THREE TERMINAL SI-SI:GE MONOLITIC TANDEM SOLAR CELLS. Lu Wang

Transcription

1 THREE TERMINAL SI-SI:GE MONOLITIC TANDEM SOLAR CELLS by Lu Wang A thesis submitted to the Faculty of the University of Delaware in partial fulfillment of the requirements for the degree of Master of Science in Electrical and Computer Engineering Summer 2011 Copyright 2011 Lu Wang All Rights Reserved

2 THREE TERMINAL SI-SI:GE MONOLITIC TANDEM SOLAR CELLS by Lu Wang Approved: Allen M. Barnett, Ph.D. Professor in charge of thesis on behalf of the Advisory Committee Approved: Kenneth E. Barner, Ph.D. Chair of the Department of Electrical and Computer Engineering Approved: Babatunde A. Ogunnaike, Ph.D. Interim Dean of the College of Engineering Approved: Charles G. Riordan, Ph.D. Vice Provost for Graduate and Professional Education

3 ACKNOWLEDGMENTS It has been an honor to conduct research under the advisement of Dr. Allen Barnett during my master and continuing Ph.D. education. He provides me with opportunities to touch silicon solar cells research, arranges skilled people to teach me solar cells fabrication and leads us to a higher level of research that is rigorous, quantitative and predictive. To be lucky as one of his students, I can stand on the shoulder of Giant and focus on meaningful research. His education let me to learn both fabrication and physics of solar cells. Thanks to Kevin Shreve, Chris Kerestes and Yi Wang. Both of Kevin and Chris taught me a lot of solar cells fabrication processing. I cannot master so many things in such a short period time without their help. My thesis is based on Chris and Yi s work, and it is impossible to reach this level without their help and previous work. I would like to thank AmberWave Inc., their collaboration makes this research possible and successful. It has been great to work with James Mutitu, Ruiying Hao, Xiaoting Wang, Xuesong Lu, Paola Murcia, Martin Diaz, Nicole Kotulak and Ken Schmieder. I love this group so much since we always share everything with each other, which makes our experiments more smoothly. And thanks to Martin, Nicole and Ken for helping improve my English skill. Many thanks to my family and friends, without their support and encouragement I cannot achieve anything. iii

4 Thanks for the funding from the U.S. Government Defense Advanced Research Projects Agency under Agreement No.: HR The views, opinions, and/or findings contained in this thesis are those of the author and should not be interpreted as representing the official views or policies, either expressed or implied, of the Defense Advanced Research Projects Agency or the Department of Defense. The author would also like to thank China Scholarship Council for providing stipends during his Ph.D. education. iv

5 TABLE OF CONTENTS LIST OF TABLES... viii LIST OF FIGURES... x ABSTRACT... xv Chapter 1 INTRODUCTION TO THREE TERMINAL SI-SI:GE SOLAR CELLS Multi-Junction Solar Cells Systems Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells Objectives of This Thesis Thesis Outline BASIC THEORY OF PHOTOVOLTAICS PN Junction Photovoltaic Effect Quantum Efficiency Solar Cell Parameters Series and Shunt Resistance Bulk and Surface Recombination DESIGN AND FBRICATION OF THREE TERMINAL SI-SI:GE MONOLITHIC TANDEM SOLAR CELLS Series Connected Two Terminal Si-Si:Ge Monolithic Tandem Solar Cell Two Terminal Si-Si:Ge Solar Cell s Structure and Equivalent Circuit Drawbacks of Two Terminal Si-Si:Ge Solar Cell Design of Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells Structure of Three Terminal Si-Si:Ge Solar Cells Equivalent Circuit v

6 3.2.3 Silicon Solar Cell Design Base Dopant Density and Thichness Emitter Formation Si:Ge Solar Cell Design Electrical and Optical Properties of Ge:Si Alloy Electrical Performance of Si:Ge Solar Cells High Ge Concentration Si:Ge Layer Growth (RPCVD) Modeling of Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells Light Path Curve in Silicon Solar Cell with Inverted Pyramids on Top Surface PERL Cell and Modified PERL Cell Performances Best Fabricated Si:Ge Solar Cell and Modified Best Fabricated Si:Ge Cell Performances Overall Performances of 3 Terminal Si-Si:Ge Solar Cell Composed of Modified Best Fabricated Si:Ge Solar Cell and Modified PERL Cell Performances of Best Measured Parameters Si:Ge Solar Cell, its Modified Cell Performances and Corresponding 3 Terminal Si-Si:Ge Solar Cell Performances of Model of Si:Ge Solar Cell, Modified Model of Si:Ge Cell Performances and Corresponding 3 Terminal Si-Si:Ge Solar Cell FABRICATION, TEST AND ANALYSIS OF THREE TERMINAL SI-SI:GE MONOLITHIC TANDEM SOLAR CELLS Fabrication of Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells Testing Equipments Introduction Test Results of Three Samples First Group Structure and Testing Second Group Structure and Testing Third Group Structure and Testing CONCLUSIONS AND FUTURE WORK Summary of This Work vi

7 5.2 Future Work BIBLIOGRAPHY Appendix PERMISSION LETTER vii

8 LIST OF TABLES Table 3-1 Physical Effect Impeding Transparency and Solutions Table 3-2 Silicon PN junction s Performances right after diffusion Table 3-3 Performance of Five Generation Si:Ge Solar Cells Table 3-4 Summary of Efficiencies of PERL Cell, Modeled 3T Si-Si:Ge Cell, Si Cell and Si:Ge Cell within the 3T Solar Cell Structure at One Sun Table % PERL Cell Parameters Table 3-5 PERL Cell and Modified PERL Solar Cell Performances with Different Thickness under Different Concentration Table 3-6 Best Si:Ge Solar Cell Performances Table 3-7 Best Si:Ge and Modified Si:Ge Solar Cell Performances Table 3-8 Compare of Performances between PERL Cell and 3 Terminal Si- Si:Ge Solar Cell Table 3-9 Best Parameters Si:Ge Solar Cell Performances Table 3-10 Best Parameters Si:Ge and Modified Best Parameters Si:Ge Solar Cell Performances Table 3-11 Performances of PERL Cell and 3 Terminal Si-Si:Ge Solar Cell with Best Parameters Modified Si:Ge Solar Cell Table 3-12 Modeled Si:Ge Solar Cell Performances Table 3-13 Modeled Si:Ge and Modified Modeled Si:Ge Solar Cell Performances viii

9 Table 3-14 Performances of PERL Cell and 3 Terminal Si-Si:Ge Solar Cell with Modeled Modified Si:Ge Solar Cell Table 4-1 Silicon Solar Cells performance of 3-T structure in the 1 st group Table 4-2 Si solar cell and Si:Ge solar cell performance Table 4-3 Si solar cell and Si:Ge solar cell performance ix

10 LIST OF FIGURES Figure 1-1 World Annual Solar Photovoltaics Production Figure 1-2 Figure 1-3 A multi-junction solar cell system with standalone Si solar cell and SiGe solar cell... 3 Spectrum from 300nm to 850nm can be absorbed by high bandgap material; Spectrum from 850nm to 1100nm can be absorbed by Si; Spectrum from 1100nm to 1800nm can be absorbed by Si:Ge... 4 Figure 1-4 Transmittance of silicon substrate with planar or texturing surfaces... 5 Figure 1-5 Integration of two stand alone cells into a 3 terminal monolithic tandem solar cell... 7 Figure 2-1 Cross Section of A Solar Cell Figure 2-2 Figure 2-3 Figure 2-4 Band Gap of PN junction, Diffusion Current Equals Drift Current at Equilibrium State, No Net Current across PN Junction Band Gap of PN Junction under Illumination and Collection of Light Generated Current QE of a silicon solar cell. As indicated in the figure, response at different wavelength reflects the response at different area of solar cell, because blue light has very high absorption coefficient leading to quick absorption at the surface area of solar cell, while red light will be absorbed at the back of solar cell due to its low absorption coefficient Figure 2-5 IV curve of a silicon solar cell Figure 2-6 Parasitic series and shunt resistance in a solar cell circuit x

11 Figure 2-7 Figure 2-8 Figure 3-1 Figure 3-2 Figure 3-3 Ideal IV curve (red) and IV curve with high series resistance and low shunt resistance (blue) of a silicon solar cell Misfit dislocations are generated at the locations where there are missing or dangling bonds in the lattice between two layers with different lattice constants. The step graded Ge:Si buffer layer can minimize negative effects of the large lattice mismatch between the Si substrate and the high Ge concentration Ge:Si layer Series Connected Two Terminal Si-Si:Ge Monolithic Tandem Solar Cell Structure (right) and It Equivalent Circuit (left) A 2D structure of three terminal Si-Si:Ge monolithic tandem solar cell A 3D structure of three terminal Si-Si:Ge monolithic tandem solar cell Figure 3-4 Equivalent circuits of an ideal three terminal Si-Si:Ge solar cell Figure 3-5 Minority carriers flows under illumination Figure 3-6 Figure 3-7 Figure 3-8 Figure 3-9 Comparison of the Si absorption coefficient with cm -3 (black curve) and cm -3 (red curve) dopant concentrations Dependence of silicon transmittance taking into account of free carrier absorption, which decreases transmittance with increasing doping concentration Spreading Resistance Profile of an N type silicon emitter diffused at 880 for 26min, and sheet resistance is 54Ω/sq Si:Ge alloy band gap decrease continuously as Ge concentration increases Figure 3-10 Absorption coefficient of Si:Ge alloy at different Ge concentration at 300K Figure 3-11 Performance of Ge:Si solar cells below Si varies with the Ge concentration xi

12 Figure 3-12 Structure of 5 th Generation single junction Si:Ge Solar Cells Figure 3-13 ASM Epsilon TM E2000 Reduced Pressure Chemical Vapor Deposition System (RPCVD) Figure 3-14 SIMS measurement results of the first generation Ge:Si solar cell grown by RPCVD, Ge concentration can be accurately controlled even in graded concentration growth. Oxygen concentration is acceptable Figure 3-15 Modeling Structure of Three Terminal Si-Si:Ge Solar Cell Figure 3-16 Efficiencies of 3T Si-Si:Ge Solar Cells Consisting of Modified PERL Cell and Modeled Si:Ge Solar Cell, Comparison with PERL Cell at One Sun Figure 3-17 Experimental and calculated reflection of DLAR and SLAR on the grooved surfaces Figure 3-18 Light Path Curves of Silicon Solar Cell with Inverted Pyramids and Double Layer AR Coatings Figure 3-19 Modified PERL Cell Current vs Thickness, Interested Photons Will Only Pass Silicon Material One Time Figure 3-20 Modified PERL Cell Current vs Thickness From 200um to 800um Figure 3-21 Efficiencies of 3 Terminal Si-Si:Ge Solar Cell with Modified PERL Cell and Modified Best Si:Ge Solar Cell Device at Different Concentrations and Different Silicon Thickness Figure 3-22 Efficiencies of 3 Terminal Si-Si:Ge Solar Cell with Modified PERL Cell and Modified Best Si:Ge Solar Cell Device at Different Concentrations and Different Silicon Thickness Figure 3-23 Efficiencies of 3 Terminal Si-Si:Ge Solar Cell with Modified PERL Cell and Modified Modeled Si:Ge Solar Cell Device at Different Concentrations and Different Silicon Thickness xii

13 Figure 4-1 A Silicon Solar Cell s Mask, Cell Area: 2mmx2mm; Busbar: Width 100um, Length 1.8mm; Finger: Width 10um, Length 1.8mm, Sf 0.39mm; Probing area: 300umx400um; Metal coverage: 9.75% Figure 4-2 An E-beam Evaporator Figure 4-3 RTP AG Associates Product Heat Pulse Figure 4-4 Figure 4-5 Semiconductor Parametric Analyzer ACS Basic Edition by KEITHLEY The standard air mass 1.5 spectrum compared with the spectrums from typical solar simulator sources ELH and Arc Lamp Figure 4-6 Sun Tracker by EKO Instruments Figure 4-7 5um Ge:Si with 92%Ge concentration was grown on Si. The N type Si cap was grown on top to create a hetero-junction and at the same time passivate the high Ge concentration SiGe surface Figure 4-8 SIMS measurement results of the first generation Ge:Si solar cell Figure 4-9 IV curve of Sample One Figure 4-10 IV curve of Si:Ge solar cell in 3T after cutting through and etching, behaves like two junctions in series connected Figure 4-11 Both Si emitter and Si:Ge epitaxial layers were grown by RPCVD Figure 4-12 Spreading Resistance Profile of silicon solar cell, its emitter doping density is about 5x10 17 cm Figure 4-13 Spreading Resistance Profile of Si:Ge solar cell, its emitter doping density is about 1x10 17 cm Figure 4-14 IV curve of a Si:Ge solar cell in 3T structure Figure 5-1 Pictures of Epitaxial Layer Grown on Unpolished Surface (left) and Polished Surface (right) xiii

14 Figure 5-2 Pictures of Epitaxial Layer Grown on Unpolished Surface (left) and Polished Surface (right) xiv

15 ABSTRACT Stand-alone transparent Si solar cell and low band gap Si:Ge solar cell can be applied together in a high performance multi-junction solar cell system to absorb the middle and long wavelength light. In this system, optics needs to be designed carefully for the back side of transparent Si solar cell to optimize the long wavelength light transmission. Photon losses cannot be eliminated in the stand-alone system. In this thesis a 3-terminal Si-Si:Ge monolithic tandem solar cell consisting of a top silicon solar cell and a bottom Si:Ge solar cell sharing a common base, which can take advantage of both the silicon solar cell and Si:Ge solar cells and overcome the current mismatch effect and photon losses between Si solar cell and Si:Ge solar cell is designed, modeled and fabricated. It can replace independent Si solar cell and Si:Ge solar cell in a multi-junction solar cell system. Modeling of 3 terminal Si-Si:Ge monolithic tandem solar cell is done to find out its maximum efficiency. A modified PERL cell is put on top of the modeled structure because it's the world record silicon solar cell. Its thickness is modified to analyze the effects of near band edge absorption. There are three cases for the bottom Si:Ge solar cell under modified PERL cell, the best fabricated device, product of the best measured parameters and the model of Si:Ge solar cell 14. Modeling shows that 3 terminal Si-Si:Ge solar cell outperforms PERL cell in most cases, and a thicker Si device leads to higher efficiency. The efficiency of 3 terminal Si-Si:Ge solar cell with 800um silicon substrate is 10% relative and 2.5% absolute higher than PERL cell at 500 suns. xv

16 Three groups of solar cells are fabricated. Performance of Si:Ge solar cells in this structure improve significantly due to better Si:Ge epitaxial layers growth conditions. The Si:Ge solar cell achieved with a Jsc of 3.26mA/cm 2, Voc of 189mV below the silicon filter under one sun without light trapping. Silicon solar cells in this structure have fine performances, the best one which has a POCl3 diffused emitter has a Jsc 22.9mA/cm 2, Voc 598.5mV and FF 77.3%. This work demonstrates that three terminal Si-Si:Ge monolithic tandem solar cell can achieve higher potential efficiency than standalone transparent silicon solar cells and Si:Ge solar cells and avoid current mismatching which exists in two terminal multi-junction solar cells. xvi

17 Chapter 1 INTRODUCTION TO THREE TERMINAL SI-SI:GE SOLAR CELLS Photovoltaics continues to be the world s fastest growing power generation technology due to increasing interests and demands of clean and sustainable energy. More than 100 countries already had renewable energy preferred policies by early The annual average increase of grid-connected PV capacity was about 60% from 2004 to Silicon dominates the materials supply markets because of its mature technology and ease of fabrication, and good efficiency. Optimized high efficiency silicon solar cells can be made both in laboratory and manufacturing plants. SunPower announced that a full-scale solar cell with an efficiency of 24.2% was produced at their manufacturing plant in the Philippines in , which is very close to the world record lab made cells (PERL) efficiency. Silicon s bandgap is 1.12eV, photons that have lower energy cannot be absorbed by silicon itself. Traditional single junction solar cells are very close to its theoretical efficiency limits due to its band gap limit. To surpass the limit, advanced concepts such as intermediate band, hot carriers, etc. are being developed 6. Another solution is tandem PV structure system which allows a wider portion of spectrum to be efficiently converted. 1

18 Megawatts 12,000 10,000 8,000 6,000 4,000 2, Figure 1-1 World Annual Solar Photovoltaics Production Multi-Junction Solar Cells Systems Multi-junction solar cells systems that have response to a broad range of spectrum are of great interest. Integrated optical system PV modules that operate at greater than 50% efficiency are being developed by the Very High Efficiency Solar Cell (VHESC) program 6. In the multi-junction system of Figure 1.2, sunlight is first split by a dichroic mirror, mid-energy photons are absorbed by InGaP and GaAs solar cells and low-energy photons are collected by Si and Si:Ge solar cells. A sum of the solar cells efficiency of 42.7% has been attained by such a system with spectrum splitting and light trapping 7. 2

19 Figure 1-2 A multi-junction solar cell system with standalone Si solar cell and SiGe solar cell 7 A standalone transparent silicon solar cell and a standalone Si:Ge solar cell are below the dichroic mirror in this system which are in charge of collecting midand low-energy photons. The transparent silicon solar cell is designed to absorb the spectrum from 850nm to 1100nm, and Si:Ge solar cell to absorb the wavelengths longer than 1100nm. 3

20 # photons 5.E+18 5.E+18 4.E+18 4.E+18 3.E+18 3.E+18 2.E+18 2.E+18 1.E+18 5.E+17 0.E wavelength nm Hi Bandgap Si Si:Ge Figure 1-3 Spectrum from 300nm to 850nm can be absorbed by high bandgap material; Spectrum from 850nm to 1100nm can be absorbed by Si; Spectrum from 1100nm to 1800nm can be absorbed by Si:Ge. To optimize the absorption of near band gap photons and the transmission of long wavelength photons, dopant density, thickness, AR coatings and surface texturing of transparent silicon solar cell need to be carefully designed. Free carrier absorption occurs when an absorbed photon fails to excite an electron from the valence band to conduction band. High dopant density will cause high free carrier absorption. To minimize the free carrier absorption effect, a base dopant density below 1x10 17 cm -3 (0.2 Ω-cm) is preferred 8. The thickness of the transparent silicon solar cell determines the amount of photons to be absorbed, more transmission means fewer electrons at the edge of its bandgap can convert at silicon voltage. Texturing helps 4

21 light trapping in silicon while reduces the transmittance of long wavelength photons. As shown in Figure 1.4 only around 60% of the photons in the wavelength range above 1200nm can transmit through the silicon solar cell with any kind of light trapping due to the reflection of the back inner surface of silicon cell. Thus optics losses here are too high to achieve high efficiency for low band gap solar cell. Figure 1-4 Transmittance of silicon substrate with planar or texturing surfaces 8 5

22 Fabrication complexity is another disadvantage of this structure. Since the transparent silicon solar cell and Si:Ge solar cell have to be fabricated separately, four surfaces have to be carefully treated. 1.2 Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells To simplify the above system, three terminal Si-Si:Ge monolithic tandem solar cells are being developed to replace standalone silicon solar cell and standalone Si:Ge solar cell in the multi-junction solar cell system as shown in Figure 1.3. This three terminal cell integrates a silicon solar cell and a Si:Ge solar cell on the same silicon substrate. A detailed description of this structure will be addressed in chapter 3. There are several potential advantages of this structure. First of all, optics losses between the silicon solar cell and Si:Ge solar cell can be avoided, since there is no reflection at the interface of silicon solar cell and Si:Ge solar cell. Thus inverted pyramids can be put on the front surface of silicon solar cell without reducing light transmitted to Si:Ge solar cell. Second, since there are only one device and two surfaces, fabrication of this solar cell is easier and its cost will be lower compared to two solar cells with four surfaces. Third, fewer surfaces mean less overall surface recombination which is a big factor that limits solar cell s performance. Fourth advantage of this three terminal structure comparing with 2 terminal tandem structure which will be explained in Chapter 3 is that it can avoid current mismatch between the upper silicon solar cell and the lower Si:Ge solar cell Thus both cells can operate at their maximum power point 9. 6

23 Figure 1-5 Integration of two stand alone cells into a 3 terminal monolithic tandem solar cell 1.3 Objectives of This Thesis 1. The primary objective is to design, model, fabricate, test and analyze three terminal Si-Si:Ge monolithic tandem solar cells so as to replace stand alone transparent Si solar cell and low band gap Si:Ge solar cell in a multi-junction solar cells system. 2. Another objective is to analyze the potential of this three terminal structure in contrast to an excellent IR sensitive Si solar cell, such as PERL cell. 7

24 1.4 Thesis Outline This thesis focuses on the design, modeling, fabrication, test and analysis of three terminal Si-Si:Ge monolithic tandem solar cells. Chapter 2 shows the basic theory of photovoltaics. PN junction, photovoltaic effects, quantum efficiency (IQE, EQE), open circuit voltage (V oc ), short circuit current (J sc ), fill factor (FF), lifetime (τ), and surface and bulk recombination (S r, S b ) will be introduced. Chapter 3 contains the design, modeling of three terminal Si-Si:Ge monolithic tandem solar cells. The design part includes the reason to design this structure, structure and equivalent circuit of 3 terminal Si-Si:Ge solar cell, and design of Si solar cell and Si:Ge solar cell. Model of three terminal Si-Si:Ge solar cell consists of a modified PERL cell on top and Si:Ge cell at bottom, modified PERL cell has different thicknesses, and Si:Ge solar cell has three cases, best device, products of best parameters and its model. Chapter 4 is the fabrication, test and analysis section. Details of fabrication process, testing method (test platform, pictures and equipments), the performance of these solar cells and analysis of this structure is described. future works. Chapter 5 is the final chapter which summarizes the work and suggests 8

25 Chapter 2 2 BASIC THEORY OF PHOTOVOLTAICS Photovoltaics is a way of converting solar power into electrical power in the form of direct current electricity. A solar cell is a semiconductor electronic device that produces both current and voltage when sunlight shines on its surface based on the photovoltaic effect, Semiconductor materials can absorb sunlight to create electron hole pairs; Light generated carriers will be collected by a PN junction formed in the material through doping; High energy electrons will move to an external circuit from the solar cell, dissipate its energy there and return to the solar cell. 2.1 PN Junction A PN junction is created by joining N type and P type semiconductors together. N type semiconductor has excess electron concentration and P type semiconductor has excess hole concentration. Electrons diffuse from N type area to P type area leaving behind positive charges in N type area near the interface of N type and P type material; Holes as the majority carriers in the P type semiconductor diffuse to N type area leaving behind negative charges in P type area near the interface N type and P type areas. An electrical field forms between these positive and negative charges, in the depletion region. The voltage due to the electrical field is built in voltage. The built in voltage will impede diffusion of both electrons and holes. Minority carriers (holes) in N type area drift to P type area through built in electrical field, while minority carriers (electrons) in P type area will drift to N type area. When at equilibrium state (no light illumination), there is no net current across the PN junction 11. 9

26 Figure 2-1 Cross Section of a Solar Cell 10

27 Figure 2-2 Band Gap of PN junction, Diffusion Current Equals Drift Current at Equilibrium State, No Net Current across PN Junction 2.2 Photovoltaic Effect When sunlight shines on the semiconductor material, photons with energy above the band gap will be absorbed and excite electrons from valence band to conduction band and corresponding holes to generate light generated carriers. The excess minority carriers generated in the P type and N type material will drift across the PN junction, light generated electrons in P type area will drift to N type area, and light generated holes in N type area will move to P type area by drift. In this way positive and negative charges are separated from each other. This is called the photovoltaic effect. This device is able to provide power once connected with an external circuit. 11

28 Figure 2-3 Band Gap of PN Junction under Illumination and Collection of Light Generated Current 2.3 Quantum Efficiency Quantum Efficiency (QE) is the ratio of carriers collected by a solar cell to the number of incident photons at a given wavelength. If incident photons at a particular wavelength are totally absorbed and all light generated carriers are collected, QE at this wavelength is 100%. QE is a powerful tool to analyze the performance of solar cells, since it allows us to locate performance of different part of solar cells. QE consists of blue response, middle energy response and red response corresponding to performance of front surface, bulk and back surface of solar cells. External quantum efficiency also includes optics losses such as reflection and transmission. 12

29 Figure 2-4 QE of a silicon solar cell. As indicated in the figure, response at different wavelength reflects the response at different area of solar cell, because blue light has very high absorption coefficient leading to quick absorption at the surface area of solar cell, while red light will be absorbed at the back of solar cell due to its low absorption coefficient Solar Cell Parameters Dark diode IV curve will shift down to the fourth quadrant to generate power when light shines on solar cell. 13

30 2.1 : dark reverse saturation current, : light generated current density. Figure 2-5 IV curve of a silicon solar cell Short circuit current (Isc) is the current when the solar cell is short circuited, which is the maximum current of a solar cell. It depends on the spectrum and intensity of the incident light, optical properties (reflection and absorption), area, absorption coefficients, collection probability of the solar cell

31 Where q is the electronic charge, W is the width of the device, coefficient, H 0 is the number of photons at each wavelength, is the absorption is the collection probability. Open circuit voltage (Voc) is the highest voltage of the solar cell, which happens when net current flow through the device is zero. The open circuit voltage equals to the amount of forward bias on the solar cell due to the bias on the solar cell junction with the light generated current At the operating points of both short circuit current and open circuit voltage, the power output is zero. Solar cells want to work at its maximum power point, and fill factor (FF) is the ratio of maximum power point to the product of short circuit current and open circuit voltage. Series resistance and shunt resistance will cause a lower fill factor Series and Shunt Resistance Parasitic resistances existing in the solar cells, like series resistance and shunt resistance, may affect the performance of solar cells. Series resistance mainly comes from resistance of the metal contacts, the contact resistance between metal and semiconductor, the resistance of bulk and emitter. High series resistance causes low FF and low current. Low shunt resistance resulting from manufacturing defects (like damage caused by cutting) may exist between emitter and base which provides an 15

32 alternate pathway for light generated current and it could cause a significant power loss. Figure 2-6 Parasitic series and shunt resistance in a solar cell circuit 16

33 Figure 2-7 Ideal IV curve (red) and IV curve with high series resistance and low shunt resistance (blue) of a silicon solar cell 2.6 Bulk and Surface Recombination There are three different mechanisms of recombination in the bulk of single crystal semiconductor: radiative recombination, Auger recombination 12, Shockley-Read-Hall recombination 13. Radiative recombination is also called band to band recombination which happens when an electron directly recombines with a hole and releases a photon. Auger recombination happens when an electron recombines with a hole in the conduction band, and released energy is give to another electron in the conduction band. Shockley-Read-Hall recombination involves defects with energy 17

34 states in the forbidden band. An electron moves down to defects level from conduction band, and a hole moves up to the same energy level and they recombine there. Bulk lifetime is directly related with these three recombination mechanisms in single crystal material. To make good solar cells long lifetime is favorable. 2.5 To make good solar cells, good bulk lifetime is required, so bulk recombination has to be minimized. If one semiconductor material is grown on another kind of semiconductor material, misfit dislocation which is another kind of defect in bulk may exist at the boundary of these two materials, because lattice constants are different between these two crystals, there will be dangling bonds in the lattice between two crystals with different lattice constant. The growth of Si:Ge epitaxial layer on top of Si crystalline will introduce lattice misfit dislocation as shown in the figure below. The more concentration of Ge in the Si:Ge layer, the bigger difference of lattice constants between Si:Ge layer and Si. To minimize misfit dislocation, a buffer layer with graded concentration can be grown, so that there is a slowly continuous change instead of a sudden change. 18

35 Germanium atom Silicon atom Figure 2-8 Misfit dislocations are generated at the locations where there are missing or dangling bonds in the lattice between two layers with different lattice constants. The step graded Ge:Si buffer layer can minimize negative effects of the large lattice mismatch between the Si substrate and the high Ge concentration Ge:Si layer 14. In addition to the bulk recombination, surface recombination is another important factor that limits solar cell performance. Periodic crystalline structure stops at its surfaces, this disruption introduces defects in the form of dangling bonds, so there will be very high recombination velocity at the surface. To minimize the surface recombination velocity, a passivation layer needs to be grown on top of the surfaces to tie up these dangling bonds, for example thermal SiO 2 is a good passivation layer for silicon surface. Si:Ge is difficult to passivate, a thin silicon layer is grown on the top of Si:Ge acting as a passivation layer which will be discussed in a later chapter. 19

36 Chapter 3 3 DESIGN AND FBRICATION OF THREE TERMINAL SI-SI:GE MONOLITHIC TANDEM SOLAR CELLS 3.1 Series Connected Two Terminal Si-Si:Ge Monolithic Tandem Solar Cell Two Terminal Si-Si:Ge Solar Cell Structure and Equivalent Circuit A series connected two terminal Si-Si:Ge monolithic tandem solar cell consists of a silicon solar cell on top which absorbs photons above its energy gap, and a low band gap Si:Ge solar cell at the bottom which absorbs low energy photons. It is a potential candidate to replace standalone transparent silicon solar cell and standalone Si:Ge solar cell. There are no optics losses between the silicon solar cell and the Si:Ge solar cell, because the Si:Ge solar cell s N type emitter layer is grown on the P type silicon substrate, light will go to Si:Ge epitaxial layer directly from silicon substrate without experiencing any optical surfaces. 20

37 Figure 3-1 Series Connected Two Terminal Si-Si:Ge Monolithic Tandem Solar Cell Structure (left ) and its Equivalent Circuit (right) Under illumination, minority holes from N silicon emitter will go through silicon PN junction, in this way they are collected by silicon solar cell. To be collected by Al contact, holes have to go across Si:Ge solar cell s PN junction. Minority electrons from P Si:Ge layer will be collected by Si:Ge solar cell once they go across Si:Ge PN junction. To power a circuit load, these electrons have to go across Si solar cell s PN junction. In other words, light generated current of the silicon solar cell has to go through Si:Ge solar cell before they can power the load, and vice versa. Light generated current of silicon solar cell has to go across the Si:Ge layer Drawbacks of Two Terminal Si-Si:Ge Solar Cell Under illumination, minority holes in Si solar cell emitter drift to Si base and become majority holes. After tunneling through the tunnel junction, they become 21

38 minority holes again in the emitter of Si:Ge solar cell. Then they have to experience another photovoltaic effect in Si:Ge solar cell before they can power a loading circuit. Thus overall current is determined by the solar cell which provides less current. Current match between two solar cells has to be considered in this case, which will limit the overall maximum power generation. This is a disadvantage of the series connected two terminal structure. While current match is considered in design of this solar cell under AM1.5G, series connected solar cells cannot always work at maximum power point, because the spectrum changes according to the weather, the location and the time of day and year. The band gap of Si:Ge has to be a set at a value so that current will match between two solar cells, in this way concentration of Ge in Si:Ge has to be a set value. While Ge concentration will determine Si:Ge solar cells performance, and growth condition and grown Si:Ge layer quality are determined by Ge concentration in Si:Ge layer, which means that good growth technology at another Ge concentration may not be an option. Due to these drawbacks, series connected two terminal Si-Si:Ge solar cell is not a good choice to replace standalone Si solar cell and Si:Ge solar cell. To overcome these problems of series connected two terminal Si-Si:Ge solar cells, A three terminal Si-Si:Ge monolithic tandem solar cell is designed in the following part, in which the silicon solar cell and Si:Ge solar cell work independently. 3.2 Design of Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells Structure of Three Terminal Si-Si:Ge Solar Cells 22

39 A three terminal Si-Si:Ge monolithic tandem solar cell is the integration of a standalone transparent silicon solar cell 14 and a low band gap Si:Ge solar cell 8. The top silicon solar cell and the bottom Si:Ge solar cell share a common silicon base and base contact. Plus separate emitter contacts of the top silicon solar cell and the bottom Si:Ge solar cell there are three contact terminals. Detailed descriptions of both top silicon solar cell and bottom Si:Ge solar cell are given in section and respectively. Figure 3.2 and 3.3 are 2D and 3D structures of the three terminal Si- Si:Ge monolithic tandem solar cell 9 ; Figure 3-2 A 2D structure of three terminal Si-Si:Ge monolithic tandem solar cell 23

40 Figure 3-3 A 3D structure of three terminal Si-Si:Ge monolithic tandem solar cell The top silicon solar cell and the bottom Si:Ge solar cell work independently. They do not need current match. This is a big advantage compared to two terminal monolithic tandem structure of which solar cells are series connected, and current match is required between solar cells to achieve high efficiency Equivalent Circuit An ideal equivalent circuit is shown in Figure 3.4 9, which does not take into consideration of both shunt and series resistance. 24

41 Figure 3-4 Equivalent circuits of an ideal three terminal Si-Si:Ge solar cell 9 Under illumination, mid- energy photons will be absorbed by the silicon material, and lower energy photons will be absorbed by Si:Ge material. Once photons are absorbed they will activate electrons from valence band to conduction band to generate light generated carriers in silicon and Si:Ge. Minority carriers (holes) generated in the emitters of both Si and Si:Ge solar cells are collected once they go across PN junction to P type silicon base or substrate. Then they will flow through the Al contact of P type Si base to load; Minority carriers (electrons) generated in the P type silicon base will be collected once they go across silicon PN junction to reach the N type silicon emitter, and they will flow to loading circuit through Ti/Pd/Ag contact; 25

42 The same way minority carriers (electrons) generated in P type Si:Ge epitaxial layer will be collected once they go across Si:Ge PN junction and reach the N type Si:Ge emitter, there they will flow to the load circuit through Ti/Pd/Ag contact, as seen in Figure 3-5. In this way, neither light generated current of silicon solar cell has to go through the PN junction of Si:Ge, nor light generated current of Si:Ge solar cell has to go through the PN junction of silicon solar cell. This is why these two solar cells can work independently. Figure 3-5 Minority carriers flows under illumination 26

43 3.2.3 Silicon Solar Cell Design Base selection and emitter formation are two important aspects in the initial stage of solar cell design. In this section, we will determine the dopant density and thickness of the P type Si substrate. Then we will discuss the emitter formation methods for both Si solar cell and Ge:Si solar cell Base Dopant Density and Thickness The design of silicon solar cells must maximize the absorption of mid energy photons and minimize the free carrier absorption to ensure maximum absorption of photons above its band gap and maximum transparency of photons below its band gap. Work is done in our group and it was found that factors can affect the transparency of silicon, such as dopant density and thickness of silicon 8,14. Table 3.1 is a conclusion of factors that affect transparency and solutions. Table 3-1 Physical Effect Impeding Transparency and Solutions 8 27

44 Free carrier absorption impedes transparency of greater wavelength. Free carrier absorption occurs when a photon is absorbed by the material but fails to activate an electron from valence band to conduction band. Studies 16,17,18 show that free carriers absorption, α fc, is related to dopant density N through fc K fc 2 N 3.1 Where K fc is a constant dependent on the free carrier effective mass and mobility and λ is the optical wavelength. Total transmission of a silicon device after taking into consideration of free carrier absorption is determined through the following relationship 3.2 device. R total reflection of silicon, α absorption of intrinsic silicon, t thickness of 28

45 Figure 3-6 Comparison of the Si absorption coefficient with cm -3 (black curve) and cm -3 (red curve) dopant concentrations 14. Kerestes 8 modeled and tested the transparency of 300um planar silicon devices with double layer AR coatings on both sides across a wide range of dopant concentration (Figure 3.7) 14. Both model and experimental results show that highly doped silicon (1x10 19 cm -3 ) has strong free carrier absorption which causes low long wavelength light transmission. However, when dopant concentration of the Si substrate is less than 1x10 17 cm -3 corresponding to resistivity 0.2Ω-cm, most long wavelength light will transmit through the silicon devices and reach the Ge:Si solar cells 8. Thus, the resistivity of the silicon substrate must be above 0.2Ω-cm, while as 29

46 discussed in Chapter 2, lower dopant concentration of the substrate is good for voltages of solar cells. A compromise between the electrical and optical performance must be considered to achieve good performance. Further analysis demonstrates that good Si solar cells can be obtained on the Si substrates with a resistivity in the range from 0.2Ω-cm to 1 Ω-cm. This resistivity will also provide good long wavelength light transmission. The resistivity of the silicon substrate we chose is Ω-cm. Figure 3-7 Dependence of silicon transmittance taking into account of free carrier absorption, which decreases transmittance with increasing doping concentration 8. 30

47 Silicon thickness will affect the absorption of mid energy photons. We want above band gap energy photons to be absorbed as much as possible so that they can convert into electrons at silicon voltage. A thicker silicon substrate has the potential to absorb more photons at first pass according to equation The thickness of silicon substrate we used is about 600um. 3.3 Where q is the electron charge, W is the thickness of device, α(λ) is the absorption coefficient, H 0 is the number of photons at each wavelength Emitter Formation Emitter of silicon solar cells can be formed in two ways, diffusion and epitaxial growth. Details of epitaxial growth (Reduced Pressure Chemical Vapor Deposition) will be addressed in the section of Si:Ge Solar Cells Design. Here we introduce our POCl3 diffusion. POCl 3 diffusion is a widely used technology for both commercial and lab solar cells to form a high performance emitter, such as the state of the art 25% efficiency PERL solar cells fabricated by UNSW 9. Phosphorus oxychloride (POCl 3 ) is a liquid phosphorus source used for diffusion of N-type regions into silicon substrate. POCl 3 oxidizes at normal process temperature and forms P 2 O 5. P 2 O 5 then reacts with silicon at the surface of the wafer to form elemental P and SiO 2. Elemental P will 31

48 diffuse into silicon wafer at a rate determined by the temperature of the furnace. The chemical reaction is as follows: 4POCl 3 + 3O 2 2P 2 O 5 + 6Cl P 2 O 5 + 5Si 4P + 5SiO Temperature, time and gas flow are three important parameters that will influence the yielded sheet resistances. Maintaining a constant temperature for the chemical and constant flow rate of the carrier gas through the bubbler, a uniform and repeatable sheet resistance can be achieved. Sheet resistance is more sensitive to temperature than time, higher temperature means higher diffusion rate of elemental P in silicon, yielding lower sheet resistance within the same period of time. Our diffusion temperature is around 880 to decrease in temperature may double or triple the sheet resistance. Carrier gas (N 2 ) flow rate and bubbler s temperature determine the amount of chemicals that can be sent into the furnace, faster carrier flow rate and higher bubbler temperature means more chemicals sent to the furnace. Oxygen also plays an importance role in determining sheet resistance, it reacts with silicon at the surface of wafers and P 2 O 5 to form a layer of SiO 2, which can prevent elemental P from diffusing into silicon. Figure 3.8 is a Spreading Resistance Profile of our typical diffusion which is performed at 880 for 26min, and the sample is P type <100>, resistivity is 0.86Ω-cm. 32

49 Concentration(cm-3) 1.00E+21 Spreading Resistance Profile 880C 26min 54Ω/sq 1.00E E E E E E depth(um) Figure 3-8 Spreading Resistance Profile of an N type silicon emitter diffused at 880 for 26min, and sheet resistance is 54Ω/sq Table 3-2 Silicon PN junction s Performances right after diffusion. Lifetime and J 0 were tested by Sinton lifetime tester, Implied Voc was tested by Sinton sun s Voc tester, and R-sheet was tested by four point probe tester. Samples Type Resistivity (Ω-cm) J 0 (A/cm 2 ) R-sheet (Ω/sq) Implied Voc(mV) Lifetime (us) FZ 200 P FZ x SEH P FZ x SEH P FZ 5-7 ~

50 Table 3.2 is the silicon solar cells performance right after diffusion. These samples are either high resistivity Ω-cm or low resistivity with very good quality (SEH). After diffusion, layers of PSG are formed on both sides of samples, which are very good passivation. During the testing, PSG was left on as surface passivation. To achieve a high performance diffusion, good quality wafers have to be selected, right cleaning steps prior to diffusion are important, diffusion recipes that can yield uniform and repeatable sheet resistance have to be developed. Both lifetime and implied Voc in Table 3.2 show that our diffusion process is successful. PSG layers have to be removed by HF dip in later steps, and thermal oxidation (SiO 2 ) can be grown on the silicon surface at high temperature which is almost as good as PSG layer in terms of surface passivation. In a word, diffusion is an accessible and good choice to form the emitter of silicon solar cell in the three terminal Si-Si:Ge monolithic tandem solar cell Si:Ge Solar Cell Design Three terminal Si-Si:Ge monolithic tandem solar cell integrates a transparent silicon solar cell 4 and a Si:Ge solar cell 14 into one structure as described previously. Si:Ge solar cell as the bottom solar cell absorbs the long wavelength spectrum as shown in Figure 1.3. The design, fabrication, test and analysis have been thoroughly studied 14 in our group. A 1.37% efficiency of Ge:Si solar cell below 200um Si at 30 suns has been achieved. The purpose of this thesis is to integrate his technology instead of improving the performance of Si:Ge solar cells. In this section, a 34

51 brief summary of design, fabrication and performance of Si:Ge solar cells in our group will be introduced Electrical and Optical Properties of Ge:Si Alloy To optimize the efficiency of Si:Ge solar cells below the silicon filter, a proper band gap of Si:Ge has to be determined. Band gap of Si:Ge alloy decreases from 1.12eV (silicon like band gap) to 0.66eV (Si:Ge like band gap) continuously as the Ge concentration increases as shown in Figure A sharp turn occurs at the Ge concentration around 85%. A higher band gap means higher potential voltage but less current density, and a lower band gap means lower potential voltage but more current density. A compromise between voltage and current has to be made so that an optimized power output can be achieved. As the Ge concentration increases, absorption coefficient curves of Si:Ge changes from silicon like curve to Ge like curve as shown in Figure

52 Figure 3-9 Si:Ge alloy band gap decreases as Ge concentration increases 21 Figure 3-10 Absorption coefficient of Si:Ge alloy at different Ge concentration at 300K 21 36

53 Electrical Performance of Si:Ge Solar Cells After finding the band gap and absorption coefficient, the performance of Si:Ge solar cells at different Ge concentration can be determined by first principles. 37

54 Figure 3-11 Performance of Ge:Si solar cells below Si varies with the Ge concentration 14 According to the performance curve of Figure 3.11, high efficiency Si:Ge solar cells can be achieved if the Ge concentration is above 85%. Five generation of Si:Ge solar cells, and the 5 th generation Si:Ge solar cells were developed and the 5 th generation achieved an efficiency 1.37% below silicon filter at 30 suns. Ge concentration of Si:Ge epitaxial layer is 88% except the first generation which is 92%. The fifth generation has a 5 um P type Si:Ge layer with a dopant concentration of 5x10 15 cm -3 and a 100 nm N type Si:Ge layer with a dopant concentration of 1x10 18 cm -3 to create a PN junction on a silicon substrate. resistivity of silicon substrate is Ω-cm. A 100 nm 5x10 18 cm -3 N type Si cap is on the top of N type Si:Ge layer as the passivation layer and conduction layer for emitter contact. A critical improvement to reach 1.37% from 0.79% (third generation) is the use of point back 38

55 contact, back inverted pyramids and back mirror in this structure, which effectively trapped light within this solar cell. Table 3-3 Performance of Five Generation Si:Ge Solar Cells 14. Significant improvement of Voc happens at the third generation, when growth condition was improved. Fourth and fifth generations implement optics to trap light. Condition Ge (%) At one sun below Si At 30 suns below Si Jsc (ma/cm2) Voc (mv) FF (%) Efficiency (%) First Generation Second Generation Third Generation Fourth Generation Fifth generation Model (Light trapping)

56 Figure 3-12 Structure of 5 th Generation single junction Si:Ge Solar Cells High Ge Concentration Si:Ge Layer Growth (RPCVD) One important reason that high efficiency Si:Ge solar cells were successfully developed in a short periods of time is the collaboration between AmberWave Inc. and our lab. AmberWave Inc. has mature technologies to grow precisely controlled, reproducible Ge:Si layer on Si with low defect densities and low thermal expansion by reduced pressure chemical vapor deposition (RPCVD). A good quality of Si:Ge layer is a key to achieving high performance. Figure 3.13 is the 40

57 equipment of RPCVD, and Figure 3.14 is SIMS measurement results of the first generation Ge:Si solar cell. Figure 3-13 ASM Epsilon TM E2000 Reduced Pressure Chemical Vapor Deposition System (RPCVD) 14 41

58 Figure 3-14 SIMS measurement results of the first generation Ge:Si solar cell grown by RPCVD, Ge concentration can be accurately controlled even in graded concentration growth. Oxygen concentration is acceptable Modeling of Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells In this section modeling of three terminal Si-Si:Ge solar cells is going to be discussed. The modeling structure consists of a modified PERL cell as a top cell and Si:Ge cell as the bottom cell. Modified PERL cell has different thicknesses, and Si:Ge solar cell has three cases, best fabricated device, products of best measured 42

59 parameters and model of Si:Ge solar cells 14. The difference between the modified PERL cell and real PERL cell is that modified PERL cell has Si:Ge layer at the backside, and its Al contact is on the front side. We assume material quality, surface passivation, light trapping, metal shading, doping density and other structure are all the same. Figure 3-15 Modeling Structure of Three Terminal Si-Si:Ge Solar Cell A summary table of efficiencies of PERL cell and 3 terminal Si-Si:Ge solar cell is shown below. This 3 terminal Si-Si:Ge solar cell consists of a PERL cell as the top cell with modified thickness, and a Si:Ge solar cell as the bottom cell. There are three cases of Si:Ge solar cell, best device, products of best parameters and model 43

60 to be the bottom cell. All three cases are from the dissertation of Si:Ge solar cells 14. This combination shows the maximum efficiency that this structure can achieve. Detailed analysis will be presented in a later part of this chapter to explain how these data are obtained. There are three cases of Si:Ge solar cell that will be investigated in the 3 terminal structure in this chapter, best Si:Ge device, products of best parameters of Si:Ge solar cells and modeled Si:Ge solar cells. Combination of PERL cell and best Si:Ge solar cell indicates the maximum efficiency based on our current technologies. Modeling of PERL cell and best parameters of Si:Ge solar cells shows the potential efficiency that can be achieved in near future using current technologies. Figure 3-16 shows that 3 terminal Si-Si:Ge solar cell can outperform PERL cell and a thicker device leads to higher efficiency. Table 3-4 Summary of Efficiencies of PERL Cell, Modeled 3T Si-Si:Ge Cell, Si Cell and Si:Ge Cell within the 3T Solar Cell Structure at One Sun Si Thickness Concentration 3T Si Cell 3T Si:Ge Cell 3T Total Eff PERL Cell 200um X % 3.07% 26.17% 24.96% 400um X % 2.75% 26.56% 24.96% 600um X % 2.59% 26.76% 24.96% 800um X % 2.49% 26.90% 24.96% 44

61 Efficiency 26.6% 26.4% 26.2% 26.0% 25.8% 25.6% 25.4% 25.2% 25.0% 24.8% Efficiency of 3T Si-SiGe Solar Cells with Different Thickness x1 3T Si-Si:Ge x1 PERL Cell Thickness (um) Figure 3-16 Efficiencies of 3T Si-Si:Ge Solar Cells Consisting of Modified PERL Cell and Modeled Si:Ge Solar Cell, Comparison with PERL Cell at One Sun Light Path Curve in Silicon Solar Cell with Inverted Pyramids on Top Surface Inverted pyramids can significantly decrease the reflection of at the front of solar cells and increase the effective path length of photons within solar cell 22. Reflection at the front surface reduces to 11% from 33% on a planar silicon surface, because inverted pyramids allow light to enter silicon more than one time. The AR coatings of PERL cell is composed of a 132 nm MgF2 layer whose refraction index is 1.383, a 56 nm ZnS layer whose refraction index is 2.33 and a 25 nm SiO2 layer is 45

62 below ZnS as the passivation layer. Optimized antireflection coatings on inverted pyramids can further minimize the surface reflection down to 1-2% 23. Figure 3-17 Experimental and calculated reflection of DLAR and SLAR on the grooved surfaces 23 46

63 Figure 3-17 shows that the reflection of PERL cell which surface has both inverted pyramids and AR coating is as low as 2% across a broad range of spectrum. Snell s law tells us reflection and refraction at the interfaces of different materials if their refractive indexes are different. Materials for antireflection coatings should be carefully chosen according to their reflective indexes; their thicknesses are also optimized so that they can minimize the reflection and maximize absorption. The following Figure 3-18 shows the light path curves, reflection and light path direction can be determined. Figure 3-18 Light Path Curves of Silicon Solar Cell with Inverted Pyramids and Double Layer AR Coatings 47

64 n 1 *sin(ө 1 )= n 2 *sin(ө 2 ) 3.6 Snell s Law: n 1, n 2 are the reflective index respectively, and Ө 1, Ө 2 are the angle of incidence and angle of refraction. When photons hit the surface of pyramids for the first time, they have two choices---refraction and reflection. Refracted light follows path curve 1 in the figure 3-18 according to Snell s Law, and an average of 90% refraction can be obtained according to both my calculation and other people s work 8,23. The angle between path curve 1 and silicon surface is After the first meet of photons and AR coating, the remaining 10% of the reflected light hits another surface of inverted pyramids. And a second time 90% refraction occurs. This leads to a total of 99% incident photons refracted in the solar cell. After entering silicon at second surface bounce, photons have two path curves to go, path curve 2 and path curve 3 as in figure But most of them (~90%) follow path curve 2 because of specific geometrical structure of inverted pyramids and AR coatings. The angle between path curve 2 and silicon surface is PERL Cell and Modified PERL Cell Performances We use world record PERL solar cell (25% efficiency under one sun) to predict J 0, which metal coverage is 4%. This J 0 is used to calculate open circuit voltage of modified PERL solar cells after light generated current is obtained use the same diode equation 2.1. We assume that fill factor holds. At this point efficiency can be calculated, which is shown in Table 3-5. Light generated currents are calculated by 48

65 adding generated current in path curve 1 and path curve 2 as shown in Figure The physical path length of path curve 1 is times of silicon thickness, and the physical path length 2 is 1.94 times of silicon thickness. The physical path length in PERL cell is multiple times of its physical thickness because light will reflect at the back surface so that silicon material has greater chance to absorb. But in the modified structure light that could be absorbed by silicon material will only pass the silicon material one time, because light will reach Si:Ge layer which has a strong absorption of infrared light. In this way no infrared light has the chance to be reflected back to silicon material. Photons ratio in path curve 1 equals (100%-metal coverage)* refraction ratio, that is (100%-4%)*90% = 86.4%. Photons ratio in path curve 2 equals (100%-metal coverage)*reflection ratio*refraction ratio = (100%-4%)*10%*90% = 8.64%. So both metal shading loss and reflection loss together is %-8.64% = 4.96%. Figure 3-19 and 3-20 show the relationship between modified PERL cells current and thickness. Thicker silicon absorbs more infrared photons to convert into current at silicon voltage. This modeling is to determine the thickness of silicon solar cell thus the amount of infrared photons that should be absorbed in order to achieve maximum overall efficiency of 3 terminal Si-Si:Ge solar cell. Table % PERL Cell Parameters 19 PERL Cell Concentration Voc(mV) Jsc(mA/cm 2 ) FF J0(mA/cm 2 ) Efficiency X % 5.575x % 49

66 Jsc current density (ma/cm2) 45 Modified PERL Cell Current vs Thickness Thickness (um) Figure 3-19 Modified PERL Cell Current vs Thickness, Interested Photons Will Only Pass Silicon Material One Time 50

67 Jsc current density (ma/cm2) 42 Modified PERL Current vs silicon thickness Thickness (um) Figure 3-20 Modified PERL Cell Current vs Thickness From 200um to 800um 51

68 Table 3-5 Silicon Cell PERL Cell 3T 200um Si Cell 3T 400um Si Cell 3T 600um Si Cell 3T 800um Si Cell PERL Cell and Modified PERL Solar Cell Performances with Different Thickness under Different Concentration Concentration Voc(mV) Jsc(mA/cm 2 ) FF Efficiency X % 24.96% X % 28.06% X % 30.63% X % 23.10% X % 25.98% X % 28.36% X % 23.81% X % 26.78% X % 29.23% X % 24.17% X % 27.18% X % 29.67% X % 24.41% X % 27.44% X % 29.96% Best Fabricated Si:Ge Solar Cell and Modified Best Fabricated Si:Ge Cell Performances The modeling of bottom Si:Ge solar cell is similar to that of top silicon solar cell. The best Si:Ge solar cell made in our group is selected, and then it s parameters are used to calculate efficiencies of modified Si:Ge solar cell performance as a bottom cell in the 3 terminal structure. As very good light trapping has been proved for Si:Ge solar cell application, perfect light trapping is assumed in this calculation, which means that all photons above its band gap filtered by silicon are absorbed by Si:Ge solar cell. Ge concentration is 88% which corresponds to 0.8eV band gap and 1550nm wavelength. The current is 54.55mA/cm 2 if all photons above 52

69 0.8eV are converted into current. This corresponding photons subtracts these reflected on the top surface and absorbed by silicon material is what Si:Ge solar cell gets. Table 3-6 Best Fabricated Si:Ge Solar Cell Performances Best Si:Ge Cell Concentration Voc(mV) Jsc(mA/cm 2 ) FF J0(mA/cm 2 ) Efficiency X % % Table 3-7 Best Fabricated Si:Ge and its Modified Cell Performances Si:Ge Cell Concentration Voc(mV) Jsc(mA/cm 2 ) FF Efficiency Best X % 0.94% Si:Ge Cell X % 1.38% 200um filter X % 1.74% Under 200um Si Under 400um Si Under 600um Si Under 800um Si X % 1.62% X % 2.34% X % 2.93% X % 1.45% X % 2.09% X % 2.63% X % 1.36% X % 1.97% X % 2.48% X % 1.30% X % 1.89% X % 2.38% 53

70 3.3.4 Overall Performances of 3 Terminal Si-Si:Ge Solar Cell Composed of Modified Best Fabricated Si:Ge Solar Cell and Modified PERL Cell After obtaining both modified PERL cell efficiencies and modified best fabricated Si:Ge efficiencies in this two junction three terminal structure with different silicon thickness and under different concentrations, overall efficiencies are easy to obtain. Comparison between PERL cell and this 3 terminal Si-Si:Ge solar cell determines the value of this structure. Table 3-8 Compare of Performances between PERL Cell and 3 Terminal Si- Si:Ge Solar Cell Si Thickness 200um 400um 600um 800um Concentration 3T Si Cell 3T Si:Ge Cell 3T Total Eff PERL Cell X % 1.62% 24.72% 24.96% X % 2.33% 28.31% 28.06% X % 2.92% 31.29% 30.63% X % 1.44% 25.26% 24.96% X % 2.09% 28.87% 28.06% X % 2.62% 31.85% 30.63% X % 1.35% 25.53% 24.96% X % 1.97% 29.15% 28.06% X % 2.47% 32.14% 30.63% X % 1.30% 25.71% 24.96% X % 1.89% 29.33% 28.06% X % 2.37% 32.33% 30.63% From the above table we can tell that efficiencies of 3 terminal Si-Si:Ge solar cell are higher than the PERL cell. Thicker silicon leads to higher overall efficiency, because more infrared photons are absorbed by silicon material which has 54

71 Efficiency higher band gap than Si:Ge. Higher concentration leads to bigger gaps between efficiencies of 3 terminal Si-Si:Ge solar cell and PERL cell. The biggest efficiency difference here is 1.7% which happens at 500suns with 800um silicon of PERL cell. 33% Efficiencies at Different Concentration with Different Si Thickness 32% 31% 30% 29% 28% 27% 26% 25% x1 modified PERL+Best Parameter SiGe x30 modified PERL+Best Parameter SiGe x500 modified PERL+Best Parameter SiGe x1 PERL Cell x30 PERL Cell x500 PERL Cell 24% Thickness (um) Figure 3-21 Efficiencies of 3 Terminal Si-Si:Ge Solar Cell with Modified PERL Cell and Modified Best Fabricated Si:Ge Solar Cell Device at Different Concentrations and Different Silicon Thickness 55

72 3.3.5 Performances of Best Measured Parameters Si:Ge Solar Cell, its Modified Cell Performances and Corresponding 3 Terminal Si- Si:Ge Solar Cell In this section best measured parameters of Si:Ge solar cell will be used to calculate its modified Si:Ge solar cell performance under different silicon thicknesses and at different concentrations. The product of best measured parameters represents the potential highest efficiency which has been fabricated so far. The calculation is the same as the above discussion. J 0 is calculated using the diode equation and best open circuit voltage and highest short circuit current. Then the current is calculated using the assumption that all photons above Si:Ge band gap are absorbed and converted into current, which is 12.85mA/cm 2 under 200um modified PERL cell. The best current of actual device so far is 8.14mA/cm 2 under 200um silicon filter. Table 3-9 Best Parameters Best Measured Parameters Si:Ge Solar Cell Performances Concentration Voc(mV) Jsc(mA/cm 2 ) FF J0(mA/cm 2 ) Efficiency X % % 56

73 Table 3-10 Best Measured Parameters Si:Ge and its Modified Cell Performances Si:Ge Cell Concentration Voc(mV) Jsc(mA/cm 2 ) FF Efficiency Best Parameter X % 1.49% Si:Ge Cell X % 1.95% 200um filter X % 2.33% Under 200um Si Under 400um Si Under 600um Si Under 800um Si X % 2.53% X % 3.27% X % 3.89% X % 2.26% X % 2.94% X % 3.49% X % 2.13% X % 2.77% X % 3.29% X % 2.04% X % 2.66% X % 3.16% Table 3-11 Performances of PERL Cell and 3 Terminal Si-Si:Ge Solar Cell with Best Measured Parameters Modified Si:Ge Solar Cell Si Thickness 200um 400um 600um 800um Concentration 3T Si Cell 3T Si:Ge Cell 3T Total Eff PERL Cell X % 2.53% 25.63% 24.96% X % 3.27% 29.25% 28.06% X % 3.89% 32.25% 30.63% X % 2.26% 26.07% 24.96% X % 2.94% 29.72% 28.06% X % 3.49% 32.72% 30.63% X % 2.13% 26.30% 24.96% X % 2.77% 29.95% 28.06% X % 3.29% 32.96% 30.63% X % 2.04% 26.45% 24.96% X % 2.66% 30.10% 28.06% X % 3.16% 33.12% 30.63% 57

74 Efficiency These performances have similar trends as what discussed in section Thicker silicon leads to higher overall efficiency, because more infrared photons are absorbed by silicon material which has higher band gap than Si:Ge. Higher concentration leads to bigger gaps between efficiencies of 3 terminal Si-Si:Ge solar cell and PERL cell. The biggest efficiency difference 2.5% happens at 500suns with 800um silicon of PERL cell. 34% 33% 32% 31% 30% 29% 28% 27% 26% 25% 24% Efficiencies at Different Concentration with Different Si Thickness Thickness (um) x1 modified PERL+Best Parameter SiGe x30 modified PERL+Best Parameter SiGe x500 modified PERL+Best Parameter SiGe x1 PERL Cell x30 PERL Cell x500 PERL Cell Figure 3-22 Efficiencies of 3 Terminal Si-Si:Ge Solar Cell with Modified PERL Cell and Modified Best Measured Parameter Si:Ge Solar Cell at Different Concentrations and Different Silicon Thickness 58

75 3.3.6 Performances of Model of Si:Ge Solar Cell, Modified Model of Si:Ge Cell Performances and Corresponding 3 Terminal Si-Si:Ge Solar Cell In this section model of Si:Ge solar cell will be used to calculate its modified Si:Ge solar cell performance under different silicon thicknesses and at different concentrations. The model represents the highest theoretical efficiency. The calculation is the same as the above discussion. J 0 is calculated using the diode equation and modeled open circuit voltage and modeled short circuit current. Then the current is calculated under the assumption that all photons above Si:Ge band gap are absorbed and converted into current. Table 3-12 Model of Si:Ge Solar Cell Performances Model Si:Ge Cell Concentration Voc(mV) Jsc(mA/cm 2 ) FF J0(mA/cm 2 ) Efficiency X % 5.597x % 59

76 Table 3-13 Model of Si:Ge and Modified Model of Si:Ge Solar Cell Performances Si:Ge Cell Concentration Voc(mV) Jsc(mA/cm 2 ) FF Efficiency Modeled X % 1.84% Si:Ge Cell X % 2.36% 200um filter X % 2.80% Under 200um Si Under 400um Si Under 600um Si Under 800um Si X % 3.07% X % 3.92% X % 4.62% X % 2.75% X % 3.52% X % 4.15% X % 2.59% X % 3.31% X % 3.91% X % 2.49% X % 3.18% X % 3.76% Table 3-14 Performances of PERL Cell and 3 Terminal Si-Si:Ge Solar Cell with Modified Model of Si:Ge Solar Cell Si Thickness 200um 400um 600um 800um Concentration 3T Si Cell 3T Si:Ge Cell 3T Total Eff PERL Cell X % 3.07% 26.17% 24.96% X % 3.92% 29.90% 28.06% X % 4.62% 32.98% 30.63% X % 2.75% 26.56% 24.96% X % 3.52% 30.30% 28.06% X % 4.15% 33.38% 30.63% X % 2.59% 26.76% 24.96% X % 3.31% 30.49% 28.06% X % 3.91% 33.58% 30.63% X % 2.49% 26.90% 24.96% X % 3.18% 30.62% 28.06% X % 3.76% 33.72% 30.63% 60

77 Efficiency 34.00% Efficienies at Different Concentration with Different Si Thickness 33.00% 32.00% 31.00% 30.00% 29.00% 28.00% 27.00% 26.00% x1 modified PERL+modified modeled Si:Ge x30 modified PERL+modified modeled Si:Ge x500 modified PERL+modified modeled Si:Ge x1 PERL Cell x30 PERL Cell 25.00% 24.00% Thickness um x500 PERL Cell Figure 3-23 Efficiencies of 3 Terminal Si-Si:Ge Solar Cell with Modified PERL Cell and Modified Model of Si:Ge Solar Cell Device at Different Concentrations and Different Silicon Thickness 61

78 Chapter 4 4 FABRICATION, TEST AND ANALYSIS OF THREE TERMINAL SI-SI:GE MONOLITHIC TANDEM SOLAR CELLS 4.1 Fabrication of Three Terminal Si-Si:Ge Monolithic Tandem Solar Cells After diffusion or epitaxial growth of both Si solar cells and Si:Ge solar cells on the silicon substrate, the rest of the processing steps are done in our lab, such as emitter patterning, contact deposition, isolation. HNA etching (HF: Nitric: Acetic 10:60:30) was used to pattern emitter areas of both Si and Si:Ge solar cells. Positive photoresist S1813 defined by photolithography and designed mask is used as protective layer against HNA solution. This etching step needs to be very careful, since S1813 cannot stand against HNA for a long time, to get a deep enough pattern etching may needs to repeat several times. Only the silicon emitter has to be defined at this moment because Al base contact has to be deposited on silicon side, Si:Ge solar cells will be separated automatically after cutting which will described later. Al contact areas are defined by thick negative photoresist NR9-3000PY to ease liftoff process, then 1.5-2um Al is deposited by E-beam evaporator. Ti/Pd/Ag (200nm/200nm/1.5um) emitter contacts are defined and deposited in the same way. Alignment of contacts is difficult due to the small size of both Si and Si:Ge solar cells, the emitter areas of silicon solar cells are 1.5mmx2mm 2mmx2mm. 62

79 Figure 4-1 A Silicon Solar Cell s Mask, Cell Area: 2mmx2mm; Busbar: Width 100um, Length 1.8mm; Finger: Width 10um, Length 1.8mm, Sf 0.39mm; Probing area: 300umx400um; Metal coverage: 9.75% 63

80 Figure 4-2 An E-beam Evaporator in Our Lab Then samples have to be cut into single devices. The final step is RTP annealing to make ohmic contact. High temperature processing has to be avoided because Si:Ge crystal quality is very sensitive to temperature. Our samples are usually annealed in the range of 300 to 500. The melting point of Al is 557, therefore neither Si-Al alloy nor back surface field will be formed during this annealing. RTP (rapid thermal processing) is a powerful tool to do annealing, which can accurately control annealing temperature, time and gas flow rate. Our samples are usually annealed in forming gas environment at 300 to 500 for around 20 to 120 seconds. 64

81 Figure 4-3 RTP AG Associates Product Heat Pulse 210 in Our Lab 4.2 Testing Equipments Introduction Solar cells can be measured both indoors and outdoors in our lab. Sunlight was simulated indoor by ELH and arc lamps which are widely used by other research groups 24. The spectrum of ELH lamp is smooth and stable, while arc lamp spectrum has too much noise, especially during 800nm to 1400nm which should be absorbed by the three terminal solar cells in the multi-junction solar cells system. Thus ELH lamp was chosen to simulate sunlight at one sun. Semiconductor Parametric Analyzer ACS Basic Edition by KEITHLEY is used to drive solar cells and obtain IV curves. 65

82 Figure 4-4 Semiconductor Parametric Analyzer ACS Basic Edition by KEITHLEY in Our Lab 66

83 Figure 4-5 The standard air mass 1.5 spectrum compared with the spectrums from typical solar simulator sources ELH and Arc Lamp used with permission 10 Sun Tracker by EKO Instruments is used for outdoor test, which can track the sun to make sure that our tested solar cells are normal to sunlight. 67

84 Figure 4-6 Sun Tracker by EKO Instruments in Our Lab 4.3 Test Results of Three Samples First Group Structure and Testing This is the first time fabrication of three terminal Si-Si:Ge monolithic tandem solar cells. The silicon solar cells emitters were diffused in our own diffusion furnace, which is capable to form very good emitters as mentioned previously, and the sheet resistance was about 45Ω/sq. Its emitter doping concentration profile should be close to the one in Figure 3.4, since they were diffused under very similar conditions. 68

85 Si:Ge epitaxial base layer was grown by RPCVD, Ge concentration is 91% in Si:Ge alloy, and P type Si:Ge base layer is about 5um thick, its emitter is 0.2 um N type silicon cap, also grown by RPCVD. The Si:Ge solar cell in this structure is the first generation Ge:Si solar cell as shown below 14. Figure 4-7 5um Ge:Si with 92%Ge concentration was grown on Si. The N type Si cap was grown on top to create a hetero-junction and at the same time passivate the high Ge concentration SiGe surface 14 69

86 Figure 4-8 SIMS measurement results of the first generation Ge:Si solar cell 14 The following table is the silicon solar cells performance in the three terminal Si-Si:Ge monolithic tandem solar cell structure. Good fill factors were a little bit surprise, since the distance between edge of emitter and base contact is 300um and distance between center of emitter and base contact is as far as 1.3mm. Good fill factors show that series resistance is actually very low. 70

87 Table 4-1 Silicon Solar Cells performance of 3-T structure in the 1 st group Samples J sc (ma/cm2) V oc (mv) FF Efficiency % 10.60% % 5.20% Figure 4-9 IV curve of Sample One The performance of Si:Ge solar cells in this sample was not good. At the beginning it was believed that there could be a problem of contact between Si:Ge and 71

88 metal. Then I made another group of three terminal structure solar cells, as well as some single junction Si:Ge solar cells. The IV curves of Si:Ge solar cells in three terminal structure were still very strange, and the IV curves of single junction Si:Ge solar cells were fine. To find out the reason of bad IV curves, some experiments were designed, one of them was to etching surrounding edge area of solar cells by HNA solution, since there would be damage after cutting. To do this experiment, a thick layer of photoresist S1813 was covered on both sides of solar cells and baked at 150 for 40mins. Then solar cells were separated with photoresist on both front and back surfaces. After that these cells were etched in HNA solution (HF: Nitric: Acetic 10:60:30) for 15 seconds. These samples were tested again, and Si:Ge solar cells in 3T structure had IV curves shown as below. It clearly showed that the IV curve behaved like two series PN junction (one is Si PN junction and the other one is Si:Ge PN junction). We got the conclusion that the etching of diffused N type silicon layer was incomplete, so the epitaxial Si:Ge PN junction was grown on a silicon PN junction instead of silicon substrate. The fabrication this time was not successful, so a second group was fabricated. 72

89 Figure 4-10 IV curve of Si:Ge solar cell in 3T after cutting through and etching, behaves like two junctions in series connected Second Group Structure and Testing The emitter of second group solar cells with a thickness of 1um and a doping density of 5x10 17 cm -3 were grown by RPCVD on a P type silicon substrate, then Si:Ge epitaxial layers were also grown at the same time on the other side of substrate, including a 5um 86.3% Ge P type Si:Ge base with doping density around cm -3, a 1um 86.3% Ge n type Si:Ge layer with doping density cm -3 and a 200nm n+ Si cap by AmberWave, Inc.. 73

90 Figure 4-11 Both Si emitter and Si:Ge epitaxial layers were grown by RPCVD Figure 4-12 Spreading Resistance Profile of silicon solar cell, its emitter doping density is about 5x10 17 cm -3 74

91 Figure 4-13 Spreading Resistance Profile of Si:Ge solar cell, its emitter doping density is about 1x10 17 cm -3 Table 4-2 Si solar cell and Si:Ge solar cell performance. These cells were tested as previously described in group one. Sample Test Condition J sc (ma/cm 2 ) V oc (mv) FF Si Solar Cell Si:Ge Solar Cell Outdoor 1 sun % Outdoor 1 sun No Data 16.7 suns % 75

92 The performance of the silicon solar cell is fair. Compared to the silicon solar cell in group one, the diffused emitter is a better choice. The performance of Si:Ge was a little bit surprise, because there were a lot of scratches on the Si:Ge epitaxial layer after growth which replicates any defect on the surface of silicon wafer. The emitter growth of this Si:Ge solar cell is good Third Group Structure and Testing The third group was fabricated to improve the performance of Si:Ge solar cells, since great improvement of efficiency of mono-junction Si:Ge solar cells 14 has taken place in our group owing to the improvement of growth condition by AmberWave Inc. as shown in Table 3.3. The 3 terminal Si-Si:Ge solar cells growth conditions, dopant concentration had been updated, according to the best single junction Si:Ge solar cell 14. The structure is the same as in group two. The Si:Ge layer was grown on the polished side of substrate. But silicon epitaxial layer was grown on the unpolished side of substrate. The emitter of silicon solar cell is 0.5 um thick with a doping density 1x10 18 cm -3, the P type Si:Ge base is 5 um thick with doping density 5x10 15 cm -3 and 88% Ge, and N type Si:Ge emitter layer is 1um thick with doping density 1x10 18 cm -3 and 88% Ge, above it there is a 200nm N type Si cap with a doping density 5x10 18 cm -3. The silicon substrate is 0.5 to 0.8Ω-cm. Rest fabrication steps and testing are the same as previously described. 76

93 Current(A) Table 4-3 Si solar cell and Si:Ge solar cell performance. These cells were tested as previously described in group one. Sample Test Condition J sc (ma/cm 2 ) V oc (mv) FF Si Solar Cell Outdoor 1 sun % One sun % Si:Ge Solar Cell Under Si filter No data 9.7 suns % A SiGe solar cell IV curve one sun 5.0E-4-2.0E Voltage (V) -5.0E-4-1.0E-3-1.5E-3 Figure 4-14 IV curve of a Si:Ge solar cell in 3T structure 77

94 Silicon solar cell s performance is not good due to the rough surface and low quality of silicon substrate on which N type silicon emitter was grown. As mentioned previously, the silicon epitaxial layer was grown on the unpolished surface of silicon substrate. The 0.5 um thin silicon emitter will duplicate the shape of the rough surface, growth across the surface is not uniform, so good growth cannot be guaranteed. Also rough surface leads to more surface area which means more recombination across surface. To find out the quality of silicon substrate, an experiment was designed to find the lifetime of both emitter and base. Lifetime of emitter and base are 3.3 us and 14.3 us respectively. Low lifetime means high recombination, short diffusion length and low collection probability. Si:Ge solar cell s performance is improved as expected. While its performance can be further improved, since the best mono-junction Si:Ge solar cell made by Yi with similar structure has a V oc of 263mV, J sc of 3.41mA/cm 2 and FF 61% below silicon filter under one sun. To make a good 3 terminal Si-Si:Ge solar cells, a good silicon substrate has to be selected, polished, high lifetime FZ material, resistivity close to 0.2Ω-cm. Polished substrate can provide good growth platform for Si:Ge epitaxial layer. Diffusion is a better choice to form PN junction once high performance diffusion is accessible, and thermal oxidation can be done right after diffusion. High open circuit voltage can achieved at this moment. Optics can be implemented in this structure, which will help light trapping thus increase current, and it will be described further in Chapter 5. 78

95 Chapter 5 5CONCLUSIONS AND FUTURE WORK 5.1 Summary of This Work This work focuses on the design and fabrication of three terminal Si-Si:Ge monolithic tandem solar cells. Three terminal Si-Si:Ge monolithic tandem solar cell combines a transparent silicon solar cell 8 and a high concentration Ge Si:Ge solar cell 14. The first fabrication yielded good silicon solar cells with a Jsc of 22.9 ma/cm 2, Voc of 598.5mV and FF of 77.3%, though the Si:Ge solar cell was not good. In the second group both silicon solar cells and Si:Ge solar cells had good performance., Si solar cell with a Jsc of 17.5mA/cm 2, Voc of 583.3mV, FF of 81.8%, Si:Ge solar cell with Jsc of 1.92 ma/cm 2, Voc of 180mV below silicon filter at one sun was achieved. The third group improved the Si:Ge solar cells performance to Jsc of 3.26mA/cm 2, Voc of 189mV below silicon filter under one sun. This work demonstrates that this structure can achieve higher potential efficiency, easier fabrication process, lower cost, lower optics losses, less surface recombination because of less surface areas, than transparent silicon solar cells and Si:Ge solar cells, and avoid current mismatching which exists in two terminal multijunction solar cells. 5.2 Future Work To further improve the performance of three terminal Si-Si:Ge solar cell, the following can be optimized including material selection, fabrication process, and optics implementation. 79

96 A good silicon substrate has to be selected which directly influences silicon solar cells performance. High lifetime FZ material with resistivity close to 0.2Ω-cm. High lifetime means less bulk recombination and higher carrier collection probability resulting in higher voltage as well as higher current density. 0.2Ω-cm is an optimized resistivity, which can lead to high potential voltage but also higher transparency of infrared light. Polished surface is a better platform for epitaxial growth than a rough surface. We have to make sure that growth happens on polished side if growth layers are in the range of nanometers. Figure 5-1 Pictures of Epitaxial Layer Grown on Unpolished Surface which is random reflection (left) and Polished Surface which is mirror reflection (right). POCl3 diffusion is a better choice to form silicon emitter as discussed in chapter 3. High performance diffusion is being developed in our labs. Thermal 80

97 passivation for silicon solar cells emitters can be done right after diffusion, which is the best passivation method accessible. Optics including antireflection (AR) coatings and pyramids are powerful tools to increase light generated current, since they can trap light within solar cells. AR coatings and pyramids can minimize the reflection at front surface and maximize the reflection at back surface, change pathways of photons inside solar cells, so that light will be trapped within solar cells which allow more absorption. A back side inverted pyramids with SiO 2 as mirror in mono-junction Si:Ge solar cell can increase the optic path length up to 17 time solar cells physical thickness 14. Figure 5-2 Configuration of Improved 3T Solar Cell 81