This thesis focuses on two areas in silicon photodetectors, the first being enhancing the sub-bandgap light absorption of IR wavelenghts in silicon, and the second being intrinsic signal amplification in silicon photodetectors. Both of these are achieved using heavily doped p-n junction devices which create localized states that relax the k-selection rule of indirect bandgap material. The probability of transitions between impurity band and the conduction/valence band would be much more efficient than the one between band-to-band transition.

The waveguide-coupled epitaxial p-n photodetector was demonstrated for 1310 nm wavelength detection. Incorporated with the Franz-Keldysh effect and the quasi-confined epitaxial layer design, an absorption coefficient around 10 cm-1 has been measured and internal quantum efficiency nearly 100% at -2.5V.

The absorption coefficient is calculated from the wave function of the electron and hole in p-n diode. The heavily doped impurity wave function can be formulated as a delta function, and the quasi-confined conduction band energy states, and the wave function on each level can be obtained from the Silvaco software. The calculated theoretical absorption coefficient increases with the increasing applied bias and the doping concentration, which matches the experimental results.

To solve the issues of large excess noise and high operation bias for avalanche photodiodes based on impact ionization, I presented a detector using the Cycling Excitation Process (CEP) for signal amplification. This can be realized in a heavily doped and highly compensated Si p-n junction, showing ultra high gain about 3000 at very low bias (<4 V), and possessing an intrinsic, phonon-mediated regulation process to keep the device stable without any quenching device required in today’s Geiger-mode avalanche detectors.

The CEP can be formulated with the rate equations in conduction bands and impurity states. The gain expression, which is a function of the primary photocurrent and related to the phonon absorption time, predicts the same trend of the gain increasing with temperature and decreasing with increasing primary photocurrent.

Melanopsin-expressing retinal ganglion cells (mRGC) are intrinsically photosensitive and combine their melanopsin-based photoresponses with rod and cone signals to convey light information to a subset of retinal brain targets. mRGC axons to non-image forming (NIF) visual centers are essential for the proper functioning of processes like circadian photoentrainment and pupillary light reflex. Surprisingly, mRGCs also send axons to image-forming regions of the brain. It is unknown how mRGCs mediate such diverse functions.

Classically, a cell’s morphology and location in a biological system is a direct reflection of its synaptic connections and, by definition, their function. mRGCs can be divided into five subtypes (M1-M5) based on morphology and dendritic stratification in the inner plexiform layer. In the classical sense, since M1s send axons to only a subset of mRGC-target regions and are the only subtype that monostratify in the OFF-sublamina, M1s likely serve a distinct function from other subtypes. However, M1s, like all mRGCs, exhibit an ON-response. This reveals a hole in what we understand about intraretinal connectivity and attenuates the weight that should be afforded to stratification in determining function. While the other mRGC subtypes have distinct morphology and branching patterns, it is unknown whether they serve specific functions. Thus, in order to explore the structure-function relationship of mRGC subtypes, we must consider connectivity. Unfortunately, the variable expression of melanopsin protein between subtypes and across the architecture of a single mRGC and the lack of unique markers for up- and downstream interactors has precluded rigorous study of mRGC connectivity in the retina and central targets.

We use a correlated light and electron microscopy label and serial blockface scanning electron microscopy to explore the architecture and synaptic partners of mRGCs in an attempt to better understand the connectivity of mRGC subtypes. We show significant differences in the ultrastructure of mRGC axonal terminals in mRGC-recipient brain regions, stratification-specific differences in mRGC dendrites, and catalog the intraretinal connections specific to mRGC subtypes.