Search

In this thesis, crystalline silicon thin films by atmospheric pressure chemical vapour deposition have been studied. These silicon films can be deposited on silicon wafers or transferred to various substrates for photovoltaic applications. One of the main advantages is the flexibility in thickness and doping concentration which allows the application of the silicon thin films in various solar cell concepts. The combination of these films with an industrial solar cell fabrication process has a high efficiency potential and offers a cost reduction as well as reduced material consumption. The key aspects of this publication were the deposition process, the characterisation of silicon thin films and the implementation of these layers into different solar cell concepts. In the following subchapters the main findings will be summarised and suggestions for further investigation will be given.Silicon Deposition by APCVD at Temperatures from 1150 °C to 850 °CThe silicon deposition by atmospheric pressure chemical vapour deposition (APCVD) using chlorosilanes is described by a set of gas phase and surface reactions. A general and a simplified model of these chemical reactions for the decomposition and the deposition process using TCS as precursor have been discussed. The identified assumptions and limitations of the simplified model underline the importance of process characterisation.Since the deposition process also depends on the reactor setup, a general description of the lab-type deposition tool RTCVD100 and RTCVD160 was given. The presented characterisation of the deposition homogeneity is crucial for the material characterisation as well as solar cell fabrication and shows the limitations and the influence of the deposition properties. Based on the deposition of cSiTF at 1150 °C an additional process at 1050 °C has been developed.The correlations between precursor composition, deposition temperature, deposition rate substrates orientation and doping incorporation show the complexity of crystalline silicon thin film (cSiTF) deposition. Therefore, detailed investigations of the deposition process at temperatures between 1150 °C and 850 °C with different gas mixtures have been carried out. The change in deposition rate from 0.5 ± 0.2 µm/min to 2.4 ±0.4 µm/min increases the throughput or allows the reproducible formation of silicon films with a thickness below 500 nm. The basic mechanism for the doping incorporation and the resulting doping range of 1x1015 1/cm³ to 2x1020 1/cm³ for boron and 1x1017 1/cm³ to 3x1020 1/cm³ for phosphorous have been explained. Additionally, the increase in doping incorporation by almost one order of magnitude depending on the Cl/H ratio has been investigated.Subsequently, the process development has been used to further optimise the deposition of the epitaxial emitters. The deposition of advanced emitter profiles with a thickness from 500 nm to 3.5 µm and the deposition on textured as well as planar wafers has been analysed. Furthermore, the reduction of contact peak thickness from 500 nm to 25-50 nm by diffusion or epitaxial deposition and the influence of the following process step have been discussed.Apart from the epitaxial deposition on silicon substrates at a deposition temperature from 1150 °C to 950 °C the direct deposition of microcrystalline silicon thin film (µcSiTF) on substrates with intermediate layers between 1050 °C and 850 °C has been realised using APCVD. In addition, the successful deposition of µcSiTF on temperature sensitive substrates like borosilicate glass has been presented. These µcSiTF can be applied as single junction thin film solar cell, as active layer in a tandem solar cell, as diffusion barrier or contact layer and as seed layer for recrystallisation. The basic understanding of this deposition process has been complemented by an investigation about the nucleation process. The experimental results show that the random nucleation density of silicon on SiNx and SiOx by PECVD is increasing with temperature and results in a homogeneous seed layer. However, the etching of thermal SiO2 in hydrogen atmosphere at temperatures above 900 °C reduces the nucleation density on this type of intermediate layer from approximately 270 to 40 counts/mm2.The presented experimental results and the identified correlations are essential for the optimisation and further development of the deposition processes by APCVD.Crystallographic Characterisation of Silicon Thin FilmsDifferent measurement methods to characterise the crystallographic properties of cSiTF have been presented. The formation of various defects like etch pits, stacking faults and spikes in the epitaxial layer depending on the deposition temperature, precursor composition as well as substrate orientation have been shown. The increasing number of stacking faults and the change in recombination intensity of those defects at temperatures below 1100 °C can result in a VOC loss of approximately 200 mV.The microcrystalline silicon films by direct deposition have been characterised with EBSD, XRD and µRaman measurements. The crystallographic properties of these layers including a grain size of 2-3 µm and the columnar growth have been determined. Various measurements to determine the electrical material quality of microcrystalline silicon films depending on deposition and post treatment parameters have also been done. These experiments lead to the conclusion that the RTA and RPHP treatments are necessary for a deposition temperature below 1000 °C because the stress and temperature sensitive defects can be reduced.The limitations of established measurement techniques for the characterisation of cSiTF and the benefit of approaches like the calibrated µRaman, µPL and µLBIC measurements have been shown. Based on a set of calibration samples it is possible to determine the doping concentration above 6x1016 1/cm3 by analysing the Fano resonance. The presented measurements of epitaxial silicon films have been used to investigate the overall material quality and the stress distribution in these layers. Furthermore, the investigation of the growth interface between epitaxial films and the silicon substrates shows no increased stress or recombination activity. However, the influence of the growth regime of epitaxial films on the reflection properties of textures samples has been identified as crucial parameter.These experiments have been necessary to explain and improve the material quality of cSiTF in general. Furthermore, these measurements prove the applicability of epitaxial emitters for wafer based solar cells.Lifetime measurements of cSiTF and wafer based solar cellsThe challenges and solutions for reliable effective minority carrier lifetime measurements on various cSiTF have been discussed. Based on the upgraded setup and an improved measurement routine the MWPCD signal intensity has been increased from approximately 4 mV to values above 35 mV. In combination with a new analysis algorithm, which incorporates the effect of carrier trapping on the decay transient, it is possible to determine reliable lifetimes below 1 µs of various cSiTF. Additional experiments to determine the influence of the excitation laser wavelength, the excess carrier density, the substrates and the passivation have been evaluated. The measurements of cSiTF deposited in the RTCVD160 show an effective diffusion length of about 90-120 µm and epitaxial cSiTF on recrystallised templates of 50-80 µm. Subsequently, MDP, LBIC and IQE measurements of cSiTF solar cells have been conducted to validate the MWPCD results and to determine the material quality of cSiTF.Furthermore, QSSPC measurements of silicon wafers with epitaxial emitters and experiments to investigate the material degradation at high temperatures have been presented. The effective lifetime and emitter saturation current measurements underline the successful optimisation of the epitaxial emitter deposition. At 1050 °C the effective minority carrier lifetimes of 200 µs for p-type and 250 µs for n-type show an increase of more than one order of magnitude compared to the standard epitaxial process at 1150 °C. The emitter saturation current of J0e < 45 fA/cm2 for p-type and J0e < 30 fA/cm2 for n-type FZ wafers show the potential for an application of epitaxial emitters into industrial and high efficiency solar cell concepts.In addition, the lifetime degradation of silicon substrates during the standard deposition process above 1100 °C has been analysed. The experiments show a minor lifetime degradation due to the formation of thermal defects by the high thermal budget and the cooling ramp. However, the diffusion of impurities depending on duration, ambient gas and temperature in the reactor has been identified as the dominating effect. Iron concentrations measurements on reference wafers of 1x108 1/cm3 increases to values of 4x1011 1/cm3 after a 4 min annealing process at 1100 °C. Alternative effects like the formation of mono vacancies and the migration of defects from the silicon surface have also been discussed.Solar cells using cSiTF and µcSiTFThe properties and possible adjustments to cSiTF solar cells have been investigated using the epitaxial wafer equivalent (EpiWE) concept. An alternative formation of the BSF by diffusion from a gaseous source with similar absorber material quality EPD < 2x104 1/cm² and the integration of a plasma texture have been carried out. An optimisation in absorber thickness to 30 µm and the decrease of open circuit voltages at deposition temperatures below 1100 °C has been presented. Furthermore, the development and a proof of concept of a solar cell process for the n-type cSiTF with p-type epitaxial emitter by APCVD with a record VOC of 655 mV and an efficiency of 13.9 % have been realised.Additional experiments have been conducted to investigate the different applications for the microcrystalline silicon layers by direct deposition using APCVD below 1000 °C. The presented experiments evaluate the solar cell properties of these layers including the parameters contact resistance, series resistance and material lifetime. The solar cell batches proof the feasibility of these concepts and show a VOC of 466 mV and a current of 14 mA without light trapping and texturing for a 10 µm thick absorber. Additional results and simulation by PC1D lead to the conclusion that the absorber layer has to be reduced to a thickness of 6 µm. Using microcrystalline silicon layers as seed layer on glass in combination with e-beam crystallisation shows voltages of 539 mV. A more industrial feasible approach is the crystallisation by ZMR resulting in voltages of 614 mV [8].These results and the characterisation of cSiTF solar cells will be used to increase the efficiency and the overall understanding of various cSiTF concepts.Wafer based solar cells with epitaxial emitterIn the last Chapter the development of n- and p-type solar cell concepts with epitaxial emitters has been presented. Solar cell simulations by PC1D and process simulations by Sentaurus Process have been conducted to determine the potential and limitations of epitaxial emitters by APCVD and to optimise the fabrication process. Based on these simulations and the improvement in material quality, the solar cell batches show record efficiencies of eta = 16.4 % for n-type solar cells and eta = 16.1 % for p-type cells on mc wafers. Optimised deposition process for mc wafers at 1000 °C and for mono crystalline wafers at 1025 °C have been established.The adjustments in metallisation and passivation lead to optimised concepts and new solar cell processes for epitaxial emitters and for cSiTF concepts as well. The record solar cells with a conversion efficiency of eta = 20.0 % on n-type, FZ and eta = 18.4 % on p-type, FZ proves that epitaxial emitters are able to compete with the state of the art diffusion process. These results are an increase in efficiency of over 2 % absolute over the last 3 years und show the successful process optimisation. The diffusion length Leff > 1500 µm and the similar open circuit voltages compared to the diffused emitter of 658 mV for n-type solar cells also underline their potential for high efficiency and industrial application.The improvement and the transfer of the epitaxial emitter from a simplified development concept to the high efficiency concepts includes the development of the plasma texture with in-situ selective emitter formation (JSC = + 2.85 mA/cm²) and an epitaxial BSF.In summary it can be stated that the advancements in process development presented in this thesis have already been used to optimise the current deposition process of cSiTF and will contribute to the ongoing development. Furthermore, the thorough characterisation and the application of alternative characterisation methods will contribute to a better understanding of cSiTF.

Differential light-biased dynamic measurements of charge carrier recombination properties in semiconductors have long been known to yield only differential rather than actual recombination properties. Therefore, the determination of injection-dependent recombination properties from such measurements was previously found to require integration over the entire injection range. Recent investigations of the phase shift between a time-modulated irradiation of silicon samples and excess carrier density reveal a striking analogy to the above findings: the phase shift is greater than the actual effective carrier lifetime in the case of a positive derivative of lifetime with respect to excess carrier density, and vice versa. This work attempts to rearrange the mentioned previous findings in a quantitative theoretical description of light-biased dynamic measurements of effective carrier lifetime. Both light-biased differential lifetime measurements as well as harmonically time-modulated methods without additional bias light are shown to represent a limiting case in a general treatment of light-biased dynamic lifetime measurements derived here. Finally, we sketch a way to obtain actual recombination properties from differential measurements—referred to as a differential-to-actual (d2a) lifetime analysis, which does not require integration over the entire injection range.

In this work, new approaches for the characterization of crystalline silicon wafers and solar cells using electroluminescence imaging (ELI) and photoluminescence imaging (PLI) were presented.Basis for the measurements was the construction of a sensitive, camera based luminescence imaging setup with a widely adjustable field of view. GaAs based optical filters were introduced for near optimal excitation light suppression. The illumination of samples for PLI is possible both with an LED panel and a homogenized solid-state laser. Both light sources have an adjustable and calibrated photon flux.A lifetime calibration method for PL images based on QSSPC measurements was implemented. To correlate the spatially resolved PLI data with integral QSSPC values, the QSSPC sensitivity distribution was characterized.In addition to the default Si CCD camera, several InGaAs cameras were evaluated and compared for the measurement of defect band luminescence. Corresponding measurements of a multicrystalline Si solar cell at different temperatures show both an anticorrelation of band-to-band PL with defect band PL and a PL increase for low temperatures.To separate the contributing factors to solar cell series resistance, a technique based on EL images was developed. A multilayered 2-D model of the solar cell is solved and fitted to p-n junction voltage maps from ELI until both voltage distributions agree. The series resistances of front metallization grid, metallization contact, emitter and base are parameters in the solar cell model and reach their respective values during fitting. The obtained results for different solar cells qualitatively respond correctly to changes, e.g. of emitter sheet resistance. However, light scattering effects lead to a systematic underestimation of the solar cell resistivity. Further refinements taking care of this issue are required to make quantitatively correct measurements.A newly developed technique for calibration-free imaging of the minority charge carrier lifetime tau_eff using time-resolved photoluminescence imaging (TR-PLI) was presented. The measurement is based on a modulated illumination of the sample, synchronized to a periodically shuttered recording of the PL emission. Three different shutter setups were described, with increasing experimental complexity and correspondingly higher time-resolution. The first setup uses a rotating shutter wheel in front of the camera lens and was able to resolve tau_eff down to ~300µs. The second setup-which was filed for patent-uses a shutter wheel in the intermediate image plane between two objective lenses, extending the measurement range down to ~5µs. The third setup is based on an electronically switched image intensifier unit, which allows reliable mapping of tau_eff from ~2.5µs to several ms. The electronic shuttering also simplifies evaluation and calibration, making this setup the preferred choice if the required image intensifier can be acquired.A comparison of the three shutter setups shows very similar results within their respective measurement range. The choice of excitation frequency has significant impact on the uncertainty of the measurement, an empirical optimum was found at f_exc = 1/tau_eff,max. Compared to the established lifetime measurement techniques QSSPC and µ-PCD, very good agreement was observed. At low excess carrier densities the measurements could not be compared as the results from QSSPC and µ-PCD were impaired by trapping and DRM artifacts, to which TR-PLI is less susceptible.To increase the dynamic range of the measurement, the TR-PLI lifetime map can be used to calibrate steady-state PL images. Also, a map of tau_eff at a laterally homogeneous excess carrier density can be obtained by the combination of several measurements.During the course of this work, several ideas to improve upon the presented techniques could not yet be realized. The series resistance characterization method would benefit from the implementation of a solar cell model including carrier diffusion, as well as the deconvolution of blurring effects in the luminescence images. The combined use of EL and PL might improve the robustness of the fitting procedure for this technique.For TR-PLI, a change of the evaluation method to use discrete Fourier transformation instead of fitting a model function is a promising approach to speed up the evaluation procedure, provided it does not degrade measurement quality. Alternatively, the procedure may be accelerated by the use of GPU computing for the evaluation algorithm. To extend the TR-PLI measurement range to even shorter lifetimes, a pulsed laser might be used for illumination. Extending the evaluation theory to account for rise and fall times of the illumination is another possible step to a greater measurement range.

This thesis focuses on three aspects of rear passivated solar cells:• The first part presented the theoretical aspects, with the analytical modeling of rear passivated solar cells.A special attention was given to analytical modeling of recombination losses and series resistance losses.• In the second part, two characterization methods for advanced solar cell structures have been resented. These methods, which were developed uring the Ph.D., are especially adapted to rear passivated structures.• In the third part, the technological aspect was developed, with the development of a deposition technique for Al2O3 passivation layers.

Der Siliziumwafer hat einen der größten Anteile an den Kosten einer auf kristallinem Silizium basierten Solarzelle.Diese Arbeit beschreibt den Einfluss verschiedenster Prozessschritte der Solarzellenherstellung auf eine Reihe neuartiger, defektreicher Silizium-Wafermaterialien, die verglichen mit herkömmlichen Silizium-Wafermaterialien gewisse Kostenvorteile bieten.Im Besonderen wird ein Hocheffizienzprozess für Laborsolarzellen vorgestellt, der sich auf eine Vielzahl unterschiedlicher Silizium-Wafermaterialien anwenden lässt und eine Bestimmung des Wirkungsgradpotentials dieser Materialien ermöglicht.

Characterization of ingots and wafers is important for quality control and evaluation of new manufacturing processes. On ingots the mechanisms of charge carrier lifetime and resistivity measurements have been investigated, involving the determination of the instrument characteristics. Scheil's equation was confirmed, the iron concentration was determined and boron oxygen defects were detected in the outer layer. The protocols SEMI MF-1188-1107 and SEMI MF-1391-0704 for determination of interstitial oxygen and substitutional carbon concentrations by infrared spectroscopy require sample thicknesses, which exceed common wafer thicknesses. Effects on thinner samples have been studied. We found a relation between resolution, sample thickness and the appearance of interference fringes. Physical and mathematical efforts to get rid of the interferences failed. A recalibration for the conversion coefficients is done for use with lower resolutions. To evaluate defect engineering of interstitial iron, an iron mapping method has been implemented for a Microwave-Photoconductance-Decay device.

The amorphous silicon carbide films investigated throughout this work were pro-duced on the basis of plasma enhanced chemical vapor deposition (PECVD). The two different PECVD reactors used for the deposition of the films were presented. The principle development of the material was conducted at the AK400M batch-type reactor from Roth&Rau and the processes were then transferred to an industrial type SINA in-line reactor from the same company. For standard passivating Si-rich a-Si1 xCx layers deposited in the AK400M reactor working with RF power only, SIMS measurements revealed rather low C-atom densities (< 10 at.%) in the matrix. Doping atom densities of 1×1021 cm-3 for phosphorous and of 5×1020 cm-3 for boron were determined for a typical gas flow of 100 sccm (B2H6 or PH3 highly diluted in H2). The a Si1 xCx films fabricated in the SINA reactor exhibit a similar composition as well as similar electrical properties as their batch-type counterparts.A parallel analysis of a-Si1-xCx/c-Si and a-Si1-xCx/c-Ge systems was performed. From literature it is known that the main differences in terms of surface passivation of crystalline silicon and germanium substrates are the instability of (thermally grown) GexOy and the limited practicability of hydrogen for the passivation of Ge dangling bonds. From a compositional viewpoint, the performed transmission electron micros-copy (TEM) studies reveal no difference in the film structure depending on the substrate type. With increasing C-content, the Si-rich a-Si1-xCx films become less dense, a fact which is attributed to an increasing microvoid density. The c-Si surface passivation quality is found to be directly correlated with the Si-H bond density in the film. The onset of Si H bond rupture at around 300°C therefore coincides with the onset of electrical degradation independently of the carbon content and the doping density in the film. The surface passivation of germanium by a-Si1-xCx demonstrates a fairly different behavior. First of all, the thermal stability of intrinsic films is significantly increased. The electrical degradation starts at temperatures as high as 450°C and therefore indicates a “decoupling” of hydrogen present in the film and passivation quality.Regarding the optimum substrate temperature during film deposition, a value of Topt=270°C was found to provide the best overall performance for Si as well as for the Ge substrate. A similar behavior for both substrate types is observed for deposition temperatures below and above this temperature. Post-deposition annealing improves the passivation quality for Tann Topt. This finding is in accordance with the role of hydrogen for the passivation of c-Si surfaces. As a result of the experiments, the carbon in the a-Si matrix is supposed to inhibit the epitaxial growth observed for a-Si depositions at increased temperatures. The latter is considered to be responsible for the electrical degradation of the film. In the case of c Ge, the level of surface passivation is assumed to be indirectly correlated with the hydrogen incorporation at increased deposition temperatures since the latter is supposed to have a direct impact on the growth kinetics of the film.From isothermal lifetime experiments, an activation energy for the degradation of the passivation quality was extracted. For the c-Si/a-Si1-xCx system this energy amounts to approx. 0.5 eV, for the c-Ge/a-Si1-xCx system a value of approx. 2 eV was found, clearly pointing to different passivation mechanisms involved. Furthermore the results of this type of experiment allow for linking the temperature stability of the electrical properties of the film to the equilibrium between Si-H bond rupture and its inverse process (Si H↔Si + H). The latter is triggered by the availability of free atomic or molecular hydrogen in the matrix.A set of a-Si1-xCx passivated silicon wafers of different thicknesses allowed for the extraction of the “pure” injection dependent effective SRVs. The fitting of the SRV data of p- and n-type silicon substrate with the extended Shockley-Read-Hall (SRH) model reveals a change of sign for the fixed charge density parameter Qf of the film when changing the doping polarity of the substrate. The existence of amphoteric interface (near) states is confirmed by SPV measurements. Several experiments performed with the a-Si1-xCx/c-Ge system point to a direct saturation of Ge dangling-bonds by Si- and/or C-atoms. The latter effectively suppresses carrier recombination at the germanium surface. The passivation of c-Si surfaces on the contrary is inherently linked to the saturation of Si dangling-bonds by hydrogen. Although the incorporation of carbon in the a-Si matrix was shown to clearly enhance the thermal stability of the c-Si passivation in the temperature range up to 500°C (as compared to pure a-Si), the stability during firing processes (750-900°C) could not be verified.The performance of intrinsic and doped a-Si1-x¬Cx as rear side passivation (in im-mediate contact to the solar cell base) was investigated. Intrinsic films preserve their excellent passivation quality at the solar cell level, in particular no indication of inversion layer shunting comparable to the phenomenon observed in SiNx is found. The performance of its doped counterparts as passivating media strongly depends on the doping level and on the applied rear contacting scheme. However, the major benefit of doped Si-rich a Si1 xCx layers is unveiled in combination with (thin) intrinsic films. This is demonstrated by a newly developed rear passivation and contacting scheme based on intrinsic and doped a Si1 xCx films in conjunction with an adequate laser process (PassDop). The fulfilled demands of the amorphous silicon carbide system on the rear of the solar cell are threefold: very low SRVs are yielded by the intrinsic film next to the crystalline silicon interface, an overlying highly doped Si-rich a Si1 xCx layer acts as dopant source during the laser process and a C-rich layer enhances the optical con-finement at the rear of the cell. On high-efficiency n-type solar cells, this approach evidenced a very high process stability and led to efficiencies of up to 22.4 % (Voc =701 mV, FF =80.1 %). On p-type cells, the equivalent process yielded efficiencies of up to 21.6 % (Voc =683 mV, FF =80.7 %) being therefore directly comparable to passivation schemes featuring high quality thermal SiO2 and to atomic-layer deposited Al2O3 in combination with the LFC process.The use of a highly doped a Si1 xCx layer as a doping source during the solar cell fabrication can result in a thermally stable high-low passivation scheme. A simplified process sequence for n-type solar cells based on the in-situ diffusion of a n+ BSF from phosphorous doped a Si1 xCx was introduced. The diffusion of the rear high-low junc-tion takes place during the high-temperature process for the formation of the boron emitter. The rear contacting is done by a laser process equal to the one presented in the PassDop approach. The introduction of an additional C-rich film on the rear makes this approach optically superior to diffused full area contacted BSFs. In combination with a well passivated front side, this concept was found to have an efficiency potential of up to 20 %.Layer structures based on amorphous silicon carbide were also applied to the solar cell front side. The former necessarily comprise at least a two layer system. A thin Si-rich film (≈ 10 nm) next to the interface is needed for the electrical surface passivation and a C-rich film of adjusted layer thickness serves as anti-reflection coating (ARC).Si-rich a-Si1-x¬Cx films exhibit good passivation properties on planar (shiny etched) phosphorous diffused n+-surfaces (J0e ≈ 40 fA/cm2), however a strongly deteriorated performance is observed on textured (random pyramids) surfaces (J0e ≈ 400 fA/cm2). Boron diffused p+-surfaces could not be satisfactorily passivated by a-Si1-x¬Cx. (J0e ≈ 900 fA/cm2). The typical procedure for the extraction of the emitter saturation current density (J0e) from lifetime measurements was found to be unfeasible for the passivation with a Si1 xCx. This finding is ascribed to the strongly injection dependent surface recombination velocity of a Si1 xCx at low injection levels independently of the surface doping polarity.C-rich a SiyC1-y (y >0.5) layers were optimized regarding their transparency, yielding anti-reflection coating properties slightly inferior to the ones of typical ARC-SiNx. A low temperature approach for the front side passivation and front contact formation (SE-PassDop) was presented which basically consists of a laser process for opening of the silicon carbide and of subsequent front metallization by Ni/Ag or Ni/Cu plating. During the laser process, additional dopant atoms incorporated in the front passivation scheme are driven into the surface and allow for the formation of a selective emitter.First p-type solar cells featuring the PassDop rear and the SE-PassDop front side approach for the passivation and contacting of respective surfaces were fabricated with planar (shiny etched) and textured (random pyramids) front side. A strong parasitic absorption on the front side is mainly attributed to a non-optimized C-rich a SiyC1-y layer. Open-circuit voltages of up to 672 mV and 623 mV evidence a high level of surface passivation on planar emitters and a strongly deteriorated performance on textured surfaces, respectively. The feasibility of the metallization concept including the laser process is impressively demonstrated by fill factors of up to 79 %. The incor-poration of a doped, Si-rich a Si1 xCx film in the emitter passivation scheme proved to be beneficial in terms of series (contact) resistance as well as in terms of front surface recombination.While the parasitic absorption in the front a-SiCx layer stack should be strongly re-duced by application of an optimized C-rich a SiyC1-y layer and by further reduction of the thickness of the passivating/dopant containing films, it is not yet clear how to over-come the problems related to the poor electrical passivation of surfaces featuring random pyramids.As a final, practical remark, passivating a Si1 xCx is more closely related to amor-phous silicon than to SiNx – from a chemical as well as from an electrical viewpoint. However, “carbon doping” of the amorphous silicon matrix clearly exhibits a beneficial impact on the (low temperature) thermal stability of the film (< 500°C) while maintaining an excellent level of electrical properties.

Chapter 1 gives an introductory overview of the current status of photovoltaics, with focus on crystalline silicon (c-Si) based technology. An essential contribution to the reduction of electricity generation costs at the solar module production level is to be expected mainly from reduced silicon consumption by using thinner wafers and/or employing cheaper silicon feedstock. Together with sufficient light trapping, the key factor to being able to exploit the combined cost reduction potential by using thin solar grade silicon wafers is the availability of very good and industrially applicable electronic surface passivation methods. That way, material quality makes almost no difference anymore when going down to wafer thicknesses of 30 µm from the current 150-200 µm. This gives the motivation for this work.Chapter 2 gives an overview of the theoretical bases of surface passivation and antireflection coating and describes the methods and equipment used to characterize the layers created in this work.While the µPCD was applied to obtain spatially resolved lifetime maps of the entire sample, the QSSPC was subsequently used to determine absolute values of the best areas that can be compared with the literature, as QSSPC is the established standard in c-Si photovoltaics.The refractive index and thickness of the investigated dielectric films were measured by spectroscopic ellipsometry, and the chemical composition was analyzed by Fourier-Transformed Infrared Spectroscopy (FTIR) to investigate relations with the surface passivation and optical properties.Chapter 3 explains the difference between growth and deposition as methods to obtain a film on top of a substrate and gives some examples of the dielectrics and their formation techniques which are most commonly used in current c-Si solar cell technology, with focus on SiOx and SiNx from thermal oxidation and Plasma Enhanced Chemical Vapor Deposition (PECVD), respectively.Subsequently, the thin film formation technology by PECVD is described more detailed, and the particularities of the low-frequency (LF) direct-plasma PECVD reactor from Centrotherm, mainly used in this work, are outlined.In chapter 4, the results of experiments with a-SiNx:H (SiNx) are presented. SiNx was solely deposited using the LF Centrotherm system, in contrast to SiCx.Regarding deposition parameters, the influence of the gas flow ratio and wafer position in the horizontal boat position on the passivation and optical properties as deposited and after simulated contact co-firing were investigated.As an approach to potentially reduce the costs of PECVD deposition by using cheaper precursor gases, the effect of a variation of the purity grade of ammonia used for the SiNx depositions was investigated on lifetime samples as well as solar cells. Finally, the long-term stability of the encapsulated solar cells was tested by temperature variation cycling as no clear difference between the different purity grades was detectable on the cell level between ammonia purity grades N50 (UHP), N36 and N20 (industrial grade, 99% purity).Chapter 5 presents the experiments and results of a-SiCx:H (SiCx) depositions from methane and silane by both 13.56 MHz high-frequency and 40 kHz low-frequency PECVD, as well as experiments with a-SiCxNy:H (SiCxNy) deposited by low-frequency PECVD from either methane, ammonia and silane or ammonia together with an alternative silicon-containing precursor.Resulting films were characterized by QSSPC lifetime measurements and thickness and refractive index evaluated by spectroscopic ellipsometry. The low-frequency deposited samples were later also subject to FTIR measurements to investigate their structure and chemical composition.The best passivation was found for layers fabricated by the high-frequency PECVD with highest Si content, i.e. these layers were stoichiometrically close to amorphous silicon, with the passivation quality continuously decreasing with Si content. This trend changed after firing, with the highest effective lifetime being reached at a GFR of 6, whereas the passivation of the 4 ms sample almost completely degraded to 4 µs, which is in accordance with a-Si:H results, not able to withstand firing.SiCx from LF-PECVD gave lower passivation for the as-deposited layers, but the best passivation after firing is better than for the films deposited by HF-PECVD within this work, and equals the best passivation by intrinsic SiCx reported by two authors.

Solarzellen aus Silizium nutzen die im Sonnenspektrum enthaltene Energie nur unvollständig aus. Diese Verluste können durch Photonen-Management reduziert werden. Photonen-Management zielt darauf, den Wirkungsgrad von Solarzellensystemen zu erhöhen, indem das Sonnenspektrum aufgeteilt oder verändert wird, bevor das Sonnenlicht von Solarzellen absorbiert wird.In dieser Arbeit werden dazu zwei Konzepte untersucht: Fluoreszenzkonzentratoren mit photonischen Strukturen und die energetische Hochkonversion von Photonen. Mit theoretischen Modellen werden die wesentlichen Wirkmechanismen untersucht, sowie experimentell wesentliche Materialeigenschaften bestimmt und komplette Solarzellensysteme realisiert. Für beide Gebiete werden konzeptionell weiterführende Systemarchitekturen entwickelt, die Perspektiven für eine erfolgreiche Anwendung der Konzepte eröffnen.

The present work was concerned with the analysis of electrically active defects in silicon for solar cells. Analysis of such defects was performed using the two different characterization techniques deep-level transient spectroscopy and lifetime spectroscopy. The challenge of the present work was the in-depth comparison of the different measurement and evaluation techniques as well as the experimental use of these techniques for accessing the defect parameters of various impurities in silicon material.

Die vorherrschende Technologie zur Kontaktierung von Silizium-Solarzellen in der Photovoltaikindustrie ist die Dickfilmmetallisierung. Die Vorteile dieser kostengünstigen Technologie liegen im hohen Durchsatz, der geringen Zahl an Prozessschritten und der Möglichkeit, von dem Erfahrungsschatz der Mikroelektronik zu profitieren. Ein Weg zur weiteren Reduktion der Kosten pro Wattpeak ist die Steigerung des Wirkungsgrades von Industriesolarzellen. Die Hauptverluste einer solchen Solarzelle sind zum einen materialinduziert, zum anderen eng mit der Dickfilmmetallisierung verknüpft.In dieser Doktorarbeit wird deshalb die Dickfilmmetallisierung von kristallinen Silizium-Solarzellen grundlegend untersucht. Konkurrierende Prozesse während der Kontaktbildung werden getrennt voneinander analysiert. Aus den daraus gewonnenen Erkenntnissen werden Modelle für die Kontaktbildung und die elektrische Leitung in einem Dickfilmkontakt erstellt. Diese Modelle werden angewendet, um eine Prozesssequenz zu entwickeln, die es ermöglicht, moderat dotierte Emitter (ND, Oberfläche = 4×10^19 cm^-3) mit einer Standard-Silber-Dickfilmpaste zu kontaktieren. Des Weiteren werden blei- und cadmiumfreie Silber- und Aluminium-Dickfilmpasten zur Metallisierung von Solarzellen entwickelt. Optimierte Pasten, die innerhalb des europäischen Forschungsprojekts EC2Contact entwickelt wurden, führen zur Herstellung von Industriesolarzellen, die den Wirkungsgrad bleihaltiger Referenzpasten erreichen. Dies ist der bisher höchste publizierte Wirkungsgrad, der mit bleifreien Dickfilmpasten erzielt wurde. Die entwickelten Pasten werden derzeit vom Projektpartner Metalor in den Markt eingeführt.Diese Arbeit ist in drei Teile unterteilt. Der erste Teil handelt von der Industriesolarzelle im Allgemeinen. In Kapitel 1 wird dieser Typ Solarzelle und sein Herstellungsprozess vorgestellt. Im 2. Kapitel werden die Hauptverlustmechanismen in typischen Industriesolarzellen identifiziert und untersucht. Typische Charakterisierungsmethoden, die in dieser Arbeit benutzt werden, werden eingeführt und angewendet. Abschließend wird der Einfluss des Kontaktfeuerungsprozesses auf die Solarzellenleistung untersucht.Die Untersuchungen des Silberdickfilmkontaktes auf Solarzellen werden im zweiten Teil vorgestellt. In Kapitel 3 wird die Entwicklung des Linienwiderstands aufgrund von Sintern der Silberpartikel im schnellen Feuerprozess mit Hilfe einer neu entwickelten Messmethode, der in-situ Linienwiderstandsmessung, untersucht. Die Mikrostruktur des Kontaktinterfaces von Silberdickfilmkontakten zu Silizium wird mit elektronenmikroskopischen Aufnahmen und Röntgenstrahlanalysen in Kapitel 4 untersucht. Die elektrische Kontaktbildung ist Inhalt des Kapitels 5. Der Fokus liegt hier auf der Rolle des Bleioxids, welches in der Glasfritte enthalten ist. Die Glasfritte ist Teil einer typischen Dickfilmpaste. Sowohl der Einfluss der Oberflächenkonzentration des Phosphors als auch der Oberflächentextur des Siliziums wird studiert. Aus den Erkenntnissen wird ein Modell der Kontaktbildung aufgestellt. In Kapitel 6 werden die elektrischen Eigenschaften des eingebrannten Silber-Dickfilmkontakts untersucht. Mit Hilfe von grundlegenden Metall-Halbleiter Gleichungen werden die Eigenschaften des Silberdickfilmkontakts simuliert und die Ergebnisse mit experimentellen Daten verglichen. Die Kinetik der elektrischen Kontaktbildung wird mit Hilfe der neu entwickelten In-Situ Kontaktwiderstandsmessung studiert. Des Weiteren wird der hilfreiche Effekt eines Forming Gas Anneals auf den Kontaktwiderstand des Silberdickfilmkontakts zum Silizium untersucht. Die Ergebnisse werden zur erfolgreichen Herstellung einer Solarzelle mit einem moderat dotierten Emitter mit einem Serienwiderstand von 0.7 Ohmcm^2 führen.Im dritten Teil wird die Entwicklung von bleifreien Dickfilmpasten zusammengefasst. Das in Kapitel 5 entwickelte Modell der Silberdickfilmkontaktbildung wird in Kapitel 7 angewendet, um geeignete Alternativen zum Bleioxid zu finden. Schließlich werden diese Untersuchungen zur erfolgreichen Entwicklung einer blei- und cadmiumfreien Silberpaste führen. In Kapitel 8 werden die Untersuchungen und Experimente zur Entwicklung einer blei- und cadmiumfreien Aluminiumpaste vorgestellt.

Photozellen mit niedriger Bandlücke, d.h. kleiner als jene von Si, können in Tandem-Solarzellen oder in der Thermophotovoltaik (TPV) angewandt werden. Niedrige Bandlücken können durch das Einbetten von Lagen mit SiGe-Nanostrukturen in Si erzielt werden. Befindet sich solch ein Stapel in der Raumladungszone einer pin Diode, erhöht sich dadurch die Infrarot-Sensibilität der Photozelle. Die Kompatibilität zur Si-Technologie wird durch die Verwendung von kristallinen Si-Scheiben als Substrat und durch Ultrahochvakuum-Epitaxie aus der Gasphase (UHV-CVD -- Ultra High Vacuum-Chemical Vapour Deposition) gewährleistet, das eine kosteneffizientere Massenfertigung im Vergleich zu anderen Materialien mit niedriger Bandlücke, wie z.B. GaSb, verspricht. Die Gitterfehlanpassung zwischen Si und Ge von ca. 4% führt zur Stauchung der SiGe Strukturen, welche die maximale Dicke, die epitaktisch abgeschieden werden kann, limitiert. Da Absorption des Lichtes eine Voraussetzung für die photovoltaische Umwandlung in Elektrizität ist, und die abscheidbare SiGe-Dicke klein im Vergleich zur Dicke einer Si Photozelle ist, wird für eine vergleichbare Quantenausbeute ein entsprechend höherer Absorptionskoeffizient der SiGe Nanostrukturen benötigt. Die örtliche Begrenztheit der Ladungsträger (Confinement) in den SiGe Nanostrukturen bewirken eine Verbreiterung ihrer Wellenfunktion im Impulsraum, die zu einer höheren Wahrscheinlichkeit eines direkten Übergangs der Ladungsträger vom Valenz- zum Leitungsband führt. Deshalb kann erwartet werden, dass diese Übergänge, die kein Phonon benötigen, die optische Absorption signifikant erhöhen. In dieser Arbeit wird die optische Absorption von SiGe Nanostrukturen zum ersten Mal in absoluten Einheiten untersucht und die Ergebnisse mit einer Photostrommessung verglichen. Darüberhinaus werden die Gestehungskosten von Elektrizität, die durch TPV erzeugt wird, abgeschätzt.Die optische Absorption der SiGe Strukturen wurde mittels interner Vielfachreflexions-Spektroskopie gemessen, bei der das Licht den Stapel von SiGe Lagen mehr als 400 mal durchläuft. Die Totalreflexion des Lichtes bewirkt eine stehende elektromagnetische Welle. Der Absorptionskoeffizient wurde aus den experimentellen Daten unter Berücksichtigung der Geometrie und der Verteilung des elektrischen Feldes gewonnen. Das Messergebnis wurde mit einem theoretischen Modell verglichen, das die Bandstruktur des gestauchten SiGe, sowie Confinement-Effekte berücksichtigt. Der Vergleich von der Photostrom- mit der Absorptionsmessung weist darauf hin, dass einige Photonen mit einer Energie < 1 eV nicht in Photostrom umgewandelt werden. Ohne Lichteinfang wären 1000 Lagen mit SiGe Strukturen notwendig, um 1% des Lichtes mit einer Photonenenegie von 1 eV zu absorbieren. Lichteinfang und eine Verbesserung der Abscheide-Technologie ist erforderlich, um SiGe Nanostrukturen zu erhalten, die für den Einsatz in der Photozellen-Produktion geeignet sind.Es wurde eine detaillierte Kostenabschätzung für TPV-Systeme angefertigt, die auf drei verschiedenen Photozell-Technologien basieren: Si Photozellen, GaSb Photozellen und Photozellen mit niedriger Bandlücke, zu einem Preis, wie er mit Si-kompatibler Technologie zu erwarten ist. Für die Berechnung der Kosten der Elektrizität wurde eine Lebensdauer von 20 Jahren, ein jährlicher Zinssatz von 4,25% und ein jährlicher Wartungsaufwand von 1% der Investitionskosten angesetzt. Für die Bestimmung der Produktionskosten eines TPV-Systems mit einer thermischen Leistung von 12-20 kW wurden die Kosten der TPV-Einzelkomponenten zuzüglich 100 EUR/kW_{el,peak} für die Fertigung und Kleinteile angenommen. Die Kosten des GaSb basierten Systems wurde von den Kosten der Photozellen und der Annahme, dass diese 35% der Gesamtkosten betragen, berechnet. Die Berechnung wurde für vier verschiedene TPV-Szenarien durchgeführt: Ein Prototyp-System, das auf existierender Si Solarzellen Technologie basiert (Systemwirkungsgrad eta_{sys} = 1,0%), ergibt Investitionskosten von 3.000 EUR/kW_{el,peak}, ein optimiertes Si Photozellen basiertes, mit verfügbarer Technologie fertigbares System (eta_{sys} = 1,5%) ergibt Investitionskosten von 900 EUR/kW_{el,peak}, ein weiter optimiertes System, das auf zukünftiger, kostengünstiger Technologie basiert (eta_{sys} = 5%) ergibt Investitionskosten von 340 EUR/kW_{el,peak} und ein auf GaSb Photozellen basiertes System (eta_{sys} = 12,3% mit Wärmetauscher) ergibt Investitionskosten von 1.900 EUR/kW_{el,peak}. Die daraus resultierenden Elektrizitätskosten betragen zwischen 6 und 25 EURcents/kWh_{el} (inklusive 3,5 EURcents/kWh für das Gas) und wurden mit jenen einer Brennstoffzelle (31 EURcents/kWh) und eines Gasmotors (23 EURcents/kWh) verglichen.

Within this work amorphous SiC is investigated for its applicability in photovoltaic devices. The temperature stability and dopability of SiC makes this material very attractive for applications in this area. Physical basics of amorphous SiC networks and plasma processes are discussed and first measurements with FTIR of the different layer types show the complexity of the network. The special features of the plasma reactor such as high temperature deposition and two-source excitation are also discussed. Beside plasma etching I furthermore tested the etching behaviour of stoichiometric SiC in different wet chemical etching solutions. The results show that etching of SiC, especially when it is already annealed, is very difficult for both etching processes. Furthermore, stress measurements of our layers deposited and annealed at different temperatures show the change of stress from compressive (