Combined Epitaxy Optimizations for GaN on Silicon Microwave Power Devices – GoSiMP

High frequency/high power microwave devices can contribute to answer to the increasing demand in terms of security, sensors and communications. The aim of this project is to demonstrate GaN based devices on Silicon substrate with high output power density and efficiency as well as reduced propagation losses opening the way to the fabrication of Monolithic Microwave Integrated Circuits (MMICs). Substrate choice is motivated by the low cost and the availability of Silicon in large size, potentially leading to process cost reduction. However, the starting material necessarily needs to satisfy requirements in terms of structural and electrical quality, which are still yet difficult to reach in production environment. Indeed, Metal-Organic Chemical Vapor deposition (MOCVD) is the preferred tool for the crystal growth of GaN. It has led to high-quality GaN, large size production tools (up to several times 8 inch wafers capacity) but the high temperature involved in the growth process is a drawback for the growth on Silicon which necessitates trade-offs depending on the target application. A first objective will be to develop such knowhow not available in the literature to demonstrate MOCVD grown GaN high electron mobility transistors (HEMTs) on Silicon for power applications at 40 GHz and beyond. First, the project will focus on the reduction of microwave propagation losses through GaN based buffers and various templates on highly resistive silicon. An optimization loop will be necessary to satisfy this criterion as well as others such as crystal quality and low strain to avoid cracks in the films. The project is leaded by CRHEA which is in charge of crystal growth. With the help of its partners GREMAN and IEMN, the critical interface between the nucleation layer and the substrate will be studied. The growth conditions for the nucleation layer will be monitored to satisfy the requirements for a high electrical resistivity interface and buffer layer. The state of the art results previously obtained with the molecular beam epitaxy (MBE) technique will be a reference for this work. More, nucleation at low temperature with MBE will be investigated as an alternative solution. Compared to Silicon, the larger band gap and the chemical inertness of cubic Silicon Carbide are advantages. The alternative solution consisting in an intercalated Silicon Carbide buffer layer grown on Silicon will be evaluated. To benefit the advantage of the small transit time of a large carrier density under a short gate (less than 100 nm) and then to get a good aspect ratio, the transistors must present a small distance separating the carriers from the gate (thin barrier) as well as small access resistances. Today, the trade-off between barrier thinning and resistance increasing has led to thicknesses of the order of 15 nm and sheet resistances around 300 Ohm/sq in the case of AlGaN barrier. This induces a noticeable access resistance in the device which becomes the main limitation for operation beyond 30 GHz. The aim of the second task of the project will be to develop solutions to mitigate this limitation. To do so, the Aluminum content in AlGaN and InAlGaN barriers will be increased to enhance the carrier density and to reduce the sheet resistance. The barriers will be protected by a GaN cap or by in-situ grown SiN able to passivate the surface. More, in order to further reduce the access resistances, N+ doped GaN contacts will be selectively regrown by MBE in growth windows after etching of the barrier. This will avoid the high temperature annealing of the contacts and then permits to get abrupt edges of the ohmic contact, facilitating the fabrication of transistors with gates very close to the source. In order to validate the HEMT structures grown with low propagation losses, IEMN will perform the technology developments and the device electrical characterizations, including the load pull measurements of power performance at 40 GHz and beyond.