Back to overview

Rapid lateral crystallization of Ge1-xSnx films, overcoming underlying substrate lattice mismatch and Sn solubility.

English title Rapid lateral crystallization of Ge1-xSnx films, overcoming underlying substrate lattice mismatch and Sn solubility.
Applicant Fontcuberta i Morral Anna
Number 190289
Funding scheme Spark
Research institution Laboratoire des matériaux semiconducteurs EPFL - STI - IMX - LMSC
Institution of higher education EPF Lausanne - EPFL
Main discipline Material Sciences
Start/End 01.12.2019 - 31.08.2021
Approved amount 98'720.00
Show all

Keywords (5)

flash lamp annealing; GeSn; crystallization; heterogeneous integration; thin films

Lay Summary (French)

Ce projet a comme but d’optimiser la cristallisation des couches amorphes de GeSn pour leur application dans des détecteurs infrarouges. Nous envisageons d’appliquer des techniques de ‘machine learning’ pour raccourcir l’optimisation. Ce projet devrait permettre d’atteindre des matériaux très prometteurs pour la technologie LIDAR et des caméras infrarouge.
Lay summary

La qualité crystalline des semiconducteurs a une influence directe sur leur fonctionnalité, donnée par leurs propriétés optiques (absorption/émission de la lumière) et transport électrique (conductivité). La fabrication de matériaux en couches minces permet d’économiser en ressources. Le dépôt de couches minces monocristallines est seulement possible lorsque le substrat a une structure extrêmement similaire. Dans le cas contraire, des défauts se développent en détriment des propriétés fonctionnelles.

L’alliage GeSn est un matériau idéal pour la détection en infrarouge moyen, ce qui peut être utile pour des caméras infrarouges et/ou des technologies LIDAR. Ce matériau est bien plus avantageux économiquement que le concurrent InGaAs, qui utilise des éléments rares sur la croute terrestre. Le ‘challenge’ de ce matériau est la faible solubilité du Sn (<1%).

Dans ce projet nous voulons cristalliser des couches amorphes des alliages GeSn en appliquant un processus de recuit ultra-rapide avec des lasers ou avec lampes flash. Ce processus permet de cristalliser ce matériau en piégeant le Sn et donc il devrait nous permettre d’obtenir des couches avec un haut pourcentage de Sn. L’optimisation des conditions de recuit se fera avec une routine ‘machine learning’. Nous pensons que les techniques développées dans le cadre de se projet pourront s’étendre à d’autres domaines d’optimisation de la fabrication/processing des matériaux.

Direct link to Lay Summary Last update: 25.11.2019

Responsible applicant and co-applicants



Fabrication of Single Crystalline InSb‐on‐insulator by Rapid Melt Growth
Menon Heera, Morgan Nicholas Paul, Hetherington Crispin, Athle Robin, Steer Matthew, Thayne Iain, Fontcuberta i Morral Anna, Borg Mattias (2021), Fabrication of Single Crystalline InSb‐on‐insulator by Rapid Melt Growth, in physica status solidi (a), xx(xx), xx.
Cubic, hexagonal and tetragonal FeGe x phases ( x = 1, 1.5, 2): Raman spectroscopy and magnetic properties
Kúkoľová A., Dimitrievska M., Litvinchuk A. P., Ramanandan S. P., Tappy N., Menon H., Borg M., Grundler D., Fontcuberta i Morral A. (2021), Cubic, hexagonal and tetragonal FeGe x phases ( x = 1, 1.5, 2): Raman spectroscopy and magnetic properties, in CrystEngComm, 23(37), 6506-6517.


The crystalline quality of semiconductors relates directly to their functional properties such as charge carrier mobility and radiative recombination efficiency. Compared to bulk substrates, thin film technology provides the path of combining different materials for an increased functionality and a strong reduction in material utilization. Monocrystalline thin films may grow seamlessly on crystalline substrates, which provide the crystallographic orientation as well as mechanical support. Defects on the film appear when lattice constants between the two become too important. To circumvent this, a buffer layer may be employed to relax the strain and limit the defect formation close to the interface with the substrate. Alternatively, limiting growth at the nanoscale also results in single crystalline structures. In a hybrid strategy, growth can start in nanoscale regions that then overgrow laterally forming a thin film in the so-called lateral epitaxy method. Alternatively, amorphous thin films can be crystallized in a heating pulse or cycle by applying a laser pulse or an ultra-fast broad-band flash lamp. This strategy is used for example to re-crystallize amorphized regions of devices after ion implantation. These processes can also be applied to amorphous layers on glass, for example in transistors addressing pixels in flat panel displays. Critical to the final quality layer is limiting the nucleation to a low density so that the crystallized grains can be as large as possible. This reduces the amount of grain boundaries, resulting in improved functional properties. The goal of this project is to obtain crystalline thin films of metastable Ge1-xSnx semiconductor alloys on a silicon substrate. Ge1-xSnx with x>0.06 exhibits a direct bandgap in the mid-infrared, rendering this material interesting for optoelectronic and photo-detection applications in LiDAR, imaging and biosensing. Our approach combines the selective initiation of crystallization at nanoscale regions and the use of ultra-rapid annealing cycles. The amorphous Ge1-xSnx layer is first deposited on a patterned oxidized Si substrate, with nanoscale openings to the crystalline Si. Prior to the Ge1-xSnx deposition, the oxide is covered by a graphene layer. A few-ms flash annealing pulse is applied to trigger the crystallization. Crystallization should preferentially start in the nanoscale openings, adopting the crystalline orientation of the substrate. We will look for conditions in which crystallization then proceeds out of the nanoscale openings to laterally crystallize the whole film. The high speed of the thermal process (few ms) should suppress the diffusion of Sn out of the structure. The graphene should foster a higher selectivity of the crystallization process, compared to just using an oxide as a mask. The crystalline properties and functionality of the layers will be first scanned by Raman spectroscopy and X-Ray diffraction. The best samples will in addition be investigated in more detail by high-resolution transmission electron microscopy. The conditions leading to the best crystalline quality will be found by a machine-learning design of experiment approach. The routine will optimize the linewidth of the (004) diffraction peak. Crystalline Ge1-xSnx layers constitute a real alternative to much more costly and less sustainable materials absorbing in the mid-infrared such as InGaAs and lead-based compounds. In addition, we believe this process will be of general nature and will open new avenues for the defect-free thin film formation of other materials systems.