Research Highlights

Venture Capital and Cleantech

From our op/ed in the Financial Times:

In 2006, Silicon Valley began to bet big on clean energy technology. Seduced by grand visions of making a fortune while saving the planet, venture capitalists invested a then-record $123m in the first round of fundraising for 16 new companies that year. In 2008, they would sink nearly $1bn in over 100 new companies.

But when these investments began to flop, the cleantech bubble abruptly popped. Since 2009, VCs have barely funded 25 new cleantech companies a year, slowing new investment to a trickle.

What went wrong? And where should cleantech go from here? To answer these questions, we compared the performance of every medical technology, software technology and cleantech company that received its first round of VC funding between 2006 and 2011. We found that betting on cleantech start-ups just did not make sense for VCs, because cleantech could not deliver the outsized returns found in other sectors.

This conclusion is alarming because new technologies are desperately needed to confront climate change. Still, guided by the lessons learnt from the cleantech VC boom and bust, new private and public funding sources may be able to better support revolutionary technologies.

More details can be found in Energy Policy and at the MIT Energy Initiative.

Optical Absorption in Aluminum Nitride

Aluminum Nitride (AlN) is an ideal material for growing high-performance ultraviolet (UV) LEDs and lasers which can be used for water disinfection. However, current growth techniques result in a substrate which re-absorbs the UV light generated in the active region. Improving the efficiency of these devices relies on finding the origin of this parasitic absorption. This project allowed us to create a rich feedback loop between our modeling efforts and advanced synthesis and characterization teams in the Wide-Bandgaps group at NC State, HexaTech, Tokyo University of Agriculture and Technology, and Tokuyama Corporation.

Schematic of an LED structure and the carbon defect.

The structure of an LED showing the AlN substrate, and an exploded view of a simulation cell that includes a carbon defect.

Experimental measurements of the absorbent substrates indicated high concentrations of carbon impurities in the samples. To determine whether carbon could cause the observed 265 nm absorption band, we calculated the formation energies of native and impurity substitutional defects. We found that carbon substituting on the nitrogen site was likely to act as a deep acceptor over a wide range of chemical conditions. Our model predicted an optical transition associated with this defect which would absorb at 4.7 eV (265 nm) and emit at 3.5 eV.

This prediction was then tested by our collaborators in Raleigh and Japan, by growing AlN samples with carefully controlled carbon concentrations. Absorption and emission measurements show that the parasitic absorption peak is eliminated as the amount of carbon is reduced. Furthermore, the emission peak associated with the same defect is also reduced.

For further details, please see our paper published in Applied Physics Letters. (Collazo, Xie, Gaddy, et al., Appl. Phys. Lett. 100, 191914 (2012).)

Smooth Growth of Polar Oxide Semiconductors

New functionality can often be achieved at the interface between different materials. Often, some of the most interesting effects are achieved when the two materials have different underlying structures. Integrating these materials with minimal distortion is a challenge that can be overcome by exploiting similar symmetries across the two crystal structures. For instance, a cubic rocksalt insulator, like calcium oxide, can be integrated with a hexagonal wurtzite semiconductor, like gallium nitride, if the rocksalt oxide can be grown in the {111} direction. With this growth mode, a clean, atomically abrupt interface has been achieved. However, after only a few layers of the material are deposited, the surface of the material facets into {100}-type microfacets, leading to a rough growth mode. These facets severely affect the performance of the devices.

Experimental investigations determined that smooth {111} surfaces could be stabilized in a humid environment, but the mechanism of stabilization was not fully understood. Our computational modeling compared the stability of CaO surfaces with varying surface structures, including reconstructions and the presence of molecules at the surface and attached to the surface. We used ab initio thermodynamics to evaluate the most stable surface for a given temperature and pressure and thus were able to predict a range of conditions of the chemical environment that could be used to stabilize the surface.

To read more about this research, see our paper in Nature Communications. (Paisley, Losego, Gaddy, et al., Nature Comm. 2, (2011).)