1. M. Dauelsberg¹, D. Brien¹, R. Pusche¹, O. Schon¹, E.V. Yakovlev², A.S. Segal², R.A. Talalaev²;
(1) AIXTRON AG, (2)STR Group - Soft Impact Ltd
Investigation of Nitride MOVPE at high pressure and high growth rates in large production reactors by a combined modelling and experimental approach
The scope of this work is the analysis of gas phase processes during MOVPE of GaN at conditions of high pressure, high growth rates and large reactor volumes by growth experiments in a production scale reactor, refine our computational model based on these data and lead the way to capacity scaling while increasing growth rate and pressure at the same time. Usually difficulties arise at the above conditions due to parasitic gas-phase processes that ultimately result in gas phase nucleation, which is critical for deposition efficiency and uniformity on large substrates. An 8x4 inch Planetary reactor was used to investigate the parametric dependencies of gas phase nucleation and its effect on GaN growth rate profiles on stalled wafers. Pressure, MO precursor flow, residence time and wall temperatures were varied independently and over a wide range. Modelling includes the computation of flow, heat transfer and reaction chemistry. High pressure depletion is critically governed by nucleation of an over-saturated low volatile species and subsequent nano-particulate growth. Among our findings there is a critical pressure at otherwise fixed conditions when radial depletion of the growth rate curve increases due to the onset of gas phase nucleation. Likewise, there is a linear relation between mean growth rate on wafer and MO flow up to a point when this relation starts to saturate. Both critical values can be pushed out by increasing gas flow. The effect of increasing wall temperature, however, turned out rather ambiguous due to the competing effects of reduced thermophoretic particle drift to the colder walls and increased parasitic reaction kinetics in the hotter ambient. The modelling approach is validated and the typically obtained prediction accuracy is discussed in view of the complexity of the process. Understanding of the processes that control gas phase nucleation and intensive use of computational modelling enabled us to increase the chamber and optimise gas injection and chamber contours. We were able to obtain growth rates of GaN up to 8 µm/h at p=600 mbar on multiple 6 inch sapphire wafers before saturation started, with thickness standard deviation of 2% and layer quality judged by XRD FWHM.

2. W.V.Lundin¹, A.E.Nikolaev¹, A.V.Sakharov¹, E.E.Zavarin¹, G.A.Valkovsky¹, M.A.Yagovkina¹, S.O.Usov¹, N.V.Kryzhanovskaya¹, V.S.Sizov¹, A.F.Tsatsulnikov¹, N.A.Cherkashin², M.J. Hytch², E.V. Yakovlev³, D.S. Bazarevskiy³;
(1) Ioffe Institute, (2) CEMES/CNRS, (3) STR Group - Soft-Impact Ltd
Single quantum well deep-green LEDs with buried InGaN/GaN short-period superlattice
In spite of the great progress in III-N technology, LEDs with wavelength >535 nm still demonstrate low efficiency comparing to blue and short-wavelength green ones. Here we report on significant improvement of deep-green LED properties by modifications of structure design. The structures were grown in AIX2000HT system with 6 X 2” planetary reactor. An optimized structure consists of 5 μm n-GaN, 12-period InGaN/GaN short-period superlattice (SPSL) with 2 nm period fabricated by InGaN-conversion technique, 25 nm n-GaN barrier grown at reduced temperature (LT GaN), 2.5 nm InGaN QW, 4 nm undoped GaN upper barrier, 15 nm p-AlGaN, and 120 nm p-GaN. It was observed that InGaN/GaN SPSL followed by LT GaN are the key elements of high-efficiency deep-green LED. If InGaN QW is grown directly on the top of high-temperature n-GaN layer, EL efficiency is 15-30 times lower and wavelength is ~ 10 nm shorter in comparison with the optimized structure. HRTEM and HR X-ray reciprocal space mapping were used for structural characterization. It was revealed that the used InGaN/GaN SPSL prevents inheritance of GaN buffer layer mosaic structure by the consequent layers. Moreover, InGaN/GaN SPSL and LT GaN barrier improve LED properties only if implemented together and does not effect if used alone. A special attention will be given to the procedure of InGaN/GaN SPSL formation by InGaN-conversion technique: repeating of 2 nm thick InGaN growth followed by growth interruption (GI) with hydrogen admixing into the carrier gas. During SPSL formation, indium concentration on the surface is governed by an interplay between InGaN decomposition at the stage of GI, indium segregation, desorption, and incorporation into InGaN during subsequent growth. Modeling has been used to study the effect of operating parameters on these processes. Thicknesses and growth conditions of the other layers forming the structure should be carefully optimized too. For example, EL efficiency is very sensitive to GaN upper barrier thickness; p-GaN contact layer should be grown in the hydrogen-free ambient. For the LEDs processed and assembled in a simple flip-chip geometry, external quantum efficiency of 16% (545 nm) and 20% (535 nm) were achieved.

3. J. Stellmach¹, O. Savas¹, J. Schlegel¹, M. Pristovsek¹, M. Kneissl¹, E. Yakovlev²;
(1) Universita"t Berlin, (2) STR Group - Soft-Impact Ltd
AlGaN growth rate and composition in a close-coupled showerhead MOVPE reactor
For deep UV light emitters AlGaN layers with high aluminium content are needed. However, metalorganic vapour phase epitaxy of AlGaN is quite challenging due to gas-phase pre-reactions and the formation of nanoparticles. Therefore, the growth rate and composition depend non-linearly on temperature and growth pressure. The underlying mechanism is the formation of AlN particles, but interactions of gallium species with the particles must be considered under high growth rate conditions. We have investigated the growth of AlGaN layers on sapphire substrates in an Aixtron 3x2” close-coupled showerhead (CSS) MOVPE reactor. In case of a CSS MOVPE reactor the chamber height, i.e. the distance between gas inlet and susceptor, provides an additional growth variable to control gas phase reaction. To prevent gas phase reactions, AlGaN is typically grown at low total pressures to obtain a high velocity of the carrier gases and thus only a short residence time in the gas phase. We found that the chamber height is also a critical parameter to minimize the parasitic reactions. Reducing chamber height from 21 mm to 6 mm results in an increase of growth rate from 0.5 µm/h to 4 µm/h and an increase of aluminium content from 15 % to 50 %. With 6 mm chamber height, the growth rate of AlGaN over the entire composition range exceeded 3 µm/h for standard fluxes. The observed composition and growth rates could be reproduced by two models. The first is an analytical model assuming fixed rate coefficients for TMAl and TMGa loss via first order particle formation, with the rate constants being fitted to reproduce the AlGaN growth rate vs pressure and chamber height. The second model uses the CVDSim software package. It assumes formation of AlN particles and their additional growth at the expense of Ga(CH3)x species. At this stage, the particle growth is kinetically limited by CH3 desorption from the particle surface. Both models fit the data well, but an increase of Al content at chamber heights larger than 18 mm (visible in XRD data) is only predicted by the numerical model including nanoparticle formation.