Considering the rapidly growing energy demand in worldwide and climate deterioration caused by fossil fuel, the remarkable potential of solar energy has captured the attention of individuals and industries alike. Among different techniques, poly-Si based passivating contacts have shown great performance on solar cell application, which enabling a high efficiency of over 26%. Plasma-enhanced chemical vapor deposition (PECVD) as one of the promising technologies in a-Si contacting layer fabrication has gradually replaced the conventional LPCVD method in industry. However, the accompanied severe ion bombardment is not negligible, especially on its underlying fragile tunnelling oxide. In this work, the impact of PECVD a-Si:H contacting layer deposition on poly-Si/SiO_x passivating contact is investigated. 
PECVD radio-frequency (RF) power for the contacting a-Si layer on the underlying SiO_x is the only variable in this project, varying from 5 W to 55 W, and pinhole density acts as a bridge to help analyse the intrinsic principles. Firstly, the results of a-Si:H thin film characterization suggest that with an increasing RF power, the a-Si:H thin film is grown at a higher deposition rate and becomes porous. In addition, the pinholes in tunnel oxide are inspected by applying the concepts of “selective etching” and “pinhole magnification”. With a two-step five-point sampling method, it is shown that the effect of RF power on the pinhole density is not monotonically increasing. The highest value is found at 25 W. To explain this, a concept of “protective layer” is proposed, which is defined as a buffer layer (contacting layer) formed at the very beginning during a-Si:H deposition. It appears to be more effective when higher RF power (> 35 W) is applied. Another influence on tunnel oxide property is discussed according to the result from XPS measurement. The percentage of Si⁴⁺ species is found in the case of 25 W, corresponding to the highest pinhole density. This proves to some extent that the severe particle bombardment brought by strong power would weaken or directly break the Si-O bonds in the PECVD substrate, that is the tunnel oxide in our case.
As a result of thin film characterization, five factors contribute to pinhole formation: (i) Defects in tunnel oxide from imperfect oxidation leave potential for pinhole formation. (ii) Severe ion bombardments in PECVD deposition are allowed to weaken or break Si-O in SiO_x. (iii) Island growth of a-Si:H contacting layer makes the exposed region in tunnel oxide continue to be damaged. (iv) “Buffer layer” formation protects the substrate from ion bombardments. (v) The tensile stress applied by a-Si:H films during annealing intensifies the formation of pinholes. 
Subsequently, an unexpected result from passivation quality assessment is that higher passivation level is presented with higher pinhole density. The best passivation quality is found in the case of 25 W, with J₀ of 3.3 fA/cm² and iV_oc of 714 mV. Further, a large optimal process window for RF power adjustment is found from 25 W to 35 W, which leads to an iV_oc over 710 mV, with single side J₀ below 3.5 fA/cm². The results from specific contact resistivity indicate that it is positively correlated to the pinhole density. Eventually, the champion passivating contact with a selectivity of 14.37 in this project is expected to yield a maximum efficiency of 28.9% in an ideal c-Si solar cell.

In the past decades, much progress has been made in the field of AI, and now many different algorithms exist that reach very high accuracies. Unfortunately, many of these algorithms are quite resource intensive, which makes them unavailable on low-cost devices.
The aim of this thesis is to explore algorithms and neural network techniques suitable for implementation on FPGAs. While FPGAs provide almost complete control over all aspects of design, allowing for the development of high-performance systems, they have not gained widespread popularity in neural network development due to their limited accessibility compared to computers and microcontrollers.
In the thesis, an inference-only 8-bit quantized neural net is designed, implemented and deployed on the Diligent Zedboard, and the performance is compared to similar networks on other devices. The thesis then focuses on two learning algorithms: Forward-Forward learning and Hebbian learning. It is shown how Forward-Forward can be seen as a way to apply Hebbian learning rules, and a simplified algorithm is proposed for use in a quantized system and implemented on an FPGA.
Although the performance of the network is quite low, reaching only 90.5% on the MNIST dataset and 74.2% on Fashion MNIST, the results are promising enough to give ground for further research and show that even very simplified versions of the Forward-Forward algorithm are capable of learning.
Moreover, it demonstrates that the Forward-Forward algorithm is suitable for FPGA implementation.
Both implementations show that the processing speed of the FPGA implementations is much faster than that of similar network implementations on other devices.

Silicon solar cells account for about 95% of the total photovoltaic market share. Poly-Si passivating contacts are promising techniques enabling high performance c-Si solar cells with conversion efficiency over 26.0%. The highly absorptive nature of poly-Si materials makes the transparent passivating contacts attractive, such as poly-Si(O_x). However, the research of p⁺ poly-Si(O_x) passivating contacts on a textured surface with an in-situ doping nature is still missing. Therefore, in this thesis, the optimization of p⁺ poly-Si(O_x) carrier selective passivating contacts on double side textured wafers are given and the application in solar cells is demonstrated.
Firstly, the influences of different interfacial tunnelling oxides fabrication methods, nitric acid oxidation of silicon (NAOS-SiO_x), plasma assisted N₂O oxidation (PANO-SiO_x), and thermal oxidation (t-SiO_x) on the passivation of p⁺ poly-Si(O_x) passivating contacts are explored. It is found that Si⁴⁺ stoichiometry in the tunnelling oxide layer is an indicator for its quality. There is a positive correlation between Si⁴⁺ and SiO_x density. The t-SiO_x can be denser with higher Si⁴⁺ compared to its counterparts NAOS-SiO_x and PANO-SiO_x. And fewer boron dopants in-diffuse phenomenon can be observed in the t-SiO_x samples. Then, different intrinsic layer deposition approaches are explored. The intrinsic layer deposited by LPCVD results in higher iV_oc compared to PECVD counterpart. The enhanced iV_oc is given by suppressing the blistering which is caused by hydrogen accumulation at the interface between intrinsic layer and SiO_x. It is assumed that less hydrogen accumulation exists in intrinsic layer deposited by LPCVD. Next, p⁺ doping layer thickness is changed from 0 nm to 200 nm to observe its effect on the passivation quality. The optimum p⁺ doping layer thickness is found to be 100 nm with the highest iV_oc. After that, the hydrogenation process is introduced to enhance chemical passivation by coating SiN_x:H and performing forming gas annealing. The highest iV_oc with the standard hydrogenation process is 674 mV. In order to improve the hydrogen level of p⁺ poly-Si(O_x) passivating contacts, AlO_x:H inserted layer is used for the hydrogen reservoir together with SiN_x:H. It results in an improved iV_oc of 685 mV. It is assumed that the hydrogen in AlO_x:H diffuses into c-Si/p⁺ poly-Si(O_x) interface and enhances the chemical passivation.

Besides, metallization methods of p⁺ poly-Si(O_x) passivity contacts are also studied. There are two approaches to complete the metallization process. Firstly, with thin p⁺ poly-Si(O_x) passivating contacts, TCO is required to provide with the lateral and vertical carrier transport as respect to the carrier collection. However, it commonly brings with the TCO introduced sputtering damage. The iV_oc losses are 70-90 mV when TCO is sputtered on p⁺ poly-Si(O_x) passivating contacts. When the thickness of p⁺ doping layer is over 50 nm, sufficient lateral conductivity can be provided with thick p⁺ poly-Si(O_x) passivating contacts. Therefore, it can directly contact with metal which is the second method of metallization. However, when utilizing this metallization method, the metal induced recombinations need to be taken into consideration when contacting with p⁺ poly-Si(O_x) passivating contacts. Thus, the plot of J_o,total along with different metal fractions is fitted to extract J_o,metal of p⁺ poly-Si(O_x) passivating contacts with 100 nm p⁺ doping layer. When contacting with evaporated aluminum, the measured J_o,metal is around 91 fA/cm². In addition, after calculating, there is 24 mV iV_oc loss when p⁺ poly-Si(O_x) passivating contacts contacting with metal. It is smaller than the loss induced by TCO sputtering. Therefore, the thick p⁺ poly-Si(O_x) passivating contacts with 100 nm p⁺ doping layer directly contacting with metal is used as metallization method for the application in c-Si solar cell. In addition, after linear fitting and calculating, ρ_c between p⁺ poly-Si(O_x) passivating contacts with 100 nm boron doped layer and c-Si is about 23 mΩ· cm².

Finally, p⁺ poly-Si(O_x) passivating contact is applied in c-Si solar cells together with n⁺ poly-Si(O_x) passivating contact as front surface field. The poly-poly solar cell of the highest quality has the following electrical performance: V_oc is 648 mV, J_sc is 35.9 mA/cm², FF is 73.1% and η is 17.0%. A roadmap to realize 22% is given by addressing the bottlenecks of poly-Si(O_x) passivating contacts based c-Si solar cells.

Interdigitated back-contacted solar cell (IBC) is a successful high-efficiency solar cell concept. Without a metal grid at the front side, the metal shading loss is eliminated. However, the fabrication process of an IBC cell is much more complicated than that of a FBC cell. Within the PVMD group, two poly-Si carrier-selective passivating contacts (CSPCs) IBC cell fabrication methods were developed, namely Self-aligned and Etch-back method. These two methods require respectively to pattern the IBC rear side twice and three times to define the emitter and BSF area. In this project, a novel poly-Si/SHJ hybrid IBC solar cell design using tunneling recombination junction (TRJ) is proposed, which only requires one pattern step to define the emitter and BSF area, thus, significantly simplifies the flowcharts of IBC cells.
The main objective of this thesis is to demonstrate the novel poly-Si/SHJ hybrid TRJ IBC cell concept by fabricating such high-efficiency hybrid IBC cells. With the proposed hybrid IBC cell design, the BSF layers are deposited on the full rear side after the emitter patterning. Thus, a TRJ is introduced at the emitter ((p+)poly-Si). The carrier tunneling efficiency across the TRJ layers should be guaranteed. The BSF layers passivation quality is also crucial for hybrid IBC cell performance, and the shunting due to the full area deposited BSF layers should be limited as well.
The TRJ proof-of-concept was firstly demonstrated in FBC cells due to their easier fabrication processes. And different materials combinations of (i)a-Si:H, (n)a-Si:H and (n)nc-Si:H were used to form TRJ with (p+)poly-SiOx. Firstly, it was found that the existence of (i)a-Si:H is detrimental to the device performance, especially for the cell FF. Secondly, for cells without (i)a-Si:H layer, the cell performance was improved by replacing (n)a-Si:H with more conductive and low activation energy (n)nc-Si:H layer. At last, the TRJ with dual-n-layer, (p+)poly-SiOx/(n)a-Si:H/(n)nc-Si:H, was found to be most promising. And the cell FF decreases with (n)a-Si:H layer thickness. However, instead of only depositing (n)nc-Si:H, the (n)a-Si:H was kept for its better passivation ability than (n)nc-Si:H, as it is directly deposited on the c-Si surface at BSF in hybrid IBC cells.
Then the hybrid design with dual-n-layer was demonstrated and optimized regarding the passivation quality, TRJ efficiency and the Rshunt in IBC cells. We firstly demonstrated that poly-Si delivers better performance than poly-SiOx due to its lower resistivity. With 18 nm(at textured BSF)(n)n-Si:H, the (n)a-Si:H layer thickness optimizes at 3 nm in poly-Si/SHJ hybrid cells. The pitch width was also found to have an influence on cell external parameters as the number of fingers decreases with pitch width. The cell FF increases and the Jsc decreases with pitch widening. The best cell obtained in this project has 3/18 nm (n)a-Si:H/(n)nc-Si:H and a medium pitch width (650 μm). It has a Voc of 665 mV, a Jsc of 39.36 mA/cm2, a FF of 74.33% and an efficiency of 19.45%.

As the efficiency of silicon (Si) photovoltaic (PV) moves ever closer to the theoretical limit, 2-junction PV becomes increasingly interesting. Since PV cells are tested under standard test conditions (STC), but real world working conditions differ, it is interesting to see how 2-junction PV performs worldwide in different climates. Worldwide spectra where simulated in SMARTS using data from NASA’s Global Land Data Assimilation System (GLDAS), Clouds and the Earth’s Radiant Energy System (CERES) and Socioeconomic Data and Applications Center (SEDAC) and from the Joint Institute for the Study of the Atmosphere and Ocean (JISAO). SMARTS provides only clear-sky spectra, to account for cloudiness, the BRL model is used. The top absorber is 1.72 eV Perovskite and the bottom absorber is 1.12 eV Si. Because Si is an indirect bandgap material, a limiting Efficiency model by Richter et al. is used that takes both Auger and radiative recombinations into account. Because this model relies on variables and constants only given for Si, and the top absorber is a direct bandgap material, the detailed limit model by Shockley and Queisser is used to calculate the performance of Perovskite. The distribution of high and low local yearly average irradiance is overall realistic, except for the north of Africa and the Middle East. Here aerosol optical depth (AOD) values are elevated. The elevated AOD can be explained by dust events, still these areas should be among the ones with the highest irradiance. The AOD effects the blue side of the spectrum mostly. All spectra where normalised using CERES’ irradiance data. The Richter model is thickness dependent. Thus the optimal thickness of Si is determined to find the optimal efficiency. To distinguish between the effects of the top layer and local climate effects on the thickness, firstly the optimal thickness of single junction Si is calculated and analysed. The optimal thickness ranges from 70 to 870 µm globally. The average is 205 µm and the irradiance weighted average is 165 µm. For double junction Perovskite - Si current matching was used to find the optimum. The thickness of Si ranges from 10 to 4500 µm. But only nine locations (of 597) have a thickness above 500 µm. The average thickness is 77 µm and the irradiance weighted average is 81 µm. Locations with a high optimal thickness had a blue-er spectrum than the locations with a low thickness. Using the optimal thickness of single junction Si and current matching, the optimal top-bandgap was determined around the world. The optimal bandgap ranges between 1.55 and 1.76 eV, the average is 1.655 eV and the irradiance weighted average is 1.663 eV. The relation between the top bandgap and optimal Si thickness was determined under STC. High bandgaps had the lowest optimal thicknesses and the thickness increased with decreasing top-bandgap. Alterations in spectrum and temperature where applied as well to find the effect on that relation. Decreasing temperature and airmass both resulted in an increase in optimal thickness per bandgap vise versa. Changing the spectrum has a greater effect than the temperature change.

Solar cells can play a key role in the transition towards a sustainable future. This transition is one of the major challenges our society faces during the coming decades. Development of high-efficiency photovoltaic solutions at reasonable costs will help accelerating the transformation of our energy system. In this respect, perovskite solar cells are very promising due to their outstanding opto-electronical properties and low-cost fabrication. Obtaining a complete understanding of the device physics and charge transfer mechanisms inside perovskites is crucial for further device improvements. This thesis focuses on the notorious hysteresis in the current-voltage characteristics of perovskite solar cells. So far, this remarkable phenomenon has usually been explained using ion migration, despite the lack of clear experimental evidence. We implement a simulation platform of perovskite solar cells to analyse the charge transfer mechanisms among energy states including those with energy within the forbidden bandgap. We evaluate transient behaviour and identify the limiting physical mechanisms. To explain anomalous hysteresis in perovskite solar cells we use a novel approach in which charge accumulates near the material interfaces due to defects with relatively low capture cross-sections. Defects in lead halide perovskites create shallow sub-gap energy states, that act as charge carrier traps. Near the interfaces, this leads to accumulation of trapped charge carriers, effectively screening the electric field inside the perovskite layer. This reduces the device performance. A slow release of trapped charge due to low capture cross-sections results in hysteresis in the current-voltage curve at commonly used scan rates. TCAD Sentaurus is used as a platform to simulate J-V scans of a planar non-inverted architecture based on the archetypal perovskite MAPbI3, with TiO2 as electron transport layer and spiro-OMeTAD as hole transport layer. This thesis presents a systematic study of different trap distributions, both in the spatial and energetic domain. The capture cross-sections, densities, energy levels and locations of traps are varied and also the effect of scan rate is analysed. This work analyses both tail state defects and deep defects, based on reported values in literature. It is found that defects near the ETL/perovskite interface potentially cause anomalous hysteresis in the current-voltage curve. These defects have their transition energy around 0.25 eV and are possibly attributed to iodine interstitials.

Silicon heterojunction solar cells employing transition metal oxides as carrier selective contact are of particular interest due to the potential of reducing parasitic absorption while featuring optimal electrical properties. Recently, a record efficiency of 23.5% was achieved by employing molybdenum oxide (MoOx) as carrier selective contact. MoOx exhibits advantageous properties with respect to the p-doped standard amorphous silicon contacts due to its lower parasitic absorption and better thermal stability. However, achieving an efficient carrier collection is challenging and not well understood yet. In this work, transport of charge is studied from drift diffusion and atomistic approach by means of numerical simulations. Two different state-of-art computational tools are employed: TCAD Sentaurus for drift diffusion simulations, and VASP for ab initio simulations. Through drift diffusion simulations, the contact formation of molybdenum oxide as carrier selective contact is consistently explored including quantum confinement and transport based in mid-gap energy states. The work function of MoOx is shown to be the core for an efficient charge collection. Thanks to experimental results, it is revealed relevant phenomenon at MoOx/intrinsic amorphous silicon (i-a-Si:H) interface which includes silicon oxide formation and charge accumulated. Therefore, a special focus at interface is here presented, in order to study the inner physics of the detrimental effects and how to avoid them. Altogether, drift diffusion simulations reveal that MoOx thickness is an essential parameter because it strongly determines the work function and hence the efficiency of the solar cell. All the knowledge acquired is used to provide guidelines on the fabrication of these type of solar cells.
Looking at MoOx/a-Si:H interface, ab initio simulations are employed to study interface properties. Accordingly, such interface is analysed using both materials in their crystalline matrix. It is demonstrated that oxygen deficiency tunes the MoOx work function, a statement which is key for the proper contact formation and consistent with drift diffusion results. Finally, the charge arrangement at interface reveals the creation of an interface dipole together with silicon dioxide interlayer which is coherent with drift diffusion simulations analysis.

Within the coming years it is expected that PV installations could be established on every possible surface and terrain, within the urban environment. Due to the complex morphology of buildings in urban environments,systems will likely be more susceptible to partial shading compared to other conventional PV systems. In particular bifacial PV modules are highly susceptible to shading since they rely on irradiance on both the front and the back surface. To better estimate the irradiance received by such systems, particularly in urban environments, much research is conducted to develop fast and accurate simulation tools. The research described in this report investigated the different simulation frameworks that have been developed for modeling bi-facial PV systems, and develop a new simulation framework capable of simulating irradiance within urban environments. Available models capable of simulating rear irradiance differ in input variables considered,simulation time required and accuracy of predictions. Empirical models, for example, can result in inaccurate predictions since not all variables affecting rear irradiance are considered. Only the 3D view factor, back-ward and forward ray tracing methods fulfill the requirements to simulate the backside irradiance/irradiation when the aim is to perform simulations in more complex urban environments. Out of the existing simulation models tested, the backward ray tracing model performed within the Radiance software proved to be the fastest simulation tool for modeling yearly irradiation or a single irradiance measurement incident under free horizon conditions. Time of use (TOU) simulations, however, results in longer simulation time for backward ray tracing performed in Radiance, since a ray tracing simulation at each time instance is required. To calculate the irradiance a receiving surface receives in an urban environment a method is worked out based on view factors and ray casting. Within the software Rhinoceros a CAD design of the surrounding environment is created, while the plug-in Grasshopper is used for the ray casting and mathematical calculations. Through a series of sanity checks, it was determined that the developed methods are reliable for calculating irradiance/irradiation when the aim is to perform simulations in more complex urban environments. It was also identified how different sky models can result in large differences in irradiance simulated. Sky models such as Isotropic, Hay and Davis or simplified Perez underestimates the irradiance, when receiving surfaces are tilted in comparison with the Perez luminance distributed sky model. The model was validated using monitoring station measurements and compared with simulations performed with other ray tracing models. DHI, DNI measurements obtained from the Solys 2 are used for replicating the irradiance measured at the dual-axis and single-axis tracker. Since their orientation was fixed throughout the measurement period they are referred to as POA 1 and POA 2 respectively. With POA 1 and 2 having a 90 and 30 degree tilt respectfully and an azimuth of 67 and 180 degree respectfully. Making POA 1 the front/back side of a typical (bifacial) east-west configuration and POA 2 the front of a tilted (bifacial) module. Two measurement days are considered, a fully overcast day and a clear sunny day. For the overcast day, an relative RMSE value of 23.09% and 11.7% are achieved for POA 1 and 2. For both cases the irradiation was mostly underestimated, considering the negative MBE and positive MAE. On the clear sunny day, relative RMSE values of 20.35% and 5.26% are achieved for POA 1 and 2. The model again mostly underestimates on the sunny day. A simple DHI correction with the factor 1 divided over the SVF at the Solys 2 is performed in order to investigate the impact of potentially corrupted DHI measurements on the irradiance simulations. With relative RMSE value of 19.96% and 5.2% recorded for POA 1 and 2 on the overcast day. While on the sunny day relative RMSE values of 19.38% and 3.77% are achieved for POA 1 and 2 respectively. Only slightly changing the model predictions. When compared to ray tracing models the model performs slightly better than the forward ray tracing model. A possible reason why the forward ray tracing was underestimated was proposed to be due too the small aim area used. However increasing the aim area would require a larger number of rays to maintain the same accuracy which results in longer simulations.Possible improvements to the model could be, adding reflected irradiance term on reflecting surfaces and the Perez luminance distribution sky model.

As solar energy finds its way into unexpected applications such as buildings or objects, the traditional rectangle-shaped module becomes inadequate. Although traditional modules have been optimized to deliver a large amount of power, they cannot be fitted easily into a façade or a wearable. This is why, new methods for integrating solar power into buildings and products have to be found. These methods should be attractive to designers and architects and must also allow solar energy to reach common people.
The fabrication of mini-modules can help with this task as they can be fabricated with the already mature technology of crystalline silicon solar cells. They can be made lightweight and in different figures.
This work aims to investigate ways to facilitate the fabrication of modules for use in Building Integrated Photovoltaics or Product Integrating Photovoltaics by means of studying the impact of cutting solar cells into different figures and sizes. To prove the suitability of tools of full-sized modules into mini modules, a Cell-To-Module analysis is performed.
The results show that crystalline silicon can be effectively used for making modules in reduced sizes tackling the full integration of solar cells into objects.

Global energy demand is on the rise. According to the International Energy Agency (IEA), in 2018, the building sector alone constituted 28 % of all energy-related CO2 emissions. In an effort to reduce the carbon footprint of the building sector, the European Union has decreed that every new building from 2021 must be a nearly zero energy building where the low energy for such buildings must come from renewable sources. While this policy works well for new buildings, there needs to be also a solution developed for existing buildings or monumental buildings to be able to easily harness renewable energy, without the need for any major and costly renovation. The concept of a flexible cloth-based solar curtain as a plug and play solution for existing buildings has been explored in the current research, with particular focus on a bifacial (two-sided) design, to assess the significance of the indoor reflected daylight and artificial lights on the rear side power output of such a semi-transparent solar curtain. An optical model representing a basic semi-transparent curtain comprising 70 % cloth of roughly 62 % transparency (based on availability), and 30 % of bifacial solar photovoltaic (PV) minimodules, is prepared as a base case. To validate the optical model, it is replicated on a smaller scale and tested using an identical experimental setup in a controlled environment using a solar simulator. Four cases are prepared for this: the curtain in an empty white-walled room, the curtain in the same room but with some furniture, the curtain in a shorter room with a white rear wall and the curtain in a shorter room with a black rear wall. An analysis of simulation and experimental results shows near likeness between the two with an error margin of 3.3 % for the base case to 10.6 % for the black rear wall case. Slight non uniformity of optical property inputs used in the simulation model and an overestimation of experimentally measured power due to a relatively lower irradiance seen by the reference cell are attributed to this difference. After successful validation of the optical layout, the model is made to scale in simulations to forma PV curtain module of 538.44 Wp DC installed capacity. Using a north-east facing room as reference, an annual simulation is found to give a total DC energy yield of 188.45 kWh or 323 kWh/kWp with a bifacial gain of 8.64 %. 10 hours daily of artificial lighting inside the office is seen to increase the annual DC energy yield by 1.33 %. With an assumption that the curtain remains closed throughout sunshine hours (since it allows diffuse light to enter through its porous cloth), the power output of the curtain is seen to meet modest DC annual load profiles of 57.76 kWh in an office room for 95 % of the year with a 145 Wh lithium ion battery bank and of 48.86 kWh in a residential room for 92%of the year, but with a much higher 290 Wh battery capacity, due to a greater mismatch between generation and demand. A sensitivity analysis for room orientation shows that for a more optimal azimuth of 202.5° (180° being south), the same curtain gives 82%more annual DC yield compared to its current orientation, but at a much lower bifacial gain of 7.8 %. When analyzed for performance in different locations across the world, the curtain intuitively gives the best yield for locations in the southern hemisphere due to its orientation towards the north east. Bifacial gain is nevertheless, most significant for temperate locations where low average annual insolation makes the gain due to bifacial more significant. In cases where the room depth is reduced by half, yield is seen to improve by 2.1 %. Halving the amount of PV in the curtain, improves the bifacial gain by nearly 0.6 % but reduces the annual yield by 49 %. Conversely, doubling the PV in the curtain, results in a dip in bifacial gain by about 3.6 %, but an increase in annual DC power by 87 %. Although this is attractive, such a case must be treated with caution since the likelihood of leaving this curtain (that is now more opaque than semi-transparent) open during sunshine hours is higher. This can result in a lower power output than the base case as well as worse aesthetics. Finally, it is concluded that a bifacial curtain is a beneficial for ‘solarizing’ existing buildings with minimal renovation. The bifacial aspect is best exploited for rooms with sub-optimal azimuths. Nevertheless, to make the most of the PV curtain, it is advised that itmust be placed in shorter rooms with large windows and ideally facing the most optimal direction.

Whilst research in solar cell materials lead to higher efficiencies every year, new difficulties arise with every breakthrough to get an even higher efficiency. Therefore, every component of a PV system should be examined, as well as how their behaviour under operation affects performance to improve the overall PV yield. In particular the temperature of a module can have a negative impact on the power output. The drop in power output is a consequence of a negative thermal coefficient that results in a decrease of open circuit voltage with increasing temperature. For silicon solar cells, efficiency drops of 0.4-0.65%/°C have been reported in literature. Moreover, daily repetition of temperature cycles can cause mechanical degradation, thereby decreasing the lifetime of PV modules. In this thesis, phase change materials (PCM) have been studied as a method to passively reduce the operating temperature of PV modules. This is based on the ability of materials to stay at a relatively stable temperature during a phase change. By placing a PCM at the back of a PV module, the temperature difference between the module and the melting PCM causes a thermal gradient, resulting in conduction of heat away from the
module. In order to find the optimal properties of a PCM, a thermal model was first developed in COMSOL Multiphysics and benchmarked with field measurements from literature. Simulations for Rotterdam, the Netherlands, revealed that an optimized PCM could increase the yearly electrical yield by 1.23% for a rack-mounted module, or 3.52% for a roof-mounted module. Furthermore, measurements were performed with commercially available PCMs under a Large Area Solar Simulator (LASS). These were able to reduce the average module temperature by 30-36°C, albeit under heavy infrared radiation coming from the simulator.

Interdigitated back contact silicon heterojunction (IBC-SHJ) solar cells have demonstrated 26.6\% world record efficiency by combining outstanding passivation of Si thin film layers with the absence of front shading contact in the crystalline silicon (c-Si) absorber bulk. Furthermore, this type of solar cell merges advantages of c-Si with Si thin film technology that allows for adjustment and tuning of bandgap and Fermi energy in deposited layers, and thus, modifying material properties that also impact the carrier collection. As heterointerfaces tailor to band offset and potential barriers for collecting carriers, the transport is described by thermionic emission and tunneling mechanisms. Moreover, Si thin film layers typically exhibit a defective matrix with characteristic traps and charge distribution that affect solar cell external parameters. Such phenomena involves the so called trap assisted tunneling (TAT) together with trap recombination mechanisms in a complex physical system.

In this work, the effect of TAT mechanisms on IBC-SHJ is studied by means of numerical simulations based on TCAD Sentaurus. Firstly, the TAT model is implemented as non-local process. It was found that TAT exhibits a negligible effect on electron collection, but dominates transport mechanisms in case of hole contact. Such an effect depends on band alignment at the doped layer/transparent conductive oxide (TCO) interface, where band-to-band tunneling in combination with TAT describe the transport of holes. In particular, TCO carrier concentration in combination with activation energy of deposited layer allows the collection by either TAT or band-to-band tunneling. In more detail, the density of traps described typically as dangling bond improves the transport of collecting carriers. Thus, the collection of carriers is evaluated in terms of energy and trap concentration, demonstrating that traps with an energy level of 0.5 to 0.7eV over the valence band enhances the FF, and thus, the solar cell performance.

Finally, the complete model is calibrated by comparing measured with simulated external parameters from a reference solar cell with and without TAT mechanisms. From this analysis, it is demonstrated that solar cell external parameters deploys more consistent values for Shockley-Read-Hall (SRH) recombination associated to bulk lifetime and surface recombination velocity using TAT model.

High-performance perovskite solar cells (PSCs) have attracted great attention from researchers around the world. Rapid researches and developments of PSCs are shown by the increase of its efficiency from 3.8% in 2009 to a new world record of 22.4% in 2017. Hysteresis phenomenon in current density voltage (J-V) is one of the challenges occurring in some specific PSCs that should be overcome to continue the improvement of PSCs. This phenomenon occurs when a voltage is swept with different scan directions from negative to positive value (forward) and positive to negative value (backward). Hysteresis effect results in a different maximum power point on the J-V characteristic leading to under- or overestimation of PSCs efficiency.

Furthermore, when the applied voltage is abruptly changed, a transient current density response is introduced implying a capacitive behaviour. Due to this behaviour, there is an indication that PSCs cannot be represented by the conventional equivalent circuit. Thus, the purpose of this project is to investigate hysteresis phenomenon in PSC by electrical modelling. In this project, a PSC sample was fabricated by Solliance Solar Research. Time-resolved J-V measurement was done to obtain more insight of J(t) as a function of applied voltage. The hysteresis phenomenon was analyzed in different voltage scan direction and various scan rates. Simulation of band diagram in dark condition was done to understand working principle of PSC device. Two predicted equivalent circuit of the cell were derived from the simulated band diagram. These equivalent circuit models considered the charge accumulation at the bulk of perovskite and at the interface between charge transport layers and perovskite. Furthermore, two additional equivalent circuit models were proposed to represent the hysteresis effect. J(t) curve fitting of measurement results and simulation was employed to verify the equivalent circuit. This insight might help to get a better understanding of hysteresis effect in PSCs.

External Quantum Efficiency (EQE) is an important performance measure parameter of a solar cell. EQE characterization gives deeper insights into the opto-electrical properties, current generation and recombination mechanisms in solar cells. Conventional EQE characterization of solar cells are done using a monochromator and lock-in amplifier. These measurements take a few minutes to completely resolve the spectral response of a solar cell in the visible and near infrared (VIS/NIR) region of the spectrum. Having faster EQE measurement setups will aid in faster solar cell characterization both in research and industrial production environments.

In this work a promising fast EQE measurement prototype called Fast Optical Measurement System (FOMS) developed by Delft Spectral Technologies (DST) is investigated. The FOMS prototype currently can operate only with a modulation scheme (a sequence of images) based on Michelson interferometer. The FOMS prototype hence cannot test other more suitable modulation schemes. To overcome this limitation a model to mimic the working of the FOMS prototype was developed by characterizing the existing setup which involved spectral distribution and intensity variations measurements. Various sub-functions were developed based on the characterization results and they were used to develop the model that could predict the spectrum on each pixel of the image loaded on the system. The developed model takes images corresponding to different modulation schemes as input and predicts the current output for each image. The model was finally validated by generating a new modulation scheme based on monochromatic EQE system. The percentage error in the current generated by the model and the prototype was around 3\% which is within permissible limits.

In the future the developed model can serve as a tool to quickly test new modulation schemes that can make the FOMS prototype faster and more accurate. Further research with respect to developing images based on more suitable modulation schemes can help avoid the mathematical complexities in the Michelson interferometer based modulation scheme. Developing appropriate post-processing steps to complement the modulation schemes will help in obtaining the EQE characteristics of the device under test from the current predicted by the model.

The electrical output of a solar panel can be optimised in several ways, through the inclusion of bypass diodes (BPDs) or to have a maximum power point (MPP) tracker or DC optimiser to ensure that the panel is always operating at its global MPP. However most of these systems have to be implemented externally to the panel. A new system, the active switching circuit was proposed as a means to actively sense shading and switch between three wiring configurations drawing inspiration from matrix converter theory. The active switching circuit prototype would operate for 16 solar cells arranged in a 4 x 4-cell solar panel and be able to change the wiring configuration of these sixteen cells using switches. The base unit of the solar panel is a substring composed of four solar cells in series and the circuit would switch between three wiring configurations. The shading would be measured through four current sense resistors and amplifiers, with one of these pairs in each substring and a differential operational amplifier system to measure the voltage across the strings. The current and voltage measurement helps determine the power output of the panel. A Simulink model of the system was first created and a panel with 4 x 4 solar cell arrangement ordered and then measurements made with shading to verify if the shading model was accurate. From these results a Simulink model for the active switching circuit was designed to test the theory of its functionality and if such a circuit could be controlled to switch between the three wiring configurations. The circuit was successfully built and was able to operate in its intended way.

Thin-film silicon solar cells make use of relatively thin layers of active material compared to wafer based solar cells. The main advantage of thin absorber layers is the possibility to fabricate flexible solar cells. However, due to limited absorber layer thickness and weak absorption coefficients at long wavelengths, light management techniques need to be implemented in order to increase the
device performance. Advanced substrate texturisation is one of the promising light management techniques that have drawn much attention recently. In state-of-the-art devices, randomly textured interfaces are often used to increase the optical performance. However, the introduction of periodic diffraction gratings resulted in world record conversion efficiencies for hydrogenated amorphous/nanocrystalline silicon (a-Si:H/nc-Si:H) tandem solar cells.

A different method of advanced texturisation is introducing modulated surface textures (MST). These MST structures combine various surface morphologies (random and/or periodic) to reach higher levels of light scattering over a broader wavelength range. Modulated surface textures based on large periodic gratings and small random textures have been successfully employed in single junction
nc-Si:H solar cells, resulting in a world record efficiency of 11.8 %. An MST structure based on large random features and small periodic gratings, however, has not been reported yet.

The aim of this work is to investigate the possibility of fabricating MST structures based on micro-textured glass with superimposed periodic gratings, and to obtain a functioning n-i-p nc-Si:H solar cell, based on an MST substrate. An optical and morphological analysis of randomly textured substrates, periodic gratings, and modulated surface textures was carried out. It was found that MST substrates can be fabricated using ITO induced, wet-etched glass substrates, and a photolithography process. MST structures resulted in light scattering into greater, less distinct angles, when compared to either one of the individual surface morphologies. Therefore, these structures show promising light scattering behaviour for application in nc-Si:H solar cells.

Hydrogenated nanocrystalline silicon solar cells were fabricated on randomly textured glass substrates and MST structures. Two-dimensional (2D) periodic gratings, with square and hexagonal lattice structures, were superimposed on randomly textured glass to obtain MST substrates. nc-Si:H solar cells based on hexagonal MST structures seemed to outperform their square lattice counterparts. A functioning device with an active area efficiency of 6.46 % and short-circuit current density (Jsc) of 19.74 mA/cm2 was fabricated. This work reports the first functioning nc-Si:H solar cell, based on a periodic-random modulated surface texture substrate. The best performing solar cell on randomly textured glass, however, exceeded this performance with an active area efficiency of 7.67 % and JSC of 23.2 mA/cm2. It can be concluded that functioning nc-Si:H solar cells can be fabricated on 2D MST substrates. Further research focused on the morphological optimisation of MST substrates could help to prevent defective material growth, to improve the electrical properties of future devices.

Photovoltaic (PV) systems have provided the world with a renewable and environment friendly energy solution. However, PV systems face various loss mechanisms at the module level as well as power electronics level which need to be addressed for further growth of the technology. Power generation from PV modules suffer from heat losses that are a function of the temperature coefficient of maximum power (γ) of the PV module. Another loss mechanism is the shading of PV module where the power output of a module and ultimately the system can be severely compromised. Although various approaches such as bypass diodes and module level power electronics have been employed to minimize the effects of shading on a PV module, so far, the performance of a PV module under shading had only been addressed vaguely and was generally described qualitatively. Therefore, a parameter called Shading Tolerability (ST) was recently developed to quantify the behavior of PV modules under all kinds of shade. However, the observations pertaining to Shading Tolerability were only carried out under Standard Test Conditions (STC). Hence, this thesis aims to investigate whether ST is an innate property of a PV module by formulating a correlation, if there’s any, between ST and ambient temperature, classify various PV technologies based on ST and γ and develop a selection map so that an optimal selection of PV modules can be made for the design of PV systems simply by knowing the specifications of the module along with the meteorological conditions of the installation location. Furthermore, different interpolation techniques were evaluated to achieve a comprehensive picture of the meteorological conditions of The Netherlands to determine the most optimal PV technology for The Netherlands using the PV selection map. Finally, Performance Ratios of various existing PV systems were calculated using suitable irradiation decomposition and transposition models as a proof for the validity of the selection map. The obtained results showed that PV systems using CIGS technologies exhibited higher performance ratios on average as was suggested by the selection map.

High efficiency silicon heterojunction (SHJ) solar cells have already reached more than 26% (Yoshikawa et al., 2017) efficiency when tunnel oxide passivated contacts solar cells just broke the 26% barrier at the beginning of February 2018 (ISFH, 2018). For such devices, major losses occur at both the front and rear contacts where parasitic recombination can be very high. Carrier-selective contacts use a special design in order to build a barrier that would block one specific charge carrier and let the other one go through. A Hybrid solar cell is a combination between a heterojunction solar cell and a TOPCon device, featuring then carrier-selective contacts at both sides. In this thesis, a p-type TOPCon structure is implemented at the rear when the front contact is made of n-type amorphous silicon (a-Si:H).
Intrinsic amorphous silicon (a-Si:H) used as front passivation layer, is deposited on top of crystalline silicon and requires an interface with as few defects as possible to minimize the parasitic recombination velocity. A new pretreatment method studied in this thesis involves the growth of a silicon oxide (SiO2) on a crystalline silicon substrate, that will allow to get rid of most of the superficial defects after etching and before a-Si:H deposition. Lifetimes of up to 6 ms and saturation current density (J0) as low as 14 fA/cm2 can be reached with a 200 nm thick oxide.
As a-Si:H presents a very low lateral conductivity, a transparent conductive oxide (TCO) is needed to transport the charge carriers towards the front metal contacts. The resistivity of such a material should be as low as possible. While increasing the deposition temperature of Indium Tin Oxide (ITO), it has been possible to decrease the resistivity up to 2.5 ·10-4 Ω.cm at a temperature of 130°C, without reducing the optical properties of such a layer.
Finally, manufacturing defects are often introduced during the fabrication process, leading to some shunt losses (low shunt resistance). As we are fabricating several solar cells per wafer, it is most of the time necessary to cut them to get rid of the shunt before doing the measurements. This sensitive step could be avoided with a better isolation of each solar cell. Patterned ITO and metal have been developed in this thesis, allowing to reduce the shunt power losses in the range of 1-2% without the need to cut the cells.

A quick-scan algorithm has been developed in order to evaluate rooftop PV potential in the Netherlands. Both its panel fitting and yield prediction functions have been validated with existing systems monitored by Solar Monkey. The calculation times of different parts of the algorithm were measured and decreased while keeping the accuracy of the algorithm in the desired range.

First, a new approach to determine the roof segment orientation was introduced, using both the normal vector and the longest side of the roof segment polygon. A visual inspection was carried out, in which recent aerial images were compared to 3D roof segments that were provided by Readaar. For 145 roofs, the roof segments on which PV was placed were manually selected, such that they could validate the quick-scan. From the visual inspection it was deduced that in-roof obstacles were often not detected. Sometimes the customer had no desire to use the full potential of the roof, or preferred a rectangular panel layout instead of fitting the maximum amount of panels. Another finding was that the distance kept from the roof segment edge was much smaller than initially expected. For pitched roofs, there was virtually no distance between the roof edge and installed panels, while for flat roofs around 20 cm was kept. Using zero distance from the roof edge, the panel placement algorithm still underestimates the roof potential by 17.5% on average, with a relative standard deviation of 46.3%.

Three yield calculation methods were compared: the Solar Monkey method, the SVF & SCF method, and the method without obstacles. The predicted performance or final annual AC yield was compared with the actual measured performance, both measured in kWh/kW_p per year. For the three methods, relative standard deviation values of 7.2%, 7.5% and 9.1% were found respectively. The three methods could generate yield predictions for 91.0%, 92.7%, and 93.1% of the 145 roofs. For highly shaded roofs with Sun Coverage Factor values above 0.25, the method neglecting obstacles performed significantly worse. Additionally, the performance of large roofs with an average segment area above 70m² was generally under-predicted, while the relative standard deviation was highest. It is expected that using one obstacle view is not accurate enough for large roof segments.

The computational speed of the panel fitting algorithm for pitched roofs without internal obstacle segments was found to be 20.1± 5.0 m²s^-1, whereas for flat roofs it was found to be 56.9±12.0 m²s^-1. In order to optimise the quick-scan algorithm in both speed and accuracy, two filtering steps were carried out. Segments with a pitch angle over 10° and an orientation between 0° to 60° or 300° to 360º were filtered out, since they would have an annual performance below 650 kWh/kW_p. Moreover, segments with an area less than 8.4 m² were filtered out, since they were observed to fit less than 2 panels. The quick-scan calculation times for different yield prediction methods were found to be 15.50±1.01, 14.58±1.13 and 2.75±0.44 seconds per roof, respectively. These times were measured for data sets that had around 2 segments per roof after filtering on segment area and pitched segment orientation. It can be concluded that the method without obstacles is preferred when the calculation time is a limiting factor, whereas for accuracy the Solar Monkey method is preferred.

One of the most important aspects to consider when a PV system will be mounted, is the PV panel’s orientation in order to receive the highest amount of solar radiation and thus produce as much energy as possible. This is achieved by tilting the solar panel to an optimal angle (which depends of the location’s latitude) or by tracking systems which follow the Sun to get the maximum of energy from it.
In this project several topologies of PV panels will be analysed and a new one will be introduced: a floating PV module offshore. The irradiance incident on the panels and how much energy reaches the module at several static and tracking topologies. For the statical PV modules, it will be analysed and discussed which is the optimal tilting angle according to its location, (the latitude is the variable to consider in this case) as well as the monthly energy yield generated on each topology.
Finally, a comparison of incident energy yield per topology (including the new offshore one) will be assessed and discussed, indicating what are the results found in this project, which tools were used to get these results and how they were obtained.