🔭 astrophysics

Evolution of Sink Pixels in ACS/WFC and Connection to Charge Transfer Efficiency

This study analyzes seven years of Hubble ACS/WFC data to characterize the creation, persistence, and spatial distribution of sink pixels, revealing that their observed density gradient and "bounce-back" effect near the chip gap are primarily driven by Charge Transfer Efficiency (CTE) losses and charge release during readout rather than intrinsic pixel failure patterns.

Original authors: Alyssa M. Guzman, Jenna E. Ryon

Published 2026-03-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Alyssa M. Guzman, Jenna E. Ryon

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

Imagine the Hubble Space Telescope's camera (specifically the ACS/WFC) as a massive, high-tech digital canvas. Over the years, this canvas has been bombarded by tiny, invisible particles from space (radiation). Most of the time, the camera handles this fine, but sometimes these particles hit a single pixel and leave a permanent "scar."

This report is about studying those scars, which the scientists call Sink Pixels (SPs).

Here is the story of what they found, explained simply:

1. What is a "Sink Pixel"?

Think of a normal camera pixel like a bucket that catches rain (light). When the camera takes a picture, it empties the bucket row by row to count how much rain fell.

A Sink Pixel is like a bucket with a hole in the bottom, or perhaps a bucket that has become "sticky." When the camera tries to empty the bucket, the water (electrons) gets stuck in the hole instead of flowing out. Because the bucket ends up emptier than it should be, the camera records a value that is way too low—so low it actually looks negative.

These pixels are "sinks" because they swallow up the charge that should have been part of the image.

2. The "Healing" Attempt (The Annealing Process)

Hubble's camera gets very cold to work best. Over time, the cold and radiation cause "warm" or "hot" pixels (defects that act like stuck lights). To fix this, engineers perform a monthly "anneal."

Think of this like baking a cake. They turn off the cooling and heat the camera up to about 20°C (68°F) to "reset" the silicon. This usually fixes the warm pixels, making them act normal again.

The Big Discovery: The scientists wanted to know if this "baking" process could fix the Sink Pixels too.

The Result: Almost never. Out of thousands of Sink Pixels, only a tiny handful (less than 0.1%) ever "healed" and went back to normal. Once a pixel becomes a Sink, it stays a Sink. It's like a permanent scratch on a vinyl record; you can't bake it away.

3. The "Traffic Jam" Effect (Charge Transfer Efficiency)

Here is where it gets interesting. The camera reads the image by shuffling electrons down a line, like people passing a bucket in a human chain to fill a pool.

The Serial Register: This is the "pool" where the electrons end up.
The Chip Gap: This is the other end of the line.

The scientists noticed a strange pattern: There were fewer Sink Pixels detected near the Chip Gap and more near the Serial Register.

The Analogy: Imagine a long line of people passing buckets.

If you are standing right next to the pool (Serial Register), you pass your bucket quickly. If your bucket has a hole (a Sink Pixel), the hole is obvious immediately.
If you are at the very back of the line (Chip Gap), you have to pass your bucket through 2,000 other people before it reaches the pool. Along the way, other people might accidentally spill extra water into your bucket to fill it up.

The scientists realized that the "Charge Transfer Efficiency" (CTE) is like those extra people spilling water. As the electrons travel from the Chip Gap to the Serial Register, the "leaky" Sink Pixels get filled in by stray electrons from their neighbors. This makes the Sink Pixel look less negative, or sometimes even normal, so the computer doesn't spot it as a defect.

The Gradient: The further a pixel is from the "pool" (Serial Register), the more likely it is to get "filled in" by this traffic jam of electrons. This is why the scientists see fewer Sink Pixels at the far end of the camera—they are being hidden by the very process of reading the image.

4. The "Bounce-Back" Mystery

There was one weird spot in the data: right near the Chip Gap, there was a sudden spike in the number of Sink Pixels. The scientists call this the "Bounce-Back" effect.

They tried to simulate the camera's behavior on a computer to see if they could recreate this. They built a virtual camera, filled it with random Sink Pixels, and let the "traffic jam" happen.

What they got right: The simulation showed the "traffic jam" gradient (fewer sinks at the far end).
What they missed: The simulation did not show the sudden spike (Bounce-Back) near the gap.

This means the "Bounce-Back" is a mystery. It's like a traffic jam that suddenly clears up right at the exit, but the scientists don't know why yet.

Summary of Findings

Creation Rate: The camera is slowly getting more damaged. About 2 new Sink Pixels are created every single day on each of the two camera chips.
Total Count: By 2021, there were over 44,000 Sink Pixels in the camera. That sounds like a lot, but it's only about 0.25% of the total pixels. The camera is still working great!
No Healing: The monthly "baking" (annealing) fixes hot pixels but does almost nothing for Sink Pixels.
The Hidden Defects: The camera actually has more Sink Pixels than we can see. The ones far away from the readout are being "filled in" by the reading process, making them invisible.

The Bottom Line: The Hubble camera is aging gracefully, but it is slowly accumulating permanent "sink" defects. While we can't fix them, understanding how they hide themselves helps scientists correct the images so we can still see the universe clearly.

1. Problem Statement

The Advanced Camera for Surveys Wide Field Channel (ACS/WFC) on the Hubble Space Telescope (HST) operates in a Low Earth Orbit radiation environment. This environment causes energetic particles to hit the detector, creating charge traps that lead to increased dark current and the formation of "warm" and "hot" pixels. While monthly annealing processes mitigate these issues, they do not eliminate them.

A specific subset of these defects, known as Sink Pixels (SPs), has been identified. SPs are pixels with values significantly lower than the local background (specifically $\le -10$ electrons) in post-flashed short darks. They result from localized excess charge traps that capture electrons during the parallel readout shift, preventing them from reaching the readout register. This causes the pixel to read out as a negative value relative to its neighbors and creates negative "trails" in the opposite direction of readout.

The report addresses three primary questions:

What is the creation rate and accumulation of SPs over time?
Can the monthly annealing process "heal" SPs (return them to normal values)?
How does Charge Transfer Efficiency (CTE) loss influence the spatial distribution and detectability of SPs?

2. Methodology

The authors analyzed data spanning from 2015 to 2021, utilizing three main data sources:

SP Reference Files (*snk.fits): Generated by the ACS/WFC reference file pipeline to flag SPs for each anneal cycle.
Post-flashed Short Darks (flt.fits): 0.5-second exposures taken after a 4.6-second LED flash (60 $e^-$ background).
CTE-Corrected Short Darks (flc.fits): Images processed to correct for parallel CTE losses.

Key Analytical Steps:

Thresholding: SPs were defined as pixels with values $\le -10 e^-$ (stricter than the reference file's $\le -7 e^-$ ) to distinguish true defects from noise.
Creation Tracking: By comparing pixel values before and after each anneal date, the team identified "creation events" where a pixel's mean value dropped by at least 5 $\sigma$ (five times the typical variability).
Healing Analysis: The team investigated if SPs could revert to positive/normal values. A "healed" pixel required the pre-break mean to be negative and the post-break mean to be positive, with a difference less than 5 $\sigma$ .
Spatial Analysis: The distribution of SPs was mapped across the X (parallel) and Y (serial) axes for both WFC1 and WFC2 chips.
Simulation: A synthetic short dark was generated with uniformly distributed SPs. This data was processed through the acscteforwardmodel to simulate CTE losses during readout, allowing the team to isolate the impact of CTE on SP detectability.

3. Key Contributions

Quantification of SP Accumulation: Provided the first comprehensive 7-year timeline of SP creation rates and total accumulation for both ACS/WFC chips.
Annealing Efficacy: Demonstrated that the standard monthly annealing process is ineffective at healing SPs, distinguishing them from transient warm/hot pixels.
CTE-SP Correlation: Established a causal link between CTE degradation and the observed spatial gradient of SPs, showing that CTE losses make SPs harder to detect further from the serial register.
Simulation Validation: Created a simulation confirming that CTE-induced charge trailing can explain the gradient in SP density, though it could not explain a specific "bounce-back" local peak near the chip gap.

4. Key Results

A. Creation Rates and Accumulation

Creation Rate: SPs are created at a steady rate of approximately 2.15 pixels/day for WFC1 and 2.23 pixels/day for WFC2 (approx. 60–62 pixels/month).
Total Accumulation: By the end of 2021, the detectors contained 44,068 SPs (21,168 in WFC1 and 22,900 in WFC2).
Impact: These SPs constitute approximately 0.25% of the science frame (excluding trails).
Persistence: SPs are permanent defects. The study found it is extremely rare for SPs to "heal." Only 11 pixels in WFC1 and 17 in WFC2 showed signs of healing, often transitioning into warm pixels rather than returning to normal.

B. Spatial Distribution and CTE Effects

Y-Position Gradient: There is a distinct non-uniform distribution of SPs along the Y-axis (serial register direction).
- Fewer SPs are detected near the chip gap (far from the serial register).
- More SPs are detected near the serial register.
- WFC1: ~40% fewer SPs near the chip gap compared to the serial register.
- WFC2: ~43% more SPs near the serial register.
X-Position Distribution: The distribution along the X-axis (parallel to readout) remains largely uniform.
Temporal Evolution: The gradient in the Y-distribution has become steeper over time (comparing 2015, 2018, and 2021 data). This correlates with the known degradation of CTE over time.
The "Bounce-Back" Effect: A local peak in SP density was observed very near the chip gap (opposite the amplifier) in both chips. This feature was not reproduced in the CTE simulation, suggesting its origin is distinct from standard CTE losses.

C. Simulation Findings

The simulation of a uniformly distributed SP population, subjected to CTE losses, successfully reproduced the gradient (fewer detectable SPs further from the serial register).
This confirms that CTE losses cause charge to "fill in" SPs during readout, making them less detectable at greater distances from the serial register.
The simulation failed to reproduce the local peak near the chip gap, indicating that the "bounce-back" effect requires further investigation.

5. Significance

This report provides critical insights for the calibration and data reduction of ACS/WFC:

Calibration Strategy: Since SPs are permanent and accumulate linearly, reference files (*snk.fits) must be updated regularly to flag these pixels. The study confirms that annealing cannot be relied upon to remove them.
CTE Correction Limitations: Current CTE correction algorithms do not fully restore the detectability of SPs located far from the serial register. The gradient in SP density is an artifact of CTE degradation, not necessarily a physical lack of defects in those regions.
Data Quality: Understanding that SPs create negative trails and affect surrounding pixels is essential for accurate photometry and source detection in deep imaging.
Future Research: The unexplained "bounce-back" peak near the chip gap suggests there are still unknown mechanisms affecting SP distribution that require further study.

In summary, the paper characterizes the lifecycle of Sink Pixels in the ACS/WFC, confirming their permanence, quantifying their growth, and attributing their spatial distribution gradients primarily to Charge Transfer Efficiency losses.