AlPN Enlarges The Nitride Family
Growth of AlPN epilayers promises better HEMTs and VCSELs
A partnership between researchers at Nagoya University and Japan's Institute for Material Systems for Sustainability claims to have broken new ground by producing the first epilayers of AlPN. This ternary, latticed-matched to GaN, promises to improve the performance of GaN HEMTs and VCSELs.
Latticed-matched AlPN could transform HEMTs by introducing a very high polarisation that leads to a high carrier concentration in the channel. Early results are very encouraging, with an unoptimised sample producing a sheet resistance of just 150 ± 50 Ω/square.
For GaN VCSELs, AlPN could be a game-changer, simplifying and improving the fabrication of the mirrors. Prior to the work of the Japanese team,
GaN and AlInN provided the only pair of nitrides that could be used for growing mirrors. With this duo, the growth of a mirror takes 12 hours or more, due to drawbacks on three fronts: there is a need for many mirror pairs; the ternary has a slow growth rate; and temperature ramping is needed between GaN and AlIN, which must be grown using a temperature deviation below 3degC.
Those issues are to blame for a growth time that is far too long for the production of VCSELs incorporating two GaN-based mirrors. Instead, devices tend to combine one GaN-based mirror with another made from a dielectric.
Switching to mirrors made from AlPN and GaN promises to slash the growth time to 2-3 hours, says the spokesman for the team that is pioneering this novel alloy, Markus Pristovsek from Nagoya University. According to him, the substantial time saving stems from the faster growth rate for the ternary and the reduction in the number of mirror pairs, realised thanks to a much larger difference in refractive index between the two nitrides.
The development of ternary nitrides has a long history, with efforts between 1996 and 2005 directed at a cousin of AlPN, GaPN. During those years researchers discovered that when the phosphor content exceeded 3-4 percent, phosphor atoms head for gallium sites, due to the shorter bond length and smaller size. Adding aluminium offered a solution. (The figure above shows how cranking up the flow of tertiary-butylphosphine increases the ratio of phosphor-to-nitrogen and reduces the density of cracks in the AlPN layer).
In 1999 Panasonic filed a patent for AlPN and AlGaPN. “However, there was never a publication,” points out Pristovsek, who reasons that either attempts failed or the patent was filed simply to expand an intellectual property portfolio.
Pristovsek started to actively pursue AlPN in 2012. “A first attempt to patent it at TU Berlin failed, because they thought there is absolutely no commercial value and the patent would not earn its fee.”
A move to the University of Cambridge enabled Pristovsek to secure funding for AlPN research. However, by the time an order had been placed for a tertiary-butylphosphine (tBP) bubbler that would provide a source of phosphor - phosphine is toxic, so forbidden in many labs - Pristovsek had an offer of a professorship at Nagoya University.
Taking that up in 2016, he took some time to find an underused reactor for his experiments and convert a metal-organic line to tBP. The first epiwafers were riddled with cracks, but cranking up the tBP flow addressed this issue.
Pristovsek and co-workers turned to X-ray diffraction to investigate the crystalline structure of a 60 nm-thick layer of AlPN, grown on a GaN-on-sapphire template. Measurements produced reflections only from GaN, sapphire, and strained AlP0.103N0.897, revealing that the ternary is pure wurtzite AlPN. Based on the position of the diffraction peak for this alloy, to ensure lattice-matching the composition of this ternary needs to be AlP0.106N0.894.
Ellipsometry measurements on samples with AlPN thicknesses of 180 nm, 315 nm and 665 nm indicate that the refractive index of this ternary, when lattice-matched to GaN, is around 1.95 to 2.05. There are Fabry-Perot oscillations associated with the 655 nm-thick sample that suggest that the bandgap for this alloy is around 5.5 eV.
One of the next goals for the team is to develop GaN HEMTs with an AlPN layer.
'Wurtzite AlPyN1−y: a new III-V compound semiconductor lattice-matched to GaN (0001)' by M. Pristovsek et al; Appl. Phys. Express 13 111001 (2020)