Community Research and Development Information Service - CORDIS


ALMA Report Summary

Project ID: 645776
Funded under: H2020-EU.


Reporting period: 2015-06-01 to 2016-11-30

Summary of the context and overall objectives of the project

In semiconductors and insulators heat transport is mainly mediated by phonons. Although for many decades most semiconductor modeling efforts have concentrated on electronic transport, in recent years phonon thermal transport modeling has become an increasingly growing priority. In many technologies, the problem of heat dissipation stands on the way to further progress. The spectrum of areas affected by heat management is very diverse. Also, in all cases, our ability to solve the challenge relies on being able to understand and predict heat flow in non-trivial systems, often involving micro and nanostructures.

In power electronic devices, such as LEDs or High Electron Mobility Transistors (HEMT), the choice of substrate is crucial for the lifetime of the device. For example, lowering the temperature of the active region in a HEMT by 50 degrees increases the device lifetime by one order of magnitude. Thus, various generations of multi-layer structured substrates have been evolving in recent times in order to increase heat flow, while also satisfying other constraints such as structural integrity and cost. The difficulty in this design process is that modeling thermal transport in these structures is a much more complex problem than for macroscopic systems. The problem in hand may contain simultaneous ballistic and diffusive flow of phonons, which are scattered by complex defect types such as vacancies, dislocations, interfaces, or nanoinclusions, and it may involve novel materials which have not yet been thoroughly explored.

At its source, this complexity stems from the fact that phonon transport is inherently a multiscale problem. This is immediately clear when one realizes that relevant contributions to heat transport in semiconductors simultaneously come from phonons whose lifetimes and mean free paths are spread across more than four orders of magnitude, and which interact amongst themselves and with the structural features in a nontrivial way. Multiscale « predictive » modeling of phonon transport is therefore a complex problem, and also a very urgent one. This motivated us to propose the ALMA project, which started in June 2015.

The originally stated objectives of the project are:
O1. Model extension to mesoscale (TRL 3)
This objective consists in being able to solve the phonon Boltzmann equation for spatially inhomogeneous systems, in a totally ab-initio fashion. This is an extension of the spatially homogeneous model that we had already implemented in our ShengBTE software. Target metrics for the achievement of this objective are predictive accuracy with less than 20% error.

O2. Integration into a modelling software (TRL 4)
This objective aims at integrating the model into a new software that can be used by industry to help in the computer design of new substrates for power electronics. This is the first software of its kind, capable of predicting heat flow in layered multiscale structures comprising novel materials (for which no experiment has ever been performed), and including ballistic and wave-like effects that are not present in standard TCAD modeling software.

O3. Design of new generation substrates for GaN-based power electronics (TRL 5).
The third objective of the project is to use our model and software to design a new generation of substrates for power electronics. The designs should enable the operation of new power transistors, under highly demanding conditions: 1200 V, 10 A, on 4 cm2. The substrate needs to ensure enough heat flow to keep the transistor temperature below 175°C.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

The main results pertaining to each of the stated objectives are:

Model extension to mesoscale:
The model extension has been attained, with the completion of WP1 and WP2 during the first 18 months. Experimental validation of the model started from month 1, and will continue until month 30. We have validated results on thin films, two- and three-dimensional materials with defects, and superlattices, amongst others. So far the model fulfills our accuracy target of less than 20% error. Six papers have been published, four submitted and four are in preparation.

Integration into a modelling software:
This objective targets a professional level software package, user interface, and documentation. These three targets have been attained within WP3 (completed) and WP4. In particular: the program is fully written in C++, currently comprising over 85000 lines of code. It comes with a full set of automated unit tests, and complete code documentation via Doxygen. There is a brief user manual, and a set of test case examples for each of the executables. The code has been tested on Linux clusters, macOS platforms, and even on Windows with the help of a Docker container. Moreover, a database of ab-initio materials input files has been made available online, containing the most relevant semiconductors used in the electronics industry. The database is in continuous expansion. The software will continue to be improved and extended as part of WP4, particularly regarding its graphical capabilities, until the end of the project.

Design of new generation substrates for GaN-based power electronics:
This is the goal of WP6, which is scheduled to begin on month 31.

The 6 main technical workpackages of the project can be grouped into three categories, regarding their scope:
• Scientific modeling development and experimental validation: WP1 (complete), WP2 (complete), and WP5 (ongoing).
• Software development: WP3 (complete) and WP4 (ongoing).
• Industrial application: WP6 (not yet started).
Another two workpackages on coordination (WP0) and dissemination (WP7) run in parallel throughout the project.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

AlmaBTE, created by this project, is the first public software of its kind, capable of predicting heat flow ab-initio in layered multiscale structures comprising novel materials (for which no experiment has ever been performed), and including ballistic and wave-like effects that are not present in standard TCAD modeling software. AlmaBTE can accelerate the design and development of new devices in technologies were thermal dissipation constitutes a bottleneck, for example power electronic devices.

We have already validated the model onto which AlmaBTE is based, by modeling various systems for the first time and comparing them with experimental measurements. This includes:

-Effect of anti-site defects on half-Heusler materials
-Influence of As defects on thermal conductivity of novel compound BAs
-Thin film thermal conductance versus thickness
-Thermal transport through suspended black phosphorus flakes
-Thermal conductivity of InGaAs superlattices
-Thermal transport through suspended germanane flakes
All the above systems are highly relevant for industrial applications.
Other systems are currently under investigation.

Related information

Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top