Fig. 13
From: An applied noise model for scintillation-based CCD detectors in transmission electron microscopy

(a) Autocovariance function of the signal according to Eq. 84. The central value, altered by Eq. 22 due to correlation, yields \(\beta \cdot g = 1.55025 \pm 0.00002\). The red box marks the nearest neighbor coefficients that were used to calculate (b), which depicts the relative change in the smoothing factor \(\beta _{conv}\) when changing the intensity of the signal relative to the signal of the gain reference. The red curve depicts a Padé approximation of order [4/5]67. It can be observed that the smoothing factor decreases sharply for very low and decreases rather linearly with increasing intensities. In (c) the PSF of the brighter-fatter effect is shown for a spike signal of 30k counts. The PSF was reconstructed by the smoothing value of \(\beta _{BF} \approx 0.995\) counts obtained in (b), by reversing Eq. 40 and utilizing Eq. 37. In (d), we show the effect under vertical summation, as it is performed e.g. in EELS. It can be seen that the effect is negligible and well below the expected Poisson noise, as only 75 counts migrate to the neighboring channels as a result of the brighter-fatter effect.