Opendata, web and dolomites

Report

Teaser, summary, work performed and final results

Periodic Reporting for period 2 - GaNonCMOS (GaN densily integrated with Si-CMOS for reliable, cost effective high frequency power delivery systems)

Teaser

The total energy consumption in the world is projected to increase by 25- 35 % over the next 15 years , as up to 3 billion people will enter the global middle class and the world economic output will double. Energy efficiency is an essential component of any strategy to...

Summary

The total energy consumption in the world is projected to increase by 25- 35 % over the next 15 years , as up to 3 billion people will enter the global middle class and the world economic output will double. Energy efficiency is an essential component of any strategy to deliver affordable and reliable energy systems. Power electronics is the key technology to control the flow of electrical energy from the source to the load for a wide variety of applications. The power electronics industry now deals with conversion and motion, and thus requires lighter/smaller, cheaper and more efficient systems. To address these requirements silicon (Si) semiconductors have traditionally been employed in power circuits. However, silicon device technology is approaching its material limitations. As alternative materials, wide bandgap semiconductors (WBG) such as GaN and SiC are used to operate at higher voltages, temperatures, and switching frequencies with greater efficiencies. GaN is expected to be the dominant WBG semiconductor replacement for silicon for applications requiring device ratings less than 600 Volts. GaN is currently widely used in LEDs and RF amplifiers, and its emergence in power electronics is relatively recent. The most cost effective method for fabricating GaN power devices is GaN on silicon semiconductors. Improvement in GaN-on-Silicon epitaxy techniques in the last years, allowed major development for the creation of substrates with less impressive properties but at orders of magnitude lower cost than GaN on bulk SiC or GaN substrates.

The overall objective of the GaNonCMOS project is to develop novel low cost and reliable GaN-based process, components, modules and integration schemes, and demonstrate their performance and economic potential on system level for significant energy reduction in a wide range of energy intensive applications such as data centres, automotive and industrial applications.

The project aims to bring GaN power electronic systems to the next level of maturity by providing the most densely integrated systems to date. The key innovations steps are to integrate GaN power switches with CMOS drivers densely together using different integration schemes. GaN materials stack and device layout will be optimized to enable fabrication of normally-off devices for low temperature integration processes. New soft magnetic core materials will be developed reaching defined switching frequencies with ultralow power losses. New materials and methods for miniaturised packages will allow GaN devices, modules and systems to operate under maximum speed and with high energy efficiency.
Finally, long term reliability improvements will be achieved over the full value chain of materials, devices, modules and systems with consortium partners that cover the entire value chain.

Work performed

The focus of the GaNonCMOS project so far was to define the system specifications for each of the different voltage regulator modules (VRM) demonstrators, based on the complete set of application requirements. To achieve this, a study of the state of the art landscape, based on these requirements, was performed. This resulted in an extended list of demonstrators to be built in this project, including their required electrical specifications.
Secondly, in order to develop a VRM system level model, this model needs to have optimised power conversion topology. The overall VRM system involves a number of components i.e. power transistors, capacitors, magnetic components and control IC. Hence, the optimization of the system performance in relation to the trade-off between efficiency and power density involves a very large number of design parameters. The approach was to use validated analytical and compact models for each of the system components so that the overall system performance could be optimised. The design of the GaN power switches, the magnetic components and the CMOS driver and control circuits was undertaken and finalized during the second stage of the project.

At the materials levels, new magnetic materials and processes to deposit such materials for application in high frequency magnetic cores were developed and tested. Thin films of new soft magnetic materials were grown and characterised. The epitaxial material to fabricate the GaN switches were successfully fabricated, processed and wafers were characterized to evaluate their static and dynamic performance.

At the components level, a functional embedded inductor demonstrator was manufactured and tested for a realistic application. Simulation, calculation and measurement results showed a good match. Furthermore, the fabrication of BiCMOS wafers is ongoing, with driver and control circuits newly developed, tested and further adapted to fit all the different demonstrators.

Final results

The GaNonCMOS project will progress the current state of the art of different technologies, which are situated at three levels, namely at VRM concept and materials and devices level.

Granular dynamically variable multiple voltage domain VRMs are needed to improve the energy efficiency of all computing systems. This has been addressed by various attempts at building on-chip VRMs. These VRMs consist of buck-converters that are built using CMOS power switches and controllers on the CPU chip with the large passives (inductors and capacitors) built off-chip. The current VRM prototypes are limited to a specific voltage conversion ratio because of the breakdown characteristics of the CMOS transistor. While CMOS device scaling is attractive from a VRM perspective, they also suffer from ever lower breakdown voltages limiting the voltage conversion ratios. In this context, GaN switch can be a game changing device that enables large voltage ratio conversion due to its larger breakdown voltage. Within GaNonCMOS we will explore: innovative fabrication techniques that bring the GaN devices close to CMOS chips or embedded within them, packaging techniques that enable passive components embedded within laminate along with new magnetic core materials.

At the materials and devices level, progress beyond the state of the art is targeted at the power switches and the magnetic materials. Regarding the power switching, GaN on Si is currently typically used for devices targeting the 650V node, operating at frequencies between 50kHz and 700kHz. The epiwafers used for the 650V node have a destructive breakdown voltage rating that is several hundreds of Volts higher than the operating voltage. The current project requires GaN switches and epiwafers that combine a medium breakdown voltage with very low sheet resistance. The GaNonCMOs consortium will make a significant step forward in the development of ultra-low loss switches for specific low voltage applications using very low sheet resistance epiwafers as well as minimized ohmic contact and pad areas. Regarding the magnetic materials, integrated passives with magnetic cores have been fabricated for a long time, however existing commercial solutions are based on air-core technology. This highlights the challenges to integrate a magnetic core using standard MEMS processing. A key aspect of this challenge is the development of high performance, reliable magnetic films with high flux density to achieve miniaturization. The GaNonCMOS project proposes to develop novel nanocrystalline soft magnetic thin film materials and new integration techniques for realising a high performance passive device.

Website & more info

More info: http://www.ganoncmos.eu.