Steps to Commercially III-Nitride Device Technology based on Si/PD Heat Extracting Substrates

Project: Research council

Project Details


The high breakdown voltage and large sheet carrier density of GaN based HEMTs provide major advantages for rf and microwave systems owing to their power handling capability. These advantages have also underpinned the emergence of GaN based components for low frequency power electronics. The latter is a major growth area as energy efficiency and sustainability become critical factors in the design of electrical systems. The overwhelming cause for reduced electrical power efficiency in active electronic components and systems is unwanted increases in operating temperature, which degrade power gain in amplifiers, the internal quantum efficiency of light emitting diodes and power conversion efficiency of diode lasers. As an example of the impact device heating on system efficiency, about 70% of the electrical power consumed by mobile telephone transmitter is wasted as heat owing to Joule heating in the electronics and consequent reductions in power gain of its constituent transistors. The most effective way to limit the temperature rise of a semiconductor device is to introduce high-thermal conductivity heat spreading layers as close as possible to its active region, for example over the top of the device or growing the device structure on a thermally conducting substrate. Typically GaN based HEMTs of the type used in rf circuits and high power electronics are grown on SiC or increasingly Si wafers substrates. Whilst SiC is a better thermal conductor than Si, polycrystalline or crystalline diamond are far superior, better even than metals. Recently GaN HEMT grown on crystalline diamond substrates have been recently demonstrated. However, the small size (5 x 5 mm) of current crystalline diamond (PD) substrates and their high cost prohibit this ideal approach. Thus, a polycrystalline diamond substrate offers the best solution. The calculations show that the larger thermal conductivity of polycrystalline diamond could bring to power HEMT performance compatible or better than that on SiC or Si substrates. To date, the most widely investigated method of exploiting PD substrates in GaN power HEMT technology has been to grow the III-Nitride layers on a Si substrate, then transfer the epitaxy to carrier substrate and finally bonding the device layers to the PD wafer. The procedure involves two wafer bonding steps, a process that requires minimal wafer bow if breakage is to be avoided, something that is difficult to achieve owing to the lattice mismatch between Si and III-Nitride materials. There is also a tendency for the final structure to delaminate and despite several years of development by companies like Group4Labs, SOITEC and Nitronex, commercial products are still not established. To overcome these difficulties, an alternative approach has been developed by the Applicants in collaboration with Element 6. Briefly, this involves forming a composite structure comprising a thin layer of Si on a thicker layer of polycrystalline diamond, intimately contacted without wafer bonding. The upper Si surface is suitable for immediate III-Nitride growth. More information is given in section 3. One patent application has already been filed (world wide) and a second is in preparation; in both instances Bath has assigned its rights to Element 6. Independently of Element 6 and other parties, Bath has developed methods for growing III-Nitride hetero-epitaxy on these complex Si/PD substrates. The results of applying these methods to realise high quality III-Nitride epitaxial layers on Si/PD substrates has recently been reported in ICNS9, critical details in the process were not disclosed and thus the opportunity exists to create an intellectual property portfolio covering the realisation of device grade III-Nitride epitaxial films on Si/PD heat extracting substrates to complement the very separate existing IP covering the manufacture of the latter. The new knowledge will be owned in its entirety with the University of Bath.
Effective start/end date1/02/1231/03/13


  • Engineering and Physical Sciences Research Council