy of PECVD a-Si:H on AIC Poly-Si Seeded Glass Substrates”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Sprain, June, 2005.

J.E. Wu and A.G. Aberle, “Characterisation of Micrometer-Sized Inversion-Layer Emitters in Crystalline Si”, Technical Digest, PVSEC-14, Bangkok, Thailand, January, 2004, pp. 439-440.

G. Yao, J. Zhao and J.E. Cotter, “Hybrid Stencil-Print/Buried-Contact Solar Cells”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

J. Zhao and A. Wang, “Stable High Efficiency Rear Boron Emitter Solar Cells on n-type Single Crystalline Silicon Substrates”, Proceedings, 8th China Photovoltaic Conference and China-Japan PV Workshop, Shenzhen, China, November, 2004, pp. 866-871.

J. Zhao and A. Wang, “High Efficiency Real Emitter PERT Solar Cells on n-type FZ Single Crystalline Silicon Substrates”, 20th European Photovoltaic Solar Energy Conference, Barcelona, Spain, June, 2005.

Muriel Watt, “Progress in Australian Photovoltaic and Hybrid Applications”, 14th International Photovoltaic Science and Engineering Conference, Bangkok, January 2004

Muriel Watt, Hugh Outhred and Iain MacGill, “Policy Support for Renewables in Australia with PV as a Case Study”, 14th International Photovoltaic Science and Engineering Conference, Bangkok, January 2004

M. Watt, “Overview of International PV Developments and Implications for Australia”, presented at Sustainable Energy 2004, the BCSE Annual Conference, 29 March-1 April 2004, Sydney.

M. Snow, D.K. Prasad and M. Watt, “The Delivery of Building Integrated Photovoltaics Best Practice Guidelines for Australian Conditions”, presented at the 19th European Photovoltaic Solar Energy Conference and Exhibition, 7-11 June, 2004, Paris.

M. Watt, S. Partlin, M. Oliphant, H. Outhred, I. MacGill, E. Spooner, “The Value of PV in Summer Peaks”, presented at the 19th European Photovoltaic Solar Energy Conference and Exhibition, 7-11 June, 2004, Paris.

M. Watt (2004) “Development of the Australian PV Roadmap”, presented on behalf of the Australian Business Council for Sustainable Energy at the IEA-PVPS Workshop for Industry at the 19th European Photovoltaic Solar Energy Conference and Exhibition, 7-11 June, 2004, Paris.

M. Watt, “Overview of Australia’s PV Industry”, presented at the IEA-PVPS Workshop, 10 September, Sydney.

M. Watt, S. Partlin, M. Oliphant, H. Outhred, I. MacGill, E. Spooner, “Analysis of PV System Output, Temperature, Electricity Loads, and National Electricity Market Prices, Summer 2003-04”, presented to Solar 2004, Annual Conference of the Australian and New Zealand Solar Energy Society, November 2004, Perth.

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

N.C. Shaw and J.E. Cotter, “Increasing the Light Generated Current In Thin Wafer-type Silicon Solar Cells Using Pigmented Diffuse Reflectors,” Proceedings of the International PVSEC-14, Bangkok, Thailand, 2004, pp. 339-340.

J.E. Cotter, et al., “Photovoltaics and Renewable Energy Engineering Degree Programs at UNSW - The First Four Years,” Proceedings of the International PVSEC-14, Bangkok, Thailand, 2004, pp. 477-479.

M.D. Abbott, L. Mai and J.E. Cotter, “Laser Texturing of Multicrystalline Silicon Solar Cells,” Technical Digest of the International PVSEC-14, Bangkok, Thailand, 2004, pp. 1015-1016.

Wenham S.R., Mai L. and Ho A. (2004), High Efficiency Silicon Solar Cells, Invited Paper for the 8th China Photovoltaic Conference, Shenzhen, China. Nov 2004.

Ho Anita W.Y., Wenham Stuart R. (2004), Buried Contact Solar Cells with Pinhole Size Rear Contacts, In Proc. of the 42nd Annual Conference of the Australian and New Zealand Solar Energy Society, Perth, Australia. Dec 2004.

P. J. Cousins, C. B. Honsberg and J. E. Cotter, “Double-sided buried contact solar cells on Czochralski wafers”, 13th Workshop on Crystalline Silicon Solar Cell Materials Processes, Vail, 2004, pp 190-193.

P. J. Cousins, N. B. Mason and J. E. Cotter, “Manufacturing and design issues for thin silicon solar cells on FZ(B), MCZ(B), CZ(Ga) and CZ(B) wafers”, 31st IEEE Photovoltaics Specialists Conference, Orlando, January 2005.

P. J. Cousins and J. E. Cotter, “Misfit dislocations generated during non-ideal boron and phosphorus diffusion and their effect on high-efficiency silicon solar cells”, 31st IEEE Photovoltaics Specialists Conference, Orlando, FL, January 2005.

J. H. Guo, B. S. Tjahjono, and J. E. Cotter, “19.2% Efficiency N-type Laser-grooved Silicon Solar Cells,” in Proc. 31st IEEE Photovoltaic Specialists Conference, Orlando, FL, 2005.

M. D. Abbott, P. J. Cousins, F. W. Chen and J. E. Cotter,”Laser-induced defects in crystalline silicon solar cells’’ 31st IEEE Photovoltaic Specialist Conference, Orlando, FL, January 2005.

J.E. Cotter, “RaySim 6.0 - A Free Geometrical Ray Tracing Program for Silicon Solar Cells”, 31st IEEE Photovoltaic Specialists Conference, Orlando, FL, January 2005.

J.E. Cotter and G. Yao, “Laser-formed Stencils for Printed Silicon Solar Cells”, 31st IEEE Photovoltaic Specialists Conference, Orlando, FL, January 2005.

Ho Anita W.Y., Wenham Stuart R. (2005), Investigation of the Surface Passivation Quality of Various Novel Low Temperature Rear Localised Contacting Schemes for PV Devices, In Proc. of 31st IEEE Photovoltaic Specialists Conference, Florida, USA. Jan 2005.

180

Ann

ual R

epor

t 2

00

4

Pub

licat

ions

181

P. J. Cousins and J. E. Cotter “Impact of thermal transients on the lifetime of upright randomly textured silicon surfaces evaluated using photoconductance lifetime measurements”, 20th PVSEC, Barcelona (accepted for poster presentation).

G. Yao, J. Zhao and J.E Cotter “Hybrid stencil-print/buried-contact solar cells”, 20th PVSEC, Barcelona (accepted for oral presentation).

F Chen, JE Cotter, A Cuevas, K Roth and S Windebaum “Anomalous Behavour of Annealed Silicon Nitride Films,” 20th PVSEC, Barcelona, 2005 (accepted for poster presentation).