Épitaxie | EPFL Graph Search

L'épitaxie est une technique de croissance orientée, l'un par rapport à l'autre, de deux cristaux possédant un certain nombre d'éléments de symétrie communs dans leurs réseaux cristallins. On distingue l'homo-épitaxie, qui consiste à faire croître un cristal sur un cristal de nature chimique identique, et l'hétéro-épitaxie, dans laquelle les deux cristaux sont de natures chimiques différentes. Étymologiquement, « épi » en grec signifie « sur » et « taxis », « arrangement ». Principe L'épitaxie est utilisée pour faire croître des couches minces, de quelques nanomètres d'épaisseur. Pour cela des atomes sont déposés sur la surface parfaitement polie d'un monocristal, le substrat. Le substrat est choisi de façon à avoir des paramètres de maille proches de ceux du cristal qui souhaite être obtenu. Il existe plusieurs techniques d'épitaxie :

l'épitaxie par jet moléculaire ;
l'épitaxie en phase liquide, dont : ** la méthode de Czochralski ; ** la méthode de Bridgman-Stoc

Selective area epitaxy of GaAs: the unintuitive role of feature size and pitch

Akshay Balgarkashi, Elif Nur Dayi, Didem Dede, Anna Fontcuberta i Morral, Martin George Friedl, Lucas Güniat, Nicholas Paul Morgan, Valerio Piazza

Selective area epitaxy (SAE) provides the path for scalable fabrication of semiconductor nanostructures in a device-compatible configuration. In the current paradigm, SAE is understood as localized epitaxy, and is modelled by combining planar and self-assembled nanowire growth mechanisms. Here we use GaAs SAE as a model system to provide a different perspective. First, we provide evidence of the significant impact of the annealing stage in the calculation of the growth rates. Then, by elucidating the effect of geometrical constraints on the growth of the semiconductor crystal, we demonstrate the role of adatom desorption and resorption beyond the direct-impingement and diffusion-limited regime. Our theoretical model explains the effect of these constraints on the growth, and in particular why the SAE growth rate is highly sensitive to the pattern geometry. Finally, the disagreement of the model at the largest pitch points to non-negligible multiple adatom recycling between patterned features. Overall, our findings point out the importance of considering adatom diffusion, adsorption and desorption dynamics in designing the SAE pattern to create pre-determined nanoscale structures across a wafer. These results are fundamental for the SAE process to become viable in the semiconductor industry.

IOP Publishing Ltd2022

Additive Fabrication of Silicon Pillars on Monocrystalline Silicon by Direct Laser Melting

Marie Camille Laëtitia Le Dantec

Additive Manufacturing (AM) is an emerging part production technology that offers many ad-vantages such as high degree of customization, material savings and design of 3D highly complex structures. However, AM is a complex multiphysics process. Therefore, only a limited number of materials can already be commercially used to produce parts and a handful of others are being studied or developed for such process. Consequently, limited knowledge on this process is available, especially concerning materials that present thermomechanical challenges such as brittle materials. The present research focuses on additive fabrication of silicon pillars on a monocrystalline silicon wafer by Direct Laser Melting (DLM) with a pulsed 1064 nm laser beam. The simple geometry of pillars allowed for the first determining steps into process understanding. Several results were achieved through this PhD work. First, crack-free silicon pillars were successfully built onto monocrystalline silicon wafers. With the help of in-situ process monitoring and sample characterization, wafer substrate temperature and laser repetition rate were found to be the main influential parameters to obtain crack-free samples, as minimum substrate temperature of 730°C and a minimum repetition rate of 100 Hz were necessary to reach this goal (for a feed rate of 15 g/min and a pulse duration of 1 ms). The influence of secondary process parameters such as feed rate and energy per pulse were also discussed. A simple Finite Element Modeling (FEM) model validated by the experiments was used to explain crack propagation in the samples. Then, process monitoring of the DLM process was realized. High-speed camera image analysis re-vealed that vertical stage speed and powder feed rate should match to obtain a constant pillar building rate. As all pillars presented necking at their base, estimations of the thermal characteristics of the pillar during growth were carried out by FEM simulations. They were additionally used to explain the pillar final shape. Finally, the microstructure of the pillars built was characterized by the Electron Back-Scattering Dif-fraction (EBSD) technique. In the conditions presented in this work, the microstructure of the pillar was found to be in the columnar growth mode. The feed rate was identified as the most influential parameter on the microstructure, followed by the stage speed, the impurity content of the powder and the crystallographic orientation of the substrate. Epitaxial growth was achieved on more than 1 mm with a feed rate of 1.0 g/min, a stage speed of 0.1 mm/s, a powder with purity of 4N and a oriented wafer substrate. This work could be further continued by making improvements to the DLM setup, studying the influence of additional process parameters on the thermomechanical behavior and the microstructure control of the pillars, and/or using these results to realize more complicated shapes, either with this setup or by using a powder bed technique.

EPFL2018

An Experimental Setup for Heterogeneous Catalysis on Atomically Defined Metal Nanostructures

Jean-Guillaume De Groot

The main objective of this PhD thesis is to link the atomic scale structure to the catalytic properties of self-assembled nanostructures. The growth and characterization of these nanostructures was carried out in an ultra-high vacuum (UHV) chamber initially designed for the study of the kinetics of epitaxial growth of thin films and nanostructures by variable temperature scanning tunneling microscopy. During this PhD thesis, we built a very sensitive and selective gas detector based on a quadrupole mass spectrometer and adapted to the geometry of this UHV system. The proper functioning of the gas detector is confirmed by temperature programmed desorptions of Xe on Ag(100) which demonstrate the very high sensitivity of the device of about

10^{-6}

~monolayer (ML). The zero-order desorption parameters obtained for the first monolayer of Xe adsorbed on Ag(100) are

E_d=0.22-0.23

~eV with a prefactor

\nu_0

between

4\times10^{12}

^{-1}

and

1 \times10^{13}

^{-1}

. An exchange process between Fe adatoms and Ag atoms from the Ag(100) surface is identified by means of temperature programmed desorption of N

_2

. Desorption spectra suggest that the exchange is complete from an annealing temperature of 140~K. Our study of the catalytic activity of metal nanostructures focuses on the ability of passivated Fe clusters to adsorb N

_2

molecules without dissociating them, which is a fundamental step in the bio-inspired heterogeneous catalysis of ammonia. The first system investigated for the growth of Fe clusters is MgO/Ag(100). Diffusion of Fe adatoms deposited on a MgO monolayer is observed from an annealing temperature of 280~K onwards. Fe clusters with an average size of

12.1\pm 0.5

~atoms are obtained on a MgO monolayer by thermal ripening of 0.072~ML of Fe deposited at 40~K. No sign of N

_2

adsorption was observed on these clusters by thermal programmed desorption investigation. The second system chosen for the growth of Fe clusters is graphene/Ir(111) obtained by chemical vapor deposition. The deposition of 0.3 and 0.4 ML of Fe on a graphene/Ir(111) sample at 50~K, followed by annealing at 300~K, generates a superlattice of Fe clusters which matches the periodicity of the moiré formed by graphene/Ir(111). The theoretically expected mean cluster size is 26 atoms for 0.3~ML of Fe, and 35 atoms for 0.4~ML of Fe. The temperature stability of the clusters obtained with 0.4~ML Fe is investigated from 300~K to 600~K. It is established that Smoluchowski ripening takes place as the annealing temperature increases and the superlattice structure has completely disappeared after annealing at 600~K. Temperature programmed desorption of molecularly adsorbed N

_2

at 50~K on a sample with 0.4~ML of Fe deposited on graphene/Ir(111) reveals 3 desorption peaks at 120~K, 92~K, and 60~K. In addition, slow dissociation of nitrogen molecules at 50~K adsorbed on the higher energy adsorption sites is demonstrated.

EPFL2021