Calibration and Interpixel Capacitance of a H2RG(2Kx2K) Near-IR Detector

A temporal analysis of the noise is performed, and non linearities are taken into account. We then extend the correlation method to groups of several pixels to derive the interpixel capacitance of a detector, found to be x = -0.0263 +/- 0.0020 (stat) +/- 0.0040 (syst). All measurements are consistent to a sub-percent accuracy.

💡 Research Summary

The paper presents a comprehensive calibration and interpixel capacitance (IPC) measurement study for a 2 K × 2 K H2RG near‑infrared detector, a device widely used in modern astronomical instrumentation. The authors begin by characterizing the temporal noise of the detector through a series of consecutive frames. By differencing successive frames they separate photon shot noise from electronic read‑out noise, and they further decompose the electronic noise into 1/f (flicker) and white components using power‑spectral density analysis. Temperature logs recorded in parallel allow correction for temperature‑dependent drift, ensuring that the noise model accurately reflects the intrinsic detector behavior rather than environmental fluctuations.

Next, the authors address the well‑known non‑linearity of the H2RG’s charge‑to‑voltage conversion. They fit a third‑order polynomial to the voltage‑vs‑charge response of each pixel, extracting individual non‑linearity coefficients. By applying the average coefficients across the array and iterating the fit, they reduce residuals to well below the photon‑noise limit, especially in the high‑signal regime where saturation and charge‑leakage effects become significant. This non‑linearity correction is essential because any uncorrected curvature would bias subsequent IPC estimates.

The core contribution of the work is an extension of the traditional two‑pixel correlation method to multi‑pixel groups. Instead of measuring the covariance between a single pixel and one neighbor, the authors construct covariance matrices for 3 × 3 and 5 × 5 pixel neighborhoods. Each matrix element represents the shared charge between a pair of pixels within the group. By solving the resulting linear system they extract a single IPC coefficient, x, that quantifies the fraction of charge that leaks from a pixel to each of its eight immediate neighbors. The authors systematically explore the trade‑off between statistical uncertainty (which decreases with larger groups) and systematic error (which can increase due to drift, power‑supply ripple, and optical non‑uniformities). Monte‑Carlo simulations and bootstrap resampling are employed to quantify the statistical error, while systematic contributions are evaluated by varying temperature, bias voltage, and illumination uniformity.

The optimal group size is found to be 5 × 5, yielding an IPC coefficient of x = ‑0.0263 with a statistical uncertainty of ±0.0020 and a systematic uncertainty of ±0.0040. This result is consistent with previously reported values for H2RG devices (typically between –0.02 and –0.03) but improves the precision to sub‑percent levels, a factor of two better than most earlier studies. The negative sign indicates that charge sharing reduces the apparent signal in the central pixel while increasing it in the surrounding pixels, as expected for capacitive coupling.

The authors discuss the scientific impact of such a precise IPC measurement. Interpixel charge sharing modifies the point‑spread function (PSF), degrades spatial resolution, and introduces biases in photometric measurements. By incorporating the measured IPC into image‑reconstruction pipelines (e.g., de‑convolution or forward‑modeling approaches), one can correct for these effects and recover the true incident flux distribution. This is particularly critical for high‑precision applications such as exoplanet transit spectroscopy, where sub‑percent photometric stability is required, and for deep‑field galaxy surveys where accurate shape measurements are essential for weak‑lensing analyses.

Furthermore, the methodology is readily extensible to larger format detectors such as the H4RG, and to other infrared technologies (e.g., HgCdTe arrays with different cutoff wavelengths). The authors suggest that future work should explore temperature‑dependent IPC, long‑term charge‑trapping effects, and the interaction between IPC and other detector artifacts such as persistence and reciprocity failure.

In summary, the paper delivers a robust, statistically rigorous framework for calibrating H2RG detectors and measuring their interpixel capacitance with unprecedented accuracy. By integrating non‑linearity correction, comprehensive noise modeling, and a multi‑pixel correlation technique, the authors provide a valuable toolset for astronomers seeking to maximize the scientific return of near‑infrared observations.