Black Silicon for Photodiodes: Experimentally Implemented and FDTD Simulated - Aix-Marseille Université Access content directly
Other Publications Year : 2010

Black Silicon for Photodiodes: Experimentally Implemented and FDTD Simulated

Adam J. Williamson


The experimental results presented here show the creation of black silicon through plasma processing and its subsequent implementation as an anti-reflective coating for photodiodes in the range of 200 nm to 800 nm. Finite difference time domain (FDTD) calculations of various nano-structure geometries are compared to integrating sphere reflection measurements of physical structures. The spectral sensitivity of manufactured black silicon photodiodes in the red region (675 nm to 750 nm) meets the set 0.5 A/W target, while in the blue region (375 nm to 425 nm) an improvement of +0.07 A/W over oxide covered diodes is achieved for center cites. Clear separation between nano-structure tips is experimentally demonstrated as the deciding factor to improve reflection suppression. The appearance of nano-structures on the silicon surface when processed in the SF6/O2plasma mixture is explained by generalizing the etch rate of silicon dependant on fluorine concentration (with a varying fluorine to oxygen gas ratio) into two regions, one of excess fluorine generation (weakly passivating) and one of excess fluorine consumption (strongly passivating). This is experimentally verified. Inherent non-uniformities of the plasma processing chamber which yield a small process window of extremely anisotropic nano-structures and a poor center-to-edge wafer uniformity are eliminated by the introduction of a nano-loading step. The silicon wafer surface is first etched with a set of highly uniform nano-pillars. The nano-loading mask is subsequently plasma etched utilizing gas compositions that do not by themselves result in the appearance of nano-structures on the silicon wafer surface. This yields new less-anisotropic geometries with properties more desirable for reflection suppression and decouples the dependence of attainable nano-structure geometry on plasma chamber geometry. An extended nano-loading approach is successfully applied to transfer plasma processes from blank wafers to masked wafers. FDTD results show that the majority of light absorption in the spectrum from 200 nm to 800 nm takes place within the silicon nano-structure antireflective coating, in contrast to the situation encountered with a standard anti-reflective coating. The periodicity of nano-structures is investigated with FDTD and clear diffraction grating properties are demonstrated. The ideal width of nano-structures for reflection suppression is calculated to be between 100 nm and 200 nm for the light spectrum of interest (200 nm to 800 nm). Any periodic structure with widths over 200 nm will begin diffracting light into non-specular angles. The ideal structure height is calculated to be (at least) 500 nm. However, the periodic structure simulations do not prove to be the most accurate way to evaluate the measured specular and diffuse components of real physical structures. The problem being that overlapping nano-structures (for example unseparated through the plasma etch process) disturb the conditions of periodicity (diffraction into component angles) corresponding to the structures overall width. FDTD simulations are best correlated to the optical properties of physical self-organized nano-structures (integrating sphere measurements of specular and diffuse components) when a non-periodic surface is simulated. Boundary conditions for the FDTD are discussed in the process of simulating non-normal incidence light on 2D structures. Dispersion in the injection angle for a simulation creates difficulties in simulating broadband spectrums in a single simulation, as only the center wavelength possesses the correct theta, while the smallest and largest wavelength in the incident experience the most extreme error in incident theta. Simulating each wavelength is not possible as it yields “incalculable” computational times. The solution provided breaks the desired spectrum into several smaller spectrums and gives satisfactory results while limiting the amount of dispersive error. Furthermore it is shown here that the implementation of gradient-index in front of perfectly matched layers (PMLs) improves their angle-dependent performance. Additionally it is experimentally demonstrated here that reflection suppression is not always correlated to electrical performance as plasma etching to modify surface topology 4drastically reduces carrier lifetimes, due to an increase in recombination resulting from lattice damage and an overall increase in surface area.
Fichier principal
Vignette du fichier
ilm1-2010000565.pdf (39.13 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03349103 , version 1 (20-09-2021)


  • HAL Id : hal-03349103 , version 1


Adam J. Williamson. Black Silicon for Photodiodes: Experimentally Implemented and FDTD Simulated. 2010. ⟨hal-03349103⟩


25 View
12 Download


Gmail Mastodon Facebook X LinkedIn More