Eth 14436

Diss. ETH Nr.
14436
Growth and Characterization of High Eciency CdTe/CdS Solar Cells
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH

For the degree of DOCTOR OF NATURAL SCIENCES presented by
Alessandro Romeo
Master of Science Degree in Physics, University of Parma Born October 6, 1969 Parma, Italy Citizen of Italy
accepted on the recommendation of Prof. Gernot Kostorz examiner Prof. Marc Burgelman co-examiner Dr. Ayodhya Nath Tiwari co-examiner P.D. Dr. Hans Zogg co-examiner
2002
iii
A mio padre
Maestro nella sica come nella vita.
La science na pas de patrie. Luis Pasteur, 14 November 1888
iv
Contents
Sommario Summary 1 Introduction 1.1 Historical overview of photovoltaics 1.2 Principles of solar cells . . . . . . . 1.3 Current status . . . . . . . . . . . . 1.4 Contents of thesis . . . . . . . . . . 2 The 2.1 2.2 2.3 2.4 2.5 2.6 CdTe/CdS solar cell Introduction . . . . . . . . . Substrate . . . . . . . . . . Front electrical contact . . . CdS window layer . . . . . . CdTe absorber layer . . . . 2.5.1 The CdCl2 treatment Electrical back contact . . . xi xv 1 1 2 7 10 11 11 11 12 12 13 14 14 17 17 17 18 25 27
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
. . . . . . .
3 Transparent conducting oxides 3.1 Introduction . . . . . . . . . . 3.2 Solar cell preparation . . . . . 3.3 Characterization of the layers 3.4 Solar cell performance . . . . 3.5 Conclusions . . . . . . . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
. . . . .
4 Inuence of CdS growth process 29 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.2 Morphology of CdS . . . . . . . . . . . . . . . . . . . . . . . . 30 v
vi
CONTENTS 4.3 Morphology and crystallization of CdTe . . . . . . . . . . . . 34 4.4 Photovoltaic properties . . . . . . . . . . . . . . . . . . . . . . 41 4.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5 Recrystallization in CdTe/CdS 5.1 Introduction . . . . . . . . . . 5.2 Recrystallization of CdTe . . 5.3 Intermixing of CdS/CdTe . . 5.4 Conclusions . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
45 45 47 50 53 55 55 55 58 61 63 64 65 65 65 68 68 69 71
6 Substrate cells 6.1 Introduction . . . . . . . . . . . . . 6.2 Experimental details . . . . . . . . 6.3 Microstructural properties of CdTe 6.4 Junction formation with CdS . . . 6.5 Solar cell performance . . . . . . . 6.6 Conclusions . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
7 Stability under proton irradiation 7.1 Introduction . . . . . . . . . . . . . . . . . 7.2 Radiation eects on semiconductor devices 7.3 Experimental details and results . . . . . 7.3.1 Front wall irradiation . . . . . . . . 7.3.2 Back wall irradiation . . . . . . . . 7.4 Conclusions . . . . . . . . . . . . . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
8 Flexible CdTe cells 73 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 8.2 Experimental details and results . . . . . . . . . . . . . . . . . 73 8.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 9 Final conclusions 9.1 Inuence of the front contact . . . . . 9.2 Inuence of the window layer . . . . . 9.3 Recrystallization of the absorber layer 9.4 High eciency CdTe cells . . . . . . . 9.5 Flexible CdTe cells . . . . . . . . . . . 9.6 Proton irradiation stability test . . . . 81 81 81 82 82 83 83
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
CONTENTS A Solar cell parameters A.1 Equivalent circuit model . . A.1.1 Open-circuit voltage A.1.2 Short-circuit current A.1.3 Fill factor . . . . . . A.1.4 Eciency . . . . . . A.2 Series resistance . . . . . . . A.3 Shunt resistance . . . . . . . A.4 Quantum eciency . . . . . A.5 Air Mass . . . . . . . . . . . Bibliography List of abbreviations Acknowledgments Publications and Award Curriculum vitae
vii 85 85 86 86 86 86 87 87 87 88 89 95 97 99 103
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
viii
CONTENTS
List of Figures
1.1 Typical I-V characteristic, of a CdTe/CdS cell (12.3 % eciency.) 1.2 Band diagrams of a homojunction. . . . . . . . . . . . . . . . 1.3 Typical band diagrams of a heterojunction. . . . . . . . . . . . 2.1 5 6 7
Basic structure of a typical cadmium telluride/ cadmium sulphide thin lm solar cell. . . . . . . . . . . . . . . . . . . . . . 12 18 19 20 21 22 23 24 26
3.1 3.2
Transmission spectra of dierent TCOs coated glass. . . . . . Transmission of annealed CdS grown on the dierent TCO layers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 SEM images of as-deposited CdTe surfaces on ITO, ZnO and FTO coated glass substrates (clockwise from top left). . . . . . 3.4 SEM images of CdCl2 treated CdTe surfaces on ITO, ZnO and FTO coated glass substrates (clockwise from top left). . . . . . 3.5 X-ray diraction pattern of CdTe deposited on CdS/FTO/glass. 3.6 X-ray diraction pattern of CdTe deposited on CdS/ITO/glass. 3.7 X-ray diraction pattern of CdTe deposited on CdS/ZnO/glass. 3.8 I-V characteristics of solar cells on dierent TCOs and with dierent amount of CdCl2 . . . . . . . . . . . . . . . . . . . . 4.1 4.2 Atomic force microscopy images of as-deposited HVE-CdS(top) and as-deposited CBD-CdS (bottom) layers on FTO substrate. Atomic force microscopy images of vac-annealed HVE-CdS (top) and vac-annealed CBD-CdS (bottom) layers on FTO substrate. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Atomic force microscopy images of vac-annealed HVE-CdS layers on FTO (top) and ITO (bottom). . . . . . . . . . . . . Atomic force microscopy images of CdCl2 treated HVE-CdS on FTO substrate. . . . . . . . . . . . . . . . . . . . . . . . . ix
31
32 33 34
4.3 4.4
x 4.5
LIST OF FIGURES Morphology of as-deposited CdTe on vac-annealed HVE-CdS, vac-annealed CBD-CdS and on HVE-CdS treated with CdCl2 (clockwise). The CdTe grains on CBD-CdS are 5 to 10 times larger than those on HVE-CdS, and on the treated CdS the grains are of intermediate size. . . . . . . . . . . . . . . . . . . Morphology of CdCl2 treated CdTe on vac-annealed HVECdS, on vacuum annealed CBD-CdS, and on CdCl2 treated HVE-CdS (clockwise). . . . . . . . . . . . . . . . . . . . . . . XRD patterns of CdTe deposited at 150 C on vac-annealed CBD-CdS (top), and on vac- annealed HVE-CdS (bottom). . . XRD patterns of the as-deposited CdTe grown at 300 C on different CdS layers on FTO (peaks are shifted for better reading). XRD patterns of CdCl2 treated CdTe on dierent CdS layers (the lines are shifted for better reading). . . . . . . . . . . . . I-V graphs of the solar cells made with three dierent CdS layers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantum eciencies of solar cells made with dierent CdS layer thickness. . . . . . . . . . . . . . . . . . . . . . . . . . .
35
4.6
36 38 39 40 42 43
4.7 4.8 4.9 4.10 4.11
5.1
Cross section of CdTe/CdS/ITO as deposited (top) and after annealing at 430 C using 60 nm CdCl2 (middle), 600 nm CdCl2 (bottom). . . . . . . . . . . . . . . . . . . . . . . . . 5.2 XRD peaks normalized to the 111 peak of CdTe treated with dierent CdCl2 amounts. . . . . . . . . . . . . . . . . . . . . 5.3 XRD peaks normalized to the 111 peak of CdTe treated with dierent temperatures. . . . . . . . . . . . . . . . . . . . . . 5.4 XRD bulk pattern (top)and XRD glancing angle pattern (bottom) of HVE-CdTe as-deposited. . . . . . . . . . . . . . . . 5.5 220 and 511 peaks of treated and not-treated CdTe on dierent window layers. . . . . . . . . . . . . . . . . . . . . . . . . . . 6.1
. 46 . 48 . 49 . 51 . 52
SEM images of CdTe surfaces. Top; as-deposited CdTe on Mo/glass (left) and on CdS/FTO (right). Bottom; CdCl2 treated CdTe on Mo/glass (left) and on CdS/FTO substrate after CdCl2 treatment (right). The substrate is soda lime glass. 56 6.2 SEM of as-deposited (top) and CdCl2 treated (bottom) CdTe layers on soda lime (left) and alkali free (right) substrates. The layers are grown on Sb2 Te3 /Mo coated substrates. . . . . 57
LIST OF FIGURES 6.3 X-ray diraction patterns of CdTe on Mo/glass (left) and CdTe on CdS/FTO/glass (right): as-deposited (top), treated (bottom) conditions. The substrate is soda lime glass. . . . . . X-ray diraction patterns of CdTe on Mo/glass: as-deposited (bottom) and treated (top). The substrate is alkali free glass. . X-ray diraction patterns of CdTe on Sb2 Te3 /Mo/glass: asdeposited (bottom) and CdCl2 treated (top). The substrate is soda lime glass. . . . . . . . . . . . . . . . . . . . . . . . . . I-V performance of superstrate and substrate solar cells. . . .
xi
59 60
6.4 6.5
6.6 7.1 7.2
62 63
7.3
Transmission of proton irradiated and unirradiated glass. . . . 67 Quantum eciency of irradiated and unirradiated cell for uence of 1013 cm2 , the triangle line is related to unirradiated cell calculated as it would have the same darkened glass. . . . 70 Quantum eciency of irradiated and non-irradiated cell for uence of 1012 cm2 . . . . . . . . . . . . . . . . . . . . . . . . 70
8.1
Transmission spectra of polyimide lms and CdS of variable thickness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 8.2 Schematic structure of the exible solar cell. . . . . . . . . . . 75 8.3 SEM images of as-deposited (left) and CdCl2 annealed CdTe layer (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 8.4 X-ray diraction patterns of as-deposited (bottom) and CdCl2 annealed (top) CdTe on exible substrate. . . . . . . . . . . . 77 8.5 I-V characteristic of 8.6 % eciency CdTe solar cell on ZnO:Al/polyimide, measured under AM1.5 illumination. . . . . . . . . . . . . . . 78 8.6 Comparison of the absolute quantum eciency of an 8.6 % cell on polyimide with another 12 % eciency cell on SnOx : F/glass substrate. . . . . . . . . . . . . . . . . . . . . . . . . . 79 8.7 I-V characteristics of 11 % eciency CdTe solar cell on ITO/polyimide measured under AM1.5 illumination. . . . . . . . . . . . . . . 79 A.1 Single diode equivalent circuit for solar cell. . . . . . . . . . . 85
xii
LIST OF FIGURES
List of Tables
3.1 5.1 XRD peak intensity ratios for CdTe/CdS/TCOs . . . . . . . . 25 Lattice parameters of CdTe after dierent post-deposition treatments and on dierent window layers. . . . . . . . . . . . . . . 53 parameters due . . . . . . . . . parameters due . . . . . . . . . to proton irradia. . . . . . . . . . . 68 to proton irradia. . . . . . . . . . . 71
7.1
Change in the photovoltaic tion from the front side. . 7.2 Change in the photovoltaic tion from the back side. .
xiii
xiv
LIST OF TABLES
Sommario
Questa tesi descrive lo sviluppo di diversi tipi di celle solari a lm sottili di CdTe/CdS e le loro caratteristiche strutturali e opto-elettroniche. La crescita di strati policristallini di CdS (strato nestra) e di CdTe (assorbitore di luce), la loro qualit` cristallina, la struttura degli strati e della a interfaccia e linuenza del trattamento di CdCl2 sono presentati e discussi. Sono stati sviluppati processi per la preparazione di celle solari di CdTe in congurazione superstrato e substrato ed ` stata ottenuta un ecienza e di 12.3 %.Viene inoltre presentata una caratterizzazione strutturale della regione di interfaccia del CdS/CdTe. Lo strato di CdS policristallino viene cresciuto per evaporazione da alto vuoto o deposizione per bagno su diversi tipi di ossidi trasparenti conduttori (TCO). Immagini da microscopio a forza atomica mostra che la morfologia dei grani di CdS va da 0.1 a 0.5 m a seconda del tipo di substrato, del ` processo di crescita e delle condizioni di annealing. E stato osservato che il trattamento con CdCl2 ricristallizza il CdS. Gli strati di CdTe policristallino sono cresciuti per evaporazione in alto vuoto (108 ) su substrati di CdS/TCO/vetro. Gli strati vengono poi ricristallizzati depositandovi CdCl2 e riscaldandoli in aria (trattamento di CdCl2 ). Immagini al microscopio elettronico (SEM) hanno mostrato che i grani di CdTe crescono con una grandezza che va da 0.5 a 2 m (a seconda del substrato: CdS e ossido trasparente conduttore) e dopo il trattamento col CdCl2 i grani possono raggiungere grandezze no a 10 m. Questi grani hanno una morfologia irregolare con larghi bordi di grano. Una analisi di dirazione di raggi X ha indicato che il CdTe cresce orientato prevalentemente lungo la direzione (111) e che questa orientazione ` fortemente dipendente dal sube strato e dalla temperatura di crescita. Inoltre dopo il trattamento col CdCl2 , xv
il CdTe perde orientazione preferenziale tendendo a riorientarsi in altre direzioni, diverse a seconda del trattamento e del processo di crescita. Lintermixing di CdS e di CdTe allinterfaccia ` stato studiato con analisi e a raggi X, calcolando il passo reticolare di strati CdTe su CdS con dierenti trattamenti di post-deposizione. E stato osservato che il processo di crescita e il trattamento di post-deposizione inuenza lintermixing. ` E stato sviluppato un processo a bassa temperatura (450 C) per la preparazione di celle solari di CdTe/CdS in congurazione superstrato. CdS e CdTe sono stati susseguentemente evaporati nella stessa camera da vuoto su vetri con TCO. Prima della deposizione del CdTe, il CdS ` stato scaldato e in vuoto a 450 C per stabilizzare lo strato e controllare lintermixing con il CdTe. Dopo il trattamento di ricristallizzazione del CdTe, viene applicato un contatto Cu/Au sulla supercie del CdTe e la cella viene scaldata per 20 minuti. Questo semplice processo ha dimostrato una alta riproducibilit` e a sono state preparate celle solari con ecienze no a 12.3 % su vetro. Sono state preparate e studiate celle solari in congurazione substrato (congurazione invertita). Gli strati di CdTe sono stati depositati su molibdeno (depositato per DC sputtering) o su strati di Sb/Mo. Dopo la ricristallizzazione del CdTe, gli strati di CdS sono stati depositati e ricristallizzati. Inne uno strato di ossido trasparente conduttore (ZnO:Al) ` stato deposie tato per RF (radiofrequenza) sputtering. Immagini al microscopio elettronico hanno mostrato che i grani del CdTe depositato su molibdeno hanno una diversa morfologia e ricristallizzazione rispetto a quelli depositati su CdS. Dirazione di raggi X su CdTe hanno mostrato che il CdTe ` orientato 111 e e che il trattamento di ricristallizazione, in questo caso e meno ecace nella riorientazione dei grani. Celle solari di CdTe/CdS sono state irraggiate con protoni ad alta energia di diverse uenze in modo da testare linstabilit` nello spazio. Carata teristiche I-V e misure di risposta spettrale delle celle solari prima e dopo lirraggiamento da protoni hanno mostrato che c` una minima degradazione e del 10 % solo nel caso di protoni con alta uenza e che non c` degradazione e ma anzi un miglioramento dei parametri fotovoltaici per uenze pi` basse. u Una alta stabilit`, una alta potenza specica (rapporto kW/kg) e un basso a xvi
costo di queste celle le rende una valida alternativa alle convenzionali celle di silicio e GaAs per applicazioni spaziali. Un processso per la preparazione di celle solari di CdTe/CdS su polimeri essibili ` stato ideato e realizzato. Un polimero realizzato nel nostro laboe ratorio ` stato depositato su vetro (precedentemente coperto da NaCl) per e spin coating, poi un ossido trasparente conduttore ` stato depositato per e RF sputtering e gli strati di CdS/CdTe sono stati depositati seguendo le procedure standard per le celle su vetro. Dopo aver depositato il contatto superiore la cella ` stata staccata dal vetro dissolvendo il NaCl in acqua. e Questi dispositivi hanno una alta riproducibilita e una ecienza dell11 % che risulta essere no ad ora la pi` alta ecienza registrata per celle solari u di CdTe/CdS essibili.
xvii
xviii
Summary
This thesis describes the development of dierent type of CdTe/CdS thin lm solar cells and their structural and opto-electronics characteristics. The growth of polycrystalline layers of CdS (window layer) and CdTe (light absorber), their crystal quality, the microstructure of the layers and interfaces, and the inuence of the CdCl2 annealing treatment are presented. Processes for the preparation of CdTe solar cells in the superstrate and substrate congurations were developed and a solar cell conversion eciency of up to 12.3 % was achieved. A structural characterization of the CdS/CdTe interfacial region is presented. Polycrystalline CdS layers were grown by high vacuum evaporation or by chemical bath deposition on dierent type of transparent conducting oxide coated glass. Atomic force microscopy images showed that the layer morphology of CdS grains ranges from 0.1 to 0.5 m depending on the substrate type, the growth process, and the annealing conditions. It has been observed that the CdCl2 treatment is eective for the recrystallization of CdS. Polycrystalline CdTe layers were grown by evaporation in high vacuum (10 Torr) on CdS/TCO/glass substrates. The layers were then recrystallized with vapor deposition of CdCl2 and annealed in air (CdCl2 treatment). Scanning electron microscopy (SEM) images showed that the as-deposited CdTe grains are from 0.5 to 2 m large (depending on the substrate: CdS and transparent conducting oxide) and after the CdCl2 treatment grains can reach sizes of up to 10 m. These grains have irregular morphology with large grain boundaries. X-ray diraction patterns indicated that as-deposited CdTe is predominantly (111) oriented and that this orientation is strongly dependent on the substrate and the growth temperature. They also indicate that after the CdCl2 treatment, CdTe looses the extent of preferred orienta8
xix
tion, tending to reorient in other directions that are dierent depending on the treatment and on the growth process. Intermixing of CdS and CdTe at the interface was studied with analysis of the X-ray diraction patterns by calculating the lattice parameters of differently treated CdTe layers. It has been observed that the growth process and post-deposition treatment inuence the intermixing. A low temperature (450 C) process for the preparation of CdTe/CdS superstrate solar cells was developed. On transparent conducting oxide coated glass CdS and CdTe were subsequently evaporated without breaking the vacuum. Prior to the deposition of CdTe, CdS was annealed in vacuum at 450 C to stabilize the layer and control the intermixing with CdTe. After recrystallization treatment of CdTe, a Cu/Au contact is applied on the top surface of CdTe and the cell is annealed for 20 minutes. This simple process exhibited a very high reproducibility and solar cells of eciencies of up to 12.3 % were prepared on glass substrates. Solar cell in substrate conguration (inverted type) were prepared and studied. CdTe stacks were deposited on direct current sputtered molybdenum layer or on stacks of Sb/Mo layers. After recrystallization of CdTe, CdS layers were deposited and recrystallized. Finally transparent conducting oxide layer (ZnO:Al) was deposited by radio frequency sputtering technique. SEM images have shown that grains and morphology of CdTe deposited on Mo have dierent morphology and recrystallization compared to the ones deposited on CdS. X-ray diraction patterns of CdTe have shown that CdTe is (111) oriented and that recrystallization treatment,in this case, is less effective for orientation loss. CdTe/CdS solar cells were irradiated with high energy protons of dierent uences in order to test the instability in space environment. I-V characteristics and spectral response measurements of as-prepared and irradiated cells showed that a degradation of less than 10 % is registered only in case of high proton uence and no degradation but an improvement of the performance is observed for lower uences. High stability, high specic power (kW/kg ratio) and low cost of CdTe thin lm solar cells make them a valid alternative to conventional Si and GaAs cells for space applications. xx
A process for the preparation of CdTe/CdS solar cells on exible polymers was established. Non-commercial polyimide layer was spin-coated on sodium chloride coated glass, as transparent conducting oxide front contact ZnO:Al layer was grown by radio-frequency sputtering and the CdS/CdTe stacks were deposited by vacuum evaporation and standard cell fabrication procedure was adopted. After deposition of back contact the solar cell stack was detached from the glass by dissolving NaCl layer. These devices have high reproducibility and eciency of 11 % which happens to be the highest eciency reported for exible CdTe/CdS solar cells.
xxi
xxii
Chapter 1 Introduction
1.1 Historical overview of photovoltaics
Edmund Becquerel, a French experimental physicist, discovered the photovoltaic eect in 1839 while experimenting with an electrolytic cell, made up of two metal electrodes placed in an electricity-conducting solution. He observed that current increased when the electrolytic cell was exposed to light [1]. Then in 1873 Willoughby Smith discovered the photoconductivity of selenium. The rst selenium cell was made in 1877 by W.G. Adams [2], and ve years later Fritts [3] described the rst solar cell made from selenium wafers. By 1914 solar conversion eciencies of about 1 % were achieved with the selenium cell after it was nally realized that an energy barrier was involved both in this cell and in the copper/copper oxide cell. The modern era of photovoltaics started in 1954. In that year Chapin [4] reported a solar conversion eciency of 6 % for a silicon single-crystal cell. In 1955 Western Electric began to sell commercial licenses for silicon PV technologies. Early successful products included PV-powered dollar bill changers and devices that decoded computer punch cards and tape. Already in 1958 silicon cell eciency under terrestrial sunlight had reached 14 %. In 1954 Reynolds [5] reported 6 % solar conversion eciency in what later came to be understood as the cuprous sulde/cadmium sulde heterojunction. This was the rst all-thin-lm photovoltaic system to receive signicant attention. In following years the eciency of Cux S/CdS increased up to 10 % and a number of pilot production plants were installed, but after several years of research it was realized that these solar cells have unsolvable 1
CHAPTER 1. INTRODUCTION
problems of stability owing to the diusion of copper from Cux S to CdS layers. By taking advantage of new technology, researchers have rapidly raised the eciency of the gallium arsenide based cell reported by Jenny [6] with 4 % eciency to present eciencies exceeding 24 % [7]. However in the last 20 years other thin lms solar cells have taken the place of the cuprous sulde/cadmium sulde, and their eciency have raised up to almost 20 %. The most predominant are two: copper indium gallium diselenide/cadmium sulde (Cu(In,Ga)Se2 /CdS) and cadmium telluride/cadmium sulde (CdTe/CdS). The rst CdTe heterojunctions were constructed from a thin lm of n-type CdTe material and a layer of p-type copper telluride (Cu2x Te), producing 7% ecient CdTe-based thin-lm solar cell [8]. However, these devices showed stability problems similar to those encountered with the analogous Cu2x S/CdS solar cell, as a result of the diusion of copper from the p layer. The lack of suitable materials with which to form heterojunctions on n-type CdTe, and the stability problems of the Cu2x S/CdS device, stimulated investigations into p-CdTe/n-CdS junctions since the early 1970s. Adirovich [9] rst deposited these lms on TCO-coated glass; this is now used almost universally for CdTe/CdS cells, and is referred to as the superstrate conguration. In 1972, Bonnet and Rabenhorst [10] reported eciencies of 5-6 %, in a graded band gap CdSx T e1x solar cell. The research for CuInSe2 /CdS started in the seventies, a 12 % eciency single-crystal heterojunction p-CuInSe2 /n-CdS cells were made by Wagner et al. in 1974 [11] and in 1976 Kazmerski presented the rst thin lm solar cells with 4-5 % eciency [12]. In the last 20 years a big development of these cells was given by the National Renewable Energy Laboratories (NREL) in U.S.A. and by the EuroCIS consortium in Europe, and in 1998 Siemens Solar presented on the market the rst CuInSe2 modules exceeding 10 % eciency.
1.2
Principles of solar cells
The photovoltaic process is based on the theory of the p/n junction. The p/n junction is a device made of two zones with dierent dopings, one of p-type (holes as majority carrier), one n-type (electrons as majority carriers). Let us suppose that two semiconductors with the same vacuum energy (dened as the energy of an electron at rest just outside the crystal and not
1.2. PRINCIPLES OF SOLAR CELLS
interacting with it) approach each other. Before electrical contact is made, we can see that the n-type region has an excess of electrons with respect to the p region and that p region has an excess of holes with respect to the n region. After contact we have than a gradient of concentration of carriers at the junction and an equilibrium state is established. A diusion of electrons and holes is caused by the respective gradients of concentration; the holes diuse into the n region and the electrons diuse into the p region. The result of this migration is the annihilation of free electrons in the n region by the holes, leaving only the positive charges of the dopant (assumed to remain xed), and of the free holes in the p region from the electrons, leaving only the negative charges xed. The resulting double layer near the junction is called space charge region (also called depletion layer, because it is depleted of all the free charges). Away from this zone we have still two regions p and n that are neutral. Moreover the creation of these two space charge regions of opposite sign gives an electric eld E known as intrinsic electric eld, directed from the n region to the p region, and that induces on the carriers two opposite and equal forces that keep the situation in equilibrium. For a solar cell there is no external polarization (no external current applied). When the junction is in equilibrium , the energy barrier between n and p zones is eV0 (where V0 is the diusion potential). Under light radiation when the junction is illuminated by photons whose energy is higher than the energy gap, it creates hole-electron pair. The minority carriers are pushed down to the energy barrier of the intrinsic eld. The result is the separation of the generated electrons from the holes into the p/n junction. These carriers generate a eld which is opposite to the intrinsic electric eld E. So that the resulting total electric eld is : E - E. There is then a voltage at open circuit Vf in the direction of direct polarization into the illuminated region. This eect is called the photovoltaic eect. If we consider that now there is a situation of non-equilibrium, starting from the general one-dimensional transport equations: 2n n (n n0 ) n = Dn 2 + n + n n + Gn (x) t x x x n
(1.1)
p p 2p (p p0 ) = Dp 2 + p + p p + Gp (x) t x x x p where n and p are the electron and hole density, Dn and Dp are the diusion coecient for electrons and holes, n and p are electron and hole mobilities, E is the electric eld, n and p are the minority carrier lifetimes for electrons and holes, n0 and p0 are the carrier densities at equilibrium, and Gn and Gp are the generation factors for electrons and holes.
(1.2)
Now trying to solve these equations for solar cells it must be considered that the generation of electrons and of holes are identical. Moreover, there can be either p n or n p, so the two equations 1.1 and 1.2 are uncoupled and the term x drops [3]. Finally considering steady-state conditions (compared with carrier lifetimes) dp/dt = 0 or dn/dt = 0 and a single transport equation suces to describe the excess of minority carrier density: Dp 2 pn pn (pn pn 0) + p + G(x) = 0 2 x x p (1.3)
This equation gives the current density of a part of the semiconductor (in this case the p doped part) but it is quite complicated to solve if suitable approximations are not taken. If we take in account the condition of constant generation rate G throughout the cell and consider only the minority carriers in depletion approximation, we obtain : p(x) p0 = Ce
xxn Lp
+ Gp
(1.4)
the general dependence of the current density can be expressed as: J = Js (e kT 1) qG(Lp + Ln )
qV
(1.5)
where Lp and Ln are the diusion length of the holes and electrons and Js is the saturation current. The dependence of the current on the junction characteristics is comprised in the generation factor G. This equation is just an approximation of the real situation but gives a very clear picture of the behaviour of a photogenerator
1.2. PRINCIPLES OF SOLAR CELLS
100
mA/cm
-1000 -500
80
60
40
20
0 0 -20 500 1000
mV
-40
Figure 1.1: Typical I-V characteristic, of a CdTe/CdS cell (12.3 % eciency.) device, and shows how the dependence of the current (I) on the voltage (V) is expressed (gure 1.1) There are two dierent types of junctions: homojunctions and heterojunctions. Homojunctions are made with the same material but dierently doped (one part is n-doped and the other one is p-doped). Heterojunctions are junctions made with two dierent materials with dierent doping characteristics. In homojunctions, since n- and p-type regions are from the same semiconductor, the energy bands of the two parts are aligned (gure 1.2 [3]). It is then quite easy to calculate the band diagrams for this case. This avoids the formation of barriers or spikes that complicates very much the understanding of the device, reducing its eciency. On the other hand, since as window layer an absorber material is used, the window layer part must be very thin to reduce photons losses. In most of the cases this loss is still considerable and anyway a very thin window layer gives place to current loss due to surface recombination.
Figure 1.2: Band diagrams of a homojunction.
Heterojunctions use dierent material for absorber and window layer, so that the most suitable material for each purpose can be chosen. If from one side the problem of photons loss is minimized on the other hand a heterojunction presents dierent band bendings on the n and on the p part, giving place to spikes (energy barriers for the ow of electron and holes) and interface states due to contaminants from diusion of the two dierent materials 1.3. For this case calculating the band bendings is much more dicult. There are some techniques for calculating the energy diagrams that in many cases are able to predict the behaviour of the device. The rst and most used is the Anderson model [13]. This model is based on the idea of incorporating the discontinuities in material properties such as dielectric costants, electron anities and enery gaps, but it does not take in account the possibility of interface states. When the heterojunction is very complex (intermediate compositions, dierent phases, mismatch and so on), as for the case of CdTe/CdS or Cu(In,Ga)Se2 /CdS heterojunctions, the available models are not sucient and more studies are
1.3. CURRENT STATUS
Figure 1.3: Typical band diagrams of a heterojunction. required [14]. For CdTe, some work was done by Jaegermann et al. [15].
1.3
Current status
In our times solar cells are available on the market especially as silicon solar cells. These cells are made with bulk crystals, often recycled from the microelectronics industry. Because of the particular process to produce them, crystalline and multicrystalline silicon cells have still too high cost to be considered as a reasonable competitor to the energy production from fossile fuels. However calculations show that with the increase in the production and with further improvement of the technology the production cost will drop to 1.15 $/W [16]. Crystalline silicon solar cells are the most widely spread on the market. They are made from a silicon single crystal grown at high temperature. Crystalline ingots are generally grown by the so-called Czochralski process. In
this process, the puried silicon chunks are melted in a quartz crucible along with small pieces heavily doped with boron. This produces a boron-doped melt into which a seed crystal is dipped and slowly withdrawn. The ingots are than sewed in wafers and partially doped in order to make the p-n junction. Eciencies of up to 25 % have been reached [7]. Microcrystalline silicon cells are obtained from silicon ingots with large columnar grains grown from solidifying molten silicon in a container. Multicrystalline silicon assure lower capital costs and higher throughput than crystalline silicon. In 1998, about 30 % of the worlds photovoltaic production was based on multicrystalline silicon wafers. Up to now the highest eciency of 19.8 % has been achieved by the group of Martin Green (University of New South Wales, Australia) [7]. III-V material based solar cells reach very high eciencies, giving the possibility to yield very high power per unit area. Substantial increases in conversion eciency can be realised by multijunction solar cells in comparison with single junction solar cells. Typical III-V multijunction cells (epitaxially grown) are AlGaAs/GaAs tandem cells, GaAs/GaAs cells, InGaP/GaAs tandem cells and InGaP/GaAs/InGaAs mechanically stacked cells. While single-junction cells may attain eciencies of 27 % at AM1.5, multijunction structures are able to reach eciencies of up to 35 % [17]. Solar cells are stacked in tandem so that cell facing the sun has the largest band gap. The next cell in the stack absorbs all the transmitted photons with energies equal to or greater than its band-gap energy, and the rest is transmitted to the downward stack. The preparation process of these cells is based on the technologies of liquid-phase epitaxy, vapour-phase epitaxy and metalorganic chemical vapor deposition, These cells are almost exclusively used for space applications. Thin lm solar cells have two advantages: the amount of material used is very small and the production processes involved are easily scalable. For these reasons it is calculated that the average cost per watt of a thin lm solar cell would be less than that of the silicon solar cell (in case of CdTe it would be 0.95 $ per watt [16]). There are four types of thin lm solar cells currently under investigation.
1.3. CURRENT STATUS
Amorphous silicon cells were reported for the rst time in 1976 by Carlson and Wronsky [18], since then a very big interest appeared among these devices because of great potential for low cost production and high eciencies. The amorphous silicon improves light absorption and increases its band gap when is hydrogenated: hydrogen binds with silicon atoms and passivates the dangling bonds defects. Moreover hydrogenated amorphous silicon (aSi:H) eectively behaves as a direct bandgap semiconductor. The presence of hydrogen is than essential for high eciency cells, but it has been observed that under light illumination hydrogen has the tendency to move and the cells degrade heavily. In the last years attempt to study and control these degradation brought to stable amorphous silicon cells exceeding 10 % eciencies (12.7 % eciency for unstabilized cells [7] ). This results together with the low cost of fabrication has brought many companies to module production (such as Fuji Electric, Sanyo, BP Solarex [19]). Stabilized modules have eciency range of 5 to 10 % [19]. Despite of low eciency due to degradation , amorphous silicon cells are very widely used for small devices such as calculators. Heterojunctions based on CdTe have been considered for use in solar cells since the early 1960s, and both bulk [10] and thin-lm devices on a glass substrate have been investigated. CdTe is a compound very easy to grow, and this is proved by the many dierent processes used to prepare this kind of cells obtaining always conversion eciencies exceeding 12 % [such as electrodeposition (BP Solar), Close Space Sublimation (CSS) (University of South Florida, Antec Solar), screen printing (Matsushita Corporation) Physical Vapor Deposition (PVD) (Zurich)] [20]. The near-optimum bandgap and high optical absorption of CdTe make it an attractive material for low-cost devices of the future, with eciencies of up to 16.4 % obtained to date [21]. The last improvement of a quasi-ohmic contact for CdTe have brought some companies to establish production plants, like Antec Solar (10 % eciency modules of 86 cm2 ), Matsushita (8.7 % eciency modules of 1200 cm2 ) and pre-production plants like BP Solarex (10.2 % eciency modules of 706 cm2 ). Chalcopyrite cells are together with CdTe/CdS solar cells the most promising devices for solar energy conversion. Cu(In,Ga)Se2 /CdS appears to be the thin lm cell with the highest eciency achieved [7], and it seems that a large area module with 10 % conversion eciency is easily possible. The growth
10
of the absorber (CuInGaSe2 ) material can be done by coevaporation of Cu, In, Ga following the selenization process [22], or by preparation of precursors (binary compounds) followed by the selenization [23] . Other dierent absorbers such as CuInSe2 and CuGaSe2 are also used. The rst cell was made of pure CuInSe2 , then the partial replacement of In with Ga [24] (increasing the band gap of the absorber from 1.04 eV to 1.1-1.2 eV) had brought to high eciency devices due to better band-gap match to the solar spectrum. The highest eciency obtained up to now by Cu(In,Ga)Se2 /CdS is 18.8 % [25]. These materials are more dicult to grow than CdTe but give the interesting possibility to produce devices without Cd compounds that, although very stable and non-toxic, are still not easily accepted on the market.
1.4
Contents of thesis
Several companies have already begun the commercial production of CdTe modules, and others are currently in the pre-production stage. Anyway there are still many research issues. In order to further increase the eciency of CdTe solar cells, and the stability which is very important for production, research is currently focused on ve main issues: (a) Understanding the mechanisms of the eect of CdCl2 treatment on CdTe and CdS recrystallization, intermixing and electronic properties. (b) Understanding and control of sulphide/telluride interdiusion at the device junction. (c) Studying the eects of the front contact and application of new materials as TCO. (d)Controlling the impurities in the material to evaporate and understanding their eect on the device. (e) Development of a stable and ecient back contact. The rst three issues will be the main topic of this dissertation. It has been shown that the optical and electrical properties of CdS and CdTe are wellsuited for use in PV, and that the exibility of the system in terms of the wide range of deposition methods available make the technology highly promising for large-scale production. However, a better understanding of the materials issues is required. Using a PVD deposition process, we have studied these issues which are outlined in the following chapters.
Chapter 2 The CdTe/CdS solar cell

2.1 Introduction
The CdTe/CdS solar cell is a device made of dierent layers (see gure 2.1). These layers can be deposited with dierent techniques on a substrate and have particular functions in the device. The solar cells can be grown in the substrate or superstrate congurations. The latter is the commonly used structure for high eciency cells.
2.2
Substrate
The choice of an appropriate substrate is very important. It should withstand the cell fabrication process temperatures and must not contaminate the layers that are subsequently grown. CdTe solar cells in superstrate conguration require a transparent substrate because the incident light has to pass through the substrate before reaching to the CdS/CdTe layers; any absorption in the substrate would be detrimental to the current generation of the cell. The general choice is glass because it is transparent, cheap and withstands relatively high temperatures. Common types used are soda-lime glass, which is inexpensive, and borosilicate glass [20]. The latter has a higher softening temperature, and for this reason is often used for the higher temperature deposition methods, but since it is more than 10 times expensive, soda lime glass is generally preferred for low-cost production. As will be shown, dierent glasses have dierent inuences on the growth of the layers and the device characteristics. 11
12
CHAPTER 2. THE CdTe/CdS SOLAR CELL
superstrate configuration
back contact CdTe(-4mm) CdS(0.1 mm) TCO(0.5 mm) glass (substrate)
substrate configuration
TCO(0.5 mm) CdS(0.1 mm) CdTe(-4mm) back contact metal (substrate)
Figure 2.1: Basic structure of a typical cadmium telluride/ cadmium sulphide thin lm solar cell.
2.3
Front electrical contact
The front contact must be highly conducting and transparent. In general transparent conducting oxides (TCO) are used as front contact [20]. For high eciency cells, it is required that the sheet resistance of the front contact is no more than 10 /square. The most widely used materials is tin oxide (SnO2x ), with a thickness of 500 nm, it is deposited onto the glass either by sputtering or atmospheric pressure chemical vapour deposition. As tin oxide has too low conductivity for making a good contact, it is then often doped either with indium, forming indium tin oxide (ITO) or with uorine giving a compound SnO2x which is called FTO. In the last years some other trasparent conducting oxides have been applied, for example aluminium doped zinc oxide (which is widely used for Cu(In,Ga)Se2 solar cells) and other stannate compounds such as CdSnOx or ZnSnOx [26].
2.4
CdS window layer
The polycrystalline CdS layer grows n-type providing the n part of the p-n junction. It can be deposited by vacuum evaporation (physical vapor deposition) [27], close space sublimation [28], chemical bath deposition [29] and radio frequency sputtering [30]. As a wide band gap material (Eg=2.42 eV at 300 K), it is largely transpar-
2.5. CdTe ABSORBER LAYER
13
ent down to wavelengths of around 510 nm. Therefore it also permits the absorber layer to receive the photons of lower energy to give rise to electronhole pairs up on illumination with the solar light. Depending on the thickness of the CdS some of the light below the 510 nm can still pass through to the CdTe giving additional current to the device. The reduction of layer thickness is then important to gain more photons in CdTe, but on the other hand the uniform coverage of the TCO and the consumption of layer into CdTe during annealing treatment implies not to reduce the thickness beyond a limit, otherwise the cell is either shunted or gives low voltage. CdS layer deposited by low temperature techniques generally requires an annealing treatment for recrystallization prior to the deposition of the CdTe absorber layer. The structural and optical properties of the layer are strongly depending on the deposition technique (temperature and energies involved have dierent values), so the properties must be matched with the characteristics of the absorber layer with which it has to intermix.
2.5
CdTe absorber layer
The polycristalline CdTe layer should be electrically p-type to form the p-n junction with n-CdS. It has an energy gap of 1.45 eV which gives the highest theoretical eciency. It is sucient to absorb the useful part of the solar spectrum. Since CdTe has lower carrier concentration than the CdS, the depletion region is mostly within the CdTe layer and in this region most of the carrier generation and collection occur. The conductivity and grain size of the layer depend on the deposition technique and post-deposition annealing treatment, which aects also the intermixing of CdS with CdTe. Typically thickness of this layer, depending on the growth method, is between 2 and 10 m, with grain sizes from 0.5 m to 8 m. As deposited CdTe layer is either n-type or highly resistive p-type, however, a suitable annealing treatment (in chlorine and/or oxygen ambient) can convert the layer to p-type and also increase the conductivity.
14
2.5.1
The CdCl2 treatment
For a reliable and reproducible process it is very important to control the intermixing between the CdS and CdTe and the size of the CdTe grain boundaries [31] - [32]. The so called CdCl2 treatment permits to do this and moreover increases the conductivity of p-CdTe. The CdCl2 is either evaporated or precipitated in alcoholic bath on the CdTe which is then annealed in oxygen at temperature in the range of 400 to 500 C. This procedure changes the structural and electrical properties of the absorber, and enables the solar cell to increase its eciency from 2-5 to 10-16 %. The CdCl2 treatment can cause a large increase in the grain size. It acts like a sintering ux in CdTe as small grains grow and coalesce together, but it also causes the widening of grain boundaries. Wider grain boundary regions are not desired because they aect the parallel conduction across the grain boundaries as well as they may also cause the shunting of solar cells by providing a conducting link between the top and bottom electrodes. For high eciency and stable cells large grained and compact CdTe is required. The causes and eects of this process will be widely discussed in the next chapter.
2.6
Electrical back contact
The role of a metal back contact on CdTe is to provide an electrical connection to the CdTe without adding excessively to the series resistance of the device. P-type CdTe is a notoriously dicult material to produce an ohmic contact: the barrier is given by bi = + Eg m (m is the work function of the metal, is the electron anity of the semiconductor) so due to the very high electron anity and the high energy band gap of CdTe a metal of high work function is required. During normal device operation (i.e. at forward bias under illumination), majority carriers will be moving from the CdTe to the metal. Most of the metals do not have these high work function. These holes will encounter a barrier over which they must pass via a thermionic emission process. The height, bi , of the barrier will be equal to the dierence of the two work functions: bi = s m [33]. The work function of a semiconductor depends on its doping concentration.
2.6. ELECTRICAL BACK CONTACT
15
The absence of any commonly available metals with a high enough work function means that a simple Schottky junction will not be sucient to produce an ohmic contact. Therefore one possibility is to prepare a p+ layer which can decrease the work function of the top part of the absorber in order to make a quasi-ohmic contact with the metal. Various chemical etchants may be used to provide the p+ layer (reduction of telluride ions to leave a tellurium rich (and hence more p-type) layer at the surface). The use of this etch followed by the evaporation of copper/gold is widely used for making contacts for research purpose. Etchants commonly used include brom-methanol and nitric/phosphoric acid solutions [34]. Another contacting process involves the deposition of a copper mixed graphite paste by painting, screen printing or spraying. When the complete structure is then annealed, the impurities diuse into the CdTe to produce a p+ layer [35]. In the last years a novel contact has been prepared at the University of Parma which consists of molybdenum with a buer layer of Sb2 Te3 [36]. The Sb2 Te3 is a semiconductor with a higher work function than CdTe and provides with Mo a more ohmic contact than CdTe. Eciencies of up to 15 % have been obtained with very highly stable performance.
16
Chapter 3 Inuence of transparent conducting oxides

3.1 Introduction
Transparent conducting oxides (TCOs) such as ITO, FTO and ZnO layers are used as transparent conducting electrodes for CdTe/CdS solar cells. These oxides have energy gap between 3 to 3.4 eV, depending on deposition processes. They have optical transmission in the range of 80-85 % in the visible-near infrared range and a sheet resistance in the range of 5-10 /cm2 . ITO and FTO window layers are commonly used by dierent groups for the preparation of CdTe/CdS solar cells [20, 30, 37, 38], whereas ZnO layers are not frequently used despite of their suitable properties.
3.2
Solar cell preparation
TCO coated soda lime or alkali free glass substrates are used as substrates to fabricate CdTe/CdS solar cells. Commercially available ITO and FTO were used while ZnO:Al layers were prepared in laboratory with RF magnetron sputtering [39]. CdS layers are grown in an ultra high vacuum evaporation chamber at a substrate temperature of 150 C and subsequently annealed at 450 C for recrystallization, then without breaking the vacuum CdTe is deposited at a substrate temperature of 300 C. Before evaporation, the uxes of CdS and CdTe are controlled by an UHV ionization gauge (Balzers IMG-U2), and after evaporation, thickness is mea17
18
CHAPTER 3. TRANSPARENT CONDUCTING OXIDES
100
80
Transmission (%)
60
40
ITO FTO ZnO
20
0 300 400 500 600 700 800
Wavelength (nm)
Figure 3.1: Transmission spectra of dierent TCOs coated glass. sured with a prolemeter (Dektak 3030). In a standard deposition CdS thickness is about 0.5 m and CdTe thickness is between 3 and 4 m. For the CdCl2 treatment, vacuum evaporation is used for the deposition of CdCl2 layers on CdTe. The stacks are annealed at 430 C for 30 minutes in air. CdCl2 layers of dierent thickness (between 60 and 600 nm) were used to optimize the CdCl2 treatment. For back contacts the CdTe surface is etched using a Br-Methanol solution followed by the deposition of Cu/Au stacks and a short annealing at 200 C.
3.3
Characterization of the layers
Commercial ITO and FTO coated glasses from Nippon sheet company and inhouse RF sputtered ZnO:Al have been characterized. Transmittance spectra of TCO layers have been measured and compared. As shown in gure 3.1, ITO, FTO and ZnO have a good transparency (about 80 % ) in the visible range. For ITO and FTO the absorption edge is around 300 nm, whereas the absorption edge of ZnO is 340 nm.
3.3. CHARACTERIZATION OF THE LAYERS
19
100
80
Transmission (%)
60
40
ITO FTO ZnO
20
0 500 600 700 800
Wavelength (nm)
Figure 3.2: Transmission of annealed CdS grown on the dierent TCO layers. It should be mentioned that the absorption edge of ZnO layers depends on the growth process [40]. The transmission spectra of CdS/TCOs/glass were compared after the samples were annealed at 450 C for 30 min. As shown in gure 3.2 the average transmission of CdS/FTO/glass is about 80 % , whereas it is slightly lower ( 75 % ) for CdS/ITO/glass and in the range of 40 to 65 % for CdS/ZnO/glass. It seems that during deposition CdS reacts with ZnO which results in a substantial lowering of the optical transmission. Scanning electron microscopy (SEM) was used to study the morphology of CdTe and cross section of CdTe/CdS stacks on dierent TCOs. As-deposited CdTe layers on ZnO have uniform morphology and larger grains with a size of 1 to 1.5 m. However, the CdTe layers on ITO have smaller grains with a size in the range of 0.5 to 1 m (gure 3.3). On FTO the layers have some very small grains with a size of less than 0.5 m and also a few big grains with a size of 2 m. After the CdCl2 treatment a large increase in the grain size is observed (gure 3.4). In this case a 600 nm of CdCl2 was used for all the CdTe layers of the same thickness. CdCl2 acts like a sintering ux in CdTe, small grains grow and coalesce to-
20
Figure 3.3: SEM images of as-deposited CdTe surfaces on ITO, ZnO and FTO coated glass substrates (clockwise from top left).
21
Figure 3.4: SEM images of CdCl2 treated CdTe surfaces on ITO, ZnO and FTO coated glass substrates (clockwise from top left).
22
25000
FTO as deposited FTO 600 nm CdCl2 treated 533 111 531
1000
800
20000
15000
600
counts
10000
400
*
5000
*
0
200
0
20 22 24 26 28
84
86
88
90
92
94
96
98
100 102 104
two theta
Figure 3.5: X-ray diraction pattern of CdTe deposited on CdS/FTO/glass.
gether leading to narrow and large grain boundaries [31]. The shapes of grains on FTO and ZnO are rectangular and well dened with a size of 3-7 m. Grains of dierent dimensions are inhomogeneously distributed in CdTe layers on ITO. Despite of the fact that in the morphology of as-deposited sample is similar among the dierent samples, in the CdCl2 treated case the morphology become very much dierent not only due to the size of the grains but also because of the shape. This suggests that the recrystallization of CdTe also depends on the substrates on which the layers are deposited. XRD measurements were performed with a Siemens D 500 diractometer with a Cu-K source (the measurements were performed at the Institute for Inorganic Chemistry, Division of Chemical Electron Microscopy, University of Zurich by M. Wagner). Diractograms of as-deposited and 600 nm CdCl2 treated samples were compared to study the recrystallization in CdTe. Asdeposited layers show textured growth and strong peaks corresponding to the (111) textured CdTe. After the CdCl2 treatment (gures 3.5, 3.6, 3.7) the intensity of the (111) peak decreases and the intensities of the other peaks are
23
120000
ITO As-deposited ITO 600 nm CdCl2 annealed 531
700
600
100000
80000
111 440
500
counts
60000
*
620 533
400
40000
300
20000
200
0 23 24
100 85 90 95 100 105
two theta
Figure 3.6: X-ray diraction pattern of CdTe deposited on CdS/ITO/glass.
24
16000 14000
ZnO as deposited ZnO 600 nm CdCl2 treated 111
1200
1000
12000 10000
531
800
counts
8000 6000 4000 2000 0 -2000 20 22 24 26 28
533
600
400
200
0 84 86 88 90 92 94 96 98 100 102 104
two theta
Figure 3.7: X-ray diraction pattern of CdTe deposited on CdS/ZnO/glass.
3.4. SOLAR CELL PERFORMANCE Table 3.1: XRD peak intensity ratios for CdTe/CdS/TCOs TCO treatment I(111)/(531) I(111)/(533) ITO as-dep 189.06 554 ITO treated 28.8 54.8 FTO as-dep 31.6 26.2 FTO treated 17.9 19 ZnO as-dep 11.11 25.2 ZnO treated 6.3 5.81 Intensities for powder 25 50
25
increased suggesting the random orientation of the crystallographic plane. This eect has been already been observed [41]. The peaks due to CdTe are labeled, while the star marked (*) are extra peaks due to CdS/TCOs. The trend is generally similar for CdTe/CdS on FTO and ZnO. One major dierence is that, in case of FTO and ZnO, (440) and (620) peaks of CdTe are not observed in the CdCl2 or as-deposited samples. In order to compare the recrystallization on dierent substrates the intensity ratio of peaks (e.g. I111 /I531 ) were calculated, as shown in table 3.1. For the random orientation I111 /I531 should be 25 and I111 /I533 should be 50, higher value implies that the (111) orientation dominates. After CdCl2 treatment the intensity ratios have decreased indicating a loss of the (111) preferred orientation. It is evident that the loss of (111) orientation is higher for ZnO and least for ITO. It is concluded that the recrystallization behaviour is similar for FTO and ZnO and dierent for ITO. After the CdCl2 treatment the XRD peaks are shifted to higher angles either due to thermal stress or due to intermixing of CdTe with CdS. This aspect will be discussed in the next sections.
3.4
Solar cell performance
Solar cell performance depends on the amount of CdCl2 and the type of TCO. Figure 3.8 shows the I-V curves of solar cells measured with an illumination intensity of 75 mW/cm2 . FTO solar cells made with 60 nm CdCl2 have low Voc (730 mV) and low ll factor ( 0.5), while 600 nm treated cells have Voc in the range of 800-836
26
FTO.600nm CdCl2 ITO.600nm CdCl2 ZnO.600nm CdCl2 ITO.60nm CdCl2
Current (mA/cm )
20
10
0 -1000 -500 0 500 1000
Voltage (V)
-10
-20
Figure 3.8: I-V characteristics of solar cells on dierent TCOs and with dierent amount of CdCl2 .
mV and the ll factor in the range of 64 % to 70 %. On ITO substrates this behavior is in contrast; cells with 60 nm CdCl2 have high Voc in the range of 810-838 mV and ll factor of 60 % compared to the Voc in the range of 710-770 mV and ll factor of 50-55 % for cells with 600 nm CdCl2 treatment. This may be due to an enhanced migration/diusion of copper in CdTe because of a less dense microstructure of CdTe on ITO as compared to that on FTO. The solar cells with ZnO(:Al) as window layer have poor performance, series resistance is very high, and the f.f. does not exceed 27 % , while Voc of 777 mV is slightly lower, The Isc is extremely low. It appears that the stability of CdTe/CdS/ZnO at high temperatures (above 400 C) is poor and CdS reacts with ZnO. As shown earlier the transmission of CdS/ZnO decreased after annealing at 450 C, therefore photon absorption loss in CdS/ZnO is a reason for low Isc .
3.5. CONCLUSIONS
27
3.5
Conclusions
In order to evaluate the merits of dierent TCOs, ITO-, FTO- and ZnOcoated soda lime glass substrates were used. At 450 C CdS reacts with ZnO and the average transmission of CdS/ZnO is less than that of CdS/FTO and CdS/ITO. Solar cell properties on ITO and FTO are sensitive to the amount of CdCl2 . CdTe/CdS on FTO requires more CdCl2 ( 600 nm) for high eciency as compared to those on ITO substrates. Solar cells with eciencies of 10 to 11 % are obtained on ITO and FTO. The ll factors of solar cells on ITO are lower than those on FTO. It can be concluded that the choice of an appropriate TCO has to be done not only in relation to its own properties (such as conductivity, transparency, etc.) but also in relation to the eect that this has on the physical and chemical properties of the CdTe/CdS layers deposited on it.
28
Chapter 4 Inuence of CdS growth process

4.1 Introduction
The growth and properties of CdTe layers and solar cells depend on CdS/TCO substrates [27]. Also a CdCl2 treatment of the CdS window layers, applied in some laboratories, has yielded high eciency cells [30]. Thin CdS layers grown by chemical bath deposition (CBD) are preferred as window layer because of low optical absorption losses and good covering properties on TCO. However pin-holes and irreproducibility is a concern for reliable industrial production. The CdS layers grown by physical vapor deposition are more often used for better reliability and compatibility with PVD-CdTe processes, despite of higher optical absorption losses in thicker (0.2 to 0.5 m) CdS layers. The properties of CdS, deposited by CBD or high vacuum evaporation (HVE) methods were studied to nd out which method is better suited and how the CdCl2 treatment aects the properties of the CdS layers. The structural properties of the CdTe layers on dierent type of CdS have been investigated and correlated with the photovoltaic properties. As already discussed in chapter 3, commercially available ITO and FTO coated soda lime glass substrates were used for the growth of solar cells. CdS layers were grown either in a high vacuum evaporation chamber at a substrate temperature of 150 C (HVE-CdS) or by a chemical bath deposition (CBD-CdS) method at 80 C using a solution of Cd salt, ammonia and 29
30
CHAPTER 4. INFLUENCE OF CdS GROWTH PROCESS
thiourea [29]. The thickness of the HVE-CdS was varied in the range of 50 to 500 nm, however the optimum thickness is about 300 nm. Thinner (70-100 nm) CBD-CdS layers were used for high optical transmittance. CdS layers were either annealed in vacuum at 450 C or treated with CdCl2 before the growth of CdTe layers. For comparison, as-deposited CdS layers were also used.
4.2
Morphology of CdS
Atomic force microscopy (AFM) (performed at Centro Interfacolt` Misure, a University of Parma, (Italy) by Carlo Vignali) was used to study the surface morphology of the CdS layers. The morphology of as-deposited CBD-CdS and HVE-CdS layers are quite dierent (see gure 4.1). The grain size of HVE-CdS is in the range of 0.1 to 0.3 m and the layers are rough. The CBD-CdS consists of clusters of up to 0.5 m but clearly these clusters are made of small grains of about 0.1 m. The CBD-CdS layers are rather smooth. As shown in gure 4.2, the grain size of vacuum annealed HVE-CdS (recrystallized) is in the range of 0.3 to 0.5 m and the layer is smooth. The eect of annealing in vacuum is dierent for the CBD-CdS. The grains, that on the as-deposited sample have a compact and smooth structure, are strongly recrystallized so that a ourished and rough structure is observed but the compactness is not lost (gures 4.1 and 4.2). The microstructure of the CdS also depends on TCO substrates. As shown in gure 4.3, the CdS grains on ITO are smaller and more homogenous than those on FTO. It should be mentioned that the morphology of CdS strongly depends on the morphology of the underlying TCO layer [27], the surface roughness of CdS may also depend on the thickness of the TCO layers. A CdCl2 treatment was applied to the HVE-CdS layers. After this recrystallization treatment a big change in the shape and size of the CdS grains is observed (gure 4.4). It appears that the CdCl2 treatment recrystallizes the CdS layers in such a way that some of the small grains coalesce together and form bigger grains of 0.5 m width. Also the shape of the grains is completely dierent, after the treatment the surface is smoother and the grains are predominantly rectangular in shape.
4.2. MORPHOLOGY OF CdS
31
Figure 4.1: Atomic force microscopy images of as-deposited HVE-CdS(top) and as-deposited CBD-CdS (bottom) layers on FTO substrate.
32
Figure 4.2: Atomic force microscopy images of vac-annealed HVE-CdS (top) and vac-annealed CBD-CdS (bottom) layers on FTO substrate.
4.2. MORPHOLOGY OF CdS
33
Figure 4.3: Atomic force microscopy images of vac-annealed HVE-CdS layers on FTO (top) and ITO (bottom).
34
Figure 4.4: Atomic force microscopy images of CdCl2 treated HVE-CdS on FTO substrate.
4.3
Morphology and crystallization of CdTe
Scanning electron microscopy (SEM) was used to study the surface morphology of the CdTe layers grown on dierent CdS layers. Due to the dierent structure of CBD-CdS and HVE-CdS layers the as-deposited CdTe layers grow with dierent morphologies (gure 4.5). The grains of CdTe on HVE-CdS are in the range of 0.5 to 1 m. The CdTe on CBD-CdS appears to consist of some small grains of about 1 m and many large grains of about 5 m width. The grains of the CdTe on CdCl2 treated HVE-CdS are larger compared to those on vacuum-annealed HVE-CdS but smaller than those of CdTe on vacuum-annealed CBD-CdS. The size and shape of the CdTe grains on CBD-CdS are similar to those on CdCl2 treated HVE-CdS. Moreover the morphology is extremely compact and the grain boundaries are minimized. Post deposition treatments with CdCl2 were performed on the CdTe layers grown on vacuum annealed CBD-CdS, HVE-CdS and CdCl2 treated HVECdS. After this recrystallization treatment distinct changes in the shape and size of the CdTe grains are observed (gure 4.6). For the CdTe on vacuum annealed HVE-CdS, a many fold increase in the
4.3. MORPHOLOGY AND CRYSTALLIZATION OF CdTe
35
Figure 4.5: Morphology of as-deposited CdTe on vac-annealed HVE-CdS, vac-annealed CBD-CdS and on HVE-CdS treated with CdCl2 (clockwise). The CdTe grains on CBD-CdS are 5 to 10 times larger than those on HVECdS, and on the treated CdS the grains are of intermediate size.
36
Figure 4.6: Morphology of CdCl2 treated CdTe on vac-annealed HVE-CdS, on vacuum annealed CBD-CdS, and on CdCl2 treated HVE-CdS (clockwise).
37
grain size and formation of wide grain boundaries are observed. The CdTe on CBD-CdS also is recrystallized in such a way that an entirely dierent microstructure and morphology of CdTe is created. Moreover it must be outlined that the size of the grains is not increased, as it used to be, after the treatment. This suggests that there is an intrinsic limit of the grain size that can not be enhanced by the CdCl2 treatment, as already observed for CSS-CdTe [31]. In that case the grains are already in relevant size for the as-deposited case and there is no increase after the CdCl2 treatment [41] [42]. There is insignicant change in the grain size of CdTe grown on the CdCl2 treated HVE-CdS but the morphology is changed: the layer is more compact and the grain boundaries are reduced. The crystallographic orientation of dierent types of CdTe/CdS stacks were investigated with X-ray diraction (Siemens D-500 diractometer and Cu-K tube) at the Nonmetallic Inorganic Materials group, Swiss Federal Institute of Technology. Figure 4.7 shows the XRD patterns of CdTe layers grown at a substrate temperature of 150 C on vacuum annealed CdS layers grown by CBD and HVE. The as-deposited CdTe layers exhibit a strong (111) preferred orientation in both the cases. With the increase in the substrate temperature the preference for the (111) orientation is decreased. Figure 4.8 shows the XRD pattern of as-deposited CdTe layers grown at 300 C on dierent CdS/FTO/glass samples (vacuum annealed HVE- and CBD- CdS and CdCl2 treated HVE-CdS). CdTe layers on vac annealed and CdCl2 treated HVE-CdS are (111) oriented. In contrast the XRD pattern of CdTe on vacuum annealed CBD-CdS exhibits (111), (311) and (422) peaks of similar intensities, indicating that the layer is not (111)-oriented. Therefore the crystallographic orientation of the CdTe layers on CBD-CdS is always inuenced by the deposition temperature of CdTe layers, which is not the case for CdTe on HVE-CdS. It seems that there is a correlation between the size of the CdS grains and the orientation of the CdTe that grows on it. The CdTe layers are more (111) oriented on highly recrystallized CdS layers of larger grain size, like the CdCl2 treated CdS. The loss in the texture of CdTe is further enhanced by annealing in air and also with the application of the CdCl2 treatment. XRD patterns of the CdCl2 treated (annealed at 430 C) CdTe deposited on dierent CdS layers were also measured (gure 4.9). The loss of the (111) preferred orientation is common to all the layers, but it is more evident for the CdTe deposited on the vacuum annealed CBD-
38
60000
111 As-dep CdTe (150 C)/CBD-CdS/FTO
intensity
50000 40000 30000 20000 10000 0 20 40000 35000 30000 25000 20000 15000 10000 5000 0 20 40
311
60
511
80 100
intensity
111 As-dep CdTe(150 C)/HVE-CdS/FTO
311
40 60
511
80 100
two theta
Figure 4.7: XRD patterns of CdTe deposited at 150 C on vac-annealed CBD-CdS (top), and on vac- annealed HVE-CdS (bottom).
39
10000 8000
on vac-annealed HVE-CdS on vac-annealed CBD-CdS on CdCl2 treated HVE-CdS
111
counts (arb. units)
6000
311
4000
422
2000 0 20 40 60 80 100
331
511
531
533
two theta
Figure 4.8: XRD patterns of the as-deposited CdTe grown at 300 C on different CdS layers on FTO (peaks are shifted for better reading).
40
3000
counts (arb. units)
2500 2000 1500 1000 500 0 20 40
on vac-annealed HVE-CdS on vac-annealed CBD-CdS on CdCl2 treated HVE-CdS 111 220 331 511 531 533 311 422
60
80
100
two theta
Figure 4.9: XRD patterns of CdCl2 treated CdTe on dierent CdS layers (the lines are shifted for better reading).
4.4. PHOTOVOLTAIC PROPERTIES
41
CdS where the (311) peak is predominant. In the case of CdTe on vacuum annealed HVE-CdS the (111) orientation is still stronger, but the (311) and (422) peaks have now a strong intensity so that the (111) preferred orientation is lost, and there is a more random orientation. The CdTe layers on CdCl2 treated CdS are also randomly oriented, many of the XRD peaks have almost similar intensities (between 800 and 1300 counts).
4.4
Photovoltaic properties
The eciency of solar cells does not only depend on the grain size of the CdTe but is inuenced also by the CdTe-CdS intermixing and pinholes/weak spots which are mainly generated in CBD-CdS/TCO substrates. Solar cells on thin (about 70 nm) CBD-CdS exhibit poor and irreproducible performance: pinholes are clearly visible after the CdCl2 treatment of CdTe/CdS stacks. The annealing treatment causes strain relaxation and creation of pinholes, it is also possible that almost all of the CdS layer is consumed (intermixed) in the CdTe during the annealing treatment.Therefore the cells have low eciency or electrical shorts. The annealing treatment is essential for the conductivity type conversion and junction activation. Solar cells were also prepared on thicker (150 nm) CBD-CdS layers. Fewer pinholes were observed after the CdCl2 treatment but the cells exhibit low eciency in the range of 5 to 6 % (typical values are Voc =720 mV, Isc = 17 mA/cm2 , f.f.=45 %). Cells with thin HVE-CdS (<100 nm) also yield low eciency with Voc in the range of 500-650 mV and f.f. of 40 %. The measurements suggest that cells with CBD-CdS should be treated with a very small amount of CdCl2 . High eciency cells are obtained when the HVE-CdS layers are about 300 nm thick. As shown in gure 10 the highest eciency obtained for a solar cell on HVE-CdS is 12.3 % (Voc =800 mV, Isc =23 mA/cm2 , f.f.=67 %). The solar cell with CdCl2 treated CdS exhibits an eciency of 11.2 % (Voc =780 mV, Isc =22 mA/cm2 , f.f.=0.65). The poor performance of the cells on CBD-CdS with 5.6 % eciency is attributed to the presence of pinholes and excessive intermixing. The thickness of CdS is an important factor for the photovoltaic performance of CdTe/CdS solar cells and in particular for the current. 300 nm of CdS are enough to avoid pinholes but on the other hand reduce the light that penetrates through the window layer and is absorbed by the CdTe reducing
42
current (mA/cm )
20
10
0 -500 -10 0 500 1000 Voc (mV)
1 2 3
-20 1 2 3 HVE-CdTe/vac-annealed CBD-CdS HVE-CdTe/CdCl2 HVE-CdS HVE-CdTe/vac-annealed HVE-CdS
-30
Figure 4.10: I-V graphs of the solar cells made with three dierent CdS layers.
4.5. CONCLUSIONS
43
1.00 0.90 0.80 0.70
Quantum efficiency
0.60 0.50 0.40 0.30 0.20 0.10 0.00 -0.10 300 400 500 600 700 800 900 1000
with with with with
1000 nm CdS 600 nm CdS 300 nm CdS 150 nm CdS
wavelength (nm)
Figure 4.11: Quantum eciencies of solar cells made with dierent CdS layer thickness. the generated current. As shown in gure 4.11 the photons of wavelength between 500 and 600 nm can reach the absorber layer only in case CdS is suciently thin.
4.5
Conclusions
The microstructure of CdS layers on TCOs depends on the method of deposition and post-deposition treatment. The crystallization and morphology of CdTe are strongly aected not only by the CdCl2 treatment but also by the CdS deposition method and the structure of the CdS window layers on the TCO substrates. On CBD-grown CdS window layers, large-grain CdTe is obtained even in the as-deposited condition. In contrast grain size of the CdTe on HVE-CdS is small and post-deposition treatment is required to increase the grain size. The CdCl2 treatment causes growth or even disintegration of grains in the CdTe layers. The as-deposited CdTe layers on HVE-CdS grow with (111) preferred orientation in the substrate temperature range of 150 to 300 C: CdTe layers grown at low temperature (150 C) are preferentially
44
(111) oriented but layers grown at 300 C are comparatively less (111) oriented. The CdCl2 treatment reduces the texture. The performance of solar cells on thin CBD-CdS and HVE-CdS is poor because of pinholes or excessive intermixing of CdS into the CdTe layers. Highest eciency of 12.3 % was obtained on vacuum annealed HVE-CdS layers of about 300 nm thick. Solar cells on CdCl2 treated CdS have slightly lower (11.2 %) eciencies, probably due to a weaker intermixing of the layers.
Chapter 5 Recrystallization in CdTe/CdS

5.1 Introduction
Recrystallization treatments are widely used for the CdS and CdTe layers grown by close space sublimation deposition (CSS) and high vacuum evaporation (HVE) techniques, they are essentially required for high eciency [43]. These recrystallization methods include a simple annealing in air (especially used for electrodeposited CdTe [44]), a CdCl2 treatment either by vapor deposition of CdCl2 on the CdTe or by dipping the layer in a CdCl2 methanol solution and subsequent annealing in air or by annealing the stacks in a CdCl2 ambient. Several groups are involved to study the eect of the CdCl2 treatment on the properties of CdTe, CdS layers and junctions [27]. It is still not very clear how the junction properties change in relation to recrystallization processes and to dierent deposition techniques of the window layer. It is however helpful to refer to the phase diagram for CdTe-CdCl2 [45] in which there is a eutectic composition of 77 % CdCl2 and a eutectic line at 508 C. Molten material will be present above this temperature. It is then reasonable to assume that at temperatures of 430-450 C(the standard annealing temperatures) the CdTe-CdCl2 is in a mixed phase in which the atoms are more free to move than a solid state, enhancing a recrystallization of the material. As described in this section, we have investigated the structural properties of CdTe deposited on dierent type of CdS (CBD-CdS and HVE-CdS) grown on ITO and FTO substrates. Dierent kinds of post deposition treatments were applied to the layers, from an air-annealing to dierent CdCl2 treatments at dierent temperatures. 45
46
CHAPTER 5. RECRYSTALLIZATION IN CdTe/CdS
Figure 5.1: Cross section of CdTe/CdS/ITO as deposited (top) and after annealing at 430 C using 60 nm CdCl2 (middle), 600 nm CdCl2 (bottom).
5.2. RECRYSTALLIZATION OF CdTe
47
5.2
Recrystallization of CdTe
The crystallographic rearrangements in CdTe are related to the structural disorder, to the stress in the layer and to the application of the CdCl2 sintering ux and high temperature annealing. The creation of new grains as a result of the disintegration of some large grains is due to the relaxation of the excessive strains in the lattice. Pin holes in CdTe/CdS and almost total intermixing of CdS-CdTe also deteriorate the performance of solar cells on HVE-CdS window layers with thickness <100 nm. In gure 5.1, SEM cross sections of three CdTe/CdS/FTO samples are presented: as-deposited, 60 nm thick CdCl2 treated and 600 nm thick CdCl2 treated. As deposited CdTe has a compact and columnar structure. In the 60 nm CdCl2 treated sample two types of grain structures are observed; the grains near the CdS interface are small and columnar, but the grains on top have recrystallized and their size has increased. In this case the amount of CdCl2 is not sucient to crystallize the whole layer. For samples treated with 600 nm of CdCl2 the CdTe layers are completely crystallized and larger grains are observed. This proves that recrystallization of CdTe starts from the top surface where CdCl2 is deposited and follows the diusion of the sintering ux to the junction region. The coalescence of small grains into larger ones is caused by the CdCl2 sintering ux and annealing at the high temperature. Because of this composite nature of the CdTe layers, XRD peaks corresponding to the crystallographic planes of the as-deposited CdTe grains and regenerated grains are observed. In gure 5.2 is shown the x-ray diraction pattern of dierently treated CdTe with dierent amount of CdCl2 . The peaks are all normalized to the (111) preferred orientation for easier comparison. As already shown in the previous chapters, the preferred orientation is tendentially lost during the recrystallization. It can be said that there is a reorientation of the grains, and this is stronger as more amount of CdCl2 is applied. Moreover, as shown in gure 5.3, the reorientation is also inuenced by the annealing temperature. The (111) preferred orientation is tendentially higher for higher temperatures, for the 400 C annealed CdTe the (111) orientation is lost and the (422) peak is the highest. After an annealing at 430 C the (111) orientation is recovered. Glancing angle XRD patterns (performed at Institute of Energy Consumption, Joint Research Center, Ispra, Italy by Neil Gibson) with varying angle of incidence (see gure 5.4 ) revealed a dierence in the crystalline state of bulk
48
1.2
1.0
111
311
counts
0.8
50 nm CdCl2 treated 600 nm CdCl2 treated as deposited powder data

422
220
0.6
0.4
331
0.2
400
0.0
511
440
531
620 533
-0.2 20 40 60 80 100
two theta
Figure 5.2: XRD peaks normalized to the 111 peak of CdTe treated with dierent CdCl2 amounts.
5.2. RECRYSTALLIZATION OF CdTe
49
1.2
111
1.0
as deposited CdS/CdTe CdCl2 treated at 400 C CdS/CdTe CdCl2 treated at 430 C powder data
422
counts
0.8
311
0.6
0.4
331 400 220 511 400 533 531 620
0.2
0.0 20 30 40 50 60 70 80 90 100
two theta
Figure 5.3: XRD peaks normalized to the 111 peak of CdTe treated with dierent temperatures.
50
and surface of the CdTe layers in relation to the post-deposition treatment. The intensity of the (220) peak of the CdTe surface is very high compared to the bulk of the layers. For example in the as-deposited case the ratio of (111)/(220) is 5 for the surface and 24 for the bulk (see gure 5.5). This is indicative of a dierent microstructure compared to the bulk of the CdTe. In gure 5.5 the 220 and 511 peaks of the treated and not-treated CdTe on dierent window layers are shown. In the post-deposition treated sample the presence of shoulder peaks (2 = 38 ) proves the formation of intermixed layer, which is not present on both the as-deposited samples (at 300 C the peaks are sharper) [42].
5.3
Intermixing of CdS/CdTe
To study the intermixing and stress in the CdTe layers, the in-plane lattice parameters were accurately determined from the Nelson-Taylor plot [31, 46]. For a cubic crystal, the lattice parameter a is related to the wavelength within the crystal of the radiation , the Bragg angle , and the indices of reexion (h, k, l), by the equation : h2 + k 2 + l2 a= (5.1) 2 sin From the XRD spectra lattice parameters of CdTe grown on dierent CdS or with dierent deposition and post deposition treatments were calculated. As given in table 5.1, the as-deposited CdTe layers have a high in-plane lattice constant compared to the recrystallized layers. As an example, the lattice parameter of as-deposited CdTe on HVE-CdS decreases from 6.499 to 6.446 after the CdCl2 treatment. This is because of the relaxation A A of the compressive stress which is generated due to the lattice and thermal mismatch between the CdTe and the underlying substrate. The lattice parameter of the CdTe layer deposited at 150 C is 6.484 A which is very near to the value obtained from the powder diraction data (6.481 ). This indicates that the stress in the low temperature grown layer A is insignicant. The lattice parameter may also decrease because of the intermixing of CdS in to the CdTe layer. These eects are also evident for CdTe on CBD-CdS. In the case of CdTe on the CdCl2 treated CdS, the lattice parameter of the as-deposited CdTe is not as high as for the one deposited on the vacuum
5.3. INTERMIXING OF CdS/CdTe
51
Figure 5.4: XRD bulk pattern (top)and XRD glancing angle pattern (bottom) of HVE-CdTe as-deposited.
52
CdTe (CdCl2 treated)/CBD-CdS(vac-annealed)/FTO

400 350
as-dep CdTe(300 C)/CdS(CBD)/FTO
220
300 300 250
220 511
Counts
200
511
100
Counts
78
200
150
100
50 0 36 600 37 38 39 40 75 76 77
0 36 1400 37 38 39 40 75 76 77 78
two theta
220
two theta
500
1200
511
1000 400 800
Counts
300
511
Counts
600
220
400
200
100
200
0 36 37 38 39 40 75 76 77 78
0 36 37 38 39 40 75 76 77 78
two theta
CdTe(air annealed)/CBD-CdS(vac-annealed)/FTO
two theta
As dep CdTe (150 C)/CdS(CBD)/FTO
Figure 5.5: 220 and 511 peaks of treated and not-treated CdTe on dierent window layers.
5.4. CONCLUSIONS
53
Table 5.1: Lattice parameters of CdTe after dierent post-deposition treatments and on dierent window layers. Window layers CdTe/CdS stacks CdTe lattice param. () A HVE-CdS(vac-annealed) as-dep at 300 C 6.499 HVE-CdS(vac-annealed) treated with 600 nm CdCl2 6.446 HVE-CdS(vac-annealed) as-dep at 150 C 6.484 HVE-CdS(CdCl2 treated) as-dep at 300 C 6.482 HVE-CdS(CdCl2 treated) treated with 600 nm CdCl2 6.477 CBD-CdS(vac-annealed) as-dep at 300 C 6.481 CBD-CdS(vac-annealed) treated with 60 nm CdCl2 6.480 CBD-CdS(vac-annealed) as-dep at 150 C 6.487 annealed CdS, and after the CdCl2 treatment the lattice parameter does not decrease so much. This is due to a change in the structure of the CdS, probably the lattice constants of vacuum annealed and CdCl2 treated CdS are dierent, because of crystallographic structural transformation of phases. It appears that the inuence of the CdCl2 treatment on the recrystallization of CdTe and CdS-CdTe intermixing is less pronounced for CdCl2 treated CdS window layers. This could be a reason for the slightly lower eciency of the devices made with a treated CdS, it probably needs a dierent optimization.
5.4
Conclusions
CdCl2 treatment increases the grain size and aects the preferred orientation of CdTe layers. The CdCl2 induced recrystallization starts from the CdTe surface and extends to the CdS, depending on the amount of CdCl2 and on the annealing conditions. For a complete recrystallization a certain quantity of CdCl2 is required. The preferred crystallographic orientation of CdTe depends on the deposition and annealing conditions. The CdCl2 treatment reduces a texture component. The post deposition treatments release the mismatch induced stress in CdTe and also inuence the intermixing of CdS-CdTe.
54
Chapter 6 Solar cells in substrate congurations

6.1 Introduction
The CdTe solar cells can be grown in a substrate or a superstrate congurations, gure 2.1 in chapter 2 shows the schematics of these congurations. In the most commonly used superstrate conguration CdTe/CdS stacked layers are grown on transparent conducting oxide coated glass substrate and the cell is illuminated from the glass substrate side. This conguration allows an easier encapsulation because the layers are grown directly on the glass cover. It is desirable to separately control the CdTe recrystallization and intermixing. This is dicult in superstrate conguration because the CdS layer is already present during the recrystallization of CdTe. However, this can easily be performed in the substrate conguration: i.e. TCO/CdS/CdTe/metalcontact/glass. In this conguration the light is incident from the top TCO side (and not from the glass substrate) [47, 48]. This chapter presents a comparative study of the photovoltaic and structural properties of CdTe layers grown on dierent layers that form back-contact substrate with CdTe.
6.2
Experimental details
For the growth of solar cells in substrate conguration the choice of an appropriate metal or semiconductor surface is critical, because this layer has to 55
56
CHAPTER 6. SUBSTRATE CELLS
Figure 6.1: SEM images of CdTe surfaces. Top; as-deposited CdTe on Mo/glass (left) and on CdS/FTO (right). Bottom; CdCl2 treated CdTe on Mo/glass (left) and on CdS/FTO substrate after CdCl2 treatment (right). The substrate is soda lime glass.
6.2. EXPERIMENTAL DETAILS
57
Figure 6.2: SEM of as-deposited (top) and CdCl2 treated (bottom) CdTe layers on soda lime (left) and alkali free (right) substrates. The layers are grown on Sb2 Te3 /Mo coated substrates.
58
form a low resistance ohmic contact with CdTe. We have evaluated Cu-Au, Mo, Te-Mo, Sb2 Te3 -Mo as back contact layers on glass substrates. First a layer of 0.5 m thick molybdenum is deposited with a DC magnetron sputtering system either on a soda lime or alkali free glass. Thin layer of Sb2 Te3 or Te were deposited in an ultra high vacuum chamber with a substrate temperature of 200 C. CdTe layers are grown at 300 C in the same chamber without breaking the vacuum. For the CdCl2 treatment, vacuum evaporation is used for the deposition of CdCl2 layers on CdTe and the stacks are annealed at 430 C for 30 minutes in air. After washing CdTe in water, CdS is then deposited in a high vacuum chamber at a substrate temperature of 150 C. The CdTe thickness is about 3 to 4 m and CdS is about 0.5 m thick in a standard deposition. After the growth of CdS layers, the CdS/CdTe stack is annealed in air at 430 C without CdCl2 or given a CdCl2 treatment by annealing at 430 C. For the front contact a layer of ZnO:Al is deposited on the CdS surface by RF sputtering at room temperature.
6.3
Microstructural properties of CdTe
As shown in gure 6.1 the as deposited CdTe layer on Mo is compact and it consists of grains of 0.5 m to 2 m size. The morphology is rough due to irregular shape and size of grains. This condition is similar to the as deposited CdTe on CdS/FTO/glass, even if in this later case the shape and size of grains are more randomly distributed. After CdCl2 treatment, the CdTe layer on Mo/ glass has grains with well dened shape but dierent grain size ranging from 3 m to 10 m. CdCl2 treated CdTe on CdS/FTO/glass is very dierent, with grains of random shape and size of 1 m to 10 m. The layers on Sb2 Te3 /Mo/glass (see gure 6.2) exhibit a dierent microstructure. In the as deposited case the grains have homogeneous shape and size of about 1 to 2 m and they are very compact. After CdCl2 treatment, the grains enlarge up to 8 m with wide grain boundaries. In gure 6.2 are also shown CdTe layers grown on Sb2 Te3 /Mo but on alkali free glass. The morphology of CdTe on alkali free glass is substantially dierent from that of CdTe on soda lime glass grown under identical deposition conditions. On alkali free glass, in the as-deposited case the grains have
6.3. MICROSTRUCTURAL PROPERTIES OF CdTe
59
1000
10000
800
8000
111
counts (arb. units)
600
counts (arb. units)
111
6000
glass/Mo/CdTe as deposited
400
glass/FTO/CdS/CdTe as deposited
4000
200
2000
311 422 331 531

80 100
220
0 20 40 60
511
0 80 100 3000 20 40
533
60
two theta
8000 7000 6000
two theta
111
counts (arb. units)
2500
111
2000
glass/FTO/CdS/CdTe CdCl2 treated

311 422
counts (arb. units)
5000 4000 3000 2000 1000 0 20
glass/Mo/CdTe CdCl2 treated

422 311 511 220
40 60 80
1500
1000
500
533
0 100 20
220
40
331 511
60 80
531
533
100
two theta
two theta
Figure 6.3: X-ray diraction patterns of CdTe on Mo/glass (left) and CdTe on CdS/FTO/glass (right): as-deposited (top), treated (bottom) conditions. The substrate is soda lime glass.
a size of about 0.1 m-0.5 m and they are homogeneous in shape. After CdCl2 treatment, the layers are very compact (narrow grain boundaries) and the size of grains is much smaller compared to the other treated layers on soda lime substrates (1 to 5 m). The inuence of substrate surface on the growth of CdTe is also evident from the X-ray diraction measurements. As shown in gure 6.3, CdTe in the substrate conguration has a pronounced (111) preferred orientation, which is not the case for CdTe grown in the superstrate conguration. Even after the CdCl2 treatment, CdTe layer in substrate conguration is still (111) oriented. While in the superstrate conguration the (111) preferred orientation is almost lost.
60
16000
counts (arb. units)
14000 12000 10000 8000 6000 4000 2000 0 20 40 60 80 100
111 CdTe treated on Mo/glass(alkali free) 422
220 311
511
two theta
700 600
counts (arb. units)
500 400 300 200 100 0 20
111 CdTe as-dep on Mo/glass(alkali free)
220
40 60
422 511
80 100
two theta
Figure 6.4: X-ray diraction patterns of CdTe on Mo/glass: as-deposited (bottom) and treated (top). The substrate is alkali free glass.
6.4. JUNCTION FORMATION WITH CdS
61
In gure 6.4 are shown the XRD spectra of CdTe grown on alkali free glass. A comparison with gure 6.3 clearly indicates that the sodium has a very strong inuence on the orientation. The as-deposited layer has a strong (111) orientation and the CdCl2 treatment does not change this preferred orientation. This is in contrast to that on soda lime substrates. From these results, together with the SEM pictures, it is clearly evident that the sodium has a strong inuence on the recrystallization of CdTe grains. It has been reported [49] that the eciency of superstrate cells on alkali free glass is higher than those on soda lime glass, a higher Isc of the cells on Corning glass was attributed to a better transparency of the alkali free glass. WE believe instead, that this dierence is more to be attributed to a dierent growth and orientation of CdTe. In addition the electrical inuence of sodium in CdTe should be considered. As described by Altosaar [50], Na on Cd site is an acceptor, while it may also form a complex donor-acceptor pair (NaCd-ClTe) after the CdCl2 treatment. Therefore the carrier concentration of CdTe on soda lime and alkali free will be dierent due to the presence of Na. In gure 6.5, the diractograms of CdTe grown on Sb2 Te3 /Mo/glass are shown. It can be clearly seen that the CdCl2 treatment has a dierent eect on this layer; in the as-deposited case the CdTe is (111) oriented and almost no other peaks are present. For the CdCl2 treated layer the (111) orientation is completely lost and the (311) and (422) peaks are the stronger. Once again, the observation suggest that the growth of CdTe is aected by the substrate. Therefore dierent back contacts provide dierent template for the growth of CdTe and the microstructure is inuenced.
6.4
Junction formation with CdS
Heat treatment of the CdS/CdTe is necessary for interfacial intermixing and junction activation. In case of the superstrate solar cells, CdS-CdTe interdiffusion already takes place during the deposition of CdTe on CdS at 300 C. However in the substrate conguration CdS is grown at low temperatures of 150 C. Therefore high temperature annealing is still required for an optimum intermixing even if the CdTe is already recrystallized in CdCl2 . The intermixing and photovoltaic properties are better when CdS/CdTe substrate cells are annealed in presence of CdCl2 .
62
counts (arb. units)
3000 2500 2000 1500 1000 500 0 20 3500
* 311 111 * *
40
Glass/Mo/Sb2Te3/CdTe treated 422 400 331

60
220 * *
511
80
531 620 533

100
counts (arb. units)
111
two theta Glass/Mo/Sb2Te3/CdTe as deposited
3000 2500 2000 1500 1000 500 0 20 40 60 80 100
511
531
two theta
Figure 6.5: X-ray diraction patterns of CdTe on Sb2 Te3 /Mo/glass: asdeposited (bottom) and CdCl2 treated (top). The substrate is soda lime glass.
6.5. SOLAR CELL PERFORMANCE
63
mA
-600 -400 -200
40
20
0 0 200 400 600 800 1000
mV
-20
-40
substrate cell with Mo back contact superstrate cell with Sb2Te3/Mo back contact superstrate cell with Cu-Au back contact
Figure 6.6: I-V performance of superstrate and substrate solar cells.
6.5
Solar cell performance
One of the dicult task in processing a CdTe solar cell in substrate conguration is to have a stable back contact. Dierent back contacts, such as Cu/Mo, Mo, Sb2 Te3 /Mo were evaluated. Solar cells with Cu/Au contact were shorted because of the diusion of Cu in CdTe during its growth and recrystallization. Cells on Mo or Sb2 Te3 also exhibited poor performance (< 2 %), as shown in gure 6.6 the Voc is in the range of 300 mV to 400 mV and the cells are obtained in superstrate conguration. Figure 6.6 shows I-V of typical cells, for superstrate conguration with a Sb2 Te3 back contact Voc =751 mV, Isc =24.8 mA/cm2 , f.f.=59 % =11 %, with a Cu-Au back contact Voc =795 mV, Isc =23 mA/cm2 , f.f.=67 % =12.3 % . The cells with Cu-Au back contact exhibit higher Voc and f.f. but they
64
have long term stability problem during migration of Cu.
6.6
Conclusions
A comparative study of the structural properties of CdTe layers on dierent substrates has been performed. The development of substrate type solar cell requires an appropriate substrate. Despite the large grain size of CdTe on Mo/glass and Sb2 Te3 /Mo/glass, the cell eciencies are rather low (< 2 %). With Cu/Mo contacts the cells eciencies are shunted. In contrast to this, the same back contact materials when used in superstrate conguration yield high eciency cells (e.g. 11 % and 12.3 %) with evaporated Sb2 Te3 and Cu/Au contacts. It has been found that sodium has an inuence on the microstructure and orientation in CdTe, either via substitutional doping at cation vacancy sites or by forming a donor-acceptor pair. The structure and electronic inuence on Na on the eciency of solar cells should be investigated further.
Chapter 7 Stability of CdTe cells under proton irradiation

7.1 Introduction
There is a recent interest to reduce the cost of solar modules for space applications. Polycrystalline thin lm solar cells of II-VI and I-III-VI compounds are potentially important because of their low cost, high eciency and stable performance. One of the important requirements for space application is the stability of solar cells against high-energy proton and electron irradiation. Recent investigations on Cu(In,Ga)Se2 (called CIGS) solar cells have shown that their stability against high-energy radiation is superior to Si or III-V solar cells [22, 51]. CdTe/CdS solar cells are also expected to exhibit good stability for space applications [52]. To determine the radiation tolerance of solar cells on glass substrates, they were irradiated with high-energy protons; this chapter will describe these experiments and the photovoltaic properties of the irradiated solar cells.
7.2
Radiation eects on semiconductor devices
It is known that high-energy particles (electrons and protons) can change the structural, electrical and optical properties of solids by causing a displacement damage or through ionization eects [22, 51, 52, 53]. Incident particles on polycrystalline material can create dislocations and/or displace atom from 65
66
CHAPTER 7. STABILITY UNDER PROTON IRRADIATION
their lattice sites, creating electronic traps into the band gap, and so altering the electronic characteristics of the crystal. There can be formation of mid-gap states (which facilitate the transition of electrons from valence and conduction band), formation of states close o the band edges (which facilitate trapping) and change in doping characteristics (donor or acceptor density) [54]. The displacement damage depends on the non-ionizing energy loss (N IEL), which is the energy and momentum transfer to lattice atoms (depending on the mass and energy of the incident quanta) [53]. It can be calculated from the evaluation of an integral:
180
N IEL = (N/A)
min
L[T ()]T ()
d() d d
(7.1)
In equation 7.1, N is Avogadros number, A is the atomic mass and T () is the energy transferred to the target nucleus by a particle scattered through an angle in the center of mass system (maximum energy transfer occurs for = 180 C), d()/d is the dierential cross section for elastic scattering of charged particles into a solid angle increment d and L[T ()] is the Linhard partition factor, which gives the fraction of transferred energy that is nonionizing. The lower limit on the integral, min , is the scattering angle for which the recoil energy equals the threshold for displacement. In the case of SiO2 (glass) not only the displacement damage but also the ionization eects are important. The energy absorbed by the electronic ionization in insulating layers (such as SiO2 ) liberates charge carriers, which diuse or drift to other locations where they are trapped, leading to unintended concentrations of charge and parasitic elds. Ionizing eects depend primarly on the absorbed energy and they are independent of the type of radiation. These eects cause a darkening of the glass that is proportional to the uence of irradiation (as shown in gure 7.1). For these measurements and also for the cell development soda lime glass was used. It should be mentioned that there is a certain type of glass (CeO2 doped glass) which remains highly transparent under space conditions. Such a glass is generally used to cover Si and GaAs solar cells in space.
7.2. RADIATION EFFECTS ON SEMICONDUCTOR DEVICES
67
100
80
transmission (%)
60
40
without irradiation Fluence 10 cm

20
13 12 11 -2 -2 -2
Fluence 10 cm Fluence 10 cm
0 200 300 400 500 600 700 800 900
wavelength (nm)
Figure 7.1: Transmission of proton irradiated and unirradiated glass.
68
Table 7.1: Change in the photovoltaic parameters due to proton irradiation from the front side. Fluence 1013 cm2 Fluence 1012 cm2 Fluence 1011 cm2 Energy Voc f.f. Isc Voc f.f. Isc Voc f.f. Isc (MeV) mV % mA mV % mA mV % mA 15 -20 -2.5 -6 +35 +3 -4 0 0 -1 10 -20 -2.5 -7 +35 +3 -4.5 0 0 -1 5 -20 0 -6.5 +25 +2 -5 0 0 -1
7.3
7.3.1
Experimental details and results

Front wall irradiation
Proton irradiation experiments were performed at the Paul Scherrer Institute (Villigen, CH) by Dr. Wojitek Hajdas. Solar cells were irradiated through the glass substrates with protons of dierent energies (5 to 15 MeV) and uences (1011 to 1013 cm2 ). These uences are higher than the typical uence on the earth orbits, as protons are trapped by the magnetic eld of the earth. Three sessions of irradiation experiments were performed. To investigate the inuences of proton irradiation, the I-V measurements were performed before and after the irradiation. The initial eciency was in the range of 8 to 10 % under AM 1.5 illumination. Table 7.1 gives the change in the PV performance of cells irradiated with dierent energies and uences. As already observed [52], one should expect some degradation of the cells because the proton bombardment may change the characteristics of CdTe. For high uence (1013 cm2 ) a small degradation in Voc and f.f. was observed. For the medium uence (1012 cm2 ), increase in Voc and f.f. are measured presumably due to passivation of some defects in CdTe layers. Such an improvement in the PV properties of cells has not been reported and more work is needed to explain this observation. In gures 7.2 and 7.3 the quantum eciencies of irradiated and non irradiated cells are compared. It can be seen that for the highest uence a loss in the response is registered only for wavelength between 700 and 850 nm, which is mostly due to less carrier collection of the modied absorber material. For medium uence there is almost no dierence between the irradiated and not irradiated cell, the damage of the irradiation is compensated by the passivation of defects produced by the proton bombardment.
7.3. EXPERIMENTAL DETAILS AND RESULTS
69
It is observed that Isc of irradiated cells appear to decrease with increasing uence. However, it should be emphasized that values of low Isc are not due to any degradation of the semiconductor or heterojunction, but the darkened glass is responsible for the absorption of photons during the measurements. Solar cells grown on space quality glass would not have problems of glass darkening. These measurements have shown that CdTe solar cells are stable against proton irradiation. The results give a good perspective for the application of CdTe solar cells in space.
7.3.2
Back wall irradiation
As already discussed, the presence of a 1 mm thick glass substrate will alter the energy and uence of protons reactivity to CdTe/CdS solar cells because of the absorption and scattering eects. Even though in superstrate cell the radiation will be incident on the glass, to make a complete analysis of the test and also for scientic curiosity some CdTe/CdS solar cells were irradiated from the back contact side. It must be mentioned that the browning of the glass in this case is much lower because of the absorption in CdTe but the eect still persists to cause the current loss. The I-V parameters of as-deposited and post-irradiated cells are shown in table 7.2. Apparently degradation seems to occur but actually it is lower than what the table shows because some of the loss is due to the browning of the glass (ascribed corresponds a maximum loss in current of about 30%), however the Voc decreases by a maximum of 2.5% and ll factor by a maximum of 6 %. From the Monte Carlo simulation we have observed that the irradiation through a glass and directly on the sample are very dierent not only for the loss of absorbed protons but also for the shape and distribution of energies of the ux, so the two tests can not be compared directly. These results conrm the high stability of CdTe solar cells, only at very high uence a modest degradation is indicated.
70
1.0 0.9 0.8 0.8
as-deposited irradiated as-dep recalculated
10 cm fluence; 10 MeV
13
-2
quantum efficiency
0.7 0.6 0.5 0.4 0.3 0.3 0.2 0.1 0.0 300 400 500 600 700 800 900 1000
wavelength
Figure 7.2: Quantum eciency of irradiated and unirradiated cell for uence of 1013 cm2 , the triangle line is related to unirradiated cell calculated as it would have the same darkened glass.
1.0 0.9 0.8 0.8
as deposited irradiated as-dep recalculated
10 cm fluence; 10 MeV
12
-2
quantum efficiency
0.7 0.6 0.5 0.4 0.3 0.3 0.2 0.1 0.0 300 400 500 600 700 800 900 1000
wavelength
Figure 7.3: Quantum eciency of irradiated and non-irradiated cell for uence of 1012 cm2 .
7.4. CONCLUSIONS
71
Table 7.2: Change in the photovoltaic parameters due to proton irradiation from the back side. Fluence 1013 cm2 Fluence 1012 cm2 Energy Voc f.f. Isc Voc f.f. Isc (MeV) mV % mA mV % mA 15 -20 -2.5 -6 -20 -1/0 -2 10 -15 -4 -5 -20 +1 -3 5 -10 -3 -5 -5 -1 -3
7.4
Conclusions
Proton irradiation stability tests on CdTe/CdS solar cells from the top and the bottom part have been described for the rst time. Despite of the very high uences used, which are exceeding some orders of magnitude to the proton uences that are actually encountered in space application, only for a high uence (1013 cm2 ) a minor degradation has been measured. For a scientic curiosity we irradiated the cells from the CdTe side and found minor degradation due to CdTe/CdS, darkening of the glass was observed also in this case. These measurements have shown that CdTe solar cells are stable against proton and electron irradiation.
72
Chapter 8 Flexible CdTe cells

8.1 Introduction
Development of exible and lightweight solar cells is interesting for terrestrial and space applications that require a very high specic power (dened as the ratio of output electrical power to the solar module weight). Thin lm solar cells on polymer lms can yield more than 2 kW/kg specic power. CIGS solar cells of about 10 to 12.8 % eciency have been developed on polymer substrates [55, 56]. However, CdTe solar cells on polymer substrates were not reported prior to this work, although solar cells of about 10 to 16 % have been obtained on glass substrates. CdTe solar cells on metal foils exhibit an eciency of 5 % [57, 58]. In this chapter, we describe an approach for the development of exible CdTe solar cells on polymer lms.
8.2
Experimental details and results
High eciency CdTe solar cells are generally grown in a superstrate conguration where CdTe/CdS stacks are deposited on transparent conducting oxide (TCO) coated glass substrates. Eorts for developing exible CdTe solar cells were, up to now, based on the substrate conguration; such solar cells were grown on metal foils and they exhibit low eciency 5 % due to non-compatibility of the CdTe with metal substrates. Their photovoltaic performance is poor because of inecient ohmic contact on CdTe [57, 58]. For exible solar cells in a superstrate conguration the choice of an appropriate substrate is crucial because the substrate should be optically transparent 73
74
CHAPTER 8. FLEXIBLE CdTe CELLS
and should withstand the high temperature deposition/processing of solar cells. Most of the solar cell fabrication processes require temperatures of about 450 to 550 C but transparent polymers are not stable at such high temperatures. However, some of the polyimides are stable at high temperature (up to 500 C) and their optical transparency could be sucient for CdTe/CdS solar cell application. As shown in gure 8.1 the thickness of polyimide can strongly inuence the transmission spectra, therefore CdTe solar cells on 50 to 100 nm thick polyimide lms will yield a low current due to a large optical absorption loss in the substrate. We have developed a process in which, instead of using a commercially available foil, a specic type of polyimide lm is prepared in-house. One of the advantages is that the thickness of the polyimide lm can be reduced to minimise the absorption loss in the substrate. The solar cell fabrication process is schematically shown in gure 8.2. A thin buer layer of NaCl was evaporated on a glass substrate (which permits to lift o the cell from the glass), then a polyimide layer was spin coated and cured at about 430 C. The thickness of the polyimide lm can be controlled by the spin coating process; we have used about 10 nm thin polyimide lms. As shown in gure 2, the average transmission of the polyimide lm is more than 75 % for wavelengths above 550 nm. There is a strong absorption of photons in the wavelength range of 400 to 550 nm, which coincides to the high absorption region of CdS window layer. Due the band gap of 2.4 eV the CdS window layer absorbs photons of wavelength shorter than 515 nm. For the front electrical contact with CdS, transparent and conducting ZnO: Al layers with an average transmission of about 80 % and sheet resistance of 20 /square were grown by RF magnetron sputtering. CdTe/CdS solar cells were developed with a process in which all the layers are grown by evaporation methods. After the complete processing, exible CdTe/CdS/ZnO: Al/polyimide stack was removed from the glass substrate by dissolving the NaCl buer layer in water. The structural properties of CdTe on polyimide are similar to that on glass substrates. As shown in gure 8.3, the as-deposited CdTe layer is homogeneously compact with grains of up to 1 m size. The CdCl2 annealed layers are crack-free and consist of large grains of up to 5 m. The as-deposited layers have a (111) preferred orientation but a loss in orientation for the CdCl2 annealed is observed, similar to the CdTe on glass (see gure 8.4).
75
100 90 80 70
transmission (%)
60 50 40 30 20 10 0 -10 400 500 600 700 800
CdS (0.25 m) Kapton (25 m) Upilex (50 m) ETHZ (10 m)
wavelength (nm)
Figure 8.1: Transmission spectra of polyimide lms and CdS of variable thickness.
Metal CdTe CdS ZnO:Al Polyimide Buffer Glass
PVD HVE HVE RF Sputtering Spin Coating PVD
Figure 8.2: Schematic structure of the exible solar cell.
76
Figure 8.3: SEM images of as-deposited (left) and CdCl2 annealed CdTe layer (right).
The solar cells were grown on 3x3 cm substrates and scribed to smaller area for measurements. No antireection coating was applied. Figure 8.5 shows the I-V characteristic of an 8.6 % eciency cell on a exible polyimide lm under AM1.5 illumination. The solar cell (see gure 8.5) has an area of 0.13 cm2 and exhibits Voc = 763 mV, Isc = 20.3 mA/cm2 , f.f. = 55.7 %. This is the highest reported eciency of a exible CdTe solar cell. A comparison of the quantum eciency measurements of a 12 % eciency cell on SnOx: F/glass and an 8.6 % eciency cell on polyimide indicates a slight dierence (see gure 8.6). The solar cell on polyimide has a slightly lower response in the wavelength range of 500 to 820 nm due to the absorption loss in the polyimide lm substrate. To understand the reasons for lower eciency of CdTe solar cells on ZnO: Al/polyimide, it is important to mention that the ZnO: Al, despite good opto-electronic properties, does not yield ecient CdTe/CdS solar cells. In the fourth chapter it has been reported that solar cells on ZnO: Al coated glass substrates exhibit 3 % eciency because of a very low ll factor of 30 %, low current density, and high series resistance; while the same process on ITO or SnOx: F/glass yields 11 to 12 % eciency solar cells. In the present work we have used ZnO: Al front contact on polyimide because of the non-availability of other TCO deposition equipment in our laboratory. In gure 8.7 the I-V characteristics of the latest 11 % exible cell are shown. This result was obtained using as front contact ITO deposited at the University of Parma. This layer provides better conductivity and stability than the in-house ZnO:Al.
77
2500 2000 1500 1000 500 0
Polyimide 111
CdCl2 annealed
Counts (arb. units)
220 311 400 331 422 511
20
8000 6000 4000 2000 0 20
111
40
60 as-deposited
80
100
Polyimide
40
60
80
100
Wavelength
Figure 8.4: X-ray diraction patterns of as-deposited (bottom) and CdCl2 annealed (top) CdTe on exible substrate.
78
mA/cm
-400 -200
20
10
0 0 200 400 600 800 1000 1200
mV
-10
-20
Figure 8.5: I-V characteristic of 8.6 % eciency CdTe solar cell on ZnO:Al/polyimide, measured under AM1.5 illumination.
79
1.0 Quantum Efficiency 0.8 0.6 0.4 0.2 0.0
cell on ZnO:Al/polyimide cell on SnOx:F/glass
400
500
600
700
800
900
Wavelength (nm)
Figure 8.6: Comparison of the absolute quantum eciency of an 8.6 % cell on polyimide with another 12 % eciency cell on SnOx : F/glass substrate.
50
Jsc (mA/cm )
40 30 20 10 0
V oc [mV]: 842 J sc [mA/cm ]: 18.5 f.f. [%]: 70.9 [%]: 11

2
-200 -10 -20 -30
200
400
600
800
1000
1200
Voc (mV)
Figure 8.7: I-V characteristics of 11 % eciency CdTe solar cell on ITO/polyimide measured under AM1.5 illumination.
80
8.3
Conclusions
Lightweight and exible CdTe/CdS solar cells in the superstrate conguration have been developed for the rst time to our knowledge. A spin coated polyimide layer is used as a substrate. Solar cells of 8.6 % eciency were obtained on ZnO: Al coated polyimide with a lift-o method. Replacing ZnO: Al with an other transparent conducting front contacts such as ITO has increased the eciency giving an 11 % exible CdTe solar cell. This result is the highest eciency for a exible CdTe solar cell reported so far.
Chapter 9 Final conclusions

CdTe/CdS thin lm solar cells of high eciency up to 12.3 % have been developed with a simple vacuum evaporation process. Properties of dierent layers, interfaces, and eects of the annealing treatment were investigated in detail.
9.1
Inuence of the front contact
Three types of transparent conducting oxides (TCO) were investigated: indium tin oxide (ITO), uorine doped tin oxide (FTO) and zinc oxide. CdS grains grown on ITO are smaller and more regular in morphology than those on FTO, and these dierences inuence the morphology of CdTe. It is shown that the solar cell properties on ITO and FTO are sensitive to the amount of CdCl2 used for the recrystallization and junction activation. CdTe/CdS on FTO requires more CdCl2 (600 nm) for high eciency than the same layers on ITO substrates. The ll factors of solar cells on ITO are lower than those on FTO. 12.3 % eciency cells were developed on FTO and it is concluded that FTO is preferred for high eciency and low cost, during the cell processing ZnO reacts with CdS and yields a semitransparent layers which is not suited for solar cells.
9.2
Inuence of the window layer
CdS is commonly used as a window layer to make p/n junctions with the CdTe. CdS layers deposited by chemical bath deposition (CBD) and high81
82
CHAPTER 9. FINAL CONCLUSIONS
vacuum deposition (HVE) were characterized, and the inuence on the growth of CdTe was investigated. It has been observed that on CBD-grown CdS, large (5 to 10 m)-grain CdTe is obtained even in the as-deposited condition. In contrast, the CdTe grain size on HVE-CdS is small (1 m) and a postdeposition treatment is required to increase the grain size. We have seen that despite the large grain size of CdTe on CBD-CdS, solar cell eciency is rather low (3-4 %) because of a pronounced intermixing of CdS-CdTe, causing almost total consumption of the thin CdS by the CdTe layers. The CdCl2 treated cells show electrical shorts. HVE-CdS layers yield 12.3 % cells but the absorption loss in CdS should be reduced to further increase the eciency.
9.3
Recrystallization of the absorber layer
CdTe layers grown on dierent CdS and TCOs were characterized. The crystal orientation of the layers depends on the growth temperature and the post deposition annealing treatment. It has been observed that CdCl2 treatment increases the grain size and aects the preferred orientation of CdTe layers;it causes a loss in the preferred orientation. For an ecient solar cell an optimum quantity of CdCl2 is required. It has also been observed that recrystallization starts from the surface of CdTe and diuses down to CdS. With XRD measurement it has been shown that the post-deposition treatments (air annealing or CdCl2 annealing) release the mismatch-induced stress in CdTe and that intermixing is larger for CdCl2 treated cells. These treatments cause growth or even disintegration of grains in the CdTe layers.
9.4
High eciency CdTe cells
Dierent processes for the growth of layers andactivation of heterojunctions were optimized based on the detailed understanding of the material properties of layers and interfaces. We have developed a reproducible and reliable process that produces cells with eciencies in the range of 10 to 12.3 %. This low temperature (< 450 C) process has a good potential for in-line production of solar cells.
9.5.
FLEXIBLE CdTe CELLS
83
9.5
Flexible CdTe cells
Lightweight and exible CdTe/CdS solar cells in the superstrate conguration have been developed for the rst time using a novel process in which spin coated transparent polyimide is used as substrate for the CdTe cells. With an optimized process a world record eciency of 11 % eciency has been achieved for exible CdTe solar cells.
9.6
Proton irradiation stability test
CdTe solar cells were irradiated with 5-15 MeV protons of uence in the range of 1011 to 1013 cm2 . The photovoltaic properties were characterized; despite the very high uences (1013 cm2 ) used (which are exceeding by some orders of magnitude the proton uences that are expected in space), only a minor degradation has been measured. For uences of 1012 cm2 , an increase in the values of ll factor and open circuit voltage was measured. This is due to the passivation of intrinsic defects in the CdTe. These investigations has proved that CdTe cells have a high stability against proton irradiation and are very promising for space applications.
84
CHAPTER 9. FINAL CONCLUSIONS
Appendix A Solar cell parameters

A.1 Equivalent circuit model
A non-ideal solar cell can be described with an equivalent-circuit model. The most simple equivalent circuit used consists of a non-ideal diode, representing the p/n junction, coupled with a current source to account for the collection of light-generated carriers. A shunt resistance Rshunt and a series resistance Rseries are also included in the circuit (their meaning is described in the following). The complete equivalent-circuit diagram is shown in gure A.1. The rst parameters which characterize a solar cell device are the ones calculated by the current-voltage measurement (as shown in appendix B). In this brief appendix these parameters will be described. The main parameters are:
Figure A.1: Single diode equivalent circuit for solar cell. 85
86 open-circuit voltage short-circuit current ll factor eciency
APPENDIX A. SOLAR CELL PARAMETERS
A.1.1
Open-circuit voltage
The open-circuit voltage is the applied voltage relative to an open circuit where no current ows through the device. This quantity is left unaected by series resistance losses in the cell, but is sensitive to shunt losses.
A.1.2
Short-circuit current
If no voltage is applied in the equivalent circuit then the current density measured is called the short-circuit current density, Jsc. This value is aected primarily by series resistance losses in the device.
A.1.3
Fill factor
The parameters described above represent the maximum values that it is possible to measure and their product is the ideal power. The ll factor is then dened as the inverse of the ratio of the ideal power to the maximum power in operating conditions. It can be dened also as the area of the maximum power rectangle to the product of the short-circuit current and the open-circuit voltage. Typical ll factors for high eciency solar cells range from 55 to 80 %.
A.1.4
Eciency
The most important parameter of a solar cell in terms of its ultimate function is the photovoltaic conversion eciency: . This is dened as the ratio of the output power (electricity) to input power (light) and can be calculated also as: Voc Isc Poutput = f.f. (A.1) = Plight Plight where Plight is the density of the incident radiation over surface and time.
A.2. SERIES RESISTANCE
87
A.2
Series resistance
The series resistance is the resistance the carriers nd on their way and it is given by the resistivity of the materials of the device. There are many eects that give rise to this resistance: Low conductivity of the window layer or of the absorber, non-ohmic contacts, recombination of carriers into the gap or into the bulk materials. The voltage drop is not only given by the junction but also by the series resistance. In open-circuit condition the current is zero and the voltage drop of the series resistance is also zero, so the open-circuit voltage is not aected. In short-circuit condition, current ows also through the series resistance and the current at short circuit is changed.
A.3
Shunt resistance
The shunt resistance is included to model the eect of low resistivity of some parts of the material, such as grain boundaries or secondary phases for example, and the eect of pinholes. This resistance is lower than the resistance of the diode and this gives place to a shunt of the current which results in a reduction of the voltage at open circuit.
A.4
Quantum eciency
The quantum eciency is dened as the ratio of the current of photogenerated carriers to the incident photon ux. The measured quantum eciency is determined by two phenomena: the generation eciency (ratio of the number of electron-hole pairs generated to the number of incident photons) and the collection eciency (ratio of the number of holes at the p contact and electrons at the n contact to the number of electron-hole pairs generated). As measured, this value is strictly referred to as the external quantum efciency. If corrections are made to allow for reection losses at the front of the cell, the quantity is then known as the internal quantum eciency. The quantum eciency of a solar cell depends strongly on the energy of the individual photon. This is due to both the wavelength dependence of the optical absorption coecients in semiconductors and the depth dependence
88
APPENDIX A. SOLAR CELL PARAMETERS
of the carrier collection probability.
A.5
Air Mass
The Air Mass (AM) value is correlated to the amount of air mass that the sunlight must cross to reach the device, dening a particular spectrum of light for a particular latitude. It is dened as the distance through the atmosphere that the light from the sun travels in order to reach the solar cell. If the sun is vertical (zenith) the path length is dened as 1 (A.M. 1). If the angle of the sun relative to the zenith increases, the air mass increases so that at 48 , A.M. is 1.5 and at 60 is 2.0. For most terrestrial applications, the generally accepted solar cell testing standard is that of A.M.1.5 conditions. In addition, it is usual to also specify the intensity of the light, integrated over the spectrum, as being 100mW/cm2 . When using solar concentrators, for example, the intensity might be increased by a factor of 1000, but the shape of the spectrum would remain A.M. 1.5.
Bibliography
[1] E. Becquerel. Comptes Rendues, pages 9, 561, 1839. [2] W.G. Adams and R.E. Day. In Proceedings of the Royal Society, pages A25, 113, 1877. [3] A.L.Fahrenbruch and H. Bube. Fundamentals of Solar Cells. Academic Press, London, 1983. [4] D.M. Chapin, C.S. Fuller, and G.L. Pearson. Journal of Applied Physics, 1954. [5] D.C. Reynolds, G. Leies, L.L. Antes, and R.E. Marburger. Physical Review, 96:533534, 1954. [6] Jenny D.A., Lofersky J.J., and Rappaport P. Physical Review, 101:1208 1209, 1956. [7] Martin A. Green, Keith Emery, David L. King, and Sanekazu Igari. Progress in Photovoltaics: Research and Application, 8(1):187195, 2000. [8] B.M. Basol. In Proceedings of 21st IEEE Photovoltaic Specialists Conference, Kissimimee., pages 588594, 1990. [9] E.I. Adirovich, Y.M. Yuabov, and G.R. Yagudaev. Poluprovodnikov, 1(3):8185, 1969. Fiz. Tekh.
[10] D. Bonnet and H. Rabenhorst. Proceedings of 9th IEEE Photovoltaic Specialists Conference, Silver Spring, pages 129131, 1972. [11] S. Wagner, J. L. Shay, P. Migliorato, and H. M. Kasper. (25):434435, 1974. 89
90
BIBLIOGRAPHY
[12] Kazmerski L.L., White F. R., and Morgan G.K. Applied Physics Letters, (29):268270, 1976. [13] R.L. Anderson. Solid State Electronics, 5:341, 1962. [14] J. Fritsche, D. Kraft, A. Thissen, A. Klein, and W. Jaegermann. In Proceedings of the Material Research Society Spring Meeting, page 668, 2001. [15] J. Fritsche, A. Thissen, A. Klein, and W. Jaegermann. Thin Solid Films, 387:158160, 2001. [16] L. Frantzis, E. Jones, A. D. Little, C. Lee, M. Wood, and P. Wormser. In Proceedings of 16th European Photovoltaic Solar Energy Conference, Glasgow, Scotland, pages 21002103, 2000. [17] Hutchby J. A., Markunas R.J., Timmons M.L., Chiang P.K., and Bedair S.M. In 18th IEEE Photovoltaic Specialists Conference, Las Vegas, NE, USA, pages 2027, 1985. [18] D.E. Carlson and C.R. Wronsky. Applied Physics Letters, (28):671673, 1976. [19] M.D. Archer and R. Hill. Clean Electricity from Photovoltaics, 2001. [20] Dieter Bonnet. Proceedings of 14th European Photovoltaic Solar Energy Conference, Barcelona, Spain, pages 26882693, 1997. [21] H. Ohyama, T. Aramoto, S. Kumazawa, H. Higuchi, T. Arita, S. Shibutari, T. Nishio, J. Nakajima, M. Tsuji, A. Hanafusa, T. Hibiro, K. Omura, and M. Murazawa. In 26th IEEE Photovoltaic Specialist Conference, Anaheim., pages 343346, 1997. [22] H.W. Schock and K. Bogus. In 2nd World Conference and Exhibition on Photovoltaic Solar Energy Conversion, pages 35863589, 1998. [23] N.Romeo, A. Bosio, and V. Canevari. In Proceedings of the 13th European Photovoltaic Solar Energy Conference and Exhibition, Nice, France, page 2011, 1995. [24] W.E. Devaney, W.S. Chen, J.M. Steward, and R.A.Mickelson. IEEE Transaction of Electronic Devices, (ED-37):428433, 1990.
BIBLIOGRAPHY
91
[25] Miguel A. Contreras, Brian Egaas, K. Ramanathan, J. Hiltner, A. Swartzlander, F. Hasoon, and Rommel Nou. In Conf. Record 26th IEEE Photovoltaic Specialist Conf., Anaheim, CA, USA., number 7, pages 311316, 1999. [26] Z. Zhao, D.L. Morel, and C.S. Ferekides. In Proceedings of 16th European Photovoltaic Solar Energy Conference, Glasgow, pages 808811, 2000. [27] A.Romeo, H. Zogg, and A.N. Tiwari. In Proceedings of the 2nd World Conference and Exhibition on Photovoltaic Solar Energy Conversion, Vienna, Austria, page 1105, 1998. [28] Bonnet D. Solar Energy Materials and Solar Cells, 361:547552, 2000. [29] A.K. Urner, J.M. Woodcock, M.E. Oszan, D.W. Cunningham, D.R. Johnson, R.J. Marshall, N.B. Mason, S. Oktik, M.H. Patterson, S.J. Ransome, S. Roberts, M. Sadeghi, J.M. Sherborne, D. Sivapathasundaram, and I.A Walls. Solar Energy Materials and Solar Cells, (35):263 270, 1994. [30] N.Romeo, A. Bosio, R. Tedeschi, A. Romeo, V. Canevari, and D. Leone. In Proceedings of 14th European Photovoltaic Solar Energy Conference, Barcelona, Spain, pages 23512352, 1997. [31] H.R. Moutinho, M.M. Al-Jassim, F.A. Abulfotuh, D.H. Levi, P.C. Pippo, R.G. Dhere, and L.L. Kazmerski. In 26th Photovoltaic Solar Energy Conference, Anaheim, CA, USA, pages 431434, September 1997. [32] K.D. Rogers, J.D. Painter, D.W. Lane, and M.Healy. Journal of Electronic Materials, (28):112117, 1999. [33] Paul Edwards. Beam-Induced Current Studies of CdTe/CdS Solar Cells. PhD thesis, University of Durham, UK, December 1998. [34] Dobson KD, Visoly-Fisher I, Hodes G, and et al. Solar Energy Materials and Solar Cells, 3(62):295325, 2000. [35] Rose DH, Hasoon FS, Dhere RG, Albin DS, Ribelin RM, Li XS, Mahathongdy Y, Gessert TA, and Sheldon P. PV Progress in Photovoltaics, 5(7):331340, 1999.
92
BIBLIOGRAPHY
[36] N.Romeo, A.Bosio, R.Tedeschi, A.Romeo, and V.Canevari. Solar Energy Materials and Solar Cells, 2(58):209218, 1999. [37] J.Britt and C.Ferekides. Applied Physics Letter, 62:28512852, 1993. [38] S. Kumazawa, S. Shibutani, T. Nishio, T. Aramato, H. Higuchi, T. Arita, A. Hanafusa, K.Omura, M. Murazano, and H.Takamura. In International Photovoltaic Solar Energy Conference-9, Miyazaki, Japan, pages 113114, 1996. [39] L. Maissel and R. Clang. Handbook of Thin Film Technology. Mc GrawHill Company, 1970. [40] Jeong WJ and Park GC. (65):3745, 2001. Solar Energy Materials and Solar Cells,
[41] H.R. Moutinho, F.S. Hasoon, F. Abulfotuh, and L.L Kazmerski. Journal of Vacuum Science and Technology, 6(A 13):28772883, 1995. [42] Moutinho M., Al-Jassim, D.H. Levi, P.C. Pippo, and .L Kazmerski. Journal of Vacuum Science and Technology, 3(A 16):12511257, 1998. [43] S. A. Ringel, A. W. Smith, M. H. MacDougal, , and A. Rohatgi. Journal of Vacuum Science and Technology, (70):881889, 1991. [44] M.E. Oszan, D. R. Johnson, S. Oktik, M. H. Patterson, D. Sivapathasundaram, and J.M. Woodcock. In 12th Photovoltaic European Conference and Exhibition, page 1604, 1994. [45] K. Durose, P.R. Edwards, and D.P. Halliday. Journal of Crystal Growth, (197):733742, 1999. [46] A.Taylor and H.Sinclair. Proceedings of Physical Society, 57:126, 1945. [47] N.Romeo, A.Bosio, and V.Canevari. In Proceedings of 11th European Photovoltaic Solar Energy Conference, Montreux, Switzerland, pages 972973, 1992. [48] X. Mathew, P.J. Sebastian, A.Sanchez, and J.Campos. Solar Energy Materials and Solar Cells, (59):99114, 1999. [49] C.S. Ferekides, D. Marinskiy, V. Viswanathan, B. Tetali, V. Palekis, P. Selvaraj, and D.L. Morel. Thin Solid Films, (361-362):520526, 2000.
BIBLIOGRAPHY
93
[50] M. Altosaar, P.E. Kukk, and E. Mellikov. Thin Solid Films, (361362):443445, 2000. [51] A. Jasenek, T. Hahn, M. Schmidt, K. Weinert, M.Wimbor, G. Hanna, K. Orgassa, M. Hartmann, H.W. Schock, U. Rau, J.H. Werner, B. Schattat, S. Kraft, K.H. Schmid, W. Bolse, G. La Roche, A. Robben, and K Bogus. In 16th Photovoltaic Solar Energy Conference and Exhibition, pages 982985, 2000. [52] R.M Burgess, W.S. Chen, W.E. Devaney, D.H. Doyle, N. P. Kim, and B.J. Stanbery. IEEE-1998, pages 909912, 1998. [53] G.P. Summers, S.R. Messenger, E.A. Burke, M. A. Xapsos, and R.J: Walters. IEEE Transactions on Nuclear Science, 40(6):1372, 1996. [54] Helmuth Spieler. Lecture notes, 1996. [55] B.M. Basol, V.K. Kapur, C.R. Leidholm, , and A. Halani. 25 IEEE PSC, page 157, 1996. [56] A.N. Tiwari, M. Krejci, F.-J. Haug, and H. Zogg. Progress in Photovoltaics; Research and Application, (7):393, 1999. [57] W. Wang, X. Wang, G. Thompson, J. C. McClure, , and V. P. Singh. page 1055, 1998. [58] A. Seth, G.B. Lush, J.C. McClure, V.P. singh, , and D. Flood. Solar Energy Materials and Solar Cells, 59:35, 1999.
94
BIBLIOGRAPHY
List of abbreviations
AFM AM CBD DC f.f. FTO HVE Isc ITO MBE PL PV PVD RF SEM TCO Voc XRD Atomic force microscopy Air Mass Chemical bath deposition Direct current Eciency Fill factor Tin oxide doped with uorine High vacuum evaporation Short circuit current Indium tin oxide Molecular beam epitaxy Photoluminescence Photovoltaic Physical vapor deposition Radio frequency Scanning electron microscopy Transparent conducting oxide Open circuit voltage X-ray diraction
95
96
Acknowledgments
My thank goes to Prof. Emeritus Hans Melchior, former head of the Institute of Quantum Electronics, for accepting me as his Ph.D. student. I would like to thank Prof. Hans-Rudolf Ott for taking care of the Thin Film Physics group after Prof. Melchiors retirement. I am very much indebted to Prof. Gernot Kostorz for having accepted to be the examiner of this thesis, his suggestions and support have been very helpful for the preparation of this work. I would like to thank Prof. Marc Burgelman for having accepted to be coexaminer of this thesis. I am very grateful to PD Dr. Hans Zogg, head of the Thin Film Physics group, for having given me the opportunity to write my thesis in his group and for his hints and useful discussions. My special thanks go to Dr. Ayodhya N. Tiwari for his encouragement and his support. This thesis was only possible due to his skills and knowledge in the solar cells eld. He is for me more than a good boss, during these years his suggestions and discussions were not only of great help but also an important professional and personal lesson. My gratitude is also addressed to: Dr. Karim Alchalabi, Dr. Yordan Athanassov, Derk Btzner, Dr. Franza Josef Haug, Dr. Joachim John, Marc Klin, Thomas Kmpfer, Klaus Kellera a mann, Michel Leopold, Dr. Martin Krejci, Dominik Rudmann, Matthias Terheggen, Dr. Silke Schn, Gerhard Schreiber, Dmitri Zimin, members and o former members of the Thin Film Physics group. I very much appreciated the exchange of ideas and their support in the lab. I would like to thank Joachim and Martin for their help during the rst year of my permanence in Zrich, Franz and Karim for the helpful hints for writu 97
ing this thesis and Derk with whom I had the pleasure to exchange ideas and discussions, sharing the interest in CdTe/CdS solar cells. Dr. Mathias Wagner from the Institute for Inorganic Chemistry, Division of Chemical Electron Microscopy, University of Zurich for the XRD measurements. Dr. Carlo Vignali from Centro Interfacolt` Misure, University of Parma, a (Italy) for the AFM measurements. The Nonmetallic Inorganic Materials group, Swiss Federal Institute of Technology for the XRD measurements. Dr. Neil Gibson from the Institute of Energy Consumption, Joint Research Center, Ispra (Italy) for the glancing angle XRD measurements. Dr. Wojitek Hajdas from Paul Scherrer Institute for the proton irradiation. It has been for me a great pleasure to work on the LACTEL and CADBACK European joint projects, exchanging ideas and knowledge with some of the best European solar cells laboratories. Last but not least, I am very much grateful to my father Prof. Nicola Romeo for his precious hints, his personal support and his incredible humanity. Thank you Silvie for having always supported me with your patience and understanding. The nancial support from the Federal Oce of Education and Science is acknowledged.
Mais uma vez obrigado Ayrton
98
Publications and Presentations

1. A.Romeo, A.N. Tiwari, H Zogg, M. Wagner, J.R. Guenter Inuence of Transparent Conducting Oxides on the Properties of CdTe/CdS Solar Cells Proceedings of 2nd World Conference and Exhibition on Photovoltaic Solar Energy Conversion 6-10 July 1997 Vienna (Austria), 1105-1108. 2. A.N. Tiwari, M.Krejci, F. -J. Haug, A.Romeo, H.Zogg. Dnnschichtsolarzellen aus Verbindunghalbleitern Proceedings of National Symposium of Photovoltaics, 5 May 1998, Hotel Schweizerhof, Bern, Switzerland. 3. A.Romeo, A.N. Tiwari, D.L. Btzner, M.Krejci, F. -J. Haug, H.Zogg. a Growth and recrystallization of CdTe on dierent transparent conducting oxides Proceedings of 15th Meeting of the Swiss Group of Surfaces and Interfaces, 28th of January 1999, Fribourg, Switzerland. 4. A.Romeo, D.L. Btzner, H Zogg, A.N. Tiwari. Recrystallization in a CdTe/CdS Proceedings of 1999 European Material Conference, June 1-4,1999, Strasbourg, France. Thin Solid Films, 361-362 (1-2) (2000), 420-425. 5. D.L. Btzner, A.Romeo, H Zogg, A.N. Tiwari. A Study of the Back a Contacts on CdTe/CdS Solar cells Proceedings of 1999 European Material Conference, June 1-4,1999, Strasbourg, France. Thin Solid Films, 361-362 (1-2) (2000), 463-467. 6. A.Romeo, D.L. Btzner, H Zogg, A.N. Tiwari, C. Vignali. Inuence a of CdS growth process on the structural and photovoltaic properties of CdTe/ CdS solar cells Proceedings of 11th International Photovoltaic Science and Engineering Conference (PVSEC- 11), September 99
20-24,1999, Sapporo, Japan (oral presentation). Solar Energy Materials and Solar Cells 58 (2) (1999) pp 209-218. 7. M. Wimbor, A. Romeo, M. Igalson Electrical Characterisation of CdTe/CdS Photovoltaic Devices Proceedings of European Conference on Photovoltaics 25-27 October 1999 Krakow Poland. Opto-Electronics Review 8(4), (2000), 375-377. 8. M. Igalson, P. Zabierowski, A. Romeo, L.Stolt Reverse-bias DLTS for investigation of the interface region in thin lm solar cells Proceedings of European Conference on Photovoltaics 25-27 October 1999 Krakow Poland. Opto-Electronics Review 8(4), (2000), 346-349. 9. D.L. Btzner, A.Romeo, H Zogg, A.N. Tiwari. Eect of back contact a metallization on the material stability of CdTe/CdS solar cells Proceedings of 16th European Photovoltaic Solar Energy Conference and Exhibition, May 1- 5,2000, Glasgow, Scotland, 353-356. 10. P.R. Edwards and K.Durose, A.Romeo, D.L. Btzner and A.N.Tiwari a Comparison of the microscopic electrical properties of sublimation and evaporation grown CdS/CdTe solar cells using lock-in EBIC Proceedings of 16th European Photovoltaic Solar Energy Conference and Exhibition, May 1- 5,2000, Glasgow, Scotland, 839-842. 11. A.Romeo, D.L. Btzner, H Zogg, A.N. Tiwari. A comparison of the a vacuum evaporated CdTe substrate and superstrate solar cells Proceedings of 16th European Photovoltaic Solar Energy Conference and Exhibition, May 1- 5,2000, Glasgow, Scotland, 843-846. 12. D.L. Btzner, A.Romeo, H Zogg, A.N. Tiwari. Development of ea cient and stable back contacts on CdTe/CdS solar cells Proceedings of 2000 European Material Conference, May 30-June 6,2000, Strasbourg, France. Thin Solid Films 387 (2001) 151-154. 13. A.Romeo, D.L.Btzner, H.Zogg and A.N.Tiwari. Inuence of Proton a Irradiation and development of exible CdTe cells on polyimide Proceedings of 2001 Material Research Society Spring Meeting, 16-20 April 2001, San Francisco CA, USA (oral presentation), to be published. 100
14. D.L.Btzner, G.Agostinelli, J. Beier, A.Romeo, H.Zogg and A.N.Tiwari. a Voltage dependent carrier collection in CdTe solar cells. Proceedings of 2001 Material Research Society Spring Meeting, 16-20 April 2001, San Francisco CA, USA, to be published. 15. A.N.Tiwari, A.Romeo, D.L. Btzner, H.Zogg. Flexible CdTe solar cells a on polymer lms Progress in Photovoltaics: Research and Applications 2001; 9:211-215. 16. D.L. Btzner, A. Romeo, H. Zogg and A. N. Tiwari. Radiation Harda ness of CdTe/CdS solar cells Accepted for publication in the Proceedings of 17th European Photovoltaic Solar Energy Conference and Exhibition, 22-26 October 2001, Munich, Germany, to be published. 17. D.L. Btzner, A. Romeo, H. Zogg and A. N. Tiwari. CdTe/CdS Solar a Cell Performance under Low Light Irradiation. Proceedings of 17th European Photovoltaic Solar Energy Conference and Exhibition, 22-26 October 2001, Munich, Germany, to be published. 18. M. Terheggen, H. Heinrich, G. Kostorz, A. Romeo, H. Zogg and A.N. Tiwari. Structural and Chemical Studies on CdTe/CdS Thin Film Solar Cells with Analytical Transmission Electron Microscopy. Proceedings of 17th European Photovoltaic Solar Energy Conference and Exhibition, 22-26 October 2001, Munich, Germany, to be published. 19. A.Romeo, D.L. Btzner, W. Hajdas, H. Zogg and A.N. Tiwari. Potena tial of CdTe Thin Film Solar Cells for Space Application Proceedings of 17th European Photovoltaic Solar Energy Conference and Exhibition, 22-26 October 2001, Munich, Germany, to be published.
Award:
Student Silver Award of the American Material Research Society (2001 Material Research Society Spring Meeting, 16-20 April 2001, San Francisco CA, USA)
101
102
Curriculum vitae
Name: Romeo Vorname: Alessandro Adresse: Nrenbergstrasse 23, 8037 Zrich u u Geboren am: 6. October 1969 in Parma (Italien) Schulen: Scuola Elementare (Parma, Italien) 1975-1980 Scuola Media (G. Ferrari) (Parma, Italien) 1980-1983 Middle School (Graham Hill), Mountain View, CA, U.S.A. 1981-1982 Liceo Scientico G. Marconi 1983-1988 Diploma di maturita scientica 1988 Hochschule: Physikstudium an der Universitt Parma 1988-1995 a Corso di specializzazione per insegnanti di matematica (Spezialkurs zum Matematik Lehrer) an der Universitt Parma 1997 a Doktorarbeit an der ETH Zrich 1998-2001 u
Wissenschaftlicher Mitarbeiter an der Universitt Parma 1996-1997 a Wissenschaftlicher Mitarbeiter an der ETH Zrich 1997 u
103

Eth 14436

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Eth 14436

Uploaded by

Copyright:

Available Formats

Diss. ETH Nr.

Growth and Characterization of High Eciency CdTe/CdS Solar Cells

A dissertation submitted to the

SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH

La science na pas de patrie. Luis Pasteur, 14 November 1888

4 Inuence of CdS growth process 29 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.2 Morphology of CdS . . . . . . . . . . . . . . . . . . . . . . . . 30 v

4.7 4.8 4.9 4.10 4.11

6.6 7.1 7.2

Principles of solar cells

1.2. PRINCIPLES OF SOLAR CELLS

1.2. PRINCIPLES OF SOLAR CELLS

0 0 -20 500 1000

Figure 1.2: Band diagrams of a homojunction.

1.3. CURRENT STATUS

1.3. CURRENT STATUS

Chapter 2 The CdTe/CdS solar cell

CHAPTER 2. THE CdTe/CdS SOLAR CELL

Front electrical contact

CdS window layer

2.5. CdTe ABSORBER LAYER

CdTe absorber layer

CHAPTER 2. THE CdTe/CdS SOLAR CELL

The CdCl2 treatment

Electrical back contact

2.6. ELECTRICAL BACK CONTACT

CHAPTER 2. THE CdTe/CdS SOLAR CELL

Chapter 3 Inuence of transparent conducting oxides

Solar cell preparation

CHAPTER 3. TRANSPARENT CONDUCTING OXIDES

ITO FTO ZnO

0 300 400 500 600 700 800

Characterization of the layers

3.3. CHARACTERIZATION OF THE LAYERS

ITO FTO ZnO

0 500 600 700 800

CHAPTER 3. TRANSPARENT CONDUCTING OXIDES

3.3. CHARACTERIZATION OF THE LAYERS

CHAPTER 3. TRANSPARENT CONDUCTING OXIDES

FTO as deposited FTO 600 nm CdCl2 treated 533 111 531

100 102 104

Figure 3.5: X-ray diraction pattern of CdTe deposited on CdS/FTO/glass.

3.3. CHARACTERIZATION OF THE LAYERS

ITO As-deposited ITO 600 nm CdCl2 annealed 531

100 85 90 95 100 105

Figure 3.6: X-ray diraction pattern of CdTe deposited on CdS/ITO/glass.

CHAPTER 3. TRANSPARENT CONDUCTING OXIDES

ZnO as deposited ZnO 600 nm CdCl2 treated 111

8000 6000 4000 2000 0 -2000 20 22 24 26 28

0 84 86 88 90 92 94 96 98 100 102 104

Figure 3.7: X-ray diraction pattern of CdTe deposited on CdS/ZnO/glass.

Solar cell performance

CHAPTER 3. TRANSPARENT CONDUCTING OXIDES

FTO.600nm CdCl2 ITO.600nm CdCl2 ZnO.600nm CdCl2 ITO.60nm CdCl2

0 -1000 -500 0 500 1000

CHAPTER 3. TRANSPARENT CONDUCTING OXIDES

Chapter 4 Inuence of CdS growth process

CHAPTER 4. INFLUENCE OF CdS GROWTH PROCESS

4.2. MORPHOLOGY OF CdS

CHAPTER 4. INFLUENCE OF CdS GROWTH PROCESS

4.2. MORPHOLOGY OF CdS

CHAPTER 4. INFLUENCE OF CdS GROWTH PROCESS

Morphology and crystallization of CdTe

4.3. MORPHOLOGY AND CRYSTALLIZATION OF CdTe