While hydrogenated amorphous silicon (a-Si:​H) solar cells have been studied extensively for the previous four decades, the low performance of the devices is still not well understood. The poor efficiency (below 10%, even in research-scale devices) is believed to be mainly due to the low hole mobility of the bulk material, but there is little is known about the physical phenomena responsible for this deficient mobility. This work explores the atomic structures causing the inefficient hole transport in a-Si:​H, as well as a novel rout toward improvement.

First, a large ensemble of computational a-Si:​H structures (216 Si atoms, with ~10% H) is created, representing over 600,000 atoms. This ensemble is analyzed using density-functional theory (DFT) calculations, and statistical correlations are made between discovered defective atomic structures and strong hole trapping. It is observed that a self-trapping defect arising from a reversible atomic rearrangement in the presence of a hole is most strongly correlated with deep trapping, followed by floating bonds, or over-coordinated silicon defects. Dangling bonds, or undercoordinated silicon defects, despite their traditional indictment for the responsibility of trapping in the literature, are found not to correlate with strong hole traps.

Experimental films are produced using plasma enhanced chemical vapor deposition (PECVD), in p-i-n solar cell device configurations. By varying the chamber pressure, a set of devices with widely ranging properties are produced, varying hydrogen content and bonding configuration, stress, and density. These devices are then characterized using time-of-flight (ToF) photocurrent-transient measurements, allowing the direct measurement of hole transport though the intrinsic layer, and thereby the calculation of the material hole mobility. It is found that the peak mobility occurs at both intermediate (compressive) stress and hydrogen contents, with rapid linear declines in mobility as this maximum is deviated from (with respect to film stress).

The computational ensemble of a-Si:​H is then extended to include the experimentally-observed variables of stress and hydrogen concentration. A second ensemble of lower hydrogen content (~5%) is created, and both hydrogen contents are relaxed at three differing stress states (-1, 0 and +1 GPa), extending the full simulation to approximately 2 million atoms across over 8 thousand structures. It is found that the modification of the stress and hydrogen content of computational samples correlates to shifting regimes of defect prevalence - increased hydrogen content and increasing compressive stress are both correlated with increased floating bond concentration. Low absolute values of stress correspond to increased ionization displacement defects. High tensile stress is observed to increase strong hole traps, without substantial increases in any of the previously explored defects, which is attributed to lattice expansion allowing further hole delocalization around trapping structures which would be otherwise less favorable due to the high kinetic penalty of strong wavefunction confinement. These relationships are then correlated to the aforementioned experimental results, and further experimentally vetted, where possible.

Finally, as the observed shifting nature of defects in a-Si:​H makes the further improvement of the bulk material untenable, methods are explored for utilizing the beneficial properties of the material (namely the strong bulk absorption and robust surface states) to achieve improved hole extraction (or effective mobility) from devices. Specifically, nanohole structured hydrogenated amorphous silicon (nha-Si:​H) devices are created as a proof-of-concept, showing up to 50% increases in efficiency over equivalent planar devices, for low-performing materials.