Impact of Charge Transfer Inefficiency on transit light-curves: A correction strategy for PLATO

TL;DR

This paper addresses CTI-induced biases in PLATO transit photometry by developing a calibration-and-correction pipeline that combines the Charge Density Model with an axisymmetric radial trap-density map. The workflow uses overscan fitting, smearing removal, and an EPER-based density fit to infer trap species, release times, and spatial trap distributions, then applies a Massey-style iterative correction to recover CTI-free photometry. In the worst-case scenario, CTI bias in transit depth is about $4\%$, which is reduced to $0.06\%$ after correction, yielding a radius ratio bias near $2\%$ that is below PLATO's requirements. The approach is demonstrated on simulated PLATO data and is broadly applicable to CTI-affected detectors, with potential extensions to PSF-fitting photometry and cross-camera validation.

Abstract

PLATO is designed to detect Earth-sized exoplanets around solar-type stars and to measure their radii with accuracy better than $2\%$ via the transit method. Charge transfer inefficiency (CTI), a by-product of radiation damage to CCDs, can jeopardize this accuracy and therefore must be corrected. We assess and quantify the impact of CTI on transit-depth measurements and develop a correction strategy that restores CTI-biased depths within the accuracy budget. Using a calibration dataset generated with PLATOSim to simulate a realistic stellar field, we model the parallel overscan signal as a sum of exponential decays and use least-squares fitting to infer the number of trap species and initial estimates for the release times $τ_{r,k}$. Smearing is modeled with an exponential-plus-constant function and removed on a column-wise basis. We model the spatial variation in trap density with a quadratic polynomial in radial distance from the focal-plane center. The polynomial coefficients $a_{p,k}$, the well-fill power index $β$, and the release times $τ_{r,k}$ are adjusted via iterative application of the Extended pixel Edge Response (EPER) method combined with a CTI correction algorithm, yielding the final calibration model. In the worst-case scenario (8-year mission, high-CTI zone), CTI induces a bias of about $4\%$ in measured transit depth, reduced to a residual of $0.06\%$ after correction - well within PLATO's accuracy requirements. From the calibrated parameters, we derive a correction scheme that brings photometric measurements within PLATO's noise budget, ensuring that the mission's precision requirements are met.

Figures (13)

• Figure 1: Effect of CTI on the recorded signal in the parallel transfer direction (readout proceeds from left to right). The solid curve shows a CTI-free signal profile, while the dashed curve illustrates the same profile after CTI degradation. Deferred charge release reduces the peak amplitude and produces a trailing toward increasing TDI lines (earlier pixels in the transfer path). Credit: Short_2013.
• Figure 2: Radiation map of Camera 1, composed of four CCD, showing the radial dependence of trap density. The colour bar indicates the TNID level (trap density) to which each pixel is exposed after 6.5 years. Adapted from ohb2024radiation.
• Figure 3: Radiation map generated using the radial dependence model (Eq. \ref{['eq:radial_relationship']}) for a 6.5-year TNID exposure. The CCD is oriented with the readout register at the bottom ($y = 0$). The simulated distribution shows strong agreement with the reference radiation map for CCD #1 (Fig. \ref{['fig:og_radiation_map']}), adapted from ohb2024radiation, under identical TNID conditions.
• Figure 4: Map of cumulative trap density derived from Eq. \ref{['eq:cnt_true']} for all species. Three positions are indicated: A (bottom left), B (top left), and C (top right). The overplotted circle with a radius of 4900 pixels represents the observable field of view.
• Figure 5: Comparison of transit light-curves for the worst-case scenario: an 8-year mission at high trap density (position C) and $m_V = 11$. The solid orange curve includes CTI effects, while the dashed blue curve shows the CTI-free case. CTI-induced charge loss introduces a photometric bias of $3.95\%$.