News Article
Shedding light on the p-type doping of GaN
UCSB theorists have revealed that magnesium in GaN is not a typical shallow acceptor
All electronic and optoelectronic devices require the ability to controllably dope the semiconductor.
Adding small amounts of impurities makes the material a conductor for electrons (n-type) or for holes (p-type).
In GaN, a material that enables the rapidly expanding technology of solid-state lighting and power electronics, n-type doping is straightforward, but adequate p-type doping is still difficult to achieve. According to UCSB researchers, intriguingly, magnesium is the only impurity that can turn GaN into a p-type semiconductor.
What's more, optical spectroscopy techniques that are routinely applied to characterise dopants in semiconductors have produced results for magnesium that are hard to reconcile with shallow-acceptor behaviour.
Magnesium-doped GaN exhibits two main photoluminescence signals; a peak in the ultraviolet (UV) region, at 3.27 eV, and a blue luminescence peak near 2.8 eV. The UV signal has conventionally been attributed to the “shallow” magnesium acceptor and the blue line to a magnesium-induced compensating donor.
These assignments, however, conflict with the behaviour observed during the post-growth annealing process that is necessary to activate magnesium acceptors. In other words, annealing causes a decrease in UV and an increase in blue intensity.
Moreover, the researcher say that no evidence of a compensating defect participating in the blue line has ever been found.
John Lyons and Anderson Janotti, working in Chris Van de Walle’s Computational Materials Group at the University of California, Santa Barbara (UCSB) have now unravelled this behaviour using state-of-the-art first-principles methods. They found that the magnesium acceptor exhibits dual character.
Figure: Magnesium in GaN exhibits behaviour characteristic of a “deep centre”. Its electrical level (solid line) lies close enough to the valence band to generate a modest concentration of holes, but in optical experiments (dashed line and arrow) a deep level is observed. The model shows the large atomic displacements that occur when a hole (spin density illustrated by the yellow isosurface) is trapped in the neutral charge state of the acceptor.
From an electrical point of view, magnesium induces an acceptor level just slightly above the valence band, with an ionisation energy small enough to allow for modest p-type doping. In all other respects, however, magnesium exhibits the features of a deep centre. This means that the hole state is highly localised, and leads to broad deep level luminescence in the blue region of the spectrum, at an energy well below that expected for a shallow impurity. The blue luminescence observed in magnesium-doped GaN is therefore caused by the magnesium acceptor itself!
“Magnesium in GaN only “accidentally” behaves as a shallow acceptor,” comments Van de Walle. “It’s truly lucky that its electrical level lies close enough to the valence band to generate enough holes for p-type conduction, otherwise blue LEDs and solid state lighting would not be possible!”
As to the UV signal, the researchers propose it originates from the Mg-H complex that is present in as-grown GaN. This assignment is consistent with the decrease of the signal observed upon annealing, which breaks up the Mg-H complexes.
All in all, the results explain a host of previously puzzling experimental observations, and pave the way towards better control of p-type doping in this key material.
This work has been described in more depth in the paper, "Shallow versus Deep Nature of Mg Acceptors in Nitride Semiconductors," by John L. Lyons et al, Physical Review Letters, 108, 156403 (2012). DOI: 10.1103/PhysRevLett.108.156403
This research was supported by the National Science Foundation and by the UCSB Solid State Lighting and Energy Centre.