Silvaco ATLAS model of ESA's Gaia satellite e2v CCD91-72 pixels

TL;DR

This paper presents a detailed 3D model of Gaia's CCD pixel, revealing systematic offsets in SBC capacities likely due to manufacturing processes, impacting calibration and radiation damage correction.

Contribution

We developed a 3D Silvaco ATLAS model of Gaia's CCD pixel and identified systematic offsets in SBC capacities related to manufacturing variations.

Findings

SBC capacities are smaller than design predictions.

Offsets in doping widths are about 2 μm.

Manufacturing errors contribute to SBC capacity variations.

Abstract

The Gaia satellite is a high-precision astrometry, photometry and spectroscopic ESA cornerstone mission, currently scheduled for launch in 2012. Its primary science drivers are the composition, formation and evolution of the Galaxy. Gaia will achieve its unprecedented accuracy requirements with detailed calibration and correction for CCD radiation damage and CCD geometric distortion. In this paper, the third of the series, we present our 3D Silvaco ATLAS model of the Gaia e2v CCD91-72 pixel. We publish e2v's design model predictions for the capacities of one of Gaia's pixel features, the supplementary buried channel (SBC), for the first time. Kohley et al. (2009) measured the SBC capacities of a Gaia CCD to be an order of magnitude smaller than e2v's design. We have found the SBC doping widths that yield these measured SBC capacities. The widths are systematically 2 {\mu}m offset to the nominal widths. These offsets appear to be uncalibrated systematic offsets in e2v photolithography, which could either be due to systematic stitch alignment offsets or lateral ABD shield doping diffusion. The range of SBC capacities were used to derive the worst-case random stitch error between two pixel features within a stitch block to be \pm 0.25 {\mu}m, which cannot explain the systematic offsets. It is beyond the scope of our pixel model to provide the manufacturing reason for the range of SBC capacities, so it does not allow us to predict how representative the tested CCD is. This open question has implications for Gaia's radiation damage and geometric calibration models.