Summary of the 3rd International Conference on Nitride Semiconductors (ICNS 3) in Montpellier, France, 5-9 July 1999

Summary of the 3^rd International Conference on Nitride Semiconductors (ICNS 3) in Montpellier, France, 5-9 July 1999.

This ICNS 3 conference was the follow up of ICNS 2 in Tokushima 1997, and attracted more than 500 registered participants. It was held in the rather new and comfortable conference center "CORUM" in the center of Montpellier. The facilities for the oral sessions were excellent, there was a good space for coffee breaks, but the poster area was unfortunately a bit crowded. Conference chairmen were B Gil, B Meyer and B Monemar; indeed the local chairman B Gil had assumed the main responsibility for the organisation. A Hoffmann was responsible for the scientific program. The scientific program was kept in workshop style with single sessions, which certainly facilitated the planning for those who wanted to follow the entire program, but also lead to a very large number of posters, more than 300 split into 3 different sessions. More than 30 exhibitors took part in the conference.

The scientific program covered the entire field from growth, characterisation to processing and devices. Several groups reported on excellent results from MBE growth of GaN structures on GaN templates, giving in some cases material properties (like electron mobilities) in excess of the best values reported from MOCVD growth. A large number of groups also reported progress in HVPE growth, in particular the successful use of buffer layers for the reduction of defects in heteroepitaxial growth. The epitaxial lateral overgrowth technique was discussed extensively, and characterisation results with high resolution CL were reported, showing in detail the variation of optical properties across an LEO profile. Interesting results were reported from the use of tungsten as a mask in HVPE LEO. Also LEO and PENDEO epitaxy on silicon substrates was reported, with encouraging results. In general, it appears like growth on silicon attracts an increased attention, the great tendency for cracking still needs to be addressed in this area, however.

The low temperature interlayer technique was reviewed, as a large area alternative to the LEO process. Dislocation density reductions to the range < 10⁷ cm^-2 was reported, i e similar to the LEO case. In fact the physical mechanism behind this process appears to be very similar to LEO, i e mass transport and lateral growth bend over the threading dislocations horisontally in the process.

There has been no significant breakthrough in the area of bulk growth sine ICNS 2. The high pressure solution growth method can provide flakes of an area up to 1 cm² nowadays. The growth of thick (about 150 µm) wafers by a combination of MOVPE and HVPE processes on LiGaO₂ substrates, which were subsequently removed, may be an interesting approach for GaN wafer production. No report on boule growth of III-nitrides was given at this conference.

The properties of bulk materials were discussed in several sessions. It was demonstrated that nominally undoped homoepitaxial MOCVD grown GaN has a typical PL linewidth of less than 0.1 meV for the bound excitons. This is encouraging, since with this kind of materials basic physical properties for defects, i e with perturbation spectroscopy, magnetic resonance techniques, etc can be studied with reliable results, which was so far not the case with heteroepitaxial material. It was also demonstrated that reflectance data give proper results for he excited states for the intrinsic excitons, while the PL spectra in addition show excited states for bound excitons, which complicates the interpretation.

The electronic properties of dislocations were discussed. While a few years ago there was a consensus among most workers in the field that dislocations were harmless for carrier recombination in GaN, i e no levels in the bandgap, the situation has now changed dramatically. There is now strong evidence that threading dislocations do have states in the bandgap, which cause carrier recombination. Theorists have adopted well to this new situation, and are now predicting the same.

Theoretical results for phonons in III-nitrides were presented, clarifying the situation of the optical modes in alloys. Experimentally phonon energies are rather well known from Raman and other optical studies, but certainly the values are affected by strain for heteroepitaxial samples, and the accuracy will be improved once data from strainfree homoepitaxial bulk materials become available.

Polarisation fields (i e spontaneous polarisation (SP) and piezoelectric polarisation (PE)) have been discussed extensively during the last couple of years, and it was agreed that at least the latter plays a strong role in governing the potential (and consequently even carrier injection) in many device structures. The conclusion from the contributions at ICNS 3 was that there seems to be a consensus that the SP fields indeed exist and are of the same order of magnitude as the PE fields in most structures. What remains is to get more accurate values for the SP and PE tensors in the III-N materials.

Quantum structures were discussed in several contributions, in particular the properties of InGaN/GaN QWs, which are the basis of most present light-emitting devices. It appears that composition inhomogeneities in InGaN occurs on several length scales (µ size, sub-µ size, few nm-size), affecting the localisation of carriers and excitons. These properties appear to be very sensitive to minor changes in growth parameters, so that rather different optical properties are reported from different research groups. The additional presence of piezoelectric fields complicates the analysis of optical data from these structures. Also both localisation potentials and the PE fields are screened (in different ways) by injected carriers or by doping. A proper description of the physics of InGaN QWs therefore requires a rather complete set of parameters.

AlGaN/GaN QWs were demonstrated by several groups, grown by MBE on both sapphire and silicon substrates. This system gives excellent opportunities to study the influence of polarisation fields. The PL linewidths for excitons in these QWs is limited by fluctuations in the interfaces, typically caused by the natural steps occurring in the lateral growth mechanisms. A problem is the defect-induced tunneling through barriers, occurring in thin AlGaN barriers with high point defect density.

The more exotic alloys GaNAs and GaInNAs were also discussed in several contributions. The band structure for these alloys is still debated, i e whether GaNAs/GaAs QWs are type I or type II. Although the experimental results reported for such structures demonstrate a high point defect density, they still show promise for ir emitters. No report was given on GaNAs with high N content (close to GaN).

The progress in devices was given proper attention in the program. LEDs and LDs were discussed, in terms of improved performance needed for many applications. For LDs the lack of a high quality GaN substrate is badly felt. UV detectors fabricated on LEO material were demonstrated, with much improved performance. The AlGaN/GaN HFETs show a steady increase in performance. Problems exist due to the high defect density in the materials, such as a high gate leakage current, lower than expected breakdown fields, etc. The already demonstrated prototype devices are impressive in terms of power handling capability, which is very encouraging, since there should still be a wide margin for improvements as the material quality is getting better in the future.

A discussion panel session was given one evening, devoted to device development in III-nitrides, and prospects for industrial production. It was clarified that the future volumes of devices are expected to be very high in the optical device area. For high power transistors the volumes are lower, but since a much higher cost per device will be tolerated the market value will still be large. The market is in general expected to grow rapidly once the capability of new devices has been demonstrated.

During the conference one session with a plenary talk was held in honour of Prof I Akasaki, recognising the outstanding contributions by his research group in the field of III-nitrides over the last 35 years. Prof Akasaki and Dr Nakamura both received a medal from the town on Montpellier. Dr Nakamura also received the Springer Verlag Prize.

Concerning the non-scientific activities everybody enjoyed the proximity to the very old center of the city of Montpellier, with ample opportunities for outdoor eating in a nice atmosphere. The same can be said about the conference banquet, which was held in an old monastery outside the city. The good food and rich supply of domestic wine contributed a to cheerful evening, with live entertainment the whole time all around the dining area. We envy the French who have so excellent venues for conference dinners, and in addition so pleasant weather conditions for outdoor eating!

The next version of this conference (ICNS 4) will take place in the USA in 2001 (probably in Denver late July). Prof J Pankove will be the Conference Chairman. We wish him good luck for the continuation of this conference series, in an exciting and still rapidly growing research field.