UC Berkeley

Two significant drivers to innovation in electronics and electronic materials in recent

history have been electronic device scaling and the pursuit of high efficiency photovoltaic

cells at low costs. The motivations of these two fields have been interestingly parallel, as

“density” has been a key metric for both – areal energy density in the case of photovoltaics,

and component density in the case of electronic devices. So strong is this motivation to

lower device costs by packing more performance into a smaller area that “laws” have been

devised to inspire innovation in each field, with Moore’s law to describe the periodic

doubling of transistor density and Swanson’s law to outline the steady drop in solar cell

module costs over time. As each law approaches a wall erected by fundamental physical

limitations, science must identify roadblocks and solutions that can allow innovation to

continue.

Semiconductor materials are a key limiting factor for each application, as their physical

properties determine ultimate functionality of a device and the challenges involved in

device design. In both electronic and optoelectronic applications, scalable manufacturing

of III-V materials has been a promising avenue to improvement, as while they are

traditionally expensive to produce, they use a larger portion of the solar spectrum for

photovoltaic devices, and are easily utilized in the fabrication of high performance optical

and electronic devices. In recent years, a more scalable method for the growth of III-V

materials without costly epitaxial substrates has been developed, by utilizing the vaporliquid-

solid growth (VLS) process to grow structures confined by a metal catalyst.

Structures such has nanowires have been fabricated with this technique and studied

extensively, but a recent expansion of the approach has also allowed for growth of highquality

thin films using planar templates for nucleation control. In this dissertation, I

discuss the use of this approach in a number of applications, including the development

of large-area photovoltaic devices and an evolution of the technique to greatly expand its

application space through lower process temperatures.

First, I will discuss an ideal preliminary application of this technique, with the

development of p-body InP photoelectrochemical cells for the direct production of

chemical fuel using sunlight. This application shows the utility of large-scale

polycrystalline growth with larger than normal grain sizes enabled by the technique, and

the fundamentals of the growth process and usable doping methods are explored in

tandem. This study also demonstrates the successful application of an efficient selective

electron contact to the poly-InP system, enabling promising device performance and

enhancing device stability under harsh photocathode operation conditions. Hydrogen

fuel production from simulated sunlight is also directly and quantifiably observed from

the device as a capstone to this experiment.

Following the investigation of larger area thin-film growth, the microscale templatedliquid-

phase (TLP) crystal growth method is explored and expanded to target a wider

range of applications. This method, a modification to the thin-film vapor-liquid solid (TFVLS)

process initially studied, has previously enabled growth of defined patterns of single

crystal domains on amorphous substrates. While this is an impressive result with great

promise for integration of III-Vs into highly scaled electronics, growth temperatures

previously explored would need to be lowered significantly for facile integration to be a

reality. Using a simple modification to the existing TLP process, I demonstrate growth

temperatures well within the silicon CMOS thermal budget, with proof-of-concept devices

fabricated at temperatures as low as 270ºC with the InP system. With applicability to a

variety of substrates, this study has neatly expanded the application space of III-Vs, with

complex methods and material requirements replaced with simple direct growth.