"The reported efficiency in our devices is the highest for bottom-up gallium arsenide nanopillar solar cells to date," team member Giacomo Mariani of UCLA told nanotechweb.org. "The work is also a significant step towards device reproducibility and controllability compared with traditional techniques that lead to random nanowire growth."

Nanostructured solar cells show much promise thanks to light-trapping effects that dramatically reduce the amount of photons reflected from a device. This ultimately enhances optical absorption. In recent years, researchers have studied structures such as nanodomes, nanocones, nanoparticles and nanowires as possible candidates for improving performance in solar cells. The high surface-to-volume ratio of these materials also increases the all-important photoactive junction area so that more photons are harnessed, something that leads to enhanced power-conversion efficiency.

Nanopillars for next-generation solar cells
Nanopillars – densely packed nanoscale arrays of electro-optically active semiconductors – could be used to make a next generation of relatively cheap and scalable solar cells, but these materials have been hampered by efficiency issues. Another problem is that growing such structures normally requires a metal catalyst but this technique produces randomly located nanopillars. The metal catalyst can also contaminate the pillars and increase leakage currents in finished devices.

The new method, developed by Diana Huffaker and colleagues, relies on a lithographically defined substrate for selective area epitaxy and the mask used is pre-defined to fix nanopillar diameter and pitch. What is more, it provides a way to make large-area nanopillar arrays.

The researchers grow their nanopillars in a metal-organic chemical vapour deposition reactor that allows both axial (core) and lateral (shell) nanopillar growth to be controlled at will. No metal catalyst is required, which means high crystal quality. Indeed, p-n junctions made from the nanopillars have a low leakage current of around just 236 nA at –1 V and the power conversion efficiency of the material is as high as 2.54%.

Silicon substrates
The team now plans to port the III-V devices to silicon substrates, because silicon is a much more cost-effective platform than the gallium arsenide used in this work. It is also looking at other materials as potential substrates. "For example, the pillars can be embedded in flexible polymers and peeled off from the growth platform to realize a flexible solar cell with the high efficiency of III-V materials," said Mariani.

"We are just beginning to develop this new class of GaAs device," he added. "Hetero-epitaxy on silicon will certainly lead to higher efficiency, low-cost solar cells that might even lend themselves to being mass produced."

The work was published in Nano Letters.