Inter-pixel capacitance: prospects for deconvolution

We provide an IDL implementation of Fourier deconvolution and apply it to data from a WFC3 H1RG near-infrared array detector that exhibits inter-pixel capacitance (IPC). The deconvolution removes the most obvious deleterious effect of IPC: the cross-like pattern (i.e. the arithematic symbol for addition, “+”) of charge around pixels with large dark current, a.k.a. “hot” pixels. We also exhibit a case in which our physical interpretation of a detector defect changed qualitatively before and after the deconvolution: in a region of multiple “scratches” on the detector, before deconvolution we thought each “scratch” was spatially resolved by the 18-micron pixels, but after deconvolution, it appears that the width of each “scratch” is much smaller than a pixel. We briefly discuss a number of potential benefits of deconvolution of data from arrays with IPC: the preliminary analysis in this report may motivate and expedite additional simulations and observations upon which a more definitive and quantitative analysis will be based. In practice, deconvolution of IPC can be performed after images have been calibrated with the nominal (i.e. existing) pipeline process, although we discuss how in principle and potentially for optimum effect, the deconvolution of IPC should be performed on all data (science images, flat-fields, etc) at the outset, after reading out the arrays and prior to calibrating the science images with darks, flat-fields, cosmic-ray rejection, etc. A subsequent report will quantify the benefits of IPC-deconvolution prior to, instead of after, pixel-by-pixel calibration; this report only outlines the potential benefits qualitatively.