The developed world is built on the fact that energy is readily available and functionally infinite. The electricity from the wall, the gas at the station, and the heat in our homes are reliable and low-cost. But this comfort is so far only possible through the combustion of fossilized carbon sources. This is not a sustainable system. Not only are these sources finite, but the long-term effects of this much combustion are catastrophic for our planet and species. Transitioning from current energy production to sustainable sources will require the development and deployment of a diverse group of technologies. Chief among these is the generation of electricity from sunlight, or photovoltaics (PV). Most modern PV is based on crystalline silicon. Silicon PV alone could be scaled to provide the necessary energy production, but this could be realized more quickly and efficiently if more advanced technologies are used. The defining metric is the cost per generated electricity. In the case of silicon, the current cost breakdown of materials, manufacturing, installation, and system costs leaves little room for further price reduction; and light-to-electricity efficiencies are near their practical limit. Of the proposed alternatives, the pairing of silicon with a thin film of lead-halide perovskite in a multijunction or "tandem" architecture is one of the most promising. This splits the solar spectrum in two, and raises the theoretical maximum efficiency from 33.3\% to 45.1\% based on each material better using its segment than any one material could use the full spectrum. The central goal of this thesis was therefore to advance the capability of this type of solar cell while remaining within the constraints of industrially-plausible production methods. Towards this goal, we first isolated perovskite layers to study what limits their performance. This led to improvements in the bulk material quality, the interfaces with adjacent materials, and the layer morphology. We reduced bulk disorder via balancing the annealing temperature, time, and environment. The interfaces were improved by selecting passivating materials which were compatible with our deposition methods. We further observed a complex interplay between the materials used and the environment of the annealing, which illuminated critical dependencies of device performance and gave us a more complete understanding of our material. Finally the morphology was improved via high-temperature annealing. This was enabled by a surface-binding additive which reduced the detrimental effects of such an anneal while preserving the benefits, resulting in a perovskite with less internal resistance. From there, we moved outside of the perovskite layer and developed a light scattering ZnO layer which increased the optical performance and absorption of our cells. Returning to the main goal of the thesis, the lessons learned from work on isolated perovskites were applied in tandems on textured silicon. We additionally increased the perovskite layer thickness and made the bottom-cell more robust to processing damages. Together, the result was a stabilized efficiency of 27.3\%. This is better than any silicon cell to date, but work remains to achieve 30\% and beyond, extend material stability, scale up the deposition area, and transfer these capabilities to industry. Together, the work done as a part of this thesis represents a step forward in the journey towards carbon-neutral energy production.