xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200409Ka 4500
controlfield tag 006 m d s
007 cr mn
008 040811s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000459
Optimization of the two stage process for Cu(In,Ga)Se solar cells
h [electronic resource] /
by Shirish Pethe.
[Tampa, Fla.] :
University of South Florida,
Thesis (M.S.E.E.)--University of South Florida, 2004.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 59 pages.
ABSTRACT: Copper Indium Gallium DiSelenide absorber layers are fabricated using a two stage manufacturing friendly process. The first step involves the sequential deposition of Copper and Gallium and co-deposition of indium and selenium at 275oC. This is followed by the second stage where the substrate is annealed in the presence of Selenium and a thin layer of copper is deposited to neutralize the excess Indium and Gallium on the surface to form the CIGS absorber layer. The top copper thickness as well as the time of deposition was varied to study the effect of Copper on the performance of the cells. Another recipe was developed for the precursor formation, where Gallium was co-evaporated with Indium and Selenium. A large bandgap shift was seen with this recipe and the open circuit voltage was increased. The performance of CIGS/CdS/ZnO solar cells thus fabricated was characterized using techniques like I-V, C-V, Spectral Response and EDS/SEM. Cells with open circuit voltages of 420-450 mV, short circuit currents of 33-38 mA/cm, fill factors of 58-62% and efficiencies of 9-11% were routinely fabricated.
Adviser: Don L. Morel.
x Electrical Engineering
t USF Electronic Theses and Dissertations.
Optimization Of The Two Stage Process For Cu(In,Ga)Se2 Solar Cells By Shirish Pethe A thesis submitted in partial fulfillment of the requirement for the degree of Master of Science in El ectrical Engineering Department of Electrical Engineering College of Engineering University of South Florida Major Professor: Don L. Morel, Ph.D. Christos Ferekides, Ph.D. Y.L.Chiou, Ph.D. Date of Approval: July 8, 2004 Keywords: cigs, photovoltaics, absorb er layer, co-evaporation, cis Copyright 2004, Shirish Pethe
ACKNOWLEDGMENTS I would like to thank my major profe ssor Dr. Don Morel for giving me an opportunity to work on this pr oject and also for his valuable guidanc e during the course of this research. I would also like to thank Dr. Chris Fereki des for agreeing to be on my committee and helping me in the lab whenever I faced problems. Also like to thank Dr. Chiou for being on my committee and for his guidance. I am also thankful to Mike and Trunk fr om the Electrical Engineering department for their timely help in all my problems. I would also like to thank the machine shop guys, Bob and others. I would like to thank my colleagues, Haris h, Venkatesh, and others in the lab that were always there to help me and made wo rking in the lab a wonderful experience. I would also like to thank my friends at USF that were always there to support me and made my stay at USF a memorable one. Last but not the least I would like to thank my family and friends back in India for their constant encouragement and support without which I would not be here.
i TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURE iv ABSTRACT vi CHAPTER 1 INTRODUCTION 1 1.1 Need for Solar Energy 1 1.2 Thin Film Solar Cells 2 1.3 CIS Thin Film Technology 3 1.4 CIGS Thin Film Technology 5 CHAPTER 2 SEMICONDUCTOR AND SOLA R CELL PHYSICS 8 2.1 Introduction 8 2.2 P-N Junctions 10 2.3 Heterojunctions 14 2.4 Solar cells 16 CHAPTER 3 DEVICE FABRICATION A ND CHARACTERIZATION 20 3.1 Device Structure 20 3.1.1 Substrate 21 3.1.2 Back Contact 21 3.1.3 Cadmium Sulfide 22 3.1.4 Front Contact 23 3.2 Device Fabrication 24 3.2.1 Back Contact Deposition 24 3.2.2 Absorber Layer Deposition 25
ii 184.108.40.206 Type I Precursors 26 220.127.116.11 Type V Precursors 28 3.2.3 Window Layer Deposition 30 3.2.5 Front Contact Deposition 30 3.3 Device Characterization 30 CHAPTER 4 RESULTS AND DISCUSSION 32 4.1 Effect of Bulk Copper 32 4.2 Effect of Venting on Jsc 34 4.3 Type V Precursors 36 4.3.1 Effect of Ga Content on Device Performance 37 4.3.2 Effect of Substrate Temperature 39 4.3.3 Effect of Selenization on Device Performance 40 4.3.4 Effect of Venting on Voc 42 4.4 ZnSe Buffer Layer 45 4.5 Effect of Silicon Nitride Layer 45 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS 47 REFERENCES 50
iii LIST OF TABLES Table 4.1 Effect of Cu Thickness on Devi ce Performance 33 Table 4.2 Overall Device Performance 34 Table 4.3 Device Performance With Respect to Run Sequence 40 Table 4.4 Overall Device Performance for Si3N4 Thickness Variation 46
iv List Of Figures Figure 1.1 Chronological Evolution of Small Area Solar Effi ciencies 2 Figure 1.2 A Ternary Cu-In-Se Phase Diagram 4 Figure 1.3 Chalcopyrite Structure of CIGS 5 Figure 1.4 Change in Bandgap as Function of Ga Content 6 Figure 2.1 Simplified Band Diagram of Semiconductor 10 Figure 2.2 Space Charge Distribution 11 Figure 2.3 Electric Field and Potential Distribution 12 Figure 2.4 Band Bending in P-N Junction 13 Figure 2.5 Energy Band Diagram of Isolated Semiconductor 15 Figure 2.6 Energy Band Diagram of Heterojunction 15 Figure 2.7 Ideal I-V Curve of Solar Cell 16 Figure 2.8 Energy Band Diagra m of Solar Cell under Illumination 17 Figure 2.9 Equivalent Circuit of Sola r Cell 18 Figure 2.10 Effect of Series and Shunt Resi stances 19 Figure 3.1 Basic Structure of CIGS Solar Ce ll 20 Figure 3.2 Wurtzite Structure of CdS 22 Figure 3.3 Source Arrangement for Type I Pr ecursors 26 Figure 3.4 Temperature Profile of Selenization for Type I Precursors 27 Figure 3.5 Source Arrangement for Type V Precursors 28 Figure 3.6 Temperature Profile of Sele nization for Type I Precursors 29 Figure 4.1 Voc Distribution in Run P044, P049, P059 33 Figure 4.2 Spectral Response for P007, P008 and H299 34 Figure 4.3 Voc and Fill Factor Distribution for P056 35 Figure 4.4 I-V Characteristics of Device P056-11 36 Figure 4.5 Spectral Response for P056 and P065 37 Figure 4.6 Voc Distribution for Ga Content 38
v Figure 4.7 Voc Distribution for Sample P071 39 Figure 4.8 Spectral Responses for Samples P068 and P071 40 Figure 4.9 Voc Distribution for Samples P071 and P072 41 Figure 4.10 Voc Distribution for Samples P067 and P069 41 Figure 4.11 Voc Distribution for Sa mples P065, P066 and P069 43 Figure 4.12 Bandgap Shift for Run Sequence 43 Figure 4.13 I-V Characterist ics of Device P072-08 44 Figure 4.14 Voc Distribution for Sample P007 45
vi OPTIMIZATION OF THE TWO STAGE PROCESS FOR Cu(In,Ga)Se2 SOLAR CELLS Shirish Pethe ABSTRACT Copper Indium Gallium DiSelenide absorber layers are fabricated using a two stage manufacturing friendly process. The first step involves the sequential deposition of Copper and Gallium and co-deposition of indium and selenium at 275oC. This is followed by the second stage where the substrate is anne aled in the presence of Selenium and a thin layer of copper is deposited to ne utralize the excess Indium and Gallium on the surface to form the CIGS absorber layer. Th e top copper thickness as well as the time of deposition was varied to study th e effect of Copper on the pe rformance of the cells. Another recipe was developed for the precursor formation, where Gallium was co-evaporated with Indium and Selenium. A la rge bandgap shift was seen with this recipe and the open circuit voltage was increased. The performance of CIGS/CdS /ZnO solar cells thus fabr icated was characterized using techniques like I-V, C-V, Spectral Response and EDS/SEM. Cells with open circuit voltages of 420-450 mV, short circ uit currents of 33-38 mA/cm2, fill factors of 58-62% and efficiencies of 9-11% were routinely fabricated.
1 CHAPTER 1 INTRODUCTION 1.1 Need For Solar Energy Development of clean energy resources as an alternative to fossil fuels has become one of the important challenges fo r modern science and technology in the 21st century. Reduction of air polluti on and preservation of the ecol ogical systems of earth are the main motivation factors behind these huge efforts. Some of the major contenders for the alte rnative clean resources are hydroelectric, tidal, nuclear and solar energy. But of all th ese renewable energy sources photovoltaic is the most promising one as a future ener gy technology. The photovoltaic effect is the process of direct conversion of sunlight into el ectrical energy. This effect was first observed by Henri Becquerel in 1839. The majo r advantage of these photovoltaic systems is that they tap an almost inexhaustible resource that is free of charge and available everywhere in the world. Even though the photo voltaic effect was first observed in 1839, the first practical cell wa s developed by Chapin et.al at Bell Labs in 1954.  1.2 Thin Film Solar Cells The main barrier impeding the expansi on of the terrestrial application of photovoltaic is the high price of the solar cell modules. The r easons for these high costs are the high processing and material costs. A recent survey suggests that a further cost
2 reduction to one fifth of the present leve l is a must for photovoltaics to match the conventional electricity price. On e of the solutions to achieve a reduction in this cost is the development of thin film solar cells. This cost reduction is due to factors like lower material use, fewer and simpler processing step s as compared to a traditional wafer based crystalline silicon solar cell. The thin film technologies use simp ler device processing techniques like evaporation, sputtering, etc. for large area m odules and arrays. The designs of most thin film technologies have common processing techniques so the ma nufacturing costs are very similar. Hence, the choice of any t echnology is based on the factors like highest achievable conversion efficiency, reliability, availability of materials and environmental concerns. At present, research is done on the alloys of amorphous Silicon, CdTe, and CIS as major contenders for large scale produc tion. Figure 1.1  shows the chronological evolution of small area solar cell effi ciencies in these three technologies. Figure 1.1 Chronological Evolutions of Small Area Solar Efficiencies
3 1.3 CIS Thin Film Technology CuInSe2 (CIS) is a ternary alloy belonging to the I-III-VI2 compound semiconductor group. Its lattice elements are te trahedrally coordinate d like all adamanite semiconductors. CIS has a bandgap of 1.0 eV a nd one of the highest reported absorption coefficient of 3.6E5 cm-1. The properties of CIS strongly depend on its composition, and the doping of CIS is controlled by the intrin sic defects. Hence, we have either p-type or n-type CIS if the material is grown Cupoor and p-type if Cu-rich film is grown. The Cu-rich films are highly conductive due to acceptor defects (c opper atoms on indium sites). The n-type In-rich f ilms are highly resistive due to the donor defects (indium on copper sites) and acceptor defects (copper vacanci es) occurring at the same time. Some of the recent studies have observed a thin surface layer of CuIn3Se5. This is termed as the ordered vacancy compound (OVC), a defect chal copyrite which is slightly In rich . The OVC has a bandgap of 1.3 eV. As compared to binary compounds the ternary compound CIS is more tolerant to small devi ations from stoichiometry. The composition needs to be slightly In rich in order to achieve good device performance. As the metal ratio (I/III) approaches unity the carrier concentration increases which results in better cell performance. This could be a ttributed to the formation of Cu2Se grains. This behavior is explained based on the Cu-In-Se ternary phase diagram  shown in figure 1.2. In Se rich films, both Cu and In vacancies yield strongly p-type films. Simultaneous formation of oppositely charged defects in the In rich films may lead to a highly compensated, highly resistive n or p-type films.
4 Figure 1.2 A Ternary Cu-In-Se Phase Diagram The first thin film CIS solar cell was produced by the evaporation of CIS as a compound . CIS films can be produced by co -evaporation or sequential evaporation of elemental sources. In co-evaporated CIS film s, the composition of the materials with regard to the metal ratios corresponds to th e evaporation rates whereas in sequentially evaporated CIS films, it corresponds to th e thickness. Co-evaporat ion of the elemental sources gives better device quality films as co mpared to sequentially evaporated films. But for commercial purposes sequential evaporation is preferred over co-evaporation because it requires lesser contro l over the evaporation rates. Some of the other techniques being researched for deposition of CIS film s are electro deposition, screen printing, particle deposition, sputtering, MBE, and MOCVD.
5 1.4 CIGS Thin Film Solar Cells If the bandgap of the thin film solar cell is increased it increases the band-bending which causes an increase in the open circuit voltage (Voc) but also decreases the short circuit current (Jsc) as fewe r photons are absorbed in the long wavelength range and so fewer carriers are generated. Thus there is a current-voltage tradeoff. If the bandgap of the absorber layer is increased, the doping of the window layer which is ZnO can be increased to reduce the resistiv e losses without the problem of free carrier absorption, as current is no longer generated in the infrar ed region. This bandgap increase can be caused by alloying a quaternary compound of Cu (In,Ga)Se2 (CIGS). The chalcopyrite structure of the CIGS compound is shown in figure 1.3. Th is is done by alloying CIS and CGS. As the bandgap of CGS (1.7 eV) is higher than that of CIS (1.0 eV) we get a bandgap increase. Figure 1.3 Chalcopyrite Structure of CIGS
6 The bandgap of the composition Cu In1-xGaxSe2 is given by  x Eg 71369 0 0032 1 (1.1) The variation of bandgap with Gallium content is shown in figure 1.3. Elemental coevaporation has been used to vary the bandga p of CIGS films by adjusting the Ga content over the full range of composition . One of the other advantages of Gallium incorporation in CIS films is better adhesion of the absorber film to the molybdenum back contact. Also Gallium incorporation can cause a graded bandgap which leads to better colle ction of charge carriers. Gallium has the tendency to segregate to the back of the devi ce which causes abrupt changes in the carrier concentration near the back contact. This lead s to an electric field which acts as minority carrier mirror and reflects all the minority ch arge carriers. This el ectric field is called a Back Surface Field (BSF). The BSF also prev ents any recombination of electron-hole pairs at the back contact. Figure 1.3 Change in Bandgap as Function of Ga Content
7 It has been observed that too much Gallium has adverse affects as there are some other changes in the material properties like electron and hole affinities, carrier concentration, defects, etc. which hinder cell performance. The best device efficiencies have been achieved for a bandgap of around 1. 1 eV and it has been found out that the device efficiencies start fa lling for bandgap greater than 1. 3 eV and x greater than 0.5 because of the reducing fill factors and JscÂ’s [9 ]. This could be due to the problems with the hetero-interface resulting in an increase in the interface recombination. The highest reported efficiency for CIGS so lar cells is 18.8% by NREL .
8CHAPTER 2 SEMICONDUCTOR AND SO LAR CELL PHYSICS 2.1 Introduction A solar cell can be defined as a p-n junction diode, with incident light being the excitation factor or the one generating th e charge carriers. Hence, semiconductor fundamentals are essential to understand th e behavior and operation of solar cells. Any element can be classified into a me tal, an insulator, and a semiconductor. A metal is one with a large number of fr ee electrons and so has high conductivity. An insulator has absolutely no free electrons i.e. th e valence band is completely filled and the forbidden energy gap is also very large. The conductivity of semiconductors is between these two. They have fully occupied valenc e band, at T=0, but the forbidden energy gap is not very large, due to this they do not conduct in normal conditions but these valence electrons are loosely bound and so any exc itation causes them to break the bond which results in conduction. Semiconductors can be classified as intrinsic and extrinsic semiconductors. An intrinsic semiconductor is extremely pure and has very small amount of impurities. When a semiconductor is doped with donor or acceptor impurities it, forms an extrinsic semiconductor. When a semic onductor is doped with a donor impurity it has an extra electron which makes it an n-type semiconductor. Similarly when an acceptor
9 impurity is incorporated; it gives rise to an extra hole which makes it a p-type semiconductor. The energy levels for an atom are discrete but they are so closely spaced that they form continuous energy bands. In any semic onductor there is a forbidden energy region in which allowed states cannot exist. The energy band above th is energy gap is called the conduction band and the one below is called th e valence band. And the forbidden energy region is called the Energy Bandgap (Eg) which is one of the most important parameters in semiconductors. The simplified energy band diagram of a semiconductor is shown in figure 2.1. The Fermi level is defined as the energy level below which all states will have at most two electrons of opposite spin according to PauliÂ’s exclusion principle . It can be also defined as the energy level where the probability of finding an electron is . The Fermi distribution function is given by ()/1 () 1FEEkTfE e (2.1) where k is the boltzmanÂ’s constant, T is the absolute temperature, EF is the Fermi energy. For an intrinsic semiconductor the Fermi le vel lies approximatel y in the middle of the bandgap. This Fermi level shifts toward s the conduction band for n-type and towards the valence band for p-type semiconductors. The Fermi levels can be calculated using the equations given below . For n-typeln(/)FCDCEEkTNN (2.2) For p-type ln(/)VFAVEEkTNN (2.3) where k is the boltzmanÂ’s constant, T is the absolute temperature, EF is the Fermi energy, EC is the energy level at the bottom of the conduction band, EV is the topmost energy
10 level of the valence band, NC and NV is the effective density of state in the conduction band and valence band respectively, ND and NA is the donor and acceptor concentration respectively. 2.2 P-N Junctions When two isolated p-type and n-type materi al are electrically connected they form a p-n junction. These p-n junctions are very important in understanding other semiconductor devices. The p-n junctions ca n be classified as homojunctions and heterojunctions. Homojunctions are the ones which have the same semiconductor with different doping, whereas heterojunctions are junctions of two different semiconductors. The p-n junctions can also be classified as abrupt junctions and graded junctions. When region there is an abrupt change in the impur ity concentration across the junction it is an
11 abrupt junction. In the graded junction, the ND-NA varies over a distance across the junction. Figure 2.2 Space charge There is an initial flow of free electrons from the n-type re gion to the p-type region and free holes from the p-type to n-t ype. This causes the formation of a depletion region which opposes the further flow of ch arge carriers. This depletion region has immobile charges gives rise to the space ch arge which is shown in figure 2.2. The resulting electrical field and th e potential distribution are shown in the figure 2.3 (a), (b). Under thermal equilibrium the electron a nd hole current densities are given by ../0 ../0nnF ppFJnEx JpEx (2.4) where n and p is electron and hole mobility respectively, EF is the Fermi energy, n and p is the electron and hole c oncentration respectively.
12 (a) (b) Figure 2.3(a) Electric field distri bution, (b) Poten tial distribution
13 Hence for zero net electron and hole current densities we require that the Fermi level should be a constant th roughout the sample. This cause s a bending of the bands of the semiconductor resulting in the potential being developed. Figure 2.4 shows the band bending of the semiconductor. This poten tial is called the built in voltage (Vbi). This potential is given by 2ln(./)biADiVkTNNn (2.5) where k is the boltzmanÂ’s constant, T is the absolute temperature, NA and ND is the acceptor and donor concentration respectively, ni is the intrinsic concentration. Figure 2.4 Band Bending in P-N Junction Again considering the thermal equilibrium c onditions, the electric field in the neutral region should be zero, which means the total negative charge on the p side, should cancel the total positive charge on the n side. Thus we get the equation  ..ApDnNxNx (2.6) where NA and ND is the acceptor and donor con centration respectively, xp and xn are widths of the depletion region on the p a nd n side of the junction respectively.
14 The total space charge region width is given by  D A bi D A sN qN V V N N W ) )( ( 2 (2.7) For an one sided abrupt junction (one side more heavily doped than the other) the above simplifies to B bi sqN V V W ) ( 2 (2.8) where NB is the impurity concentrati on of the lightly doped side. 2.3 Heterojunctions A heterojunction is a junction formed between two dissimilar semiconductor materials. Heterojunctions are of two type s, isotype and anisot ype heterojunctions. Isotype heterojunctions are the one formed with materials of same conductivity and anisotype are formed with materials of di fferent conductivities. The band bending of heterojunctions is not as smooth as the homoj unctions because of difference in electron affinities, work functions, etc. These cause interface states or band discontinuities which form the recombination centers. The ener gy band diagram of two semiconductor materials prior to forming a junction is show n in figure 2.4. The two semiconductors have different bandgaps Eg1 and Eg2, different work functions m1 and m2, different electron affinities 1 and 2 and different permittivities 1 and 2. The energy band diagram of a heterojunction is shown in figur e 2.5. The band discontinuities are given by 12CE (2.7) 21()VCEEgEgE (2.8)
15 Figure 2.5 Energy Band Diagrams of Isolated Semiconductors The total built in potential, Vd, is equal to the sum of partial built in voltages Vd1 and Vd2, where Vd1 and Vd2 are the electrostatic po tentials of the two semiconductors. Most of the thin film solar cells are heterojunction based. Figure 2.6 Energy Band Diagram of Heterojunction
162.4 Solar Cells When a solar cell is exposed to a solar spectrum, the photons with energy greater than Eg are absorbed and the material transmits those with energy less than Eg. Hence if we know the energy bandgap of the semic onductor then we can know the wavelength range of light that will be absorbed by the semiconductor using the equation 1.24/ Eg (2.9) where Eg is in eV and is in microns. These absorbed photons result in the gene ration of electron-hole pairs (EHP); these charge carriers diffuse to the space charge region and are th en swept across the junction if not recombined. These diffused carriers give ri se to photocurrent. The ideal I-V curves of a solar cell under dark and illuminated conditions are shown in figure 2.6. Figure 2.7 Ideal I-V Curve of a Solar Cell
17 In the dark the solar cell is a simple p-n junction diode and can be given by the diode equation  /qVkTIIoe (2.10) Where Io is the diode saturation current, V is the applied bias, k is the boltzmanÂ’s constant, T is the absolute temperature. Under illuminated conditions ther e is an extra curre nt due to the photo generated current; therefore the above equation changes to  /qVkT L I IoeI (2.11) where IL is the photo generated current. As can be seen in the figure 2.7, the built in voltage of the p-n junction changes when light is incident on the solar cell. Th is is due to excess carriers being generated which result in net charges on the n and p side. Figure 2.8 Energy Band Diagram of Solar Cell under Illumination This gives us the open circuit voltage (Voc) which appears at the contacts and can be found from the above current equation 
18 ln~lnLL OOkTIkTI Voc qIqI For I=0 (2.12) The equivalent circuit of an id eal solar cell as well as in the presence of series and shunt resistances acting on the cell under illuminatio n is shown in the figure 2.8. Here the current source is equal to IL which is the photo generated current. The series resistance, Rs, is the combination of the bulk resistance of the semiconductor, the bulk resistance of the meta llic contacts and the contact resistance between the contacts and the probe. The shunt resistance, Rsh, reduces with the leakage current in the p-n junction. The series resistan ce is given by the reciprocal of the slope of the I-V curve when the solar cell is forward biased. The s hunt resistance is found out by taking the reciprocal of the sl ope of the I-V curve when the solar cell is reverse biased. (a) (b) Figure 2.9 Equivalent Circuit of Solar Cell (a ) Ideal, (b) with Series and Shunt Resistance
19 Figure 2.10 Effect of Seri es and Shunt Resistance The effect of the series and shunt resi stance is shown in figure 2.9. The I-V characteristics are given by  () ln[()1]Lss OOsh I VIRqVIR I IIRkT (2.13) The short circuit current, Isc is defined as the current fl owing through the junction at zero bias. The fill factor is the measure of th e squareness of the I-V curve at the maximum power point and is given by  mm ocscVI FF VI (2.14) where Vm and Im is the voltage and current of the maximum power rectangle. The conversion efficiency of the solar cell is given by /(..)/minscocinPPFFIVP (2.15) where Pm is the maximum power and Pin is the incident power.
20CHAPTER 3 DEVICE FABRICATION AND CHARACTERIZATION 3.1 Device Structure The CIGS solar cell fabricated for this research has the structure Soda lime glass/CIGS/CdS/ZnO. This structure is s hown in figure 3.1. The substrate and the fabrication processes used here are those which make the sola r cells cost effective. The general process temperat ures are around 550-600 C. Figure 3.1 Basic Structure of the CIGS Solar Cell
213.1.1 Substrate The substrate used in this research is soda lime glass (SLG) for the main reason that it is cheap and easily available. Glass is corrosion resistant and another reason that SLG is used is the hypothesis that the di ffusion of sodium from the glass seems to improve the cell performance. It improves the fill factors and the Voc. But this incorporation of sodium adds an uncontrollable factor in determining the reproducibility of the cell performance. The other issue with th e soda lime glass substr ates is that, at high temperatures glass softens and finally warps. 3.1.2 Back Contact The most widely used back contact in CIGS solar cells is molybdenum (Mo). Mo forms a good ohmic contact and offers hi gh resistance to selenium corrosion. The thickness of the Mo layer is around 1 micron and is de posited using DC magnetron sputtering. The sputtering pressures determin e the resistivity and adhesion of the films . Films deposited at high pressures have poor resistivity due to a rougher surface which results in the growth of rougher absorber layers. But th ey adhere to the glass much better. Films deposited at lower pressures ha ve a smoother surface which results in lower resistivity but they suffer from adhesion problem due to compressive stresses. Due to these reasons, a bilayer process is used in th e deposition of the Mo film. The first layer is deposited at higher pressures for better adhesi on of the film to th e glass and the second layer is deposited at lower pressures for better electrical resistivity. The absorber layer is grown on the Mo film. The properties of the ab sorber layer which is CIGS are discussed in chapter 1.
223.1.3 Cadmium Sulfide (CdS) CdS forms the n-type partner for the p-t ype CIGS absorber layer in forming a p-n junction. Polycrystalline CdS has a wurtzite structure as shown in figure 3.2. CdS is a direct bandgap semiconductor with a bandga p of 2.4 eV. Due to this high bandgap, it absorbs the high energy photons of the visi ble spectrum. Basically the CdS acts as a window layer for the low energy photons for the absorber layer to absorb them. CdS can be deposited using different pr ocessing techniques like chemical bath deposition (CBD), closed space sublimation (C SS), and RF sputtering. CBD is the most widely used technique for CdS deposition as the films grown by CBD are much more uniform and smoother as compared to the other techniques. The CBD process uses cadmium acetate as the cadmium source, thiourea as the sulphur source, and ammonium hydroxide as a complexing agent. The depositi on temperature and the pH of the solution Figure 3.2 Wurtzite Structure of CdS determine the growth rate of the film  Heterogeneous nucleation gives uniform films whereas homogeneous nucleation gives powdery deposits.
23 As some of the light is absorbed in the CdS, a current loss of approximately 6mA/cm2 is estimated . Cd is a toxic material causing environmental concerns which has led to the investigation of other mate rials to replace CdS as a window layer. The other materials considered are ZnSe, ZnInxSey and InxSey . Also ZnO has been tried as a window layer . 3.1.4 Front Contact The main requirements for a front contact are high conductivity and good transparency. ZnO is the most widely used ma terial for front contact in CIGS solar cells as it has very high bandgap of 3.3 eV and has g ood optical and electrical properties. It is typically n-type and can be deposited by diffe rent processing techniques like sputtering, CVD, etc. The bilayer ZnO is deposited using RF sputtering using a ceramic target. The first layer is an intrinsic ZnO of about 500 thickness deposited in Ar/O2 ambient. The second layer is doped ZnO deposite d in Ar ambient from a ZnO:Al2O3 target. The amount of doping is a tradeoff between the electrical and optical properties. The transmission drops off at higher wavelengths due to the free carrier absorption which is a function of doping concentration. A transmission of 90% in the wavelength range of 4001000 nm with electrical resistivity of 9E-4 has been achieved . Anti-reflecting coating can be applied over th e front contact to reduce refl ection losses and aid in better trapping of light.
243.2 Device Fabrication The sequential steps followed in device fa brication are briefly discussed in this section. A thin layer of Mo is deposited on a glass substrat e using DC sputtering followed by the CIGS absorber layer using a two stage process. A thin layer of CBD CdS is grown over this absorber layer followed by a bilayer ZnO using RF sputtering. 3.2.1 Back Contact Deposition The cleaning of the substrate is a very cri tical step in device fabrication to ensure the removal of contaminants. Defects like scra tches and contaminants find their way into the device thereby hurting the device performan ce. The substrate is soaked in a detergentDI (deionized) water bath for about 3 hours and scrubbed to remove any particulates. This is followed by several rinses in DI wa ter followed by an ultrasonic treatment using 2Propanol. Several other soluti ons were tried but 2Propanol seems to give the best results. Soaking of the glass substrates in a hot water bath for around 20 minutes follows this ultrasonic treatment. The glass substrates are then stored in a DI water bath. A 1 micron thick layer of Mo is depos ited using DC magnetron sputtering. The glass is heated to 150C under vacuum before deposition to remove any moisture. The glass substrate is heated in the load-lock and the deposition is done in the main deposition chamber. The chamber is then pumped to a pres sure of around 5 torr. The first layer of 3000 is deposited at an argon pressure of 5 mT and the second layer of 7500 at an argon pressure of 1.5 mT. The resultant Mo fi lm has a smooth surface and resistivities of the order of 5E-5 ohm-cm.
253.2.2 Absorber Layer Deposition Co-evaporation of metals from elemen tal sources is the most widely used processing technique for high efficiency cells. NREL has developed a complex three stage process for their high efficiency devices . A high degree of control is required in this process as it has a di rect effect on the device performance. Hence a more suitable process for commercial producti on is a two stage process. This process involves the sequential depos ition of the metal layers followed by the high temperature anneal step in Se va por. This process requires a small degree of control because only one metal has to be c ontrolled at a time and because of the self controlling nature of Se flux at high temperat ures. The two stage process consists of the first stage of precursor formation followed by selenization at high temperature of about 550C. In the first stage, precursors are formed by sequential deposition of metals Cu, Ga, In with or without Se. In th e selenization step, these metals are reacted with Se at high temperatures to form the CIGS compound. Any excess In and Ga if deposited is removed from the surface during the selenization step which otherwise might precipitate around the grain boundaries and hurt th e device performance. The othe r technique to compensate for the excess In and Ga is addition of extra Cu during the selenization step to form the CIGS compound. Shields were introduced between the metal sources to minimize the cross contamination of sources during depositi on. The precursor was formed using two different configurations and they were na med as Type I and Type V precursors. The selenization step was also modified for some of the devices that were fabricated using the Type V precursors.
218.104.22.168 Type I Precursors The absorber layer was deposited in a new vacuum system which is a two chamber load-locked system. In the type I precursor, Cu and Ga are deposited in the first chamber and In and Se are deposited in the second chamber. The r eason for building this new system was to do the Cu and Ga depositions in a Se free atmosphere as compared to the old system in which all the metal deposit ions were done in a Se background. The Se flux during the absorber layer formation is high and once the system temperatures come down the extra Se in the atmosphere sticks to the walls of the system. So during the sequential deposition of the metals, this Se gets free and hence the metals are deposited in Se background. In the new system the two ch ambers are isolated from each other using an isolation valve and care is taken that none of the free Se enters the first chamber. The metals are deposited at relatively low temp erature of 275C to complete the precursor formation. A constant and high Se flux is maintained during the In2Se3 deposition. The substrate temperature ensures that no ex cess Se sits on the surface. The source arrangement and the thickness gradient for the metals are shown in figure 3.3. Figure 3.3 Source Arrangement for Type I Precursor
27 A constant rate of 0.6 /sec is main tained during the copper deposition. The thickness on the thickness mon itor corresponds to around one-th ird that on the substrate. A Constant rate of 0.7 /sec is maintained during Ga deposition and the thickness on the substrate is one and half times that on the thickness monitor. During In2Se3 deposition the In rate is maintained at 5.3 /sec and the Se flux is maintained at around 9.0 /sec. The In thickness on the substrate is about half that on the thickness monitor, this is because the source is placed at an angle to the substrate and so all the material that the crystal sees does not reach the substrate. The Se flux vari es from 19 /sec near the source to 16 /sec away from the source. The Se flux is mainta ined at least three times that needed for stoichiometric formation. This is necessary for good films . Figure 3.4 Temperature Profile for Selenization for Type I Precursor The precursor formation is followed by the selenization step in which the substrate temperature is ramped upto 450C in about 4 minutes which is then maintained
28 for about 7 minutes. The substrate temperatur e is again raised to 550C in another 4 minutes remaining there for about 7 minutes. The substrate temp erature is then turned off but the substrate is exposed to Se flux till the substrate temperature comes down to about 400C. The Se flux is maintained at about 9.0/sec during the whole of the selenization process. A thin layer of copper is deposited to convert any excess In, Ga into CIGS compound. The temperature profile is shown in figure 3.4. 22.214.171.124 Type V Precursor Figure 3.5 Source Arrangement for Type V Precursor In the type V precursor the Ga source is transferred from the first chamber to the second chamber. With this new configuration co -evaporation of In, Ga and Se is possible. This was also done to see whether the deposit ion of Ga in a Se background improves the device performance. The source arrangement changes from the type I precursor and is shown in figure 3.5 along with the thickness gradient. It was observed that the metal ratios, which are maintained near unity, were more uniform over the sample. The Cu was deposited as in the type I precursor. Some fr action of the total Ga was deposited first at
29 about 0.7/sec for better adhe sion of the films. The remaining Ga was co-evaporated with In and Se where a consta nt Ga rate of 0.3/sec was maintained. The actual thickness on the substrate for Ga is again one and half times that on the thickness monitor The In and Se rates were the same as in the type I precursor. Some experiments with the substrate temperature were also carried out. The selenization profile for the type V precursor was altere d from the type I precursor. The 450C step in the original selenization was removed and the substrate temperature was maintained at 550C for about 16 minutes before turning the substrate temperature off. The temperature profile for the altered selenization step is shown in figure 3.6. Different temperature profiles were tried for the type V precursors and will be discussed in the following chapter. Figure 3.6 Temperature Profile for Selenization for Type V Precursor
303.2.3 Window Layer Deposition A thin layer of CdS of about 300-500 is grown on the absorber layer by chemical bath deposition (CBD). The soluti on consists of a 0.015 molar cadmium acetate solution, 0.15 molar solution of thiourea, 0. 15 molar of ammonium hydroxide and DI water. The substrate is immersed in the solution when the water bath temperature is around 80C with constant stirring of the solution. The deposition takes place for around 5-7 minutes approximately. If th e substrate is kept immersed in the solution for a longer time, then homogeneous nucleation of CdS is formed on the surface which is undesirable. 3.2.4 Front Contact Deposition After the CdS deposition the substrate is bl ow dried using dry ni trogen gas. It is then loaded in a vacuum chamber for the RF sputtering from the ZnO target. The substrate is heated to around 125C to rem ove any moisture absorbed during the CdS deposition. The substrate is maintained at this temperature during the full front contact deposition. A thin first layer of in trinsic ZnO is deposited in an Ar-O2 ambient followed by the doped layer ZnO in Ar ambient. The su bstrate is masked us ing a stainless steel mask, to form small dots of area 0.1 cm2, to understand the effect of thickness gradients on the device performance. 3.3 Device Characterization After the devices were fabricated, I-V m easurements were done on them. The I-V measurements were done using the Kiethley 2410 source meter. This source meter was interfaced to the computer using a Labview program which also determines the Voc, the
31 FF and the Isc. The AM 1.5 global spectrum conditions are simulated using a lamp with the intensity of one sun. A four-probe measurement is done on the 25 dots (masked during the ZnO deposition). Before the measurements are done the absorber layer is scrapped off to make contact with the Mo la yer. Two probes are placed on the Mo back contact and two probes are place on the ZnO front contact. The solar cell is biased over a range of -0.2 to 0.5 V. Spectral response was done on these devices to know the current density of the devices over the wavelengt h range of 400-1400 nm. The instrument used for these measurements was an Oriel Spectrometer. This was also interfaced to the computer using a Labview program. Silicon and germanium standards were used to compare with the sample. The silicon standa rd was used over the wavelength range of 400-900 nm. The germanium standard was used over the wavelength range of 900-1400 nm. The quantum efficiency (QE) of th e device was calculated and the QE vs. wavelength graph was plotted.
32CHAPTER 4 RESULTS AND DISCUSSIONS The performance of CIGS solar cells, fa bricated by the process discussed in the earlier chapter is analyzed in this chapter. The main aim of this work is to optimize the performance of these solar cells with regards to the parameters like Voc, Jsc and fill factor. To do so a new process recipe will be developed to utilize the capabilities of the in-line processing unit. The devices were analyzed using the I-V measurements and spectral response. The main parameter used for analyz ing the influence of process details on the performance of the solar cell is the open circuit voltage. The short circuit current and the fill factor are the additional parameters considered. 4.1 Effect of Bulk Copper Because of new insights regarding fabri cation in the in-line processing unit we start by first re-optimizing the standard reci pe. The bulk copper thickness was varied to see its effect on device performance. Samp le P044 had copper th ickness of 594, P049 had 523 and P059 had 550 . Figure 4.1 s hows the Voc distribution for the three samples. Due the thickness gradient of all the metals the metal ratios are not same throughout the sample. We can see that the Vo c distribution is much more uniform in sample P059 which can be attributed to the fact that the metal ratios are nearer to unity, in the range of 1.0 to 0.94, th roughout the sample. Some of th e bad spots in these samples
33 could be due to the basic smudges and defects on the glass substrate. Table 4.1 shows the effect of bulk copper thickness on the overall pe rformance of the solar cells. As can be seen from the table, the average Voc is better for sample P059 and so is the Isc. The fill factors are also better for P059. The sa mple P044 has the worst device performance because the metal ratios are great er than unity, in the range 1.1 to 1.03. As can be seen, good VocÂ’s are attained in the lower left corn er where the metal ratio is close to unity. For sample P044 the metal ratios are in the range 0.97 to 0.91 and so the device performance is not uniform. Figure 4.1 Voc distributions for P044 (594 ), P049 (523 ) and P059 (550 ) Table 4.1 Effect of Cu Thickness on Device Performance Sample Cu thickness () Avg. Voc (mV) Avg. Isc (mA) Avg. F.F (%) P044 594 350 2.8 45 P049 523 390 2.9 48 P059 550 420 2.7 50
344.2 Effect of Venting on Jsc Selenium is used in large quantities for each run so the source has to be filled for every five runs. The system earlier had the problem that to open chamber 2, chamber 1 Figure 4.2 Spectral Response for Samples P007, P008, and H299 had to be vented first. It was observed that the Jsc of the device fabricated after the system was vented was affected, and the Jsc improved with the run sequence. As can be seen from table 4.2 the current density increa ses with the run sequen ce. It was speculated that this could be due to Gallium oxidation or due to presence of water vapor. Figure 4.2 shows the spectral response for the two devices P007 and P008. The 3rd device is fabricated by my colleague. The VocÂ’s are lowe r in these series or runs because the bulk copper thickness was not optimized till then.
35 Table 4.2 Overall Device Performance Sample Run No. Best Jsc (mA/cm2) Best Voc (mV) P007 2nd 30.6 370 H299 3rd 31.6 380 P008 5th 38.3 370 The highest current density obtained was for the device P008-13 and was 38.3 mA/cm2. The highest average Voc and fill factor was for sample P056. P056 was fabricated with bulk copper thickness of 550 and top copper thickn ess of 33 . This resulted in the metal ratios in the range 1. 0 to 0.94. Figure 4.3 s hows the Voc distribution for the sample P056. the light I-V curve for P056-11 is shown in figure 4.4. from these results it was concluded that performance wa s a function of the vacuum environment. Thus this must be taken into consid eration in ongoing optimization efforts. Figure 4.3 (a) Voc Distribution (b) Fill Factor Distribution for sample P056
36 -0.02 0 0.02 0.04 0.06 0.08 0.1 0.12 -0.3-0.2-0.100.10.20.30.40.5 P056-11 Figure 4.4 I-V Characteristics of Device P056-11 4.2 Type V Precursors In the type V precursor the Gallium is deposited in the second chamber. Here the idea was to do co-evaporation of In, Ga and Se but it was realized that the films were having an adhesion problem if Ga was co-eva porated with In and Se. This could be because, when co-evaporated enough Gallium does not segregate to the back of the device, thus causing adhesion problems. Ther efore around 25% of the total Ga to be deposited was deposited by itself before the co -evaporation. Now the films were adhering to the Mo back contact. In type V precursor s copper was deposited first in chamber 1. The substrate was then transferred to the 2nd chamber where about 25% of the total Gallium was deposited first followed by the co-evaporation of In, Ga, and Se. The
37 bandgap shift for Type V device P066 when comp ared to a Type I device P056 is evident in the figure 4.4. Since for our type I proce ss we are not able to shift the bandgap unless Cu is deposited in the 2nd chamber, this indicates a new controlling mechanism. This could be cause because Gallium requires a CuxSey species to form CIGS. Figure 4.5 Spectral Response for Samples P056 And P066 4.2.1 Effect of Ga Content on Device Performance Sample P064 had about 12% and sample P 066 has the Ga content of 20% of total group III content. All other pr ocessing details were kept the same. We can see from figure 4.5 that the overall Voc has impr oved for the sample P066. The device performance of P064 is poor because when Gall ium is co-evaporated with In and Se then enough Gallium does not segregate to the back of the device. Thus when the Gallium 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 400 600 800 1000 1200 1400 P066 P056
38 content is increased to 20 % we see the im provement in the open circuit voltages as Gallium can segregate to the back of the device with enough Gallium left to bond in the space charge region. The other reason for the p oor performance of P064 could be that it was fabricated immediately after the system was vented. However, Voc would normally not be affected this dramatically. It was s een that the short circuit current reduced as compared to the type I devices; this could in part be attributed to the bandgap shift in the type V devices. The bandgap shift results in fewer photons being absorbed and so fewer electron-hole pairs generated. But the drop in Jsc is much greater than expected from the bandgap shift, so other loss mechanisms are op erative. This is no t unlike the losses we observe in Jsc when extra Ga is used in the type I process. In those case we attribute the losses to improper bonded Ga, and that is likely the case here. P064 P066 Figure 4.6 Voc Distribution for Ga Content
394.2.2 Effect of Substrate temperature The substrate temperature during the precursor formation was changed from 275 C to 350 C for sample P071 keeping all other pa rameters the same. All the samples had 20% Ga content of the total group III content. The Voc distribution for sample P071 is shown in figure 4.6. The Voc for this sa mple has gone down and it is speculated that it could be due to the fact that the bonding of Gallium in the bulk of the film was not good and there could be some Gallium presen t in the film in the form of GaxSey. The bandgap shift for the samples P071 and P068 are comparab le as can be seen from figure 4.7. So we can say that enough Gallium is st aying in the space charge region. 340 390 390 380 370 360 380 260 360 180 350 340 290 310 380 300 X 330 340 360 310 250 X 300 290 Figure 4.6 Voc distribution for sample P071
40 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 400600800100012001400 P068 P071 Figure 4.8 Spectral Response of Samples P068 And P071 4.2.3 Effect of Selenization Pr ofile on Device Performance The selenization profile was change d for samples made under different conditions. We will be considering two sets of samples here one with the substrate temperature of 275C and one with 350C. Th e change in the selenization profile was that the 450C step was eliminated. Samp les P067 and P072 had the new selenization profile. Samples P071 and P072 have the s ubstrate temperature of 350C and samples P067 and P069 have the substrate temperature of 275C. The Voc distribution for samples P071 and P072 is shown in the figure 4.8.
41 P071 P072 Figure 4.9 Voc Distribution for Samples P071 and P072 The open circuit voltage for sample P072 has improved and is one of the highest for the new configuration indicating a better surface as compared to sample P071. The Voc distribution of samples P067 and P069 is shown in figure 4.9. We can see that the device performance of P067 is poor. It is sp eculated that the phas es formed with the substrate temperature of 275C need the 450 C step during selenization for proper bonding of CIGS. P067 P069 Figure 4.10 Voc Distribution of Samples P067 and P069
424.3.3 Effect of Venting on Voc It was observed that with the type V precursors the Voc is affected with the chamber conditions. The Voc decreases when the second chamber is vented and it increases with the run sequence. The table 4. 3 shows the overall performance of the three devices P065, P066 and P069. As the Gallium source is in the s econd chamber, it can now be stated that the oxidation of Gallium source seems to be affecting the device performance as it did for Ga deposited in ch amber 1 for the type I process. Figure 4.11 shows the Voc distribution for the three devices. The same effect was observed when copper was deposited in the se cond chamber for the type I de vice by my colleague. This suggests that Gallium is not getting incorporated in the front of the device thus affecting the bandgap shift as can be s een from figure 4.12 . We can speculate the same reason for the type V devices. Table 4.3 Device Performance W ith Respect To Run Sequence Sample Run No. Best Voc (mV) Avg. Voc (mV) Best F.F (%) Avg. F.F (%) P065 1st 410 350 49% 43% P069 2nd 440 410 42% 38% P066 3rd 480 440 51% 45%
43 P065 P069 P066 Figure 4.11 Voc Distribution of Samples P065, P066 and P069 Figure 4.12 Bandgap Shift with Run Sequence The best device fabricated using the type V precursor is P072-08 with the highest open circuit voltage of 500 mV, fill factor of 63% and Isc of 2.65 mA. The I-V curve for this
44 sample is shown in figure 4.13. The sample P072 was type V device with 20% Gallium content and the substrate temperature wa s maintained at 350C during precursor formation. The selenization process was alte red with the 450C step being eliminated. The current density for the type V precursor devices was low as compared to type I devices. This again appears to be due tot eh negative effects of excess Ga that accompany increased Ga incorporation and bonding. Thus co-evaporation of Ga with In and Se has not solved this problem for this device process. -0.003 -0.0025 -0.002 -0.0015 -0.001 -0.0005 0 -0.3-0.2-0.100.10.20.126.96.36.199 P072-07 Figure 4.13 I-V Characteri stics of Device P072-07
454.4 ZnSe Buffer Layer Additional experiments were conducted w ith the window layer to determine their role in device made with the processes di scussed above. ZnSe was tried as a window layer. ZnSe was evaporated on one half of sample P007 and CdS was deposited on the other half. The substrate temperature duri ng ZnSe deposition was maintained at around 200C. About 300A of ZnSe buffer layer was deposited. Bilayer Zn O was then deposited on both halves of the device. VocÂ’s were found to be compar able on both halves, but the current density was very poor for the ZnSe half. Figure 4.14 show s the Voc distribution for sample P007. This result suggests that Js cÂ’s are not just a function of bulk absorber properties, but also of interf ace effects. Thus shortcomings in Jsc for the type V process have to be further studied to determine wh ere the Jsc losses are occurring so that the process can be improved. Figure 4.14 Voc Distribution for Sample P007 4.5 Effect of Silicon Nitride Layer The influence of Sodium on the growth of CIGS films was observed some years ago . The Na from the soda lime glass substrate diffuses into the absorber and it
46 effects in better film morphology and higher conductivity of the film But there is no control over the amount of Na that diffuses. Table 4.4 Overall Device Performance for Si3N4 Thickness Variation Sample No. Si3 N4 Thickness Voc (mV) Isc (mA) P015 200 350 6.0 P016 800 300 2.5 P017 1600 290 0.3 Hence to better understand the role of Na in the CIGS films, a Si3N4 barrier layer was deposited on the glass substrate before b ack contact deposition by a colleague who was studying this effect . Table 4.4 shows the overall device performance for type I devices for different thickness of Si3N4 barrier layer. We can see from the table that as the barrier layer thickness increases Voc and Jsc reduces. But Voc reduction is not as dramatic as compared to the devices from the old system. This seems to be due to a reduction in formation of CuxSey in the in-line chamber a nd is the subject of ongoing studies.
47 CHAPTER 5 CONCLUSION AND RECOMMENDATIONS In this study it is demonstrated that good device performance can be obtained for the CIGS films fabricated by the two stage pr ocess. But there is still some scope for improvement in the device performance. Type I devices with open circuit voltages of around 450 mV and fill factors of around 60% were fabricated regularly. The highest open circuit voltage obtained was 480 mV. The short-circuit currents of around 35 mA/cm2 with highest of 38mA/cm2 were fabricated. It was seen that the bulk copper level ma ttered for the overall performance of the sample. With bulk copper thickness of 550 the metal ratios were nearer to unity, in the range of 1.0 to 0.94, th roughout the sample thus making th e open circuit voltages uniform even with the thickness gradient. This wa s seen even without adding the usual small amount of top copper. If the copper thickness was increased to 594 then it adversely affected the device performance, as the Cu/I n ratio was greater than unity throughout the sample. It was also observed that the chamber conditions affected the short circuit current densities. The current densities increased as the chamber remained under vacuum for longer time. This was due to a possible oxidat ion effect with Gallium or the presence of water vapor in the chamber.
48 In the type I devices, Gallium has the te ndency to segregate to the back, and so there is little Gallium present in the space charge region. Hence there is no bandgap shift relative to CIS. The open circ uit voltage increases with in crease in the bandgap as the band bending increases. But the current dens ity goes down with in creasing bandgap. In the voltage-current trade-off higher voltages a nd lower currents are preferred as the total efficiency is not affected and the I2R losses are reduced. For this reason the type V precursor was evaluated to see if the open ci rcuit voltage could be increased by forcing the incorporation of additional Ga llium in the space charge region. In the type V precursor it was seen that about 25% of the total Gallium to be deposited had to be evaporated by itself first for better adhesion of the CIGS films to the Mo back contact. Also it was seen that the Gallium content had to be increased as compared to the type I devices for improvi ng the device performance. The open circuit voltages increased as compared to the type I devi ce. This can be attributed to the fact that enough Gallium was bonded in the space charge region to result in a bandgap shift. Substrate temperature also proved to be an important parameter along with the selenization profile for better device perfor mance. It was observed that the substrate temperature of 350C and the selenization prof ile without the 450C step gave the best results for type V devices. It can be speculated that the 450C step results in formation of secondary phases which hurt device performan ce. Also if the substrate temperature is kept at 275C, then the 450C step in the selenization profile is necessary as it probably forms the required phases during the 450C ste p. With the type V precursor the chamber conditions seem to be affecting the device performance and the open circuit voltages in particular. This could be due to the possi ble oxidation of the Gallium source when the
49 chamber is vented. It was observed that the Vo c increased with every run from the time it is vented. Some of the experiments that need to be done to improve the device performance are trying argon annealing of the samples with the type V precursors. Another thing that needs to be tried is fabricating the type I precursor but with Gallium deposited in the second chamber in selenium atmosphere as it is speculated that the Gallium will be incorporated into the bulk of the film better improving the current density without reducing the open circuit voltages too much. This is speculated because it has been observed that when copper is deposited in selenium atmosphere the bandgaps open .
50 REFERENCE  D.M.Chapin et al,Â”A New Silicon PN Junction Photo-cell for Converting Solar Radiation into Electrical Po wer,Â” J.Appl. Phys., 25, 676, (1954)  Y. Hamakawa et al,Â” Thin Film Solar Cells: Next Generation Photovoltaics and its Application,Â” Springer, Germany (2004)  W.Horig et al,Â” The Optical Properties of CuInSe2 Thin Films,Â” Thin Solid Films, 48, 67, (1978)  J.Kessler et al,Â” Electro-optical a nd Photoelectrochemical Studies of CuIn3Se5 Chalcopyrite Films,Â” Proceedings of the 23rd Photovoltacis Specialist Conference, 549, (IEEE, 1993)  A.Rockett and R.W.Birkmire,Â” Current St atus and Issues in Polycrystalline CuInSe2 for Photovoltaics application,Â” Proceedi ngs of the MRS Meeting, V426, 143, (1996)  L.L.Kazmerski, M.S.Ayyagi, Y.J.Juang, R.P.Patterson,Â” Growth and Characterization of Thin Film Compound Se miconductor Heterojunctions,Â” J.Vac.Sci.Tech., 13, 65, (1977)  D.S.Albin et al,Â” A St udy on the Optical and Microstr uctural Characteristics of Quaternary CIGS Polycrystalline Th in Films,Â” Proceedings of the 21st Photovoltacis Specialist Conference, 562, (IEEE, 1990)  W.S.Chen et al,Â” Development of Thin Film Polycrystalline CuIn1-xGaxSe2 Solar Cells,Â” Proceedings of the 23rd Photovoltacis Specialist Conference, 577, (IEEE, 1993)  W.N.Shafarman et al,Â” Characterization of CuInSe2 Solar Cells with High Ga Content,Â” Proceedings of the 25th Photovoltacis Specialist C onference, 898, (IEEE, 1995)  M.A.Contreras et al Progress in Photovoltaics, Res.Appl., 7, 311, (1999)  S.M.Sze,Â” Physics of Semiconductor Devices,Â” Wiley, New York, (1981)  J.H.Scofield et al,Â” Sputte red Mo Bilayer Contact for CuInSe2 Based Polycrystalline Thin Film Solar Cells,Â” Thin Solid Films, 260, 26, (1995)
51  N.H.Schultz,Â” Solution Grown CdS Films for Photovoltaics,Â” M.S Thesis, University of South Florida, Tampa, (1991)  R.H. Mauch et al,Â” Optimization of Windows in ZnO/CdS/CIS Heterojunctions,Â” Proceedings of the 22nd Photovoltacis Specialist Conference, 8989, (IEEE, 1991)  Y.Ohtake et al,Â” CIGS Thin Film Sola r Cells with Continuously Evaporated Cd-Free Buffer Layer,Â” Proceedings of the 25th Photovoltacis Specialis t Conference, 549, (IEEE, 1995)  R.Bhatt et al,Â” Dependa nce of Reactively Sputtered ZnO Electronic Properties on Growth Parameters for Use as a Buffer Layer in CIGS Solar Cells,Â” Proceedings of the 26th Photovoltacis Specialist Conference, 383, (IEEE, 1997)  H.Sankaranarayanan, Â“ Effective Proces sing Conditions for CI GS Solar Cells Using an Additional Cu Layer,Â” M.S. Thesis, University of South Florida, Tampa, (1998)  M.A.Contreras et al,Â” High Efficiency CIGS Based Thin Film Solar Cells: Processing of Novel Absorber Layer,Â” Proceedings of the 24th Photovoltacis Specialist Conference, 68, (IEEE, 1994)  S.A. Zafar,Â”Growth and Characteri zation of Thin Film ZnO and CIGS For Photovoltaic Applications,Â” Ph D Dissertation, University of South Florida, Tampa, (1997)  H.Sankaranarayanan, Â“Processing of CIGS Solar Cells using a Two-Step Process,Â” PhD Dissertation, University of South Florida, Tampa, (2004)  V. Mohankrishanaswamy,Â” Processing an d Characterization of CIGS Solar CellsÂ”, M.S. Thesis, University of South Florida, Tampa, (2004)