6.3.2 CCD cosmetics

The calibration of the CCDs is done using the on-ground measurements taken during the pre-flight tests. The CCDs have the following cosmetics:

• •

  Defects: The Gaia CCDs were delivered by the manufacturer (then called e2v, now called Teledyne e2v) with a pre-launch list of 23 defects and their column number. This list was used to flag and exclude windows that included these columns in the DR2 RVS pipeline. For DR3, the Scatter pipeline was used offline to create column-resolution ScatterMaps. Visual inspection of these validated the pre-launch list and identified new defects. A total of 66 defects and their column number were supplied to the DR3 RVS pipeline (see Table 6.2) to flag and exclude windows that included these columns. The largest increase in the number of new defects is columns at higher and adjacent AC positions to existing hot columns. Spurious flux from hot columns has been transferred into these columns via serial charge transfer inefficiency during the read out process. The next largest increase in the number of new defects is new hot columns, followed by new defective columns read out prior to hot columns but with lower levels of spurious flux.

• •

  Dark current: Dark current was measured by the spacecraft manufacturer (then called EADS-Astrium, now called AirbusD&S) as the mean CCD dark current at 193 K, measured on the Flight Model CCDs, and then extrapolated to 163 K (Gaia’s operating temperature). This value is 2.80 $\times$ 10${}^{-4}$ electrons pixel${}^{-1}$ s${}^{-1}$.

• •

  Saturation: RVS CCDs have TDI gates (Table 1.3) but they are used only when acquiring bias non-uniformity calibration data. Because TDI gates are not used in nominal RVS observations (due to lower light levels than the other Gaia instruments), the saturation level of each column does not need to be measured as input for when to trigger TDI gates for bright observations to prevent saturation. Instead, the few brightest stars in the sky saturate pixels within the 10 columns that make up a 2D RVS window. Observations with on-board $G_{\mathrm{RVS}}$  mag have 2D windows where every pixel is read out as a sample (Table 1.2). Observations with on-board $G_{\mathrm{RVS}}$ $>7$ mag have 1D windows, where 10 pixels in the AC direction are binned at read-out as a sample. The saturated samples are flagged in the RVS pipeline. The windows including saturated samples are excluded from Calibration and from Scatter, while up to 100 saturated samples are allowed in FullExtraction and 40 in STAMTA, where these samples are masked.