Point-spread function ramifications and deconvolution of a signal dependent blur kernel due to interpixel capacitive coupling

TL;DR

This paper investigates how interpixel capacitance (IPC) causes signal-dependent blurring in detector arrays, affecting image quality and flux estimation, and proposes an iterative deconvolution method to correct for these effects.

Contribution

It introduces a novel iterative non-stationary deconvolution technique that accounts for variable IPC effects, improving accuracy over traditional fixed PSF methods.

Findings

IPC varies with signal and background levels, affecting PSF.

Degradation of PSF FWHM from 2.080 to 2.186 pixels due to IPC.

Flux estimation errors can reach 1.5% with fixed PSF fitting.

Abstract

Interpixel capacitance (IPC) is a deterministic electronic coupling that results in a portion of the collected signal incident on one pixel of a hybridized detector array being measured in adjacent pixels. Data collected by light sensitive HgCdTe arrays which exhibit this coupling typically goes uncorrected or is corrected by treating the coupling as a fixed point spread function. Evidence suggests that this IPC coupling is not uniform across different signal and background levels. This variation invalidates assumptions that are key in decoupling techniques such as Wiener Filtering or application of the Lucy- Richardson algorithm. Additionally, the variable IPC results in the point spread function (PSF) depending upon a star's signal level relative to the background level, amond other parameters. With an IPC ranging from 0.68% to 1.45% over the full well depth of a sensor, as is a reasonable range for the H2RG arrays, the FWHM of the JWSTs NIRCam 405N band is degraded from 2.080 pix (0".132) as expected from the diffraction patter to 2.186 pix (0".142) when the star is just breaching the sensitivity limit of the system. For example, when attempting to use a fixed PSF fitting (e.g. assuming the PSF observed from a bright star in the field) to untangle two sources with a flux ratio of 4:1 and a center to center distance of 3 pixels, flux estimation can be off by upwards of 1.5% with a separation error of 50 millipixels. To deal with this issue an iterative non-stationary method for deconvolution is here proposed, implemented, and evaluated that can account for the signal dependent nature of IPC.