As smartphones become the dominant mechanism for information transfer and processing in modern society, our expectations on what we hope to achieve with them also increases proportionally. In particular, the smart phone has become our portal to the internet, replaced our television, radio and music devices, and also serves as our credit card and personal guide (GPS). We also expect our mobile phones to work seamlessly as we travel across international borders. All of this is enabled by the separation of the various functions into different wireless (RF) frequency bands, and the development of sophisticated analog and digital circuitry, that enables the phone to simultaneously carry out these communications. As we move towards 5G and other technologies that increase the data throughput available, these channels must increase. While on the digital signal processing side, the steady advance of Moore's law and microelectronic integration has enabled silicon technology to keep up with the demand, this is not the case for the RF front-end circuitry, which is primarily analog. The RF front-end circuit, receives the signal from the antenna and separates it into different channels (based on RF filters), amplifies it with a low noise amplifier (LNA) and then hands it over to the DSP for baseband signal processing. Currently, RF filters and LNAs are primarily discrete devices that are co-packaged together. While this hybrid approach has certain advantages (mainly the choice of piezoelectric materials for the filter), as demand for filters continuously rises, it is known that a co-packaging approach will not scale. The main reason is that the available smartphone footprint (in terms of chip area) for the RF front-end has remained roughly the same across generations, while the filtering demand has continuously increased. As the microelectronics industry has repeatedly taught us, monolithic integration is the only long-term solution to address these problems. In this project, we will demonstrate that gallium nitride (GaN) is the ideal platform for achieving monolithic integration by exploiting a key advantage that GaN provides over traditional solutions: acoustic waveguiding. GaN allows us to guide high-frequency sound on the surface of chip with low acoustic attenuation. By routing sound in nanoscale waveguides and localising it in micron-scale resonators, one can re-design RF system components from the ground up realizing a massive reduction in component footprint, which is key to enabling monolithic integration. By applying ideas from integrated photonics to high-frequency acoustics, we hope to realize for RF systems the same benefits (in terms of size, weight and performance) that silicon photonics has achieved for optical telecommunication systems. We will show that high quality RF passive devices (in particular, piezoelectric resonators and filters) can be built on the same GaN substrate as the active transistor devices. We will implement a process flow and design the associated process development kit to implement these ideas in commercial GaN RF foundries (for ex: the Newport wafer fab) in collaboration with our project partners.

AlGaN/GaN high electron mobility transistors (HEMTs) are a transformative technology for high-power density radio frequency applications, including radar, satellite and mobile communications. In addition, efficient power conversion systems based on GaN devices are a key enabling technology for the low carbon economy, including renewable energy generation and transport electrification. However, their full potential has not yet been realised because performance is de-rated to ensure stable long-term device operation. Experimental characterisation of the electric field distribution in these devices has been lacking, despite being identified as a primary driver of degradation phenomena including breakdown, charge trapping and self-heating. These processes occur in and around the device channel and particularly the sub-micron region under the gate and field plate where peak electric fields are located. The aim of this proposal is a step-change in electric field imaging of semiconductor devices, by developing an optical three dimensional (3-D) device analysis technique with nanometre-scale spatial resolution. The primary focus will be on electric field induced second harmonic generation (EFISHG) combined with solid immersion lenses (SILs). This will enable us to investigate key performance and reliability challenges including (i) the effect of buffer doping on the dynamic distribution of charge in the device layers which causes an undesirable memory effect, (ii) optimization of field plate geometry to manage peak electric fields, (iii) comparing electric field distributions during RF and DC operation to improve reliability forecasts. These are on the critical pathway to achieving a high performance reliable GaN HEMT device technology which exploits the full benefits of the material properties of GaN.