Skip to main content

Optimal SIC substR ates for Integrated Microwave and Power CircuitS

Periodic Reporting for period 3 - OSIRIS (Optimal SIC substR ates for Integrated Microwave and Power CircuitS)

Reporting period: 2017-05-01 to 2018-11-30

OSIRIS project, Optimal SIC substRate for Integrated Microwave and Power CircuitS, a Research and Innovation Action (RIA), aims at improving substantially the cost effectiveness and performance of gallium nitride (GaN) based millimetre wave components. It will also study the interest for SiC power electronics applications. Indeed power electronics and microwave components based on SiC and GaN are more and more used by end-users and customers as they offer high electrical efficiency. This interests for instance e-cars or hybrid automotive, solar cells or wind turbine and emerging 5G telecom markets. Improving cost and efficiency will help going to a greener society.
At the dawn of the project, it is proposing to elaborate innovative SiC material using isotopic sources. Initial data showed that this material would offer thermal conductivity improvement of 30%, which is important for devices dissipating a lot of power, in particular in SiC power electronics and in microwave device using GaN high electron mobility transistors (HEMT) grown on SiC semi-insulating (SI) substrates. For microwave GaN/SiC HEMT this isotopic approach could create a complete shift in currently used substrate / GaN epi-wafer technology; it intended to grow high thermal conductivity (+30%) semi-insulating SiC on top of low cost semiconducting SiC substrates (widely used by the power electronics and LED industries). Reduced layer thickness is necessary as only the top 50 to 100µm SiC wafer are useful as expensive substrates are currently thinned to realize microstrip waveguided microwave circuits.
For power electronics isotopic innovation is focused on thermal improvement, i.e. better electron mobility at a given power dissipation as mobility and drift mobility decrease with temperature and also better carrier transport thanks to lower scattering rates. Schottky and p-i-n diodes will be tested using this material.
Improved thermal SiC properties will be obtained by using single isotopic atoms for silicon and carbon, namely 28Si and 12C. The SiC wafer size has been targeted to 100mm (4-inches) which is today widely used on industry.
The consortium is highly complementary, covering all skills required to achieve the project objectives, from the growth of crystal materials to the assessment of devices.

A project website is available at http://osiris-ecselju.eu
The main project final achievements are the following:
* Isotopic separation has been studied and suitable 28Si separation cost effective technique was determined at lab scale,
* Excellent SiC epitaxial layers were obtained on off-axis SiC substrates with high electrical resistivity,
* Development of high quality SiC epitaxy on to on-axis substrates without 3C inclusion for GaN epitaxial growth was achieved,
* High electrical resistivity of on-axis SiC epilayers and tailored resistivity for PiN and Schottky diodes with or without isotopic SiC have been obtained,
* Isotopic thermal conductivity: Following initial value obtained by Linköping University, isotopic SiC thermal characterisation showed an improvement of 23.6 % in a-direction and 16.7 % in c-direction instead of the initially foreseen 30%, It was determined that the isotopic heavy Si atom purity has the highest impact on thermal conductivity while lighter C atoms have a lower impact,
* The full chain of metrology was used and very good coherence was obtained between thermal conductivity assessment, device thermal characterisation by Raman as well as Infrared microscopy by simulations,
* OSIRIS project has allowed to improve the material and device characterization by different means especially for Intraspec Technologies, CIMAP, Linköping University and STUBA,
* Device processing shows that new epitaxial SiC material is compatible with SiC diodes and GaN HEMT. Manufacturing yield has not been impacted strongly by using isotopic material, especially in case of SiC diodes. Additional work would have to be carried out on a much larger set of samples to go to obtain statistical figures,
* GaN thermal resistance of 0.15µm gate length HEMT devices grown on semi-insulating 70 µm thick isotopic SiC for 30GHz telecom application has improved by 10%, which is in line with thermal modelling. The GaN buffer acts as a thermal barrier which screened the 23.6% isotopic SiC thermal improvement,
* Devices grown on isotopic SiC emitting CW 2.8W/mm at 30GHz with 32% power added efficiencies were obtained. The maximum output power reached 10W/mm which are record values. The power performance improvement is finally marginal in agreement with limited 10% thermal resistance improvement,
* GaN/isotopic HEMT reliability data is more favorable as 10% thermal improvement could give rise to 7 times longer lifetime,
* 2.5A 10kV pin diodes have been obtained with both natural and isotopic SiC material. No advantage but thermal one (16% for the vertical thermal conductivity and 33% for the lateral conductivity) was observed. Packaging itself is a limiting factor to get all benefits of the isotopic approach,
For the dissemination activity, STUBA has designed and implemented an e-learning site supporting OSIRIS project and CIMAP has developed a Website for OSIRIS project (http://osiris-ecselju.eu). 11 articles have been published. OSIRIS project dissemination get a steep increase to 2290 views at project completion at the end of 2018, which shows the great interest for this research,
* Main Osiris objectives have been fulfilled. Additional work would have to be undertaken for industrialization, especially by producing at competitive cost 28Si isotopes.
The novelty of the proposed project is to utilize an ingenious way of increasing thermal conductivity of substrate and hence power efficiency of diodes and HEMT by removing all but one of the isotopes for Si and C, respectively. Thermal conductivity of SiC substrates will make a significant leap by about 23.6%.
Another novelty of this proposal lays in fabrication of isotope enriched thick SI and doped epitaxial layers on on-axis host SiC substrates which have never been done before.
For microwave application, host substrate will later be polished away leaving only enriched SiC layer with processed GaN device on top. This may seem an awkward manner of doing it, but typical practice today when manufacturing GaN HEMT devices is to polish off most of the substrate to define microstrip waveguide. Considering the difficulty to produce SI substrates, this technique of depositing a high quality SI layer on a low-cost – high volume since n-type wafers would perhaps not revolutionize but definitely induce a paradigm shift for high frequency industry in general as money can be saved. If successfully implemented, this technique is intended to override the SI substrate business for high frequency components.
Finally, isotope enrichment of silicon is not an established industry. In this proposal a more production friendly method at low cost will be developed.
OSIRIS will contribute to develop innovations that meet the needs of European and global markets; in particular:
• Higher efficiency energy conversion (solar, automotive, wind turbine, base station),
• Higher bit rate data transmission, telecom (e.g. 5G market),
• Transport (e.g. radar applications).
GaN MMIC circuit