chaired by Wilfred van der Wiel & Willem Vos
Using silicon for electronics and optics
Boolean logic with disordered dopant systems in silicon
Tao Chen (NE)
Fabrication of three-dimensional silicon nanostructures for nanophotonics
Cock Harteveld (COPS)
Light from silicon
Satadal Dutta (SC)
Epitaxial integration of perovskite oxides on silicon
David Dubbink (IMS)
Boolean logic with disordered dopant systems in silicon, Tao Chen (NE)
Artificial neural networks (ANNs) are computing models inspired by biological neural networks, such as the human brain. With a proper training strategy, ANNs can outperform program-based computing in certain tasks such as pattern recognition. ANNs are mostly emulated on conventional, Von Neumann computers, which are intrinsically not suitable for this task and hence energy inefficient. Therefore, hardware implementations are sought for to overcome such shortcomings1,2.
We have found that charge transport in disordered boron dopant networks in silicon is highly tunable by peripheral control voltages at 77K. All Boolean logic gates can be evolved with a genetic search algorithm, similar to those Boolean logic evolved at 0.3K in a cluster of ligand-covered gold nanoparticles demonstrated by our group before3. Temperature-dependent measurements indicate Efros-Shklovskii variable range hopping mechanism at 77K4. The recent findings enable batch fabrication, and facilitate interfacing with CMOS technology and scaling up the networks for more advanced tasks.
1. Merolla, P. A. et al. A million spiking-neuron integrated circuit with a scalable communication network and interface. Science 345, 668–673 (2014).
2. Yao, P. et al. Face classification using electronic synapses. Nat. Commun. 8, 15199 (2017).
3. Bose, S. K. et al. Evolution of a designless nanoparticle network into reconfigurable Boolean logic. Nat. Nanotechnol. 10, 1048–1052 (2015).
4. Efros, A. L. & Shklovskii, B. I. Coulomb gap and low temperature conductivity of disordered systems. J. Phys. C Solid State Phys. 8, L49–L51 (1975).
Fabrication of three-dimensional silicon nanostructures for nanophotonics, Cock Harteveld (COPS)
The fabrication of nanostructures in silicon usually is done by etching, using a planar etch mask. Photonic bandgap crystals demand three-dimensional structures. For three dimensional structures the etch masks has to be defined on two oblique planes of the substrate and these masks should be aligned to each other.
Here we present a method to define the etch masks for three-dimensional nanostructures where the alignment of the mask is built in the design and is written by focused ion beam milling in a chromium hard mask. For this method we need substrates with known dimensions, a sharp 90 degree’s edge and an optical polished surface quality for projecting the perfect mask. We make the bar-shaped substrates by anisotropic wet etching completely through a 500micron thick 110 oriented silicon wafer. In each of these bars, using the deep reactive ion etching technique twice under oblique angle, we are able to fabricate 2D and 3D structures with chosen dimensions and configuration. We fabricate photonic bandgap crystals with a volume of 10x10x10 micron, having pores with an aspect ratio bigger than 40.
Light from silicon, Satadal Dutta (SC)
Silicon (Si) is by far, and for years to come, commercially the most successful semiconductor for a wide range of applications including CMOS integrated electronics, photovoltaics, detectors, and passive optical devices for integrated infrared photonics. However, from the purview of an optical source, Si is often sidelined being an indirect band-gap material. This makes photon emission, i.e. inter-band radiative recombination between electrons and holes, a less probable process involving the emission/absorption of a phonon. This reduces the electroluminescence (EL) efficiency of Si LEDs significantly as compared to direct band-gap semiconductors (e.g. III-Vs). Is this stop sign for pursuing silicon LEDs? The answer depends on the application. In this talk I will give a summary of the physics behind light emission from Si p-n junctions, techniques to increase efficiency and why pursuing Si LEDs is interesting from the viewpoint of CMOS-integrated optoelectronics and photonics. When a Si p-n junction is forward biased, the EL-spectrum is typically in the NIR range ~ 1120 nm. Things become more exciting when the junction is reverse biased and operated in avalanche breakdown. The high electric field broadens and shifts the carrier distribution function to higher energies, leading to wide-spectrum EL (400-900 nm). This brightens the prospect of new monolithic applications. Let us make silicon “visible” again!
Epitaxial integration of perovskite oxides on silicon, David Dubbink (IMS)
The perovskite oxides are an interesting class of materials due to its diverse range of functional properties, such as ferroelectric or magnetic. These properties are often highly anisotropic due to coupling to the structure. Preparation of fully epitaxial thin films is necessary in order to make use of these properties. These epitaxial films should be grown on silicon wafers in order to integrate these materials with current silicon based microelectronics. Growth of epitaxial perovskite oxides on silicon is not straightforward. Most of the perovskite oxides are chemically unstable on silicon. Furthermore, the unavoidable presence of an amorphous silicon native oxide prevents the silicon lattice to influence the orientation of the growing film. In this work, several strategies were studied to prepare buffer layers on silicon which allowed subsequent growth of epitaxial perovskite oxides.