Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

On the material structure of hydrogenated nanocrystalline silicon Schmidt, Kathrin Jessica 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2014_spring_schmidt_kathrin.pdf [ 1.21MB ]
Metadata
JSON: 24-1.0074324.json
JSON-LD: 24-1.0074324-ld.json
RDF/XML (Pretty): 24-1.0074324-rdf.xml
RDF/JSON: 24-1.0074324-rdf.json
Turtle: 24-1.0074324-turtle.txt
N-Triples: 24-1.0074324-rdf-ntriples.txt
Original Record: 24-1.0074324-source.json
Full Text
24-1.0074324-fulltext.txt
Citation
24-1.0074324.ris

Full Text

On the material structure ofhydrogenated nanocrystalline siliconbyKathrin Jessica SchmidtB.A.Sc., The University of British Columbia (UBC), 2011A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE COLLEGE OF GRADUATE STUDIES(Electrical Engineering)THE UNIVERSITY OF BRITISH COLUMBIA(Okanagan)February 2014c? Kathrin Jessica Schmidt, 2014AbstractThe dependence of the crystalline volume fraction on the mean crystal-lite size, for three hydrogenated nanocrystalline silicon-based photovoltaicsolar cells, is examined. For each photovoltaic solar cell considered, X-raydiffraction and Raman spectra are determined. Through the application ofScherrer?s equation, the X-ray diffraction results are used in order to deter-mine the corresponding mean crystallite sizes. Through peak decomposition,the Raman results are used in order to estimate the corresponding crystallinevolume fraction. By plotting the crystalline volume fraction as a functionof the mean crystallite size, it is found that larger mean crystallite sizestend to favor reduced crystalline volume fractions. The ability to randomlypack smaller crystallites with a greater packing fraction than their largercounterparts is suggested as a possible explanation for this observation.iiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . xixList of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . .xxiiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xxivChapter 1: Introduction . . . . . . . . . . . . . . . . . . . . . . . 1Chapter 2: Background . . . . . . . . . . . . . . . . . . . . . . . 102.1 On the use of nc-Si:H . . . . . . . . . . . . . . . . . . . . . . 10iiiTABLE OF CONTENTS2.2 A brief history of thin-film nc-Si:H technology . . . . . . . . . 122.2.1 The material structure of nc-Si:H-based photovoltaicsolar cells . . . . . . . . . . . . . . . . . . . . . . . . . 172.3 X-ray diffraction . . . . . . . . . . . . . . . . . . . . . . . . . 202.4 Raman spectroscopy and the crystalline volume fraction . . . 26Chapter 3: Experiment . . . . . . . . . . . . . . . . . . . . . . . 323.1 The analysis of nc-Si:H . . . . . . . . . . . . . . . . . . . . . . 323.2 The nc-Si:H-based photovoltaic solar cell samples . . . . . . . 343.3 X-ray diffraction spectroscopy . . . . . . . . . . . . . . . . . . 373.4 Raman spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 43Chapter 4: Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 494.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494.2 XRD analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 504.3 Raman analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 554.4 Experimental resolution limitation . . . . . . . . . . . . . . . 634.5 The dependence of the crystalline volume fraction on themean crystallite size . . . . . . . . . . . . . . . . . . . . . . . 65Chapter 5: Conclusions . . . . . . . . . . . . . . . . . . . . . . . 68ivTABLE OF CONTENTSReferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70vList of TablesTable 2.1 Calculated Bragg?s reflection peak angles, ? and 2?. . . 24Table 4.1 Summary of the mean crystallite sizes and the crys-talline volume fractions found by XRD and Ramanspectroscopy, respectively. . . . . . . . . . . . . . . . . 54viList of FiguresFigure 1.1 The number of transistors on a microprocessor as afunction of year. A least-squares fit is depicted withthe red dashed line. This figure is modified from Hadiet al. [6]. The online version of this figure is in color. 3Figure 1.2 A sampling of large area electronic device applica-tions. These images are from O?Leary [9]. The onlineversion of this figure is in color. . . . . . . . . . . . . 4Figure 1.3 The conductivity of a-Si:H as a function of time. Thisplot is after Figure 1 of Staebler and Wronski [18]. . . 7viiLIST OF FIGURESFigure 2.1 A sample cross-sectional dark-field TEM image ofan n-i-p nc-Si:H-based photovoltaic solar cell. Thebright areas, which correspond to crystallites of thesame crystallographic orientation, form a structureparallel to the growth direction. The a-Si:H incuba-tion layer, appearing overall grayish, can be observedat the onset of the intrinsic nc-Si:H layer. This imageis from Droz et al. [64]. . . . . . . . . . . . . . . . . . 19Figure 2.2 Several crystal planes and an interacting incoming X-ray beam are depicted. In this figure, ? is the anglebetween the incident beam and the crystal planes andd is the interplanar separation for the crystal struc-ture. This figure is after Cullity and Stock [67]. . . . 22Figure 2.3 The unit cell associated with c-Si. This figure is afterFranssila [68]. . . . . . . . . . . . . . . . . . . . . . . 23Figure 2.4 A representative Raman spectrum for the mineral re-algar depicting the Rayleigh, Stokes, and anti-Stokesscattering ranges. The actual intensity of the Rayleighscattering was suppressed. This image is from Van-denabeele [74]. . . . . . . . . . . . . . . . . . . . . . . 28viiiLIST OF FIGURESFigure 2.5 An illustration of a standard confocal Raman mi-croscopy setup is depicted in the figure. This figure isfrom Hollricher[75]. The online version of this figureis in color. . . . . . . . . . . . . . . . . . . . . . . . . 29Figure 2.6 Typical Stokes scattering in a Raman spectrum fornc-Si:H and a-Si:H. The online version of this figureis in color. . . . . . . . . . . . . . . . . . . . . . . . . 31Figure 3.1 A cross-section of the nc-Si:H-based n-i-p photovoltaicsolar cells considered in this analysis. The dimensionsof the n-, p-, and intrinsic layers are indicated. Thebuffer and highly crystalline seed layers are assumedto be of smaller dimensions than the n- or p-layers.All of the other dimensions are unknown. The ITOcontacts are only present for the case of sample 21886.The online version of this figure is depicted in color. . 36ixLIST OF FIGURESFigure 3.2 These photographs correspond to the actual nc-Si:H-based photovoltaic solar cell samples which are con-sidered in this analysis, i.e., these are the samples thatare considered in the XRD and Raman analysis con-sidered for the pupose of this study. As may be read-ily seen, the second sample, i.e., sample 21886, hasITO contacts deposited on top of the boron-dopedtop p-layer. The online version of this figure is de-picted in color. . . . . . . . . . . . . . . . . . . . . . . 38Figure 3.3 The XRD spectrum corresponding to sample 21821.A basic high-rate scan, corresponding to the entirespectral range, is depicted with the light solid line.High-quality scans, corresponding to the individualpeaks, are depicted with the heavier solid lines. Thehigher-quality scan results are vertically offset fromthe other results for the purposes of greater clarity.The peaks observed, and their origin, are indicated inthe figure. . . . . . . . . . . . . . . . . . . . . . . . . 40xLIST OF FIGURESFigure 3.4 The XRD spectrum corresponding to sample 21886.A basic high-rate scan, corresponding to the entirespectral range, is depicted with the solid line. Thepeaks observed, and their origin, are indicated in thefigure. . . . . . . . . . . . . . . . . . . . . . . . . . . . 41Figure 3.5 The XRD spectrum corresponding to sample 21916.A basic high-rate scan, corresponding to the entirespectral range, is depicted with the solid line. Thepeaks observed, and their origin, are indicated in thefigure. . . . . . . . . . . . . . . . . . . . . . . . . . . . 42Figure 3.6 A representative Raman spectrum corresponding tonc-Si:H with the different modes identified. Note thatthis model omits the intermediate or grain boundarycomponent, located at approximately 510 cm?1. Thisimage is after Wei et al. [84]. . . . . . . . . . . . . . . 45xiLIST OF FIGURESFigure 3.7 The entire unprocessed Raman spectrum for the 21821nc-Si:H sample, from 20-2400 cm?1. For this analysis,particular attention is paid to the two predominantpeaks at 480 and 520 cm?1, these peaks correspond-ing to the amorphous and crystalline silicon peaksof this material, respectively. Further peaks, whichcorrespond to the different modes of amorphous sil-icon, are also observed, at approximately 150, 310,and 380 cm?1. The results are not accurate for thelower wave-numbers, i.e., less than 150 cm?1, due toinput laser interference. . . . . . . . . . . . . . . . . . 46xiiLIST OF FIGURESFigure 3.8 The entire unprocessed Raman spectrum for the 21886nc-Si:H sample, from 20-2400 cm?1. For this analysis,particular attention is paid to the two predominantpeaks at 480 and 520 cm?1, these peaks correspond-ing to the amorphous and crystalline silicon peaksof this material, respectively. Further peaks, whichcorrespond to the different modes of amorphous sil-icon, are also observed, at approximately 150, 310,and 380 cm?1. The results are not accurate for thelower wave-numbers, i.e., less than 150 cm?1, due toinput laser interference. . . . . . . . . . . . . . . . . . 47xiiiLIST OF FIGURESFigure 3.9 The entire unprocessed Raman spectrum for the 21916nc-Si:H sample, from 20-2400 cm?1. For this analysis,particular attention is paid to the two predominantpeaks at 480 and 520 cm?1, these peaks correspond-ing to the amorphous and crystalline silicon peaksof this material, respectively. Further peaks, whichcorrespond to the different modes of amorphous sil-icon, are also observed, at approximately 150, 310,and 380 cm?1. The results are not accurate for thelower wave-numbers, i.e., less than 150 cm?1, due toinput laser interference. . . . . . . . . . . . . . . . . . 48Figure 4.1 A subset of the XRD spectrum corresponding to sam-ple 21821. Baseline correction has been employed.The visible peaks, and their corresponding origin, areidentified in the figure. The inset focuses on the peakfound for the silicon (220) peak. The correspondingpeak fit is depicted with the red solid line. Care wastaken in order to ensure that the adjacent ZnO-basedpeak did not influence the obtained peak fit. Theonline version is depicted in color. . . . . . . . . . . . 51xivLIST OF FIGURESFigure 4.2 A subset of the XRD spectrum corresponding to sam-ple 21886. Baseline correction has been employed.The visible peaks, and their corresponding origin, areidentified in the figure. The inset focuses on the peakfound for the silicon (220) peak. The correspondingpeak fit is depicted with the red solid line. Care wastaken in order to ensure that the adjacent ZnO-basedpeak did not influence the obtained peak fit. Theonline version is depicted in color. . . . . . . . . . . . 52Figure 4.3 A subset of the XRD spectrum corresponding to sam-ple 21916. Baseline correction has been employed.The visible peaks, and their corresponding origin, areidentified in the figure. The inset focuses on the peakfound for the silicon (220) peak. The correspondingpeak fit is depicted with the red solid line. Care wastaken in order to ensure that the adjacent ZnO-basedpeak did not influence the obtained peak fit. Theonline version is depicted in color. . . . . . . . . . . . 53xvLIST OF FIGURESFigure 4.4 A subset of the Raman spectrum corresponding tosample 21821. The experimental data points are in-dicated with the open points. Linear baseline correc-tion has been employed between 420 to 540 cm?1. . . 56Figure 4.5 A subset of the Raman spectrum corresponding tosample 21886. The experimental data points are in-dicated with the open points. Linear baseline correc-tion has been employed between 420 to 540 cm?1. . . 57Figure 4.6 A subset of the Raman spectrum corresponding tosample 21916. The experimental data points are in-dicated with the open points. Linear baseline correc-tion has been employed between 420 to 540 cm?1. . . 58xviLIST OF FIGURESFigure 4.7 A subset of the Raman spectrum corresponding tosample 21821. The amorphous, grain boundary, andcrystalline peak components have been highlightedwith thin solid lines, the corresponding peaks beingaround 480, 510, and 520 cm?1, respectively. Theexperimental data points are indicated with the openpoints. The fit with experiment, obtained throughthe addition of all of the peaks, is indicated with thesolid red line. Baseline correction has been employed.The online version is depicted in color. . . . . . . . . 60Figure 4.8 A subset of the Raman spectrum corresponding tosample 21886. The amorphous, grain boundary, andcrystalline peak components have been highlightedwith thin solid lines, the corresponding peaks beingaround 480, 510, and 520 cm?1, respectively. Theexperimental data points are indicated with the openpoints. The fit with experiment, obtained throughthe addition of all of the peaks, is indicated with thesolid red line. Baseline correction has been employed.The online version is depicted in color. . . . . . . . . 61xviiLIST OF FIGURESFigure 4.9 A subset of the Raman spectrum corresponding tosample 21916. The amorphous, grain boundary, andcrystalline peak components have been highlightedwith thin solid lines, the corresponding peaks beingaround 480, 510, and 520 cm?1, respectively. Theexperimental data points are indicated with the openpoints. The fit with experiment, obtained throughthe addition of all of the peaks, is indicated with thesolid red line. Baseline correction has been employed.The online version is depicted in color. . . . . . . . . 62Figure 4.10 The attenuation of the X-ray source and the Ramanlaser source as a function of the depth into the mate-rial. A profile of the underlying nc-Si:H-based photo-voltaic solar cell device structure is depicted beneath.The online version of this figure is depicted in color . 64Figure 4.11 The crystalline volume fraction as a function of themean crystallite size for the three nc-Si:H-based pho-tovoltaic solar cells considered in this analysis [94].The error bars, corresponding to each data point, aredepicted. The online version is depicted in color . . . 66xviiiList of SymbolsSC silane concentration[SiH4] silane flow[H2] hydrogen flowR hydrogen dilution ratio? wavelength of the lightl pathlength differencen integer? angle between the incident beam and the crystal planesd interplanar separationdhkl interplanar separation in the silicon cubic diamond structureh, k, l Miller indicesa cubic unit cell dimensiondXRD mean crystallite sizeB full-width-at-half-maximumxixList of SymbolsK Scherrer constant?B Bragg reflection peak positionXC crystalline volume fractionXGB grain boundary volume fractionXA amorphous volume fractiony(L) scattering cross-section ratioxxList of Acronymsc-Si crystalline siliconSiH4 silane gasa-Si:H hydrogenated amorphous siliconnc-Si:H hydrogenated nanocrystalline siliconTFT thin-film transistorXRD X-ray diffractionPECVD plasma enhanced chemical vapor depositionVHF-PECVD very-high-frequency plasma enhanced chemical vapor depositionHWCVD hot-wire chemical vapor depositionSC silane concentrationTEM transmission electron microscopyFWHM full-width-at-half-maximumAg/ZnO silver/zinc-oxideITO indium-tin-oxidexxiList of AcronymsS/N signal-to-noiseTA transverse acousticLA longitudinal acousticLO longitudinal opticalTO transverse opticalxxiiAcknowledgementsThis thesis would not have been possible without the help and supportof my family, colleagues, and friends. I am heartily thankful to my su-pervisor, Dr. Stephen O?Leary, for both inspiring and teaching me. Hisconstant guidance, support, and encouragement helped me throughout mystudies. I would also like to extend my gratitude to Drs. Mario Beaudoinand Guangrui Xia and to Mr. Yiheng Lin, all of The University of BritishColumbia, for their support that they provided me during my experimentalexplorations. I would also like to thank Mr. Shamsul Chowdhury, my fellowgraduate colleague, who has been a great resource in answering questionsregarding the editing of my thesis. A special thanks is also dedicated toMr. Ken Guido, President of EMPAC Engineering Ltd., for giving me anopportunity, during my graduate studies, to develop practical engineeringskills.xxiiiDedicationTo my parents, Erhard and Gertrud Pietrzak especially, and my family in particular.This thesis would have not been possible without your support and love.xxivChapter 1IntroductionThe microelectronics revolution, which found its genesis with the devel-opment of the first transistor in 1947, continues to this very day [1]. Thetransistor is the elemental component from which modern microelectronicsystems are fabricated. Transistors are ubiquitous in modern societies, suchas Canada?s. They are found in telephones, televisions, automobiles, coffeemachines, and even vacuum cleaners. They allow for the processing of dataat previously unforeseen and unparalleled speeds. They have also allowed forthe introduction of mobile computing and personal communication devices,technologies that had not even been envisioned at the time of their devel-opment. They have played an essential role in the development of modernsociety, and most likely will continue to do so for the foreseeable future [2].Crystalline silicon (c-Si) is the electronic material which has been at theheart of the microelectronics revolution since the mid-1950s; crystalline ger-manium was used for the first generation of transistors, but its small bandgap, i.e., 0.66 eV at 300 K, limited its thermal robustness [3]. Accordingly,there has been a great amount of effort invested into developing a quanti-tative understanding of the properties of this material. This understandingof the material properties of c-Si has allowed for the development of many1Chapter 1. Introductionnew types of electron devices. This, in turn, has allowed for progress withinthe field of microelectronics.Progress in microelectronics is often described in terms of the so-calledMoore?s law. In 1965, Gordon E. Moore, a founder of Fairchild Semicon-ductor and a former member of Shockley Semiconductor Laboratory, sug-gested that the number of transistors on a microprocessor will double every18-months up to 1975 [4]; interestingly, Gordon E. Moore is also one ofthe ?Traitorous Eight? who all left Shockley Semiconductor Laboratory in1957 [5]. Plotting the number of transistors on an Intel microprocessor, from1970 through to 2010, as shown in Figure 1.1, reveals that Moore?s law hasmore-or-less held up to this very day [6]. Projections suggest that this lawwill hold for at least another decade [7, 8].Progress in conventional electronics has been achieved through reduc-tions in the dimensions of the transistors employed. There are, however,electronic devices that require size in order to function properly. Displaysand scanners, providing a human interface with the electronic world, mustbe of sufficient scale in order to operate effectively. Digital X-ray imagersand photovoltaic solar cells also require size in order to function. These de-vices are collectively referred to as large area electronic devices. A samplingof large area electronic device applications is depicted in Figure 1.2 [9].In large area electronics, the focus is on depositing thin-film materialsuniformly and inexpensively over large areas, while in conventional micro-electronics, the focus is on achieving sub-micron device features. Regret-tably, c-Si, that workhorse of the conventional electronics field, cannot bedeposited as a thin-film. Thus, alternate electronic materials must be sought2Chapter 1. Introduction1970 1975 1980 1985 1990 1995 2000 2005 2010103104105106107108109Number of transistors on a microprocessorYearFigure 1.1: The number of transistors on a microprocessor as a function ofyear. A least-squares fit is depicted with the red dashed line. This figure ismodified from Hadi et al. [6]. The online version of this figure is in color.3Chapter 1. IntroductionFigure 1.2: A sampling of large area electronic device applications. Theseimages are from O?Leary [9]. The online version of this figure is in color.4Chapter 1. Introductioninstead for large area electronic device applications.Initial efforts, aimed at developing a form of thin-film silicon for use inlarge area electronic device applications, focused on the use of sputteringand thermal evaporation. Unfortunately, the materials obtained possessedextremely high defect concentrations, preventing doping and limiting theobtained photoconductivity. Chittick et al. [10] were the first to employglow-discharge in the deposition of thin-film silicon. In their process, silanegas (SiH4) was dissociated into its constituent components through the useof an electrical plasma. The dissociated gas molecules then grew a form ofthin-film silicon on a heated substrate. This work demonstrated that theconcentration of defects can be substantially reduced when contrasted withthe case of materials deposited through sputtering and thermal evaporationprocesses. Unfortunately, a lack of sustained research funding terminatedthe efforts of Chittick et al. [10], in spite of the value of their pioneeringcontributions.Spear and Le Comber [11] continued the work of Chittick et al. [10].Initially, they focused on a detailed materials characterization of this newform of thin-film silicon [12]. Later, they demonstrated that doping canbe achieved using this material, astounding theorists who had previouslybelieved that this was in fact not possible. The fabrication of the first thin-film silicon based transistor was achieved shortly thereafter. Carlson andWronski [13] demonstrated the use of this material for the fabrication of aphotovoltaic solar cell. New applications for this form of thin-film siliconcontinue to be devised with each passing year.The copious amounts of hydrogen that are present during the deposi-5Chapter 1. Introductiontion process, i.e., for every silicon atom there are four hydrogen atoms insilane, suggested that hydrogen may play an important role in determiningthe properties of this material. Paul et al. [14, 15] were the first to explorethis possibility, and they did find that the presence of hydrogen was cru-cial for the favorable properties exhibited by this material, although, at thetime, their findings were considered very controversial; at the time, Spearand Le Comber [16] themselves emphatically rejected the notion that hy-drogen is present within this form of thin-film silicon. Eventually, however,the presence of hydrogen within this form of thin-film silicon came to beaccepted by the scientific community, and the material came to be knownas hydrogenated amorphous silicon (a-Si:H).For many years, a-Si:H has served as the electronic material of choice forlarge area electronics. The presence of hydrogen within this material wasfound to reduce the concentration of dangling bonds. Hydrogen atoms, beingvery light and mobile, attach themselves to the otherwise dangling siliconbonds [17]. This reduces the effective defect concentration found within thismaterial, thus leading to the favourable properties exhibited by this material.Unfortunately, hydrogen atoms are also believed to be responsible for theinstabilities that are observed within a-Si:H. The Staebler-Wronksi effect,in which the conductivity is found to reduce when the material is exposedto light, has been directly linked to the presence of hydrogen atoms withinthese thin-films; see Figure 1.3 [18]. This has motivated researchers to seekanother form of thin-film silicon, with the advantages of a-Si:H without itsdisadvantages.Hydrogenated nanocrystalline silicon (nc-Si:H), a material comprised of6Chapter 1. Introduction0 2 4 6 8 1010?1210?1010?810?610?410?2Time (h)Conductivity (? cm)?1 DARKAT = 25 ?COPTICALEXPOSURE DARKBFigure 1.3: The conductivity of a-Si:H as a function of time. This plot isafter Figure 1 of Staebler and Wronski [18].7Chapter 1. Introductionsilicon crystallites embedded in a a-Si:H tissue, is one possible form of thin-film silicon that has been successfully used in thin-film photovoltaic solarcell device applications since the early 1990s [19, 20]. It has been found thatthis material does not exhibit significant light-induced degradation whencontrasted with the case of a-Si:H [21]. Additionally, in terms of its potentialfor photovoltaic solar cell device applications, nc-Si:H has a lower band gapthan that associated with a-Si:H, i.e., approximately 1.1 eV for the case ofnc-Si:H as compared to approximately 1.7 eV for the case of a-Si:H [17, 22].Therefore, nc-Si:H can be used as a red-light absorber bottom cell in multi-junction thin-film technologies [23].Another attractive feature of nc-Si:H is its electron mobility, which ismuch higher than that associated with a-Si:H [24]; this suggests the presenceof high-mobility silicon crystallites embedded in a low-mobility a-Si:H tissue,the electron mobility associated with a-Si:H being limited by the band-tailstates that are present, these band-tail states acting as electron/hole trap-ping centers within the material [25], thus rendering the material unsuitablefor fast switching thin-film transistor (TFT) device applications. For pho-tovoltaic solar cell device applications, however, in order to produce highquality nc-Si:H, the greatest care has to be exercised when growing thenc-Si:H layers, as the structural properties of this material directly impactupon its electrical and optical properties [26]. This will impact upon thecorresponding photovoltaic solar cell device performance.In this thesis, the structural properties of a number of n-i-p nc-Si:H-based photovoltaic solar cells is investigated. In particular, through the useof X-ray diffraction (XRD), the mean crystallite sizes associated with these8Chapter 1. Introductionsamples will be determined through the use of Scherrer?s equation. Then,through the use of Raman spectroscopy, the corresponding crystalline vol-ume fractions will be evaluated. The dependence of the crystalline volumefraction on the mean crystallite size will then be explored. Using these re-sults, conclusions, related to the material structure of nc-Si:H, will be drawn.The XRD and Raman results, employed by the author for the purposes ofthis study, were obtained through experimental measurements performed bythe author in conjunction with Dr. M. Beaudoin (XRD) and Mr. Y. Lin(Raman), both of The University of British Columbia.This thesis is organized in the following manner. In Chapter 2, thebackground for this work is provided. Then, in Chapter 3, the experimentaltechniques, employed for purposes of this analysis, are detailed. In addition,the samples investigated in this analysis are discussed. In Chapter 4, theresults of these experimental explorations are analyzed, and the dependenceof the crystalline volume fraction on the mean crystallite size is explored.Finally, in Chapter 5, the conclusions of this analysis are presented, somesuggestions for further exploration being provided.9Chapter 2Background2.1 On the use of nc-Si:HAt this point, nc-Si:H is viewed as a complex two-phase material, com-prised of silicon crystallites embedded in an amorphous silicon tissue. Thepresence of grain boundaries, located at the edges of the crystallites, hasalso been observed, further complicating the picture that has emerged [27].Voids are also found within this material, as with the case of a-Si:H [28].The a-Si:H regions exhibit properties that are similar of those found withinthin-films of pure a-Si:H. In particular, the short-range order and long-rangedisorder present within conventional a-Si:H is also found within the a-Si:Hphase of nc-Si:H, i.e., the bond angle and bond length deviations foundwithin conventional a-Si:H are also found within the a-Si:H regions of nc-Si:H [17, 29].In addition to these a-Si:H regions, nc-Si:H also exhibits regions of highlycrystalline order, i.e., small silicon crystallites. These small silicon crystal-lites are embedded in the amorphous tissue and provide some unique and,from the point of view of photovoltaic device applications, beneficial proper-ties. The formation and structure of these small silicon crystallites dependson the deposition process parameters, as will be discussed in subsequent sec-102.1. On the use of nc-Si:Htions. Before going into the material structure and the inherent advantagesoffered by nc-Si:H over a-Si:H, the evolution of nc-Si:H into its current formis presented.Owing to its considerable potential for device applications, nc-Si:H hasattracted a great deal of attention in recent years [30?32]. Thus far, nc-Si:H has been considered for TFT [33] and photovoltaic solar cell deviceapplications [34]. New applications are emerging with each passing year [35,36]. In addition to its great potential for device applications, owing to itscomplexity as a material system, nc-Si:H also offers great opportunities forfundamental discovery in terms of the role that disorder plays in shapingthe material properties of this form of thin-film silicon.As material properties lie at the heart of many of the proposed de-vice applications envisaged for this material, the literature abounds withstudies into the material properties of nc-Si:H [28, 37?41]. Finger providesan excellent, but by no means exhaustive, summary of relevant studies onthe material properties of nc-Si:H in the textbook ?Thin-film silicon solarcells? [28].In this chapter, background material, related to the study of nc-Si:H, willbe provided, this material being drawn upon in the subsequent chapters.Initially, the focus will be on nc-Si:H itself, the discussion ranging frommeans of deposition to the general material properties, with a particularemphasis on those material properties that are particularly important forphotovoltaic solar cell device applications. Then, a brief introduction to theanalytical tools used for the purposes of this study, i.e., XRD and Ramanspectroscopy, is provided. As each analytical technique employed in this112.2. A brief history of thin-film nc-Si:H technologystudy represents vast areas of inquiry in their own right, the primary focusof this discussion will be on how these scientific instruments are used in thisparticular study.This chapter is organized in the following manner. In Section 2.2, theorigins of nc-Si:H, as a potential material for electron device applications,are discussed. Then, in Section 2.3, the basic material properties associatedwith nc-Si:H are outlined. The use of XRD spectroscopy, for the purposesof characterizing nc-Si:H, is featured in Section 2.4. Finally, the use ofRaman spectroscopy, for the purposes of characterizing nc-Si:H, is describedin Section 2.5.2.2 A brief history of thin-film nc-Si:HtechnologyBy the late 1960s, there was a general recognition that forms of thin-filmsilicon could potentially be used for a variety of electron device applications.Amorphous selenium, and other forms of thin-film materials, had alreadybeen employed for such applications, and had clearly demonstrated theirworth. Sputtering and evaporation processes, which where commonly em-ployed for the fabrication of thin-film silicon in the late 1960s, producedthin silicon films that are riddled with defects, thereby rendering the ma-terial electronically useless. The main benefit of the deposition techniqueof Chittick et al. [10] and Spear and LeComber [11], in which silane gas iscracked into its chemical components through the use of a plasma, is the en-hanced electronic properties of the resultant material; this process has been122.2. A brief history of thin-film nc-Si:H technologymore recently referred to as plasma enhanced chemical vapor deposition(PECVD).These enhanced electronic properties are believed to be related to thepresence of hydrogen atoms within a-Si:H. In addition, a-Si:H offers someinherent economic and manufacturing advantages when contrasted with thecase of c-Si, i.e., thin-film a-Si:H-based photovoltaic solar cells require lessmaterial then c-Si-based photovoltaic solar cells and the deposition of a-Si:H can be achieved at relatively low temperatures. As a result of theseadvantages, within the field of large area electronics, a-Si:H is one of themore commonly found materials.The first recorded deposition of nc-Si:H dates back to 1968 when Vepr?ekand Marec?ek [42] successfully deposited this material, a hydrogen plasmabeing used for this purpose, crystallites being observed within the resultantmaterial. One of the very first comprehensive reports on the deposition of nc-Si:H using PEVCD was conducted by Usui and Kikuchi in 1979 [43]. Theseresearchers found that the transition from an amorphous to nanocrystallinestructure within the resultant thin-film material can be achieved thoughvariations in the deposition parameters, such as the reactor pressure, thepower, the substrate temperature, and the hydrogen-to-silane dilution ratio.Over the next twenty years, most of the research performed on the de-position of nc-Si:H has focused on the use of PEVCD deposition. Throughthe use of hydrogen dilution, i.e., introducing hydrogen gas in addition tosilane in the deposition chamber, nc-Si:H may be produced. By adjustingthe hydrogen-to-silane dilution ratio, the material properties of the resultantnc-Si:H may be altered in such a way that smaller to larger silicon crystallites132.2. A brief history of thin-film nc-Si:H technologywill form within the resultant material. A large number of research groupshave contributed to what is currently known about the growth of nc-Si:H.These developments have been amply chronicled in the scientific literature;see, for example, Vepr?ek et al. [44], Matsuda [45], and Tsai et al. [46].Pioneering studies on nc-Si:H reported very slow growth processes, i.e.,approximately 1 A?/s [44, 45]; by comparison, it is possible to deposit a-Si:H at around 5 A?/s [47, 48]. Through advancements and improvements inthese deposition processes through the years, it has become possible to grownc-Si:H through the use of very-high-frequency PECVD (VHF-PECVD) atmuch higher rates, i.e., about 20 A?/s [49]; this deposition rate is fast enoughso that it can be readily used for the fabrication of the intrinsic absorberlayer of a thin-film nc-Si:H-based photovoltaic solar cell, the thickness ofthis layer typically being between 1 and 3 ?m [50?53].Although VHF-PECVD has found its way into large scale manufactur-ing, research has not stopped into seeking alternate means of depositingnc-Si:H. In recent years, for example, hot-wire chemical vapor deposition(HWCVD) has been used in order to deposit nc-Si:H [54, 55]. At the presentmoment, however, HWCVD can not be used for the large scale productionof nc-Si:H-based photovoltaic solar cells owing to its slow deposition rate,i.e., 1 A?/s [56], when compared with the case of VHF-PECVD. It shouldbe noted, however, that HWCVD produces nc-Si:H-based photovoltaic solarcells with excellent electronic properties; it is speculated that this occurs asthe ion bombardment of the growing films using HWCVD is less than thatassociated with VHF-PECVD, i.e., there is less surface damage [57, 58]. Ifthere is an improvement in the rate of deposition of nc-Si:H using HWCVD,142.2. A brief history of thin-film nc-Si:H technologyit may be possible for this deposition technique to become the standardmeans of fabricating nc-Si:H. Of course, what the future holds remains un-known at the present moment.In the 1990s, many research groups focused on improving the quality ofthe nc-Si:H-based absorber layers found within p-i-n or n-i-p single junc-tion nc-Si:H-based photovoltaic solar cell devices; see, for example, Faraji etal. [19], Meier et al. [20], and Torres et al. [59]. Finger reports [28] that aresearch group from the Institute of Microengineering at the University ofNeucha?tel, Switzerland were the first to demonstrate the use of nc-Si:H asa bottom layer in tandem nc-Si:H-based photovoltaic solar cells [20]. Addi-tional research has focused primarily on the material structure of nc-Si:H,and explored its impact on the corresponding electrical properties. Meier etal. [20], for example, finds that nc-Si:H exhibits only negligible light-induceddegradation when contrasted with the case of a-Si:H, suggesting an inherentadvantage of nc-Si:H over a-Si:H for photovoltaic solar cell device applica-tions, the conversion efficiency reducing over time with light exposure forthe case of a-Si:H-based photovoltaic solar cells.Another critical focus of research into the material properties of nc-Si:His the concentrations of impurities found within this material. Oxygen im-purities are a focus of particular attention, as oxygen can act as a dopant.As a result, an oxygen impurity distribution within nc-Si:H can turn anintrinsic region into an n-type region, with a corresponding loss in perfor-mance [59, 60]. The presence of oxygen atoms within this material may alsolead to some peculiar oxygen-related phenomena inside the voids associatedwith nc-Si:H. For example, the presence of oxygen atoms within the material152.2. A brief history of thin-film nc-Si:H technologycan have a direct impact upon the surface potential, and hence, play a rolein altering the conductivity of the nc-Si:H layers within a photovoltaic solarcell [61, 62]. The study of oxygen contamination within nc-Si:H, and thedesire to reduce the concentration of the oxygen impurities present withinnc-Si:H, is still ongoing, these impurities limiting the photovoltaic solar celldevice performance expected of nc-Si:H-based photovoltaic solar cells. Itshould be noted that other researchers, such as Kilper et al. [63], have foundthat other contaminants, such as nitrogen, also play a role in influencingthe material properties of nc-Si:H. This work is adequately addressed in thescientific literature [28, 59?63].Understanding how the growth conditions of nc-Si:H influence the cor-responding photovoltaic solar cell device performance has been a priorityfor researchers for many years. Understanding how these growth conditionsshape the underlying physical structure of nc-Si:H has also been a prior-ity for researchers since the early days of nc-Si:H. The material structure,and hence, the electrical properties of nc-Si:H can be tailored through vari-ations in the hydrogen dilution ratio, this being one of the most importantdeposition parameters. Increasing the proportion of the silane within thedeposition chamber results in a shift in the nature of the resultant nc-Si:H,from a highly crystalline nc-Si:H material to an almost completely amor-phous material. The transition between nc-Si:H and a-Si:H is found to berather abrupt [38]. Vetterl et al. [38] and Droz et al. [64] find that a thin-film photovoltaic solar cell, with a nc-Si:H absorber layer, exhibits optimalphotovoltaic solar cell device performance if the material is grown close tothis transition. Finger et al. [65] explains that, in order to reduce the de-162.2. A brief history of thin-film nc-Si:H technologyfect density, the highly crystalline structure needs to be encapsulated by athin-film amorphous layer, which serves as a thin passivation layer.Over the past decade, research from many groups has focused on im-proving photovoltaic solar cell device performance and in investigating thecorresponding structural characteristics of nc-Si:H, this being vital for deter-mining the electrical and optical properties of nc-Si:H [26]. A more detaileddiscussion, on the structural attributes of nc-Si:H, is provided in the nextsection.2.2.1 The material structure of nc-Si:H-based photovoltaicsolar cellsThe material structure of nc-Si:H is critically dependent upon manyfactors in the deposition:? the silane concentration,SC =[SiH4][SiH4] + [H2], (2.1)where [SiH4] is the silane flow and [H2] represents the hydrogen flowinto the deposition chamber. Alternatively, this concentration may becharacterized through the hydrogen dilution ratio, R, which is definedas follows,R =[H2][SiH4], (2.2)which suggests thatSC =11 + R(2.3)172.2. A brief history of thin-film nc-Si:H technology? the plasma chamber pressure? the gas flows? the excitation power? the excitation frequency? and the presence of impuritiesThe most significant contributing factor to the presence of nc-Si:H isthe SC, however, as will be seen subsequently, it is not the only factor thatplays a role in determining the material structure of this material [28]. Incharacterizing the nc-Si:H structure, three analytical tools are commonlyused in order to define its structural properties; (1) transmission electronmicroscopy (TEM), (2) XRD spectroscopy, and (3) Raman spectroscopy.TEM images are usually taken of cross-sections of the nc-Si:H films underconsideration. A representative TEM image, taken of the cross section of ann-i-p nc-Si:H-based photovoltaic solar cell by Droz et al. [64], is depicted inFigure 2.1. Note that there is a a-Si:H incubation layer that grows prior tothe growth of the nc-Si:H intrinsic layer1. Visualizing the structure of theindividual silicon crystallites within a highly crystalline nc-Si:H material canbe very useful for the purposes of material structural analysis.Although TEM imaging is a useful technique, the sample preparation re-quired of such imaging is extremely labor intensive. In addition, the equip-ment required for TEM imaging is expensive and demanding of continual1Detailed analyzes of the growth processes have indicated that the growth of a nc-Si:H thin-film is preceded by the growth of a a-Si:H incubation layer, the thickness ofthis incubation layer depending upon the substrate employed and deposition conditionsused [66].182.2. A brief history of thin-film nc-Si:H technologyFigure 2.1: A sample cross-sectional dark-field TEM image of an n-i-p nc-Si:H-based photovoltaic solar cell. The bright areas, which correspond tocrystallites of the same crystallographic orientation, form a structure parallelto the growth direction. The a-Si:H incubation layer, appearing overallgrayish, can be observed at the onset of the intrinsic nc-Si:H layer. Thisimage is from Droz et al. [64].192.3. X-ray diffractionmaintenance, the skill-set required of TEM imaging being quite rigorous.For all of these reasons, TEM imaging is not typically used for routine ex-amination. In order to determine the mean crystallite size from a TEMimage, one has to individually determine the size of a large number of crys-tallites in order to produce a sufficiently accurate mean crystallite size, orcrystalline volume fraction [28]. Therefore, TEM imaging has not been usedin this study of the determination of the mean crystallite size, as well as thecorresponding crystalline volume fraction.XRD and Raman spectroscopy are non-destructive analytical tools oftenused in the analysis of the structural characteristics of nc-Si:H. The devicesrequired in order to perform these experiments are instruments that are partof the standard laboratory inventory. Furthermore, in contrast with the caseof TEM, the samples require little to no preparation for such measurements.In general, from the XRD spectra, the mean crystallite sizes can be deter-mined. From the Raman spectra, the crystalline volume fractions can beascertained.2.3 X-ray diffractionXRD spectroscopy is a non-destructive analytical technique that maybe used in order to determine the crystalline structure as well as the latticeparameters. Exposing a crystalline sample to a monochromatic X-ray beam,distinct fingerprints, also referred to as Bragg?s reflections, are produced inthe form of an XRD spectrum2. These unique reflections provide insights2Generally, diffraction refers to the interaction of light with objects. Special cases ofdiffraction include reflection, transmission, and refraction.202.3. X-ray diffractioninto the underlying structure of the crystalline material. A Bragg reflec-tion can only occur if the interference that occurs is constructive in nature.Constructive interference occurs only if Bragg?s law is satisfied.The conditions required for Bragg?s reflections may be obtained throughconsideration of Figure 2.2 [67]. This figure depicts several crystal planesof a crystallite structure and an incoming monochromatic beam of X-rays,of wavelength ?. For constructive interference to occur, the pathlengthdifference, l, between beams reflected off the various planes must be equalto a multiple of the wavelength ?, i.e., l = n?, where n is an integer. Fromthe geometry depicted in this figure, it can be seen thatl = 2d sin ?, (2.4)where ? is the angle between the incident beam and the crystal planes andd is the interplanar separation in a crystal structure. Thus, the conditionfor Bragg?s reflections reduces to? = 2d sin ? (2.5)Eq. (2.4) is referred to as Bragg?s law, and it provides the basis for XRDspectroscopy.The interplanar spacings in a crystal can be calculated if the overallcrystal structure is known. The case of c-Si is instructive. Silicon exhibits aface-centered cubic diamond structure, as is seen in Figure 2.3 [68], and itsunit cell contains 8 atoms. The spacings between the silicon crystal planes,212.3. X-ray diffractionABCINCOMINGX?RAYSA?B?C?DIFFRACTEDX?RAYSddCRYSTALPLANES2???? ?DEFGFigure 2.2: Several crystal planes and an interacting incoming X-ray beamare depicted. In this figure, ? is the angle between the incident beam and thecrystal planes and d is the interplanar separation for the crystal structure.This figure is after Cullity and Stock [67].222.3. X-ray diffractionFigure 2.3: The unit cell associated with c-Si. This figure is afterFranssila [68].232.3. X-ray diffractionat a specific orientation, can be calculated as follows:dhkl =a?h2 + k2 + l2, (2.6)where a is the crystal lattice parameter (for silicon, a = 5.43 A?) and h, k,and l are the Miller indices associated with the lattice planes.The X-ray beams that are reflected off the various lattice planes willinterfere with each other. Only those X-ray beams that satisfy Bragg?s law,i.e., Eq. (2.4), will be observed. For the case of c-Si, the following Bragg?sreflection peaks are typically found: (111), (220), (311), (400), and (331),as was noted by Cullity and Stock [67]. Other Bragg?s reflection peaks arealso sometimes observed. The commonly used Cu K? radiation source forXRD spectroscopy produces monochromatic X-rays with a wavelength, ?,of 1.54 A?. With this value of ?, Eqs. (2.4) and (2.5) allow one to determinethe angles at which Bragg?s reflection peaks occur. For the aforementionedBragg reflection peaks, these angles are tabulated in Table 2.1.Table 2.1: Calculated Bragg?s reflection peak angles, ? and 2?.h k l a (A?) dhkl (A?) ? (A?) ? (?) 2? (?)1 1 1 5.43 5.43 1.54 14.2 28.42 2 0 5.43 1.92 1.54 23.6 47.33 1 1 5.43 1.64 1.54 28.0 56.14 0 0 5.43 1.36 1.54 34.6 69.13 3 1 5.43 1.25 1.54 38.2 76.4For the case of nc-Si:H, this knowledge can be, and has been, applied inorder to characterize the structural orientation and mean crystallite size of242.3. X-ray diffractionthe silicon crystallites found within this material. In particular, the loca-tions of the Bragg?s reflection peaks may be studied in order to determinethe mean crystallite size associated with the nc-Si:H. In c-Si, very distinct?sharp? peaks are present within the XRD spectrum. The peak broadeningthat is observed for this case is mostly caused by instrumentation limitsand by the microstrain that may be present within a given material[67]. Ina material for which small crystallites are present, however, further peakbroadening does occur.If the pathlength difference is not a multiple of the wavelength, ?, theX-rays reflected from the various lattice planes will destructively interferewith each other. If atoms are periodically distributed, when Bragg?s law issatisfied, the X-rays reflected from the various lattice planes will construc-tively interfere, and a Bragg reflection peak will occur. For the case in whichperiodicity only occurs over a limited volume, i.e., the crystallites are finite,the observed Bragg reflection peaks will be broadened [67, 69].Scherrer analyzed the additional peak broadening that occurs relatedto the mean crystallite size and found that, for submicron particles, it ispossible to estimate the mean size of the crystallites, dXRD in nm, using thefollowing relationship:dXRD =K?B cos ?B, (2.7)where K is a unitless shape correction factor, referred to as the Scherrer con-stant, ? is the monochromatic excitation beam wavelength, B correspondsto the full-width-at-half-maximum (FWHM), in radians, and ?B is the angleat which the maximum is observed [69]. The value of the Scherrer constant,252.4. Raman spectroscopy and the crystalline volume fractionK, is taken to be between 0.9 and 1; the exact value for the Scherrer constantappropriate for use is material dependent and three factors can impact itsselection, i.e., how the reflection ?breadth? is defined, the shape of the crys-tallites, and how the mean crystallite size varies within the material [70].For the purposes of this study, in the interests of being consistent with otherresearchers in this field, K is set to 0.9. Peak broadening related to instru-mentation limitations is also observed but found to be relatively minor. Itwill be neglected in the subsequent analysis.2.4 Raman spectroscopy and the crystallinevolume fractionRaman spectroscopy is another non-destructive analytical technique thatmay be used in order to characterize the material properties of nc-Si:H. Aswith XRD spectroscopy, in Raman spectroscopy a monochromatic excitationsource, i.e., a laser, produces a beam of light which interacts with the mate-rial. This beam of monochromatic light is then either inelastically scattered,i.e., energy is either lost or gained, or elastically scattered, i.e., no energyloss in the scattered light occurs.Processes that lead to elastic scattering are generally referred to asRayleigh scattering processes, these being characterized by experimentallynegligible changes in energy, these corresponding to wave-number changesless than 10?5 cm?1. Some of the processes that lead to inelastic scatteringare referred to as Raman scattering as long as the difference in energy cor-responds to wave-number differences in excess of 1 cm?1 [71]. Although the262.4. Raman spectroscopy and the crystalline volume fractiontheory behind Raman spectroscopy had been introduced by ChandrasekaraRaman in 1928 [72], Raman spectroscopy did not become a standard toolfor sample characterization until the introduction of laser technologies inthe 1960s [73].Figure 2.4 [74] depicts an entire Raman spectrum of a sample material.As may be seen, the intensity for the anti-Stokes Raman scattering is sig-nificantly lower than that of the Stokes Raman scattering, i.e., at ambienttemperature, it is more likely for the molecules to be in a low vibrationalstate and for only a few of the molecules to be in a high vibrational state.Although the intensities are different, the Stokes and anti-Stokes scatteringspectra are symmetric and carry the same material structure information inboth regions[74, 75].For this study, confocal Raman microscopy is used in order to retrievethe Raman spectra for the nc-Si:H-based photovoltaic solar cells. The dif-ferences between standard Raman spectroscopy and confocal Raman spec-troscopy is in the objective lens which is being used to focus a beam ofmonochromatic light onto a sample. Therefore, with confocal Raman spec-troscopy, it is possible to attain a high level of depth resolution, as well as thesuppression of unwanted interference signals from a fluorescent or substratebackground. A representative schematic for a confocal Raman spectroscopysystem is depicted in Figure 2.5[75]. Raman spectroscopy is often used bync-Si:H researchers and it is a very important tool to analyze the materialcharacteristics. Generally, it is common in the nc-Si:H research, to onlyconsider the Stokes scattering, as this scattering carries all of the informa-tion needed to determine the crystalline volume fractions in the material272.4. Raman spectroscopy and the crystalline volume fractionFigure 2.4: A representative Raman spectrum for the mineral realgar de-picting the Rayleigh, Stokes, and anti-Stokes scattering ranges. The actualintensity of the Rayleigh scattering was suppressed. This image is fromVandenabeele [74].282.4. Raman spectroscopy and the crystalline volume fractionFigure 2.5: An illustration of a standard confocal Raman microscopy setupis depicted in the figure. This figure is from Hollricher[75]. The onlineversion of this figure is in color.292.4. Raman spectroscopy and the crystalline volume fraction[22, 28, 36, 41, 64].Figure 2.6 depicts the Stokes scattering in a typical Raman spectrum fora-Si:H and nc-Si:H. The a-Si:H spectrum exhibits a distinct, but broadened,peak at approximately 480 cm?1, while c-Si exhibits a single distinct ?sharp?peak at approximately 520 cm?1. It is also noted that nc-Si:H exhibits twodistinct, but merged, peaks at the same two peak shifts corresponding tothe amorphous and c-Si components of the material. By decomposing theRaman spectrum for nc-Si:H into its individual peaks and integrating thearea under each peak, the volumetric contributions of the amorphous andc-Si phases can be determined. Vetterl et al. [38] found that the electronicproperties seem to optimize at the transition region from amorphous tohighly crystalline silicon. Yue et al. [76, 77] have found that nc-Si:H-basedphotovoltaic solar cells with a crystalline volume fraction of approximately50 % exhibit optimal photovoltaic solar cell performances.302.4. Raman spectroscopy and the crystalline volume fraction200 300 400 500 600 700 800a?Si:Hnc?Si:HRaman shift (cm?1)Intensity (a.u.)Figure 2.6: Typical Stokes scattering in a Raman spectrum for nc-Si:H anda-Si:H. The online version of this figure is in color.31Chapter 3Experiment3.1 The analysis of nc-Si:HAs the low bandgap component of multi-junction thin-film silicon pho-tovoltaic solar cells, nc-Si:H has attracted a considerable amount of atten-tion in recent years owing to its improved long wavelength response andlower light induced degradation when contrasted with the case of a-Si:H.Record solar cell and module efficiencies have been attained using a-Si:H/nc-Si:H/nc-Si:H triple-junction photovoltaic solar cells [22]. Owing to its com-plexity as a material system, the material structure of nc-Si:H, and the rolethat impurities and the substrate play in influencing these material prop-erties, are the subject of considerable current investigation. It has beenobserved that nc-Si:H-based photovoltaic solar cells are more sensitive toimpurities than a-Si:H-based photovoltaic solar cells and the reasons forthis have also been the focus of extensive study. There is no doubt thatinterest in the material properties of nc-Si:H will continue for many years tocome.In this chapter, the means of preparing the nc-Si:H-based photovoltaicsolar cells, used in this experimental study, are described. Then, unprocessedXRD and Raman spectra, corresponding to the nc-Si:H-based photovoltaic323.1. The analysis of nc-Si:Hsolar cells under consideration in this analysis, are presented. Informationon the material structure of nc-Si:H, gleaned from these spectra, is thenhighlighted and explained. The experimental methodologies employed inthe processing of these spectra are also detailed.As noted previously, XRD is a non-destructive back-scattering techniquein which an X-ray producing, wavelength-specific, excitation source is usedin order to probe the nc-Si:H present within the photovoltaic solar cells.The resultant spectra exhibit peaks, often referred to as Bragg?s reflections,which can provide information about the crystallite structure, the latticeparameters, and the mean crystallite size [67]. For the three nc-Si:H-basedphotovoltaic solar cells considered in this analysis, the obtained XRD spectraare used in order to determine the mean crystallite size corresponding to eachphotovoltaic cell.Raman spectroscopy is a non-destructive back-scattering technique inwhich the sample is exposed to a certain excitation wavelength, which maybe altered through interactions with the vibrational modes within the nc-Si:H. The corresponding back-scattered spectra can be either Stokes or anti-Stokes scattered, i.e., the back scattered light can either lose or gain energy,respectively. As it is more probable for the material to be at a lower energystate, the intensity of the Stokes scattering spectrum is significantly higherthan that of the anti-Stokes spectrum scattering. In the Raman spectroscopyexperiments that are performed on the nc-Si:H-based photovoltaic solar cells,Stokes scattering is considered in order to determine the crystalline volumefraction of the nc-Si:H within these photovoltaic solar cells.This chapter primarily focuses on the presentation of unprocessed XRD333.2. The nc-Si:H-based photovoltaic solar cell samplesand Raman spectra results corresponding to the nc-Si:H-based photovoltaicsolar cells considered in this analysis; these samples are 21821, 21886, and21916, this being the notation introduced by Yue et al. [50] in order to de-scribe these samples. First, an introduction to the fabrication processes,used to manufacture these photovoltaic solar cells, is provided, a structuraloverview of these samples also being presented. The unprocessed XRD spec-troscopy results, corresponding to the data acquired for all of the samplesconsidered in this analysis, is then provided. Finally, unprocessed Ramanspectra, corresponding to the photovoltaic solar cell samples considered inthis analysis, are shown and commented upon.This chapter is organized in the following manner. In Section 3.2, thefabrication processes, used to determine the nc-Si:H-based photovoltaic so-lar cells, are described, the layout of these nc-Si:H-based photovoltaic solarcells being provided. Then, in Section 3.3, unprocessed XRD results, corre-sponding to the photovoltaic solar cell samples under consideration, in thisanalysis are presented. Finally, in Section 3.4, unprocessed Raman results,corresponding to the photovoltaic solar cell samples under consideration, arefeatured.3.2 The nc-Si:H-based photovoltaic solar cellsamplesFor the purposes of this study, three n-i-p nc-Si:H-based photovoltaic so-lar cells are analyzed using XRD and Raman spectroscopy. The photovoltaicsolar cells were manufactured by a Michigan-based company named United343.2. The nc-Si:H-based photovoltaic solar cell samplesSolar Ovonic, LLC, more often referred to simply as Uni-Solar. By 2008,Uni-Solar was the second largest manufacturer of thin-films solar cells in theworld, First Solar being the largest [78, 79]. Uni-Solar?s parent company wasEnergy Conversion Devices, whose founder was the renowned American in-ventor and entrepreneur and holder of many patents, the late Mr. StanfordR. Ovshinsky. Unfortunately, in 2012, Energy Conversion Devices, as wellas United Solar Ovonic, LLC, filed for bankruptcy protection. Sadly, Mr.Stanford R. Ovshinsky also passed away in 2012.Uni-Solar used a modified VHF-PECVD approach for the fabrication ofthe three photovoltaic solar cells in order to deposit all of the required a-Si:Hand nc-Si:H layers. Limited details on the growing conditions employed, forthree nc-Si:H-based photovoltaic solar cell samples considered in this study,are provided elsewhere in the literature [50]; the primary difference betweenthese depositions was related to the hydrogen dilution ratios and depositionconditions. The proprietary nature of these deposition processes, and thecompetitive nature of the photovoltaic industry, accounts for this paucity ofinformation concerning how the photovoltaic solar cells were prepared.The structural layout, for three n-i-p nc-Si:H-based photovoltaic solarcells considered in this analysis, can be seen in Figure 3.1. A stainless steelsubstrate provides the basis for these photovoltaic solar cells. A silver/zinc-oxide (Ag/ZnO) back-reflector is deposited upon this stainless steel sub-strate. The back-reflector is used in order to reflect the light back into theabsorber layer of the nc-Si:H-based photovoltaic solar cell, i.e., the back-reflector works as a light trapping layer. Next, a phosphorous-doped a-Si:Hn-layer is deposited, followed by a a-Si:H n/i buffer layer. The buffer layer353.2. The nc-Si:H-based photovoltaic solar cell samplesFigure 3.1: A cross-section of the nc-Si:H-based n-i-p photovoltaic solar cellsconsidered in this analysis. The dimensions of the n-, p-, and intrinsic layersare indicated. The buffer and highly crystalline seed layers are assumed to beof smaller dimensions than the n- or p-layers. All of the other dimensionsare unknown. The ITO contacts are only present for the case of sample21886. The online version of this figure is depicted in color.363.3. X-ray diffraction spectroscopyserves as a barrier, so that the phosphor atoms, contained within the a-Si:Hn-layer, can not diffuse into the absorber layer. This is followed by a highlycrystalline seed layer of nc-Si:H3.After these initial layers, the photovoltaic solar cell is ready to have itsintrinsic nc-Si:H absorber layer deposited. This layer is vital for electron-hole pair generation, and it is the thickest layer in the photovoltaic solarcells, i.e., around 3 ?m in thickness. Another buffer layer, i.e., the i/pa-Si:H buffer layer, encapsulates this intrinsic absorber layer in order toprevent boron diffusion into the intrinsic absorber layer. Lastly, a thinboron-doped nc-Si:H-based p-layer is deposited, as depicted in Figure 3.1. Itshould be noted that sample 21886, one of the three analyzed nc-Si:H-basedphotovoltaic solar cells, has indium-tin-oxide (ITO) contacts deposited ontop of its nc-Si:H p-layer, as seen in Figure 3.2. As the results of the Ramanspectroscopy measurements may be impacted by the presence of these ITOcontacts, special care was employed in order to ensure that the Raman lightsource was cast upon the exposed nc-Si:H p-layer only, i.e., away from theseITO regions.3.3 X-ray diffraction spectroscopyI obtained the Powder ? - 2? XRD scans using a Philips X?Pert MRDsystem, with a Cu K? radiation source (? = 1.54 A?). The focus of thisparticular aspect of this study are the crystalline silicon related peaks. For3It is widely held that nc-Si:H exhibits an initial incubation layer, which is highlyamorphous in nature. In order to reduce the thickness of this amorphous incubationlayer, a highly crystalline nc-Si:H layer may be grown [38].373.3. X-ray diffraction spectroscopyFigure 3.2: These photographs correspond to the actual nc-Si:H-based pho-tovoltaic solar cell samples which are considered in this analysis, i.e., theseare the samples that are considered in the XRD and Raman analysis consid-ered for the pupose of this study. As may be readily seen, the second sample,i.e., sample 21886, has ITO contacts deposited on top of the boron-dopedtop p-layer. The online version of this figure is depicted in color.383.3. X-ray diffraction spectroscopythis wavelength, silicon-based (111), (220), and (311) peaks are observedat around 28, 47, and 56?, respectively; these values correspond to the 2?angles, as was shown in Table 2.1. Accordingly, the XRD spectra were mea-sured between 20 and 70?, i.e., the 2? ranges between 20 and 70?. In order toincrease the signal-to-noise (S/N) ratio, the acquisition of the XRD spectrarequires sufficient scanning time. Unfortunately, the long scans demanded ofsuch a spectra acquisition could not be accommodated, and instead, fasterscans of the entire spectra were obtained. High quality scans, focused solelyon the aforementioned silicon based peaks, were also employed.In Figures 3.3, 3.4 and 3.5, the full XRD spectra, corresponding to sam-ples 21821, 21866, and 21916, respectively, are depicted. The origin ofeach identified peak is indicated on each figure. The peak locations asso-ciated with the Si, Ag/ZnO, and ITO peaks are as determined from thedatabase of the Institute of Experimental Mineralogy Russian Academy ofSciences [80], Li et al. [81], and Akkad et al. [82], respectively. In order toidentify the peaks related to the stainless steel substrate, the XRD spectrum,corresponding to such a substrate, was obtained, i.e., one of the samples?substrate surface being measured, which provided peak information corre-sponding to the stainless steel. As can be seen in the three spectra, only the(220) silicon peaks, located around 47?, can be consistently observed for allof the XRD spectra for all samples. The (111) silicon peak, observed around28?, and the (311) silicon peak, observed around 56?, have also been notedin some of the spectra, but are at much lower intensities, and thus, corre-spond to very poor S/N ratios. Therefore, for the purposes of this analysis,the focus is on the (220) silicon peaks.393.3. X-ray diffraction spectroscopy20 25 30 35 40 45 50 55 60 65 702? (?)Intensity (a.u.)Sample 21821ITO (211)Si (111)ITO (400) and ZnO (002)ZnO (101)ITO (431) and Stainless SteelSi (220) Zn0 (102)Stainless SteelFigure 3.3: The XRD spectrum corresponding to sample 21821. A basichigh-rate scan, corresponding to the entire spectral range, is depicted withthe light solid line. High-quality scans, corresponding to the individualpeaks, are depicted with the heavier solid lines. The higher-quality scanresults are vertically offset from the other results for the purposes of greaterclarity. The peaks observed, and their origin, are indicated in the figure.403.3. X-ray diffraction spectroscopy20 25 30 35 40 45 50 55 60 65 702? (?)Intensity (a.u.)Sample 21886ITO (211)Si (111)ITO (222) and ZnO (100)ITO (400) and ZnO (002)ZnO (101)ITO (431) and Stainless SteelSi (220)Zn0 (102)undeterminedpossibly Si (311), ITO (611), and ZnO (110)Stainless SteelFigure 3.4: The XRD spectrum corresponding to sample 21886. A basichigh-rate scan, corresponding to the entire spectral range, is depicted withthe solid line. The peaks observed, and their origin, are indicated in thefigure.413.3. X-ray diffraction spectroscopy20 25 30 35 40 45 50 55 60 65 702? (?)Intensity (a.u.)Sample 21916ITO (211)Si (111)ITO (400) and ZnO (002)ZnO (101)ITO (431) and Stainless SteelSi (220)Zn0 (102) Stainless SteelFigure 3.5: The XRD spectrum corresponding to sample 21916. A basichigh-rate scan, corresponding to the entire spectral range, is depicted withthe solid line. The peaks observed, and their origin, are indicated in thefigure.423.4. Raman spectroscopyCullity and Stock [67] suggest that the dominance of the (220) siliconpeaks in the XRD spectra is related to a (220) preferential growth directionfor the nc-Si:H samples. A preferential growth direction implies that thecrystallites are ordered in such a way that the (220) planes are parallel withthe surface in contrast to the case of randomly distributed planes [39, 50].Other research has suggested that by varying the deposition parameters, thesilicon XRD peaks, found by scanning nc-Si:H-based material, can shift froma (111) dominant intensity peak to a (220) dominant intensity peak [39, 44,83].3.4 Raman spectroscopyI obtained the Raman spectra from a high-resolution confocal micro-Raman system, i.e., a LabRam HR by Horiba Scientific. The excitationwavelength used is 442 nm (blue light). The power density of the excitationsource is unknown. The laser spot diameter is estimated to be approxi-mately 2 ?m for a 10? objective lens, this being used for all of the ex-perimental measurements of the Raman spectrum that were carried out forthe purposes of this analysis. Before each measurement, careful calibrationprocedures were carried out, i.e., a Raman spectroscopy scan is acquiredusing a monocrystalline silicon sample, its sharp peak being offset into itstheoretically expected position of approximately 520.7 cm?1.A representative Raman spectrum is depicted in Figure 3.6, this spec-trum corresponding to a nc-Si:H sample [84]. This Raman spectrum can bedecomposed into four peaks corresponding to five different phonon modes,433.4. Raman spectroscopyi.e., a transverse acoustic (TA) mode, a longitudinal acoustic (LA) mode,a longitudinal optical (LO) mode, and two transverse optical (TO1, TO2)modes, these peaks being centered around 150, 310, 380, 480, 510 cm?1,respectively, as shown in Figure 3.6 [84]. In contrast, other studies definethe TO2 mode as being at approximately 520 cm?1, an intermediate modebeing present between 500 to 510 cm?1, this intermediate mode representingthe grain boundaries [76, 85, 86].Figures 3.7, 3.8, and 3.9, show the unprocessed Raman spectra corre-sponding to the samples 21821, 21886, and 21916, respectively. All of theRaman measurements were carried out over the range 20 to 2400 cm?1. Dueto input laser interference, wave-numbers less then approximately 150 cm?1are of limited experimental value. All of the prior mentioned modes can beobserved at their respective positions. Significant peak shifts are not ob-served. In the analysis shown in the subsequent chapters, only the spectrabetween 420 and 540 cm?1 are considered in this analysis in order to beconsistent with the prior research conducted by Yan et al. [22] of Uni-Solar.443.4. Raman spectroscopyFigure 3.6: A representative Raman spectrum corresponding to nc-Si:H withthe different modes identified. Note that this model omits the intermediateor grain boundary component, located at approximately 510 cm?1. Thisimage is after Wei et al. [84].453.4. Raman spectroscopy200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400200400600800100012001400Raman shift (cm?1)Intensity (counts)Sample 21821? = 442 nmFigure 3.7: The entire unprocessed Raman spectrum for the 21821 nc-Si:Hsample, from 20-2400 cm?1. For this analysis, particular attention is paid tothe two predominant peaks at 480 and 520 cm?1, these peaks correspondingto the amorphous and crystalline silicon peaks of this material, respectively.Further peaks, which correspond to the different modes of amorphous silicon,are also observed, at approximately 150, 310, and 380 cm?1. The resultsare not accurate for the lower wave-numbers, i.e., less than 150 cm?1, dueto input laser interference.463.4. Raman spectroscopy200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400200400600800100012001400Raman shift (cm?1)Intensity (counts)Sample 21886? = 442 nmFigure 3.8: The entire unprocessed Raman spectrum for the 21886 nc-Si:Hsample, from 20-2400 cm?1. For this analysis, particular attention is paid tothe two predominant peaks at 480 and 520 cm?1, these peaks correspondingto the amorphous and crystalline silicon peaks of this material, respectively.Further peaks, which correspond to the different modes of amorphous silicon,are also observed, at approximately 150, 310, and 380 cm?1. The resultsare not accurate for the lower wave-numbers, i.e., less than 150 cm?1, dueto input laser interference.473.4. Raman spectroscopy200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400200400600800100012001400Raman shift (cm?1)Intensity (counts)Sample 21916? = 442 nmFigure 3.9: The entire unprocessed Raman spectrum for the 21916 nc-Si:Hsample, from 20-2400 cm?1. For this analysis, particular attention is paid tothe two predominant peaks at 480 and 520 cm?1, these peaks correspondingto the amorphous and crystalline silicon peaks of this material, respectively.Further peaks, which correspond to the different modes of amorphous silicon,are also observed, at approximately 150, 310, and 380 cm?1. The resultsare not accurate for the lower wave-numbers, i.e., less than 150 cm?1, dueto input laser interference.48Chapter 4Analysis4.1 OverviewTwo critical shortcomings of a-Si:H are its low short-circuit current den-sity and its poor long-wavelength response. Light-induced degradation isanother widely recognized shortcoming of this material. These weaknessesprompted researchers to consider other forms of thin-film silicon for largearea electron device applications. The interesting material properties pos-sessed by nc-Si:H make it ideally tailored for photovoltaic and thin-filmtransistor device applications. Accordingly, there is a great deal of currentinterest into understanding the material properties of this material. It iswell accepted that nc-Si:H is a fundamentally phase heterogeneous mate-rial, comprised of silicon crystallites embedded in an amorphous tissue. Therole that the crystallite sizes plays in determining the material properties ofnc-Si:H remains an open problem in this field.This particular analysis aims to ascertain the role that the mean crys-tallite size plays in determining the crystalline volume fraction, many ofthe properties associated with nc-Si:H being determined by the crystallinevolume fraction. Through a study, performed on three nc-Si:H-based photo-voltaic solar cells, it is hoped that one can determine whether or not such a494.2. XRD analysisrelationship is even plausible. For each nc-Si:H-based photovoltaic solar cellconsidered in this analysis, the mean crystallite size is determined throughthe application of Scherrer?s equation on the corresponding XRD spectra.The crystalline volume fraction corresponding to each photovoltaic solarcell is then determined through a peak decomposition of the Raman spec-tra corresponding to these samples. Plotting the crystalline volume fractionas a function of the mean crystallite size, the relationship between theseparameters, if it exists at all, will be rendered transparent.This chapter is organized in the following manner. In Section 4.2, theXRD analysis is presented, the peak fits, the mean crystallite size deter-minations, and the corresponding error analysis, being discussed. Then, inSection 4.3, the Raman analysis is depicted, the peak decomposition pro-cess being detailed. In Section 4.4, the experimental resolution limitation,for both XRD and Raman spectroscopy, is addressed. Finally, in Section 4.5,the dependency of the crystalline volume fraction on the mean crystallitesize is explored.4.2 XRD analysisIn Figures 4.1, 4.2, and 4.3, subsets of the XRD spectra, correspond-ing to samples 21821, 21886, and 21916, are depicted, the correspondingoverall XRD spectra being plotted in Figures 3.3, 3.4, and 3.5, respectively.The dominant (220) silicon peak, located at about 47?, is depicted in eachsubset [76, 77, 87, 88]. A peak fitting software, Origin Pro 9, is used inorder to fit a Lorentzian curve to this (220) peak; researchers have used504.2. XRD analysis30 32 34 36 38 40 42 44 46 48 500100200300400500600700800900100011002? (?)CountsSample 21821ITO (400) and ZnO (002) ZnO (101)ITO (431) and Stainless SteelSi (220) Zn0 (102)42 44 46 48 50 522? (?)Counts (a.u.)Figure 4.1: A subset of the XRD spectrum corresponding to sample 21821.Baseline correction has been employed. The visible peaks, and their corre-sponding origin, are identified in the figure. The inset focuses on the peakfound for the silicon (220) peak. The corresponding peak fit is depictedwith the red solid line. Care was taken in order to ensure that the adjacentZnO-based peak did not influence the obtained peak fit. The online versionis depicted in color.514.2. XRD analysis30 32 34 36 38 40 42 44 46 48 500100200300400500600700800900100011002? (?)CountsSample 21886ITO (400) and ZnO (002)ZnO (101)ITO (431) and Stainless SteelSi (220) Zn0 (102)42 44 46 48 50 522? (?)Counts (a.u.)Figure 4.2: A subset of the XRD spectrum corresponding to sample 21886.Baseline correction has been employed. The visible peaks, and their corre-sponding origin, are identified in the figure. The inset focuses on the peakfound for the silicon (220) peak. The corresponding peak fit is depictedwith the red solid line. Care was taken in order to ensure that the adjacentZnO-based peak did not influence the obtained peak fit. The online versionis depicted in color.524.2. XRD analysis30 32 34 36 38 40 42 44 46 48 500100200300400500600700800900100011002? (?)CountsSample 21916ITO (400) and ZnO (002)ZnO (101)ITO (431) and Stainless SteelSi (220)Zn0 (102)42 44 46 48 50 522? (?)Counts (a.u.)Figure 4.3: A subset of the XRD spectrum corresponding to sample 21916.Baseline correction has been employed. The visible peaks, and their corre-sponding origin, are identified in the figure. The inset focuses on the peakfound for the silicon (220) peak. The corresponding peak fit is depictedwith the red solid line. Care was taken in order to ensure that the adjacentZnO-based peak did not influence the obtained peak fit. The online versionis depicted in color.534.2. XRD analysisboth Lorentzian and Gaussian fits for this peak, but it was found, for thiscase at least, that the Lorentzian peak fit is more representative of the XRD(220) peak, i.e., it captures the spirit of the obtained experimental spectra.For the baseline correction, which is employed prior to these fits, the stan-dard Origin Pro 9 baseline correction algorithm is employed; Origin Pro9 uses, by default, the Levenberg-Marquadt algorithm baseline correctiontechnique [89], i.e., an exponential baseline correction approach.It is noted that, in all cases, the ZnO (102) peak is directly adjacent tothe silicon (220) peak. Accordingly, care was taken in order to correct forthe presence of this peak during the analysis of the silicon (220) peak for allof the spectra considered in this analysis. The resultant fits of the baselinecorrected silicon (220) peak are depicted in the insets of Figures 4.1, 4.2,and 4.3. The corresponding fitting parameters are tabulated in Table 4.1.Table 4.1: Summary of the mean crystallite sizes and the crystalline volumefractions found by XRD and Raman spectroscopy, respectively.Sample dXRD (nm) XC (%) XA (%) X GB (%)21821 32? 3 nm 26? 1 % 74? 1 % 5? 1 %21886 35? 4 nm 24? 0.5 % 76? 0.5 % 4? 0.5 %21916 40? 4 nm 23? 1 % 77? 1 % 3? 1 %Through a determination of the XRD peak widths, these being deter-mined from the Lorentzian fits, Scherrer?s equation is then applied in orderto determine the corresponding mean crystallite sizes, i.e., Eq. (2.7). Peakbroadening associated with the instrumentation limits and by the micros-train are assumed to be negligible when contrasted with the peak broadening544.3. Raman analysisassociated with the small crystallites. The obtained mean crystallite sizesare found to vary between 35 and 40 nm for the three nc-Si:H-based pho-tovoltaic solar cells considered in this analysis. Scherrer?s equation, whichis derived through consideration of the constructive and destructive inter-ference over parallel planes of atoms [67], asserts that the peaks associatedwith a diffraction pattern are broadened by an amount that is inverselyproportional to the crystallite size.While nc-Si:H is a very complex material, with crystallites, an amor-phous tissue, and grain boundaries, Scherrer?s equation is often used in theanalysis of this material [39, 85], despite the potential for interpretationalchallenges. The error associated with the determination of the mean crys-tallite sizes is determined by systematically assessing the variations in themean crystallite sizes obtained from the various possible peak fits to thesilicon (220) peaks. The mean crystallite size errors are also depicted inTable 4.1.4.3 Raman analysisIn Figures 4.4, 4.5, and 4.6, linear baseline corrected Raman spectra,corresponding to samples 21821, 21886, and 21916, are depicted, the corre-sponding overall unprocessed Raman spectra being plotted in Figures 3.7, 3.8,and 3.9, respectively. For the purposes of this analysis, the focus is on thesespectra between 420 and 540 cm?1. The crystalline volume fraction of thesesamples is determined through the decomposition of these Raman spectrainto their constituent peaks.554.3. Raman analysis420 440 460 480 500 520 540Raman shift (cm?1)Intensity (a.u.)Sample 21821? = 442 nmFigure 4.4: A subset of the Raman spectrum corresponding to sample 21821.The experimental data points are indicated with the open points. Linearbaseline correction has been employed between 420 to 540 cm?1.564.3. Raman analysis420 440 460 480 500 520 540Raman shift (cm?1)Intensity (a.u.)Sample 21886? = 442 nmFigure 4.5: A subset of the Raman spectrum corresponding to sample 21886.The experimental data points are indicated with the open points. Linearbaseline correction has been employed between 420 to 540 cm?1.574.3. Raman analysis420 440 460 480 500 520 540Raman shift (cm?1)Intensity (a.u.)Sample 21916? = 442 nmFigure 4.6: A subset of the Raman spectrum corresponding to sample 21916.The experimental data points are indicated with the open points. Linearbaseline correction has been employed between 420 to 540 cm?1.584.3. Raman analysisPeaks in the neighborhood of 480, 510, and 520 cm?1 are assumed,these peaks corresponding to the amorphous, grain boundary, and crystallinecomponents of the Raman spectra, respectively. Following Tay et al. [90],symmetrical Gaussian/Lorentzian peaks are assumed for the purposes ofthis analysis. Substantial a-Si and c-Si components in these Raman spectraare observed for all cases. Building upon the approaches of Tsu et al. [27]and Bustarret et al. [91], Droz et al. [64] and Han et al. [86] suggest thatthe crystalline volume fraction,Xc =Ic + IGBIc + IGB + y(L)Ia, (4.1)where Ic+IGB represents the integrated components of the 510 and 520 cm?1peaks and Ia corresponds to the integrated 480 cm?1 peak, the approach ofDroz et al. [64] and Han et al. [86] being adopted for the purposes of thisparticular analysis. The scattering cross-section ratio, y(L), which was firstmodeled by Bustarret et al. [91], is set to unity for the sake of simplicity andin order to be consistent with recent literature concerning the material prop-erties of nc-Si:H [22, 64, 76, 92]. The peak decompositions, correspondingto the nc-Si:H-based photovoltaic solar cells, i.e., samples 21821, 21886, and21916, are depicted in Figures 4.7, 4.8, and 4.9, respectively. The resultantcrystalline volume fractions are depicted in Table 4.1. The errors in thecrystalline volume fraction, determined through a systematic determinationof the range of possible peak fits corresponding to the given Raman spectra,are also depicted in Table 4.1.594.3. Raman analysis420 440 460 480 500 520 540Raman shift (cm?1)Intensity (a.u.)Sample 21821? = 442 nm~480 cm?1~510 cm?1~520 cm?1Figure 4.7: A subset of the Raman spectrum corresponding to sample 21821.The amorphous, grain boundary, and crystalline peak components have beenhighlighted with thin solid lines, the corresponding peaks being around 480,510, and 520 cm?1, respectively. The experimental data points are indi-cated with the open points. The fit with experiment, obtained through theaddition of all of the peaks, is indicated with the solid red line. Baselinecorrection has been employed. The online version is depicted in color.604.3. Raman analysis420 440 460 480 500 520 540Raman shift (cm?1)Intensity (a.u.)Sample 21886? = 442 nm~480 cm?1~510 cm?1~520 cm?1Figure 4.8: A subset of the Raman spectrum corresponding to sample 21886.The amorphous, grain boundary, and crystalline peak components have beenhighlighted with thin solid lines, the corresponding peaks being around 480,510, and 520 cm?1, respectively. The experimental data points are indi-cated with the open points. The fit with experiment, obtained through theaddition of all of the peaks, is indicated with the solid red line. Baselinecorrection has been employed. The online version is depicted in color.614.3. Raman analysis420 440 460 480 500 520 540Raman shift (cm?1)Intensity (a.u.)Sample 21916? = 442 nm~480 cm?1~510 cm?1~520 cm?1Figure 4.9: A subset of the Raman spectrum corresponding to sample 21916.The amorphous, grain boundary, and crystalline peak components have beenhighlighted with thin solid lines, the corresponding peaks being around 480,510, and 520 cm?1, respectively. The experimental data points are indi-cated with the open points. The fit with experiment, obtained through theaddition of all of the peaks, is indicated with the solid red line. Baselinecorrection has been employed. The online version is depicted in color.624.4. Experimental resolution limitation4.4 Experimental resolution limitationBefore determining the dependence of the crystalline volume fraction onthe mean crystallite size, it is instructive to consider what region of the nc-Si:H-based photovoltaic solar cells is being probed during each experimentalmeasurement. A detailed analysis of the attenuation that occurs, for boththe X-rays generated by the XRD measurement system and the Ramansource, demonstrates that it is the nc-Si:H intrinsic layers associated withthe nc-Si:H-based photovoltaic solar cells that are being probed [93]. Thep-type nc-Si:H doping layer, located at the surface of these cells, is only15-20 nm thick, while the penetration depths, corresponding to the incidentX-rays (used for the XRD measurements) and the Raman laser are foundto be 15 and 0.125 ?m, respectively, the intrinsic nc-Si:H layer underneaththis top contact layer being about 3 ?m in thickness; these thicknesses areknown from the impurity concentration profile results of Yue et al. [50].A pictorial, depicting the attenuation of the X-rays and the Raman lasersource, is depicted in Figure 4.10. It is seen that the majority of the attenua-tion, in both cases, occurs in the intrinsic nc-Si:H region itself. In the case ofthe Raman source, however, it is noted that some attenuation occurs acrossthe top p-layer, while for the X-rays, very little attenuation occurs acrossthis top p-layer. If the properties of the nc-Si:H are uniform across the in-trinsic nc-Si:H layer, these measurements correspond to the same material.Inhomogeneities in the intrinsic nc-Si:H region, below the first 125 nm, willnot be detectable with the Raman light source that is being used for theseexperiments. A longer wavelength Raman source, with a greater penetration634.4. Experimental resolution limitationFigure 4.10: The attenuation of the X-ray source and the Raman laser sourceas a function of the depth into the material. A profile of the underlying nc-Si:H-based photovoltaic solar cell device structure is depicted beneath. Theonline version of this figure is depicted in color644.5. The dependence of the crystalline volume fraction on the mean crystallite sizeinto the material, would allow for the detection of these inhomogeneities, asthe penetration depth would be deeper in such a case.4.5 The dependence of the crystalline volumefraction on the mean crystallite sizeWith the mean crystallite sizes determined, and the corresponding crys-talline volume fractions assessed, the role that the mean crystallite size playsin determining the corresponding crystalline volume fraction can now beprobed. In Figure 4.11, the crystalline volume fraction is plotted as a func-tion of the mean crystallite size for the three nc-Si:H-based photovoltaicsolar cells considered in this analysis. An increase in the crystalline volumefraction is noted as the mean crystallite size is diminished. Error bars, cor-responding to samples 21821, 21886, and 21916, are depicted. While thereis a lot of experimental error, it is noted that the crystalline volume fractionseems to increase as the mean crystallite size diminishes. It may be opinedthat smaller mean crystallite sizes will lead to a greater crystalline vol-ume fraction as smaller crystallites may be randomly packed with a greaterpacking density than those that are larger; this is an edge effect [93]. It isinteresting to note that a similar trend was observed by Funde et al. [89],although a very different range of mean crystallite sizes and crystalline vol-ume fractions is reported in their analysis; Funde et al. [89] performed theirXRD analysis on the crystalline silicon (111) peaks. The reasons for thisdifference remain unknown at the present time. A recent analysis, presentedby Schmidt et al. [93], appears to confirm this trend over a much broader654.5. The dependence of the crystalline volume fraction on the mean crystallite size28 30 32 34 36 38 40 42 44222324252627Crystallite size (nm)Crystalline volume fraction (%)Figure 4.11: The crystalline volume fraction as a function of the mean crys-tallite size for the three nc-Si:H-based photovoltaic solar cells considered inthis analysis [94]. The error bars, corresponding to each data point, aredepicted. The online version is depicted in color664.5. The dependence of the crystalline volume fraction on the mean crystallite sizerange of mean crystallite sizes and crystalline volume fractions; in this study,eight additional nc-Si:H-based photovoltaic solar cells were considered, thedetails being further discussed in the literature [93].67Chapter 5ConclusionsThe dependence of the crystalline volume fraction on the mean crys-tallite size, for three nc-Si:H-based photovoltaic solar cells, was examined.For each photovoltaic solar cell considered, XRD and Raman spectra weredetermined. Through the application of Scherrer?s equation, the XRD re-sults were used in order to determine the corresponding mean crystallitesizes. Through peak decomposition, the Raman results were used in orderto estimate the corresponding crystalline volume fraction. By plotting thecrystalline volume fraction as a function of the mean crystallite size, it wasfound that larger mean crystallite sizes tend to favor reduced crystallinevolume fractions. The ability to randomly pack smaller crystallites with agreater packing fraction than their larger counterparts was suggested as apossible explanation for this observation. Further analysis, however, will berequired in order to confirm this speculation.There are a number of possible further extensions related to this workthat can be considered in the near-term future. The consideration of abroader range of nc-Si:H-based photovoltaic solar cells is a priority item, asit is difficult to claim a statistically significant trend based on only threenc-Si:H-based photovoltaic solar cell samples. A detailed critical compari-68Chapter 5. Conclusionsson with the results of others would also strengthen the claims made withinthis analysis. The XRD scans that were performed for the purposes of thisanalysis corresponded to the whole XRD spectra, i.e., over all possible an-gles. By just scanning in the neighborhood of the silicon (220) peak itself,for substantial periods of time, the mean crystallite size error that is foundcan be substantially reduced. The use of a longer wavelength for the Ramanlaser source would also help, i.e., the penetration depth would be increased.Finally, a theoretical analysis of the dependence of the packing fraction onthe sphere diameter for the case of random packing, would provide some fur-ther theoretical heft to this analysis. These topics will have to be addressedin the future.69References[1] C. Weiner, ?How the transistor emerged,? Spectrum, IEEE, vol. 10,no. 1, pp. 24?33, 1973. ? pages 1[2] J.-C. Boudenot, ?From transistor to nanotube,? Comptes RendusPhysique, vol. 9, no. 1, pp. 41 ? 52, 2008. ? pages 1[3] M. Riordan, ?The lost history of the transistor,? IEEE Spectrum,vol. 41, no. 5, pp. 44 ? 49, 2004. ? pages 1[4] G. E. Moore, ?Cramming more components onto integrated circuits,?Electronics, vol. 38, no. 8, pp. 114?ff, 1965. ? pages 2[5] B. Lojek, ?Shockley semiconductor laboratories,? in History of semi-conductor engineering, B. Lojek, Ed. Berlin Heidelberg: Springer-Verlag, 2007. ? pages 2[6] W. A. Hadi, Ph.D. dissertation, University of Windsor, 2014. ? pagesvii, 2, 3[7] K. Ahmed and K. Schuegraf, ?Transistor wars,? IEEE Spectrum,vol. 48, no. 11, pp. 50?66, 2011. ? pages 2[8] B. G. Streetman and S. K. Banerjee, ?Integrated circuits,? in Solid70Chapter 5. Referencesstate electronic devices, 6th ed., B. G. Streetman and S. K. Banerjee,Eds. Upper Saddle River: Pearson Prentice Hall, 2006. ? pages 2[9] S. K. O?Leary, ?Presentation,? presented at the University of Toronto,2005. ? pages vii, 2, 4[10] R. C. Chittick, J. H. Alexander, and H. F. Sterling, ?The preparationand properties of amorphous silicon,? Journal of The ElectrochemicalSociety, vol. 116, no. 1, pp. 77?81, 1969. ? pages 5, 12[11] R. C. Chittick, W. E. Spear, and P. G. L. Comber, ?Substitutionaldoping of amorphous silicon,? Solid State Communications, vol. 17,no. 9, pp. 1193?1196, 1975. ? pages 5, 12[12] W. E. Spear and P. G. L. Comber, ?Electronic properties of substitu-tionally doped amorphous si and ge,? Philosophical Magazine B, vol. 33,no. 6, pp. 935?949, 1976. ? pages 5[13] D. E. Carlson and C. R. Wronski, ?Amorphous silicon solar cell,? Ap-plied Physics Letters, vol. 28, no. 11, pp. 671?673, 1976. ? pages 5[14] W. Paul, D. K. Paul, B. von Roedern, J. Blake, and S. Oguz, ?Preferen-tial attachment of h in amorphous hydrogenated binary semiconductorsand consequent inferior reduction of pseudogap state density,? PhysicalReview Letters, vol. 46, no. 15, pp. 1016?1020, 1981. ? pages 6[15] G. Moddel, J. Blake, R. W. Collins, P. Viktorovitch, D. K. Paul, B. vonRoedern, and W. Paul, ?Comparison of a-si:h produced by rf sputter-71Chapter 5. Referencesing and plasma decomposition methods,? AIP Conference Proceedings,vol. 73, no. 1, pp. 25?30, 1981. ? pages 6[16] S. Zukutynski, Private communication. ? pages 6[17] R. A. Street, Hydrogenated amorphous silicon. New York: CambridgeUniversity Press, 1991. ? pages 6, 8, 10[18] D. L. Staebler and C. R. Wronski, ?Reversible conductivity changesin discharge-produced amorphous si,? Applied Physics Letters, vol. 31,no. 4, pp. 292?294, 1977. ? pages vii, 6, 7[19] M. Faraji, S. Gokhale, S. M. Choudhari, M. G. Takwale, and S. V.Ghaisas, ?High mobility hydrogenated and oxygenated microcrystallinesilicon as a photosensitive material in photovoltaic applications,? Ap-plied Physics Letters, vol. 60, no. 26, pp. 3289?3291, 1992. ? pages 8,15[20] J. Meier, R. Fluckiger, H. Keppner, and A. Shah, ?Complete microcrys-talline p-i-n solar cell-crystalline or amorphous cell behavior?? AppliedPhysics Letters, vol. 65, no. 7, pp. 860?862, 1994. ? pages 8, 15[21] B. Yan, G. Yue, J. M. Owens, J. Yang, and S. Guha, ?Light-inducedmetastability in hydrogenated nanocrystalline silicon solar cells,? Ap-plied Physics Letters, vol. 85, no. 11, pp. 1925?1927, 2004. ? pages8[22] B. Yan, G. Yue, L. Sivec, C. Jiang, Y. Yan, K. Alberi, J. Yang, andS. Guha, ?On the bandgap of hydrogenated nanocrystalline silicon thin72Chapter 5. Referencesfilms,? in Photovoltaic Specialists Conference (PVSC), 2010 35th IEEE,2010, pp. 003 755?003 760. ? pages 8, 30, 32, 44, 59[23] J. Meier, S. Dubail, R. Platz, P. Torres, U. Kroll, J. A. A. Selvan,N. P. Vaucher, C. Hof, D. Fischer, H. Keppner, R. Fliickiger, A. Shah,V. Shklover, and K.-D. Ufert, ?Towards high-efficiency thin-film siliconsolar cells with the ?micromorph? concept,? Solar Energy Materialsand Solar Cells, vol. 49, no. 1, pp. 35?44, 1997. ? pages 8[24] X. Y. Chen, W. Z. Shen, and Y. L. He, ?Enhancement of electronmobility in nanocrystalline silicon crystalline silicon heterostructures,?Journal of Applied Physics, vol. 97, no. 2, p. 024305, 2005. ? pages 8[25] J. Liang, E. A. Schiff, S. Guha, B. Yan, and J. Yang, ?Hole-mobilitylimit of amorphous silicon solar cells,? Applied Physics Letters, vol. 88,no. 6, p. 063512, 2006. ? pages 8[26] B. Yan, G. Yue, X. Xu, J. Yang, and S. Guha, ?High efficiency amor-phous and nanocrystalline silicon solar cells,? Physica status solidi. A,vol. 207, no. 3, pp. 671?677, 2010. ? pages 8, 17[27] R. Tsu, J. Gonzalez-Hernandez, S. S. Chao, S. C. Lee, and K. Tanaka,?Critical volume fraction of crystallinity for conductivity percolation inphosphorusdoped si:f:h alloys,? Applied Physics Letters, vol. 40, no. 6,pp. 534?535, 1982. ? pages 10, 59[28] F. Finger, ?Basic properties of hydrogenated microcrystalline silicon,?in Thin-film silicon solar cells, A. Shah, Ed. Boca Raton: CRC Press,2010. ? pages 10, 11, 15, 16, 18, 20, 3073Chapter 5. References[29] R. Singh, S. Prakash, N. N. Shukla, and R. Prasad, ?Sample depen-dence of the structural, vibrational, and electronic properties of a-si:h:A density-functional-based tight-binding study,? Physical Review B,vol. 70, no. 11, p. 115213, 2004. ? pages 10[30] K. Yamamoto, A. Nakajima, M. Yoshimi, T. Sawada, S. Fukuda,T. Suezaki, M. Ichikawa, Y. Koi, M. Goto, T. Meguro, T. Matsuda,M. Kondo, T. Sasaki, and Y. Tawada, ?A high efficiency thin film sil-icon solar cell and module,? Solar Energy, vol. 77, no. 6, p. 939?949,2004. ? pages 11[31] A. Nakajima, M. Gotoh, T. Sawada, S. Fukuda, M. Yoshimi, K. Ya-mamoto, and T. Nomura, ?Development of thin-film si hybrid solarmodule,? Solar Energy Materials & Solar Cells, vol. 93, no. 6, pp.1163?1166, 2009. ? pages[32] H.-Y. Mao, S.-Y. Lo, D.-S. Wuu, B.-R. Wu, S.-L. Ou, H.-Y. Hsieh,and R.-H. Horng, ?Hot-wire chemical vapor deposition and character-ization of p-type nanocrystalline si films for thin film photovoltaic ap-plications,? Thin Solid Films, vol. 520, no. 16, pp. 5200?5205, 2012. ?pages 11[33] M. Hara, ?High mobility bottom gate nanocrystalline-si thin-film tran-sistors,? Thin Solid Films, vol. 519, no. 11, p. 3922?3924, 2011. ?pages 11[34] B. Yan, J. Yang, and S. Guha, ?Amorphous and nanocrystalline siliconthin film photovoltaic technology on flexible substrates,? Journal of74Chapter 5. ReferencesVacuum Science and Technology A, vol. 30, no. 4, p. 04D108, 2012. ?pages 11[35] L. Y. Zhu, X. F. Huang, W. B. Fan, X. W. Wang, W. Li, L. Wang, andK. J. Chen, ?Microstructures of patterned nc-si superlattices made bypulsed laser interference crystallization,? Superlattices and Microstruc-tures, vol. 31, no. 6, pp. 285?288, 2002. ? pages 11[36] H. Tan, E. Psomadaki, O. Isabella, M. Fischer, P. Babal, R. Vasude-van, M. Zeman, and A. H. M. Smets, ?Micro-textures for efficient lighttrapping and improved electrical performance in thin-film nanocrys-talline silicon solar cells,? Applied Physics Letters, vol. 103, no. 17, pp.173 905?173 905?5, 2013. ? pages 11, 30[37] C. Wronski, B. V. Roedern, and A. Ko lodziej, ?Thin-film si:h-basedsolar cells,? Vacuum, vol. 82, no. 10, p. 1145?1150, 2008. ? pages 11[38] O. Vetterl, F. Finger, R. Carius, P. Hapke, L. Houben, O. Kluth,A. Lambertz, A. Mu?ck, B. Rech, and H. Wagner, ?Intrinsic micro-crystalline silicon: A new material for photovoltaics,? Solar EnergyMaterials and Solar Cells, vol. 62, no. 1, pp. 97?108, 2000. ? pages16, 30, 37[39] E. Vallat-Sauvain, U. Kroll, J. Meier, A. Shah, and J. Pohl, ?Evolutionof the microstructure in microcrystalline silicon prepared by very highfrequency glow-discharge using hydrogen dilution,? Journal of AppliedPhysics, vol. 87, no. 6, pp. 3137?3142, 2000. ? pages 43, 5575Chapter 5. References[40] W. Collins, A. S. Ferlauto, G. M. Ferreira, C. Chen, J. Koh, R. J. Koval,Y. Lee, J. M. Pearce, and C. R. Wronski, ?Evolution of microstructureand phase in amorphous, protocrystalline, and microcrystalline siliconstudied by real time spectroscopic ellipsometry,? Solar Energy Materialsand Solar Cells, vol. 78, no. 1, pp. 143?180, 2003. ? pages[41] M. Luysberg, P. Hapke, R. Carius, and F. Finger, ?Structure andgrowth of hydrogenated microcrystalline silicon: Investigation by trans-mission electron microscopy and raman spectroscopy of films grownat different plasma excitation frequencies,? Philosophical Magazine A,vol. 75, no. 1, pp. 31?47, 1997. ? pages 11, 30[42] S. Vepr?ek and V. Marec?ek, ?The preparation of thin layers of ge andsi by chemical hydrogen plasma transport,? Solid State Electronics,vol. 11, no. 7, pp. 683?684, 1968. ? pages 13[43] S. Usui and M. Kikuchi, ?Properties of heavily doped gd-si with lowresistivity,? Journal of Non-Crystalline Solids, vol. 34, no. 1, pp. 1?11,1979. ? pages 13[44] S. Vepr?ek, Z. Iqbal, R. O. Ku?hne, P. Capezzuto, F.-A. Sarott, andJ. K. Gimzewski, ?Properties of microcrystalline silicon: Iv. electricalconductivity, electron spin resonance and the effect of gas adsorption,?Journal of Physics C: Solid State Physics, vol. 16, pp. 6241?6262, 1983.? pages 14, 43[45] A. Matsuda, ?Formation kinetics and control of microcrystallite in ?c-76Chapter 5. Referencessi:h from glow discharge plasma,? Journal of Non-Crystalline Solids,vol. 59, pp. 767?774, 1983. ? pages 14[46] C. C. Tsai, G. B. Anderson, R. Thompson, and B. Wacker, ?Controlof silicon network structure in plasma deposition,? Journal of Non-Crystalline Solids, vol. 114, pp. 151?153, 1989. ? pages 14[47] Y. Hishikawa, S. Tsuda, K. Wakisaka, and Y. Kuwano, ?Principlesfor controlling the optical and electrical properties of hydrogenatedamorphous silicon deposited from a silane plasma,? Journal of AppliedPhysics, vol. 73, no. 9, pp. 4227?4231, 1993. ? pages 14[48] H. Curtins, N. Wyrsch, and A. V. Shah, ?High-rate deposition of amor-phous hydrogenated silicon: effect of plasma excitation frequency,?Electronics Letters, vol. 23, no. 5, pp. 228?230, 1987. ? pages 14[49] H. Takatsuka, M. Noda, Y. Yonekura, Y. Takeuchi, and Y. Yamauchi,?Development of high efficiency large area silicon thin film modulesusing vhf-pecvd,? Solar Energy, vol. 77, no. 6, pp. 951?960, 2004. ?pages 14[50] G. Yue, B. Yan, L. Sivec, Y. Zhou, J. Yang, and S. Guha, ?Effect ofimpurities on performance of hydrogenated nanocrystalline silicon solarcells,? Solar Energy Materials and Solar Cells, vol. 104, pp. 109?112,2012. ? pages 14, 34, 35, 43, 63[51] S. Oda and M. Yasukawa, ?High quality a-si:h films and interfaces pre-pared by vhf plasma cvd,? Journal of Non-Crystalline Solids, vol. 137,pp. 677?680, 1991. ? pages77Chapter 5. References[52] F. Finger, P. Hapke, M. Luysberg, R. Carius, H. Wagner, andM. Scheib, ?Improvement of grain size and deposition rate of microcrys-talline silicon by use of very high frequency glow discharge,? AppliedPhysics Letters, vol. 75, no. 20, pp. 2588?2590, 1994. ? pages[53] R. E. Hollingsworth and P. K. Bhat, ?Doped microcrystalline silicongrowth by high frequency plasmas,? Applied Physics Letters, vol. 75,no. 5, pp. 616?618, 1994. ? pages 14[54] H. Matsumura, ?Formation of polysilicon films by catalytic chemical va-por deposition (cat-cvd) method,? Japanese Journal of Applied Physics,vol. 30, no. Part2/8B, pp. L1522?L1524, 1991. ? pages 14[55] J. K. Rath, ?Low temperature polycrystalline silicon: a review on de-position, physical properties and solar cell applications,? Solar EnergyMaterials and Solar Cells, vol. 76, no. 4, pp. 431?487, 2003. ? pages14[56] Y. Mai, S. Klein, R. Carius, H. Stiebig, L. Houben, X. Geng, andF. Finger, ?Improvement of open circuit voltage in microcrystalline sil-icon solar cells using hot wire buffer layers,? Journal of Non-CrystallineSolids, vol. 352, no. 9, p. 1859?1862, 2006. ? pages 14[57] Y. Mai, S. Klein, R. Carius, J. Wolff, A. Lambertz, F. Finger, andX. Geng, ?Microcrystalline silicon solar cells deposited at high rates,?Journal of Applied Physics, vol. 97, no. 11, p. 114913, 2005. ? pages1478Chapter 5. References[58] M. N. van den Donker, S. Klein, B. Rech, F. Finger, W. M. Kessels,and M. C. M. van de Sanden, ?Microcrystalline silicon solar cells withan open-circuit voltage above 600 mv,? Applied Physics Letters, vol. 90,no. 18, p. 183504, 2007. ? pages 14[59] P. Torres, J. Meier, R. Flu?ckiger, U. Kroll, J. A. A. Selvan, H. Keppner,A. Shah, S. D. Littelwood, I. E. Kelly, and P. Giannoule`s, ?Devicegrade microcrystalline silicon owing to reduced oxygen contamination??Applied Physics Letters, vol. 75, no. 10, pp. 1373?1375, 1996. ? pages15, 16[60] T. Kamei and T. Wada, ?Oxygen impurity doping into ultrapure hydro-genated microcrystalline si films,? Journal of Applied Physics, vol. 96,no. 4, pp. 2087?2090, 2004. ? pages 15[61] F. Finger, R. Carius, T. Dylla, S. Klein, S. Okur, and M. Gu?nes, ?Sta-bility of microcrystalline silicon for thin film solar cell applications,?IEE Proc.-Circuits Devices Syst., vol. 150, no. 4, pp. 300?308, 2003. ?pages 16[62] V. Smirnov, S. Reynolds, F. Finger, R. Carius, and C. Main,?Metastable effects in silicon thin films: Atmospheric adsorption andlight-induced degradation,? Journal of Non-Crystalline Solids, vol. 352,no. 9, p. 1075?1078, 2006. ? pages 16[63] T. Kilper, W. Beyer, G. Bra?uer, T. Bronger, R. Carius, M. N. van denDonker, D. Hrunski, A. Lambertz, T. Merdzhanova, A. Mu?ck, B. Rech,W. Reetz, R. Schmitz, U. Zastrow, and A. Gordijn, ?Oxygen and79Chapter 5. Referencesnitrogen impurities in microcrystalline silicon deposited under opti-mized conditions: Influence on material properties and solar cell per-formance,? Journal of Applied Physics, vol. 105, no. 7, p. 074509, 2009.? pages 16[64] C. Droz, E. Vallat-Sauvain, J. Bailat, L. Feitknecht, J. Meier, andA. Shah, ?Relationship between raman crystallinity and open-circuitvoltage in microcrystalline silicon solar cells,? Solar Energy Materialsand Solar Cells, vol. 81, no. 1, pp. 61?71, 2004. ? pages viii, 16, 18,19, 30, 59[65] F. Finger, L. B. Neto, R. Carius, T. Dylla, and S. Klein, ?Paramagneticdefects in undoped microcrystalline silicon,? Physica Status Solidi (c),vol. 1, no. 5, p. 1248?1254, 2004. ? pages 16[66] A. L. Donne, S. Binetti, G. Isella, B. Pichaud, M. Texier, M. Accia-rri, S. Pizzini, and S. Klein, ?Structural characterization of nc-si filmsgrown by low-energy pecvd on different substrates,? Applied SurfaceScience, vol. 254, no. 9, p. 2804?2808, 2008. ? pages 18[67] B. D. Cullity and S. R. Stock, Elements of X-Ray Diffraction, 3rd ed.Upper Saddle River: Prentice Hall, 2001. ? pages viii, 21, 22, 24, 25,33, 43, 55[68] S. Franssila, ?Silicon,? in Introduction to Microfabrication, 2nd ed.,S. Franssila, Ed. Hoboken: Wiley, 2010. ? pages viii, 21, 23[69] P. Scherrer, Nachrichten von der Gesellschaft der Wissenschaften zu80Chapter 5. ReferencesGo?ttingen, Mathematisch-Physikalische Klasse, vol. 1918, pp. 98?100,1918. ? pages 25[70] J. I. Langford and A. J. C. Wilson, ?Scherrer after sixty years: A surveyand some new results in the determination of crystallite size,? Journalof Applied Crystallography, vol. 11, no. 2, pp. 102?113, 1978. ? pages26[71] S.-L. Zhang, Raman spectroscopy and its application in nanostructures.John Wiley and Sons, 2012. ? pages 26[72] C. V. Raman, ?A new type of secondary radiation,? Nature (London),vol. 121, no. 3048, pp. 501?502, 1928. ? pages 27[73] T. H. Maiman, U.S. Patent 3,353,115, Apr. 13, 1961. ? pages 27[74] P. Vandenabeele, ?Theoretical aspects,? in Practical Raman spec-troscopy - an introduction, 1st ed., P. Vandenabeele, Ed. West Sussex:John Wiley & Sons, 2013. ? pages viii, 27, 28[75] O. Hollricher, ?Raman instrumentation for confocal raman mi-croscopy,? in Confocal Raman microscopy, 1st ed., T. Dieing, O. Holl-richer, and J. Toporski, Eds. Berlin Heidelberg: Springer Berlin Hei-delberg, 2011. ? pages ix, 27, 29[76] G. Yue, B. Yan, G. Ganguly, J. Yang, S. Guha, and C. W. Teplin,?Material structure and metastability of hydrogenated nanocrystallinesilicon solar cells,? Applied Physics Letters, vol. 88, no. 26, p. 263507,2006. ? pages 30, 44, 50, 5981Chapter 5. References[77] G. Yue, L. Sivec, B. Yan, J. Yang, and S. Guha, ?High efficiency amor-phous and nanocrystalline silicon based multi-junction solar cells de-posited at high rates on textured ag/zno back reflectors,? MRS Pro-ceedings, vol. 1153, no. A10-05, 2009. ? pages 30, 50[78] B. von Roedern and H. S. Ullal, ?The role of polycrystalline thin-filmpv technologies in competitive pv module markets,? in PhotovoltaicSpecialists Conference, 2008. PVSC ?08. 33rd IEEE, 2008, pp. 1?4. ?pages 35[79] S. Price and R. M. et al., ?2008 solar technologies market report,? U.S.Department of Energy - Energy Efficiency and Renewable Energy, 2008,Tech. Rep. 46025, 2013. ? pages 35[80] A. Chichagov. (2012) Institute of experimental mineralogy russianacademy of sciences. [Online]. Available: http://database.iem.ac.ru/mincryst/ ? pages 39[81] H. Li, E. Liu, F. Y. F. Chan, Z. Lu, and R. Chen, ?Fabrication ofordered flower-like zno nanostructures by a microwave and ultrasoniccombined technique and their enhanced photocatalytic activity,? Ma-terials Letter, vol. 65, no. 23-24, pp. 3440?3443, 2011. ? pages 39[82] F. E. Akkad, A. Punnoose, and G. Prabu, ?Properties of ito films pre-pared by rf magnetron sputtering,? Applied Physics A, vol. 71, no. 2,pp. 157?160, 2000. ? pages 39[83] A. Matsuda, K. Kumagai, and K. Tanaka, ?Wide-range control ofcrystallite size and its orientation in glow-discharge deposited ?c-si:h,?82Chapter 5. ReferencesJapanese Journal of Applied Physics, vol. 22, no. 1, pp. L34?L36, 1982.? pages 43[84] W. Wei, G. Xu, J. Wang, and T. Wang, ?Raman spectra of intrinsicand doped hydrogenated nanocrystalline silicon films,? Vacuum, vol. 81,no. 5, p. 656?662, 2007. ? pages xi, 43, 44, 45[85] G. Yue, J. D. Lorentzen, J. Lin, D. Han, and Q. Wang, ?Photolu-minescence and raman studies in thin-film materials: transition fromamorphous to microcrystalline silicon,? Applied Physics Letters, vol. 75,no. 4, pp. 492?494, 1999. ? pages 44, 55[86] D. Han, J. D. Lorentzen, J. Weinberg-Wolf, L. E. McNeil, and Q. Wang,?Raman study of thin films of amorphous-to-microcrystalline siliconprepared by hotwire chemical vapor deposition,? Journal of AppliedPhysics, vol. 94, no. 5, pp. 2930?2936, 2006. ? pages 44, 59[87] U. Kroll, J. Meier, P. Torres, J. Pohl, and A. V. Shah, ?From amorphousto microcrystalline silicon films prepared by hydrogen dilution using thevhf (70 mhz) gd technique,? Journal of Non-Crystalline Solids, vol. 227,pp. 68?72, 1998. ? pages 50[88] A. V. Shah, J. Meier, E. Vallat-Sauvain, N. Wyrsch, U. Kroll, C. Droz,and U. Graf, ?Material and solar cell research in microcrystalline sili-con,? Solar Energy Materials and Solar Cells, vol. 78, no. 1, pp. 469?491, 2003. ? pages 50[89] A. M. Funde, N. A. Bakr, D. K. Kamble, R. R. Hawaldar, D. P. Amal-nerkar, and S. R. Jadkar, ?Influence of hydrogen dilution on structural,83Chapter 5. Referenceselectrical and optical properties of hydrogenated nanocrystalline sili-con (nc-si:h) thin films prepared by plasma enhanced chemical vapourdeposition (pe-cvd),? Solar Energy Materials and Solar Cells, vol. 92,no. 10, p. 1217? 1223, 2008. ? pages 54, 65[90] L.-L. Tay, D. J. Lockwood, J.-M. Baribeau, M. Noe?l, J. C. Zwinkels,F. Orapunt, and S. K. O?Leary, ?Influence of growth temperature onorder within silicon films grown by ultrahigh-vacuum evaporation onsilica,? Applied Physics Letters, vol. 88, no. 12, p. 121920, 2006. ?pages 59[91] E. Bustarret, M. A. Hachicha, and M. Brunel, ?Experimental deter-mination of the nanocrystalline volume fraction in silicon thin filmsfrom raman spectroscopy,? Applied Physics Letters, vol. 52, no. 20, pp.1675?1677, 1988. ? pages 59[92] H. L. Z. L. C. Wen, H. Xu and W. Shen, ?Passivation of nanocrys-talline silicon photovoltaic materials employing a negative substratebias,? Nanotechnology, vol. 24, no. 45, p. 455602, 2013. ? pages 59[93] K. J. Schmidt, Y. Lin, M. Beaudoin, G. Xia, S. K. O?Leary, G. Yue,and B. Yan, ?The mean crystallite size within a hydrogenated nanocrys-talline silicon based photovoltaic solar cell and its role in determiningthe corresponding crystalline volume fraction,? Canadian Journal ofPhysics, in press (2014). ? pages 63, 65, 67[94] ??, ?The dependence of the crystalline volume fraction on the crys-84Chapter 5. Referencestallite size for hydrogenated nanocrystalline silicon based solar cells,?in MRS Proceedings, vol. 1536, 2013. ? pages xviii, 6685

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0074324/manifest

Comment

Related Items