Characterization of the CSST Survey Camera CCDs: I. basic electro-optical performance

Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.

Vera: We're looking at the paper: "Characterization of the CSST Survey Camera CCDs".

Jocelyn: Here's the gist of the paper. The Chinese Space-station Survey Telescope (CSST) is a flagship mission designed to carry out a large-area high-resolution multiband imaging and slitless spectroscopic survey over a decade This paper is the first in a series that present ground-based characterization of the 31 flight-model charge-coupled devices (CCDs) of the CSST Survey Camera Basic Electro-optical Performance Baseline This research establishes a performance baseline for all 31 flight-model CCDs from their bias frames, dark frames, and flat fields Key results include a full-well capacity of ≥ 80 ke− (limited by the test electronics), non-linearity < 1.4% in the range of 7-70 ke−, median readout noise ∼ 3.6 e−, dark current ≤ 0.02 e− pix−1 s−1 at the operating temperature of −85◦C, high quantum efficiency across the survey’s spectral range (255–1000 nm), typical pixel response non-uniformity 99.9994% at the flat-field level of 10 ke− All 31 CCDs meet or exceed the Survey Camera’s specifications.

Vera: Detector Architecture and Testing Setup The CSST-SC mosaic focal plane consists of 31 9k × 9k charge-coupled devices (CCDs) with a pixel size of 10µm × 10µm Two types of CCDs are installed: GM1201A from China Electronics Technology Group Corporation No.

Jocelyn: 44 Research Institute and CCD290-99 from Teledyne e2v GM1201A uses standard silicon with different coatings for near-ultraviolet (NUV) and visible (VIS) subtypes, which we refer to as 44NUV and 44VIS, respectively CCD290-99 has more variants: standard silicon with NUV coating (e2vNUV), standard silicon with astro-multi 2 coating (e2vVIS), and deep-depletion silicon with astro-multi 2 coating (e2vDD), which is significantly more sensitive at wavelengths above 650 nm than standard silicon CCDs.

Vera: The test bench comprises a flat-field illumination system and various imaging test apparatuses All measurements are carried out at the working temperature of −85◦C except for dark current, which is also measured at −75◦C and −95◦C.

Title and authors: Vera: Thinking about the title of this paper, "Characterization of the CSST Survey Camera CCDs: I. basic electro-optical performance," it really frames what we're looking at—it’s not a deep dive into the science yet, but just making sure the camera hardware is actually capable of doing its job.

Jocelyn: And you see that this work is coming from a team working on this flagship mission, which means they’re focused on making sure the detectors are reliable for ten years of continuous operation in space.

The paper's summary: Vera: The paper summarizes their testing by detailing how they derived several key performance indicators from the PTC flat-field images, which is basically using those calibration images to calculate things like quantum yield and charge transfer efficiency.

Jocelyn: They explain how the PTC slope relates to the quantum yield alpha divided by conversion gain G in units of electrons per ADU, and that conversion gain needs to be calibrated individually for every single channel of each CCD they tested.

Subrahmanyan: The paper also discusses a "brighter-fatter effect" which bends down the PTC more as flat-field intensity increases, and they say that you need the correlation between three neighboring pixels to recover linearity without adding too much random noise.

Vera: They also detail how they measured readout noise by looking at the standard deviation of pixel values in electrons from bias frames taken at negative eighty-five degrees Celsius, and then taking the median of those noises for each channel.

Jocelyn: And they quantified charge transfer efficiency, or CTE, which is given by one minus Ntrap over Nflatntrans—they found that the CTE values were greater than nine point nine nine nine four six for the e2vVIS CCD at ten kilo-electrons parallel.

Subrahmanyan: That high CTE value is critical because if charge transfer isn't efficient, you lose signal when stacking images, which impacts your ability to reconstruct faint sources or measure subtle structures in the sky.

Vera: They also measured the PRNU or pixel response non-uniformity, typically finding it less than one percent across the whole system.

Jocelyn: So in short, they’ve characterized how well each detector handles light before we even see what those detectors are seeing from deep space observations.

The paper's improvements: Vera: One thing the paper points out is that several detector effects require further study, specifically noting that the e2vNUV CCD shows a marked gradient in its flat field at three hundred nanometers but not at four hundred seventy nanometers.

Jocelyn: That means there are subtle differences between the silicon coatings, and they suggest they need to investigate those specific spectral dependencies for each detector type more closely.

Subrahmanyan: From a modeling perspective, this gradient suggests that our current models might oversimplify the response of certain detector subtypes across different wavelengths, so we need to refine those models.

Vera: Then there’s the suggestion to use semi-physical models to effectively reconstruct the brick wall pattern in NUV because that’s a dominant non-uniformity for some of these CCDs.

Jocelyn: That sounds like a practical step; if you can model that known pattern, you can correct for it and get a cleaner image, especially since they're looking at the two hundred fifty nanometer to one thousand nanometer range.

Subrahmanyan: And in terms of noise modeling, incorporating those median readout noise estimates from the test results—which they found to be about three point six electrons root mean square—will help AI systems better estimate the true signal-to-noise ratio when processing flight model data.

Vera: That's a practical application; knowing that the median readout noise is that specific number helps us build more accurate noise models for our actual science pipelines.

Conclusion: Jocelyn: So to wrap up, the paper confirms that the basic electro-optical performance of the CSST-SC flight model CCDs meets all the specified requirements, with full-well capacity exceeding eighty kilo-electrons.

Vera: And they've verified that their response non-linearity stays below one point four percent over a specific range and less than one percent over another, which is exactly what we needed to see for the survey’s science drivers.

Subrahmanyan: The dark current being no more than zero point zero two electrons per pixel per second at negative eighty-five degrees Celsius confirms that the hardware is stable enough for long-term operation in space environments.

Jocelyn: They also found the median readout noise is about three point six electrons root mean square with those test electronics, and the quantum efficiency covers two hundred fifty nanometers to one thousand nanometers.

Vera: And they verified that the PRNU is typically less than one percent, and charge transfer efficiency was greater than nine point nine nine nine four six for the e2vVIS CCD at ten kilo-electrons parallel.

Subrahmanyan: Overall, this characterization of the CSST Survey Camera CCDs: I. basic electro-optical performance provides a solid foundation that validates the hardware before we start pouring massive amounts of data into the survey over that decade.

Jocelyn: It’s a necessary first step to ensure the science we get out of this telescope is as clean and reliable as possible.

Vera: It's just good to see these numbers solid, and then we can move on to the next set of papers that will show us what actual data looks like coming from this instrument.