BTU overview
Faculty 1
Chair of Computer Engineering Substitute Professor Dr.-Ing. habil. Christian Herglotz

Project description

We are developing wavelength-selective photodetectors built from Si/Ge heterostructure metasurfaces—ultra-thin, nanopatterned layers that act like on-chip optical filters and light harvesters at the same time. Each pixel is a vertical Ge PIN diode grown on a silicon-on-insulator (SOI) wafer; the top Si and Ge layers are etched into a periodic “forest” of 400–900 nm pillars. This all-dielectric metasurface traps light at a single, designer wavelength and funnels it straight into the absorbing Ge layer.

  • Why it matters The metasurface boosts absorption to ≈45 % at 1.4 µm with only a 150 nm-thin Ge layer—performance normally requiring four times the material—while keeping the full width at half maximum (FWHM) only a few tens of nanometres . Because the resonance position shifts monotonically with pitch and pillar diameter, we can lithographically dial any target line between 1.2 µm and 1.55 µm.

  • Fabrication highlights All structures are made in a single top-down CMOS-compatible flow: epitaxial Ge on SOI, plasma etching of the nanopillars, and a blanket dielectric cap for passivation and future transparent contacts . The Si/Ge/Si stack also provides a continuous, low-resistance contact plane—no tricky nanoscale wiring required.

  • From pixels to spectrometers By placing 9–25 different metasurface pixels side-by-side, each tuned to a slightly offset resonance, the chip becomes a snapshot spectrometer. A lightweight Gaussian-matrix algorithm converts the set of photocurrents into a 300-point spectrum; the team demonstrates accurate reconstruction of plant-leaf water-stress signatures across 1.3–1.6 µm .

  • Key advantages

    • Spectral purity & tunability without external filters

    • Robustness against etch-depth errors and side-wall angle variations, thanks to the Si/Ge heterostructure design

    • Seamless integration with silicon photonics and CMOS read-out, enabling low-cost hyperspectral cameras, fibre-optic monitors and environmental sensors.

In short, the project turns a single photodiode into a precision optical filter—then multiplies it into a compact array that delivers full near-infrared spectra on-chip.

Objectives

the main purpose of this project is the reconstruction of sensivity index in the range of different wavelenght by utilizing machine learning methods with available dataset.