diffusion_validation

Effect: Diffusion

During the propagation of electrons inside the silicon, they are repeatedly scattered that leads to their displacement in the pixel plane.

Contact person(s) if any:

Craig Lage, Andrei Nomerotski/SAWG

Reference Material:

Description of theoretical model used to estimate electron diffusion amplitude is given in Craig Lage' document on Poisson_CCD22 low-level simulator. Discussion of whether this model properly describes the diffusion of Fe66 hits seen in the experiments is presented in the following talks by Sergey Karpov - 1, 2.

Data Provenance:

Data for validation are taken from the following sensors tested at BNL under different bias voltages:

• ITL-3800C-092
• ITL-3800C-095
• ITL-3800C-307
• E2V-CCD250-107

Model Details:

The amount of diffusion to apply is defined in GalSim:galsim/sensor.py, inside _calculate_diff_step function, as

# 0.026 is kT/q at room temp (298 K)
diff_step = np.sqrt(2 * 0.026 * CCDTemperature / 298.0 / Vdiff) * SensorThickness

It implements the diffusion model explained in detail in Appendix E of this document.

This model is known to underestimate the diffusion by ~2 times in respect to experimental data on Fe55 hits, so the use of an effective electron mass correction from Green (1989) has been proposed by Craig Lage. However, as of Jul 20, 2018, this correction is still not incorporated into GalSim.

Validation Criteria:

The amount of diffusion under different experimental setups is measured by fitting the Fe55 hit marks in acquired images with Gaussian model using NGMIX code by Erin Sheldon. The validity of fitting procedure for significantly undersampled case of Fe55 hits has been specifically studied by generating and pixelizing the clouds of electrons with different sizes and sub-pixel center positions, fitting them in the same way as Fe55 hits, and then comparing the results with original sizes. The procedure has been found to be unbiased in the parameter region of interest for Fe55 fitting, and therefore applicable for the validation.

The following distributions have been obtained for the sizes of Fe55 hits under different bias voltages on different CCDs:

Validation Results:

Current (as of Jul 20, 2018) GalSim version uses the value of diffusion inconsistent with the Fe55 results shown above.

Relevant Project Team for input if any:

Camera

Release and approval log:

07/20/18 - Initial Version - Sergey Karpov