🔭 astrophysics

Radiation damage to the Hubble Space Telescope during two Solar cycles, and correction of Charge Transfer Inefficiency using ArCTIc

This paper documents the radiation-induced degradation of the Hubble Space Telescope's Advanced Camera for Surveys over two solar cycles and introduces the open-source ArCTIc v7 algorithm, which effectively corrects up to 99.9% of the resulting Charge Transfer Inefficiency trailing in astronomical images.

Original authors: Richard Massey, Jacob A. Kegerreis, Juan Paolo Lorenzo Gerardo Barrios, James W. Nightingale, Richard G. Hayes, David Lagattuta, Zane D. Lentz, Gavin Leroy, Jesper Skottfelt, Felix Vecchi, Maximilian

Published 2026-02-04

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Richard Massey, Jacob A. Kegerreis, Juan Paolo Lorenzo Gerardo Barrios, James W. Nightingale, Richard G. Hayes, David Lagattuta, Zane D. Lentz, Gavin Leroy, Jesper Skottfelt, Felix Vecchi, Maximilian von Wietersheim-Kramsta

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Space Camera Getting "Sunburned"

Imagine the Hubble Space Telescope's camera (specifically the Advanced Camera for Surveys, or ACS) as a very sensitive digital camera that has been taking photos of the universe for over 20 years. It lives in space, high above Earth's atmosphere.

While this is great for seeing deep space, it's terrible for the camera's health. Up there, it is constantly bombarded by invisible "bullets" from the Sun (solar protons) and deep space (cosmic rays). These particles smash into the camera's silicon sensors, creating tiny holes and cracks in the material's structure.

Over time, these tiny cracks act like sticky traps. When the camera tries to read a photo, it has to shuffle electrons (the digital data) across the sensor to be counted. As they move, some electrons get stuck in these "sticky traps" and then released a split second later. This creates a ghostly trail behind every star or galaxy in the image, making the photo look blurry or smeared.

This paper is about two things:

Tracking the damage: Measuring exactly how much the camera has been damaged over two full 11-year cycles of solar activity.
Fixing the photos: Creating a new, super-smart software tool called ArCTIc (Algorithm for Charge Transfer Inefficiency correction) that can mathematically "un-smeared" these photos, restoring them to near-perfect clarity.

1. The Solar Connection: The Camera's "Sunburn" Cycle

The authors tracked the damage from 2002 to 2025. They found that the rate at which the camera gets damaged isn't constant; it follows the Solar Cycle.

The Analogy: Think of the Sun like a weather system. Sometimes it's calm (Solar Minimum), and sometimes it's stormy (Solar Maximum).
The Finding: The camera gets damaged fastest when the Sun is stormy. However, there is a delay. The damage peaks about 430 days (roughly 14 months) after the Sun is at its calmest.
The Magnitude: The intensity of this radiation damage fluctuates by about 18.5% over the 11-year cycle.

They also noticed a weird recent trend: in 2024/2025, the damage rate dropped unexpectedly, even though the Sun was active. This suggests something about the radiation environment is changing in a way they don't fully understand yet.

2. The Problem: "Charge Transfer Inefficiency" (CTI)

In a perfect camera, electrons move smoothly from one pixel to the next, like a bucket brigade passing water. In Hubble's damaged camera, the "buckets" have holes.

The Metaphor: Imagine a line of people passing a ball down a line. If the floor is sticky (the radiation damage), some people accidentally drop the ball, pick it up a second later, and throw it. The ball ends up landing slightly behind where it should have.
The Result: In a photo, a bright star doesn't just look like a dot; it looks like a dot with a long, faint tail behind it. This is called CTI trailing.

3. The Solution: ArCTIc (The "Digital Eraser")

The authors built a new software tool called ArCTIc (pronounced "Arctic"). Think of it as a sophisticated digital eraser that doesn't just wipe the image clean, but actually reverses the physics of the damage.

How it works:

The Model: The software simulates exactly how the electrons get stuck and released. It knows there are three main "types" of sticky traps (like three different types of glue) with different release speeds.
The Calibration: To teach the software how bad the damage is, the authors looked at "Hot Pixels." These are single pixels on the camera that are permanently "leaking" electrons due to radiation damage. Because they are so bright and isolated, they act like perfect test subjects to measure exactly how much trailing occurs.
The Correction: The software runs a simulation in reverse. It looks at the smeared photo, calculates where the electrons should have been, and moves them back.

The Results:

For photos taken recently (when the damage is worst), ArCTIc fixes 99.9% of the smearing.
For the camera's entire 23-year history, it fixes 99.5% of the smearing on average.
This level of precision is so high that it meets the strict requirements for future space missions (like the ESA's Euclid mission).

4. Why This Matters

Before this paper, scientists could fix about 97-98% of the smearing. The remaining 2-3% was a "noise floor" that limited how precise their measurements could be.

The Analogy: Imagine trying to weigh a feather on a scale that has a little bit of dust on it. You can clean off most of the dust, but if you want to know the exact weight of the feather, that tiny bit of remaining dust ruins your measurement.
The Breakthrough: ArCTIc removes almost all the "dust." This allows astronomers to measure the brightness, position, and shape of distant galaxies with extreme precision, which is crucial for studying things like dark matter and the expansion of the universe.

Summary of Key Findings

Damage Rate: The camera's damage goes up and down with the Sun's 11-year cycle, peaking about 14 months after the Sun is quietest.
The Fix: A new software tool (ArCTIc) can reverse the damage with 99.5% to 99.9% accuracy.
The Method: They used "hot pixels" (leaking pixels) as a ruler to measure the damage and built a model of how electrons get stuck in three specific types of silicon defects.
The Future: This tool allows Hubble to continue doing high-precision science for years to come, effectively extending its life by fixing its "sunburn."

Technical Summary: Radiation damage to the Hubble Space Telescope during two Solar cycles, and correction of Charge Transfer Inefficiency using ArCTIc

Problem Statement
The Hubble Space Telescope's (HST) Advanced Camera for Surveys (ACS) has operated in the harsh radiation environment of Low Earth Orbit (LEO) since 2002. High-energy solar protons and galactic cosmic rays have damaged the silicon lattice of its charge-coupled device (CCD) sensors, creating charge traps. These traps capture photoelectrons during readout and release them after a characteristic delay, creating spurious trails behind astronomical objects. This Charge Transfer Inefficiency (CTI) is a non-linear effect dependent on object brightness, shape, and prior pixel illumination history. While previous algorithms have corrected the majority of this trailing, residual CTI limits high-precision science (e.g., photometry, astrometry, and morphology) requiring accuracies better than 1%. Furthermore, the rate of radiation damage fluctuates with the Solar cycle, necessitating a robust understanding of how damage accumulates over time to maintain correction accuracy.

Methodology
The authors present ArCTIc v7 (Algorithm for Charge Transfer Inefficiency correction), an open-source, parallelized C++ tool with a Python wrapper. The algorithm models the physical process of CTI by simulating the movement of electron packets through a CCD, tracking the occupancy of charge traps.

Physical Model: ArCTIc adopts a "volume-driven" model where electron packets occupy a fraction of a pixel's volume. It assumes three distinct species of trap configurations (topological defects in the silicon lattice), each characterized by a density ( $\rho_i$ ) and a release time ( $\tau_i$ ). The model assumes instantaneous charge capture and homogeneous release for the dominant species, though it supports non-instantaneous capture and inhomogeneous release timescales.
Efficiency Optimization: To handle the millions of pixel-to-pixel transfers required for readout, the algorithm uses an "express" matrix approach. Instead of simulating every single transfer, it calculates the effect of a single transfer and multiplies it by the number of transfers a pixel undergoes, significantly reducing runtime (approx. 1 second per CCD quadrant).
Correction Strategy: Since CTI cannot be analytically inverted, the authors use an iterative forward-modeling approach. They add modeled CTI trails to a corrected image estimate and subtract the difference from the observed image. This process is repeated (typically 5 iterations) until residuals are negligible.
Read Noise Mitigation: A specific challenge addressed is the overcorrection of read noise, which is not trailed but is treated as such by the algorithm. The authors implement a method to isolate and estimate the read noise component, correcting the sky signal while preserving the noise statistics.
Calibration: The model parameters were calibrated using 178 epochs of data from 2002 to 2025. The authors utilized "warm pixels" (radiation-induced defects that leak charge, appearing as delta-functions) to measure trail profiles. By fitting the observed trails of these warm pixels across various brightness levels and distances from the readout register, they derived the time-evolving trap densities.

Key Results

Radiation Damage Evolution: The study tracks the accumulation of charge traps over two full Solar cycles. The rate of damage is modulated by 18.5 $^{+4.5}_{-0.5}$ % over an 11-year cycle. The damage rate peaks 430 $^{+11}_{-5}$ days after the Solar minimum (as recorded by sunspot numbers). The authors note a recent, unexplained decrease in the damage rate in 2024/2025 compared to previous cycles.
Trap Characteristics: The damage is consistent with three dominant trap species. The release times for these species are stable, and the capture process is effectively instantaneous relative to the dwell time. The relative densities of these species remain constant over time, while the total trap density increases linearly with superimposed Solar-cycle modulation.
Correction Performance:
- Using the calibrated ArCTIc model, the algorithm corrects 99.5% of CTI trailing averaged over the camera's lifetime.
- For the most recently acquired, worst-affected data, the correction removes 99.9% of the trailing.
- The residual error is approximately 0.2 electrons per pixel (or ~2% of the trail amplitude).
Model Robustness: The authors demonstrate that fixing physically motivated parameters (such as trap release times and well-filling parameters) to constant values derived from the full dataset significantly reduces scatter in the total trap density estimates compared to fitting all parameters independently for each epoch.

Significance and Claims
The paper claims that ArCTIc v7 represents a significant milestone in HST data processing. By achieving 99.9% correction in recent data, the residual CTI meets the stringent requirements of the ESA Euclid mission (specifically for weak lensing and shape measurements), a benchmark previously difficult to achieve with HST data alone.

The authors emphasize that their success relies on a rigorous separation of CTI calibration (determining the physical model parameters) from CTI correction (applying the model to science data). They highlight two critical improvements over previous methods:

Background Subtraction: Properly accounting for uncertainty in background subtraction prevents the need for ad-hoc reweighting of data points.
Pre-CTI Modeling: Adding the few trailed electrons back into the model of the "true" warm pixel before fitting prevents a dominant calibration bias.

The work provides a flexible, open-source tool capable of correcting CTI for a wide variety of applications and establishes a physically motivated framework for understanding radiation damage evolution in space-based detectors. The authors note that while the current model is highly effective, small residuals may persist due to imperfect background subtraction, electronic readout effects (e.g., bias striping), or limitations in modeling the behavior of charge clouds with very few electrons.