Amine-thiol solution processing of chalcogenide photovoltaics

Global energy demands have been increasing dramatically since the industrial revolution, and with a deluge of policy aimed at electrification, electricity is likely to be an ever-growing proportion of energy consumption. Solar photovoltaics is perhaps the most versatile source of electricity generation, working at scales ranging from calculators to domestic to utility scale. This versatility is reflected in its wide array of use-cases, including building and vehicle-integrated photovoltaics.

Mono and poly-crystalline Si dominates the photovoltaic market and has reached grid-parity in many parts of the world. However, second and third generation photovoltaics offer the promise of reduced material usage and cost, shorter energy payback time, and faster processing.

Copper indium gallium diselenide (CIGS) is a second generation photovoltaic material that has the rare combination of a high percentage conversion efficiency and commercial deployment. However, unlike Si technologies, there is a large discrepancy between laboratory efficiencies and module efficiencies.

The majority of high efficiency CIGS is produced using vacuum processing methods, which has a high capital costs and typically has a lower material utilisation. Additionally, high performance solution processed CIGS typically uses hydrazine as a solvent, which is highly toxic and explosive. Therefore in this work, CIGS processing was performed using an amine-thiol based solvent. An automatic ultrasonic spray deposition system was initially used to deposit the CIGS solution, but this was changed for spin coating following reliability concerns.

High efficiency CIGS relies on alkali metal doping, typically either from a soda-lime glass substrate or from a more intentional post-deposition treatment. Post-deposition treatments are more controllable and are compatible with flexible substrates such as polyimide, and so are generally preferred. However, these alkali post-deposition treatments are typically performed using alkali fluorides, which are expensive and toxic. Therefore, in this work the doping of Na and Rb was performed using alkali chlorides, which are significantly cheaper and less toxic. This was highly successful, with the NaCl treatment producing an increase in the doping density (10¹⁴ ⇒ 10¹⁶ cm⁻³) and short circuit current (10.2 ⇒ 27.5 mAcm⁻²), resulting in a CIGS cell with a percentage conversion efficiency of 8.43 %. A post-selenisation RbCl treatment, in addition to the N a treatment, lead to an 11 % increase in the open circuit voltage.

Antimony selenide (Sb₂Se₃) was also investigated in this thesis. Sb₂Se₃ is a third generation chalcogenide photovoltaic material with a unique 1-D ribbon structure. This results in anisotropic charge transport properties, with the potential for highly efficient charge transport if the orientation is favourable. In this work, Sb₂Se₃ absorbers were deposited using spin coating and spray deposition using a handheld chromatography atomiser. These absorbers were then fabricated into devices. Percentage conversion efficiency was increased dramatically from the first Sb₂Se₃ device produced in this laboratory (2.7 × 10⁻⁴ %) to the champion device produced as part of this work (1.6 × 10⁻¹ %). Interesting phenomena was reported, such as the dendritic formation of macroscopic needle-like Sb₂Se₃ structures, which were identified using light beam induced current measurements.

Both CIGS and Sb₂Se₃ devices in this work were fabricated on top of a Mo coated substrate. The Mo used in this work consists of a 4 layer stack - a layer deposited at a high sputtering pressure (to produce superior adhesion properties), a layer deposited at a low sputtering pressure (to produce a lower resistivity), a barrier layer consisting of MoN_x species (to prevent the excessive selenisation of the back contact), and a sacrificial layer of Mo (to facilitate the formation of MoSe₂ to produce an ohmic contact). In this work, the layers were characterised, and the effect of thermal processing of the layers was investigated. Thermal processing, in the form of either a heated deposition or an anneal at vacuum following the deposition, resulted in increased crystallinity and reduced resistivity for all layers. However, the efficacy of the MoN_x barrier layer suffered as a result of the increased crystallinity and a reduction in the amount of N incorporated into the film.

Finally, the way that data was stored and handled in this group was revised. Over the course of my PhD, I wrote a bespoke data-handling and analysis program in Python. The aims of the program were to solve the inefficiencies of traditional computer-based data storage solutions, to leverage the power of modern computing, and automate data analysis. The program combines the concurrent visualisation of thousands of photovoltaic data files with automated parameter extraction, and remains in use within the research group. This thesis provides an overview of the features of that program, as well as pertinent code snippets that explain how the program works.