Hot electrons are an excited distribution of electrons that emit phonons – vibrations of the quantum lattice – to cool down to temperatures approaching that of the surrounding lattice. The energy lost by an electron is directly passed onto a phonon and such a mechanism is crucial for understanding how electronic devices work.

The new experimental result from Matt Graham and colleagues backs up a recent theory that predicts that electrons in graphene cool down in an additional way – by undergoing supercollisions. This process involves the electrons colliding with disorder or defects in the crystal lattice, thereby transferring their momentum. The electrons thus cool much faster than if graphene were a perfect crystal without any defects.

Photocurrent “thermometer”

The researchers obtained their results by making measurements on a graphene p-n junction photodetector that they had fabricated. They fired very short, 100 femtosecond-long, laser pulses at the device and determined the temperature of the junction as the electrons cooled. In these experiments, the photocurrent generated in the junction is effectively used as a “thermometer” of the hot electron temperature near the Fermi level in graphene, explains Graham.

“By heating the junction with an initial laser pulse and then again with a second pulse just 100 fs later and comparing the crossover of temperatures, we were able to measure the temperature of the p-n junction in our device with sub-picosecond time resolution and within a few kelvins of accuracy,” he said. “We found the resulting cooling rate and electron temperatures were all well predicted by the supercollision model proposed in 2012 by Justin Song of Leonid Levitov’s theory group at the Massachusetts Institute of Technology.”

It is worth highlighting that our results and Levitov's prediction are both radically different from what the graphene community had previously predicted, he adds, namely a very slow relaxation process via the simple acoustic phonon emission route described above. Here electron relaxation takes as long as nanoseconds owing to energy-momentum conservation. Instead, we find that supercollsions dominate and the electron relaxation takes only picoseconds.

More efficient way to relax

Most graphene-based devices under development, from photodetectors to non-silicon field-effect transistors, depend on an accurate model for hot electron cooling rates, he explained. “Our experiments provide the first real justification for a fundamentally new model to determine this rate in graphene, and future device development will need to consider the new accelerated hot electron relaxation pathway predicted by Levitov and corroborated by our work,” he told nanotechweb.org. “We have confirmed that cooling is greatly accelerated by supercollisions in graphene, where electrons scatter with disorder in a system to find a much more efficient way to relax.”

The Cornell researchers, who report their current work in Nature Physics, now plan to determine the electronic cooling rates in other novel nanosystems using their time-resolved cooling rate measurements.

Graphene shows great promise for next-generation electronic devices thanks to its numerous unique electronic and mechanical properties that include near-perfect conductivity and high tensile strength. The fact that it is transparent to light over a wide range of wavelengths in the electromagnetic spectrum means that it could also be ideal for optoelectronics applications.