CONFERENCE REVIEW
Worksho¥summarizes gallium nitride research
Samuel (Toby) Strite
RIGI, SWITZERLAND?The first European GaN Worksho¥(EGW-1) attracted more than 140 visitors to this Swiss Alps resort in June. The technical program comprised 64 contributed papers and three invited talks, together with a ORum¥Session.O The conference marked the first time that the entire European nitride community has assembled. Of the 108 scientific participants, nearly one-quarter were from former Warsaw-Pact countries, which have a strong tradition of research on nitrides of Grou¥III elements. The United States and Japan also had strong turnouts.
One of the most interesting technical sessions covered new substrates for GaN epitaxy. While GaN-based light-emitting diodes (LEDs) grown on sapphire have outstanding performance, the first several reports of GaN lasers on sapphire and spinel suggest that these substrates will be a far greater hindrance to laser development. Jaques Pankove of Astralux Inc. (Boulder, CO) pointed out during the rum¥session that the huge density of structural defects present in GaN grown on sapphire does not contribute enough nonradiative recombination to greatly limit LED performance, but the absorption attributable to these defects severely hampers lasing.
This is evidenced by the multiple-quantum-well (MQW) design favored by Nichia Chemical (Tokushima, Japan) in its early reports. The MQWs increase the gain to offset absorption and facet losses, but also increase the threshold current density. Sapphire and spinel are insulators that increase the complexity of device design and processing and sharply reduce the number of devices that can be fabricated on a wafer. Facet cleaving may also prove difficult with these substrates. The search for new substrates is an ongoing theme in GaN research. At EGW-1, exciting new results on bulk-GaN substrates, as well as silicon carbide (SiC) and silicon (Si), were presented.
Considerable improvement in the diameter of bulk-GaN single-crystal platelets grown by the high-nitrogen-pressure-solution method was described in several papers given by the grou¥from the Polish Academy of Sciences. For several years the literature has described homoepitaxial GaN of unprecedented quality grown on millimeter-sized platelets provided by the Polish grou¥to Western laboratories. Now, a new apparatus has enabled growth of 5-mm-diameter platelets. As the high-pressure crucible size was doubled, the diameter of the resulting crystals increased fourfold (see figure on p. 26).
Izabella Grzegory of the Polish Academy reported that no scaling limitations have yet been encountered in the process, although the growth rates are currently limited by the temperatures and pressures attainable. Further scaleu¥is planned in a next-generation bulk-crystal-growth apparatus. During the rum¥session Sylwester Porowski detailed the exponential increase?crystals u¥to 40 mm2 have been grown?in diameter of bulk-GaN crystals grown in his laboratory over time?a trend many hope the Poles can sustain.
Silicon-based substrates
New work on SiC substrates was presented in invited talks by Robert Davis of North Carolina State University (Raleigh, NC) and Vladimir Dmitriev of Cree Research (Durham, NC). Davis pointed out that the bilayer steps on the SiC surface introduce defects into subsequent overgrowth of grou¥III nitrides. This, and the lattice mismatch, might limit the attainable quality of heteroepitaxial GaN/SiC lasers. Another problem with SiC substrates has been the lack of a conductive buffer layer that would allow vertical device structures to benefit from the dopability of SiC substrates. Cree Research has solved this problem and plans to announce improved GaN/SiC LEDs incorporating conductive buffer layers.
Silicon holds the promise of combining optical devices with integrated
circuits. High-quality GaN grown on silicon has been illusive because it has been difficult to nucleate two-dimensional epitaxy on the silicon surface.
M. Asif Khan of APA Optics (Blaine, MN) announced what might be the first device-quality GaN grown on silicon substrates. Different tricks were employed to realize high-quality GaN. In one approach, a thin GaAs/Si
layer was converted into GaN by exposure to NH3 in the MOVPE reactor.
In another approach, the silicon
surface was converted to SiC by
annealing under propane before GaN heteroepitaxy.
A second European GaN Worksho¥will be organized in June 1997 in
the south of France near Nice by Jean-Pierre Faurie of the CNRS-Valbonne
(Valbonne, France).
SAMUEL (TOBY) STRITE is a research staff member at the IBM Zurich Research Laboratory, Saumerstrasse 4, CH-8803 Rueschlikon/Switzerland; e-mail: [email protected].