Recent null results in searches for expected dark matter candidates have spurred interest in looking for lighter mass particles. Experiments searching for these alternatives require low energy thresholds and increased sensitivity since energy deposits can be of O(eV). An experiment uniquely suited to this parameter space is DAMIC (Dark Matter in CCDs) which uses scientific grade silicon charge-coupled devices (CCDs). These pixelated devices have an excellent read noise (2e-) and a very low threshold of 50 eV. However, searching for DM in this regime requires a good understanding of: a) detector backgrounds b) a model of the ionization response at these energies and c) perhaps some way to extend the reach of the experiment below threshold. This work explores and provides answers for these three challenges. The first major contribution in this thesis is in accurately measuring the background from small-angle scattering of environmental gamma-rays, considered a dominant background for solid state detectors. By exposing a CCD setup at the University of Chicago to Co-57 and Am-241 radioactive sources, we measured the electron recoil background between 60 eV and 4 keV. The observed spectra were found to agree with theoretical predictions and we report for the first time a measurement of "Compton Steps", a series of step-like spectral features associated with the atomic structure of the silicon target. We provide a parametrization of this Compton background applicable to background estimations for any silicon based dark matter detection experiment. Secondly, in order to reconstruct the energy of any event in a silicon ionization detector, we need an accurate charge yield model that converts between the energy deposited and the number of charge carriers ionized. We reviewed existing literature measurements of charge yield in silicon and identified a calibration gap between 12-50 eV, referred to as the UV-gap. We synthesized a phenomenological model of impact ionization with a simplified model of silicon band structure to provide ionization probability curves, arguing that these are more appropriate than the single number calibration constant and Fano factor traditionally used to describe ionization in detector target materials. Thirdly, we set constraints on unexplored parameter space for dark matter recoiling off e- with masses between 0.6 and 100 MeV and hidden photon dark matter with masses in the range 1.2-9 eV by exploiting the impressively low leakage current (<10^-21 A/cm^2} of DAMIC CCDs. By interpreting the dark current present in our setup as coming instead from flux of light species that deposits energy below our reconstruction threshold, we can trade off between a model of this dark matter and a model of our leakage current to resolve the relative contributions of both. This method is shown to be competitive with traditional dark matter limit setting approaches. Finally, this work introduces and reports on R&D progress at UChicago in using "Skipper'" instrumented CCDs - a novel readout technique that allows for counting of individual charge pairs, with a demonstrated resolution of 0.07 e-, which ushers in a new era of sensitivity to low-energy interactions.