1. Introduction
As device structures of GaN based materials are produced for commercial application in LEDs and Lasers Reference Nakamura, Mukai and Senoh[1], tools for industrial mass production become necessary. These machines are required to provide material with state-of-the-art characteristics while maintaining high throughput, high reproducibility, high growth efficiency and good uniformity of individual layers for maximal yield of the production line. The AIXTRON multiwafer MOVPE systems with the Planetary® Reactor design are uniquely suited to meet these specifications. The growth processes are continuously optimised regarding total pressure, growth temperature and precursor flows to meet the material quality requirements while the reactor design ensures high yield.
2. Sample Preparation
All GaN based structures discussed here were grown with a GaN nucleation layer on sapphire substrates in the (0001) surface orientation. The layers were produced in a AIX 2000 HT Planetary® Reactor where the uniformity of the layer characteristics is ensured by a two fold rotation of the substrates: up to 7 sattelite disks carrying 2" substrates are rotated by the Gas Foil Rotation® principle while the main disk is turned by mechanical drive. The precursors NH3, TEGa, TMGa, TMAl, TMIn, SiH4 and Cp2Mg are injected in the center of the main disk together with N2 or H2 as carrier gas, seperated in an upper and a lower flow for MO and hydride precursors respectively. Reactor temperatures up to 1200°C and total pressures between 50 and 1000 mbar were used for the growth processes.
3. Electrical characterization
Non intentionally doped GaN layers are semi-insulating or lightly n-type with background electron concentrations below 5·1016 cm−3. Intentional n-type doping is obtained by introducing SiH4 into the gas phase. Free electron concentrations of up to 1·1020 cm−3 have been obtained. Figure 1 shows a topology of the sheet resistance of a highly n-doped 0.5 µm thick GaN layer mapped by a Lehighton inductive meassurement setup. The average sheet resistance is 15.75 Ω/square with a standard deviation of 0.86 %. This indicates excellent doping uniformities taking into account that both thickness distribution and doping homogeneity contribute to this value (see Section 4). Intentional p-type doping has been achieved by using Cp2Mg as precursor for the acceptor. With Mg concentrations of more than 1·1020 cm−3 hole concentrations up to 1·1018 cm−3 are obtained. Uniformities for p-type carrier density with a standard deviation around 10 % have been found, revealing incorporation uniformity as well as influences of the activation process.
Figure 1. Sheet resistance mapping of a 0.5 µm GaN:Si layer on c-plane sapphire: average resistance is 15.75 Ω/square with a standard deviation of 0.86 %
4. Thickness measurement
The layer thickness of single layer GaN on sapphire structures are routinely examined by white light interference evaluation on a Waterloo PLM 100 wafer mapper system. Figure 2 shows a mapping of a single layer GaN on sapphire wafer - the average thickness is 2.44 µm with a standard deviation of 0.75 %. Usual variations for all nitride based materials are in the range of 2 % standard deviation.
Figure 2. Thickness mapping of a GaN layer on sapphire substrate: average thickness is 2.44 µm with a standard deviation of 0.75%.
5. Composition homogeneity of ternary compounds
The Waterloo PLM 100 photoluminescence (PL) mapper was used to study the distribution of the peak wavelength of the bandgap related emission line of ternary layers at room temperature. Structures of GaInN and AlGaN layers on a GaN buffer on sapphire have been thus examined. Figure 3a shows the wafer mapping (left) and wavelength distribution (right) for a GaInN layer with an estimated average of 6 to 8 % InN in the lattice - the average wavelength is 382.44 nm with a standard deviation of 0.94 nm. For a AlGaN layer with roughly 8 to 12 % AlN (S) an average peak wavelength of 340.31 nm and a standard deviation of 0.26 nm were found - see Figure 3b. The standard deviation for emitted intensity is typically less than 10 % for both ternary systems.
Figure 3a. RT PL peak wavelength mapping of a full 2″ wafer GaInN/GaN heterostructure and wavelength distribution: average wavelength is 382.44 nm with a standard deviation of 0.94 nm (distribution bin size is 1 nm)
Figure 3b. RT PL peak wavelength mapping of a full 2″ wafer AlGaN/GaN heterostructure and wavelength distribution: average wavelength is 340.31 nm with a standard deviation of 0.26 nm (distribution bin size is 1 nm)
6. Material property qualification by device fabrication
A simple double-hetero (DH) structure consisting of a 10 nm GaN cap, 50 nm GaInN with ca. 16 % InN (S) and 1.6 µm GaN buffer was cleaved into roughly 1 mm wide pieces and exposed to high intensity optical pumping with a pulsed N2 laser. Reference Woelk, Woelk, Woelk and Woelk[2] At excitation densities up to 1 MW/cm2 the emission from that sample was found to concentrate into a single lasing mode at 78 K, as shown in Figure 4. This lasing action without a Fabry-Perot cavity is evidence of the high optical quality and high gain coefficient of the sample.
Figure 4. stimulated emission and LT laser action by optical pumping with increasing excitation intensity of a simple GaN/GaInN DH structure Reference Yablonskii, Lutsenko, Marko, Schoen, Heuken, Schineller, Gutzeit, Schwambera, Peng Huei and Heime[3]
7. Conclusion
Electrical and optical measurements prove our GaN material and its ternary alloys to be of high quality. Doping and ternary compositions have been shown to have state-of-the-art homogeneity, except for p-type GaN where the growth process and activation of the acceptors require further optimisation. These characterization results combined with the high yield due to the unique design of the AIXTRON Planetary® Reactors prove these systems to be an optimal tool for GaN based LED and - in future - laser mass production.
Acknowledgments
We hereby gratefully acknowledge the work of G. Yablonskii et al., Institute of Physics, Minsk, on the optical excitation experiments quoted above.
