Summary from the Sixth Wide Bandgap III-nitride Workshop

Richmond, Va, 12-15 March, 2000

by B Monemar

The workshop was organized by a group led by Cole Litton, with H. Morkoc responsible for the local arrangements. The venue was the Omni Richmond Hotel. The local arrangements were excellent.

In the following some selected topics treated in the workshop will be highlighted, we do not intend to provide a full coverage of all presentations and discussions.

Bulk growth and HCVD.

An update report was given on bulk growth from solution under slight overpressure. A GaN boule size of 20-mm length was reported. Growth on single crystalline GaN seeds is now pursued, and the produced material is on the way to being single crystalline. No further details were provided, neither on growth conditions (solvent used) nor on properties of the produced material. Growth of bulk AlN with sublimation transport was discussed. Up to 13 mm dia boules were produced, so far polycrystalline. The dislocation density was claimed to be below 5 10 ^-4 cm ^-3. Preliminary results were presented from low temperature ammono-thermal growth of GaN and AlN. Small mm size crystals were obtained, but so far no seeded growth. Several reports were given on the growth of thick epilayers with the HVPE technique. By growing very thick GaN layers on sapphire a dislocation density of about 3 10 ⁶ cm ^-2 at the top surface was reported. Production of thick freestanding layers by growth on LGO substrates and subsequent etching was reported, a size of 2" was predicted soon. There was a rumor that a company in Japan will soon offer thick 2" freestanding GaN wafers, from growth on GaAs.

MOVPE growth

The LEO technique was discussed, and the growth of LEO-PENDEO GaN has now been successfully demonstrated on silicon substrates. Another study reported on in situ XRD experiments monitoring the development of tilt during LEO growth of GaN on sapphire with a SiO ₂ masking. Clearly the tilt does develop during growth, only a very small part of it has to do with cooldown stress. The temporal development of tilt during growth was displayed. The growth conditions may be optimized to minimize this tilt, in order to avoid a large dislocation density in the coalescence region of the overgrown layers.

Properties of GaN.

MOVPE grown GaN buffer layers on sapphire were shown to have a resistivity that depended on the dislocation density. Varying the growth conditions (such as reactor pressure) the dislocation density could be systematically controlled, with a strong correlation with the resistivity of the layer. Acceptor states related to the threading dislocations were held responsible for this effect, which is important for FET devices grown on GaN. Schottky barrier measurements of the vertical transport properties in MBE grown GaN layers were presented. It was concluded that the vertical mobility in such layers is less affected by the dislocations, i e while the lateral mobility was 200 cm ²/Vs the vertical mobility was in the range 1000 cm ²/Vs at room temperature. A careful study of Mg doped GaN layers was presented, comparing SIMS, Hall data, EPR and ODMR. The Hall concentration tracks well with the uncompensated Mg concentration found in EPR (4 10 ¹⁹ cm ^-3). A concentration of compensating donors in the 10 ¹⁸ cm ^-3rangewas found, of unknown origin (not Si or O). Interestingly the blue PL emission in this material was suggested to be connected with a shallow donor from ODMR data, i e not a deep donor as commonly believed. Profiling studies of point defects in thick (about 50 μm) HVPE grown GaN layers were reported. While the concentration of Ga vacancies strongly increased towards the substrate (positron annihilation data) the yellow luminescence (YL) intensity appeared to have a strong opposite trend. This is in disagreement with the previous wisdom from MOVPE layers.

QW structures

Theoretical estimates of the exciton binding energy in AlGaN/GaN QWs were presented. It was concluded that the polarization fields as well as the screening effects by photo-induced carriers in optical experiments have a dramatic effect on the exciton binding energy, which may be reduced to about 10 meV. Under these conditions it is questionable whether the room temperature PL emission is of excitonic character, it should rather be free carrier recombination. Similar arguments would apply to InGaN/GaN QWs. The electron mobility for the 2DEG in AlGaN/GaN structures grown on low dislocation density (<10 ^-4 cm ^-3) GaN substrates showed a record value of about 60.000 cm ²/Vs at low temperatures. Intersubband electron transitions were studied in AlGaN/GaN MQWs. Absorption data for structures grown with 0.45 < x < 0.8 showed absorption bands in the range 1.8 - 4 μm. Such structures might be of interest for THz optical modulators.

Devices

Status reports were given for several devices, including lasers, HEMTs, HBTs and photodetectors. We shall not give details here. It appears like high performance HEMTs may be produced at moderately high dislocation densities, but the device characteristics are influenced by defects, and possible long-term degradation problems have not yet been much studied. PNP HBTs were reported, these are easier to make (compared to NPN) since the p-doping bottleneck is avoided. A future design with a transferred substrate bottom collector was suggested. HBTs will be more sensitive to the dislocation density than HEMTs. Solar blind UV detectors showed very promising data, the performance was already rather close to the stringent specifications for military use.

Bo Monemar