Charge Summing Mode
Every photon counts – and counts only once!
CONTACT
X-Spectrum GmbH
Luruper Hauptstraße 1
22547 Hamburg
Germany
Phone: +49 (0) 40 401 130 40
E-Mail: info@x-spectrum.de
The small pixel size of the LAMBDA detectors enables best-in-class resolution, making the X-Spectrum detectors ideal for applications in synchrotrons, non- destructive testing, and electron microscopy. However, due to this small pixel size, the charge clouds generated in the X-ray sensor are prone to spread across several pixels. This phenomenon, known as charge sharing, can result in a significant overcounting of photons, especially when the detector is exposed to high-energy X- rays.
To address this issue, the LAMBDA detectors are equipped with Charge Summing Mode (CSM). In this mode, the charge shared among neighboring pixels is accounted for, ensuring that every photon counts – and counts only once!
Below, we highlight the key aspects of this feature and address the most frequently asked questions to help our customers get the best out of our detectors.
How does CSM work?
When the CSM is activated and a photon hits the detector, the photon is only counted if two conditions are fulfilled:
1. Local Maximum Condition:
Suppose that a photon hits near a pixel’s corner, resulting in some of its charge spreading to neighboring pixels. LAMBDA detectors are equipped with arbitration circuitry that ensures only one pixel registers the hit. For an event to be counted in a pixel, the signal in this pixel must be larger than the signals of the 8 adjacent pixels, making it a local maximum. This is achieved by comparing the pulse durations in the pixels. This condition prevents a photon’s hit from being registered in multiple pixels at the same time.
2. Charge-Summing Condition:
Each 2×2 pixel block is equipped with a charge-summing node, located at its corners. These nodes measure the total signal within the block and compare the summed energy to a predefined threshold (upper threshold), which represents the minimum energy expected from a photon under the given experimental conditions. For a pixel to register a hit in CSM, at least one of the nodes in the surrounding 2×2 blocks must exceed this threshold. Since each pixel belongs to four overlapping 2×2 blocks, it will register a hit if the signal in at least one of these blocks surpasses the threshold. Furthermore, each pixel is equipped with a lower threshold, which serves to distinguish true signal from noise and suppress low-energy background radiation.
These conditions prevent multiple pixels from registering the same photon and ensure that the charge is accurately detected across neighboring pixels. A schematic representation of the CSM is presented in Figure 2.
.
Figure 2 – A photon (black arrow) that hits a pixel corner has its charge cloud (orange circle) spread among different pixels. The CSM nodes (red circles) ensure that if the Local-maximum and Charge- Summing conditions are fulfilled, the count is allocated to the pixel with the highest fraction of the total charge (blue pixel).
When is CSM recommended?
1. Improved spatial resolution:
The reduction of the charge sharing phenomenon leads to an improved spatial resolution.
In Figure 3, we present a simple example in which a slit with a 50 μm opening was measured under identical experimental conditions, first using SPM and then CSM. The X-ray source was polychromatic, operating at 60kV and 30 mA.
The images recorded are shown on the top of Figure 3. Ten lines perpendicular to the slit were analyzed in each case, as indicated by the blue and red lines, to determine the spatial resolution obtained in each case. The higher counts observed with SPM compared to CSM are attributed to double-counting caused by charge sharing when using the SPM.
By plotting the normalized intensity per pixel across the ten lines for each configuration (Figure 3, bottom), it becomes evident that CSM provides higher spatial resolution compared to SPM. In this context, resolution refers to how sharply the slit is observed in the measurement. The higher the resolution, the narrower the gradient region over which counts originating from the slit are distributed.
The measured resolution improved from 88.6 μm with SPM to 82.4 μm with CSM, representing an improvement of approximately 6 μm solely from the use of CSM. Note that the measured spatial resolution in this experiment was limited by the large beam focus size employed (0.2 mm × 12 mm). In case of high-resolution X-ray setups, such as synchrotron sources or X-ray tube sources with micro/nano focus, the CSM mode can significantly enhance the quality of the experimental data.
Figure 3 – Comparison of spatial resolution achieved with SPM and CSM. Both modes were used to measure a 50 μm slit under identical experimental conditions.
2. Separating a Signal from Background:
When background radiation is present in a measurement, CSM can significantly improve the separation of the signal of interest from unwanted contributions. To demonstrate the advantage of CSM over SPM in such scenarios, we conducted the following experiment:
Using an X-ray source, we positioned two different fluorescent foils in front of the detector to simulate signal and background components. The foils and their respective fluorescence energies were:
- Mo (Molybdenum) – 17 keV (acting as background)
- Ag (Silver) – 22 keV (acting as the signal)
The energy separation between background and signal is only 5 keV. This narrowgap requires a careful threshold selection to isolate the Ag signal while minimizing Mo contribution, without excessive signal loss.
The experimental setup is illustrated in Figure 4:
Figure 4 – Experimental setup for measuring a mixed signal composed of a signal of interest and a background signal using Molybdenum and Silver reflective foils. (a) Background measurement: only shutter a is open, allowing detection of the Mo signal. (b) Signal measurement: only shutter b is open, allowing detection of the signal of interest, represented by Ag. (c) The two shutters are opened simultaneously, mixing the signals.
First, only shutter a was opened, and the detector collected data corresponding to the background. In the second step, only shutter b was opened, to acquire the pure signal. A foil with a pinhole was placed between shutter b and the detector to enhance the visual distinction between the signal and the background. Finally, both shutters a and b were opened simultaneously, and measurements were acquired using both SPM and CSM modes. A comparison of the results is presented below.
In the first step, only the SPM was activated, and different energy thresholds were tested to determine which setting provides the highest overall counts with minimal noise and which threshold achieves better signal separation while maintaining a high count rate. Figure 5 presents the best-performing threshold values.
Figure 5 – (a) Background measured using a low threshold to eliminate noise while maintaining high counts, (b) Combined signal (Ag + Mo) measurements acquired using SPM with low energy threshold, (c) Combined signal (Ag + Mo) measurements acquired using SPM with high energy threshold and (d) comparison of the three configurations presenting signal over background (S/B) values.
A region of interest (ROI) of 250 pixels in length and 16 pixels in width, centered on the pinhole opening, was selected for further analysis (red rectangle in Figure 5b). For each line of the detector, the counts from the central 16 pixels were averaged. The resulting profiles were then plotted for the different acquisition scenarios.
Figure 5d presents the results obtained from this analysis. The mean intensity is shown as a function of pixel line position for the SPM under three conditions: (a) background only, (b) background plus signal with a low threshold, and (c) background plus signal with a high threshold.
It should be noted that the high threshold energy lies between the Mo and Ag fluorescence energies. As a result, the Mo signal is excluded from the measurement, leading to a lower total count. Nevertheless, although the total number of counts decreases with the high threshold, the signal-to-background ratio (S/B) is improved compared to the low-threshold case, allowing for the separation of Ag and Mo signals.
A similar analysis was performed using CSM. In this case, two thresholds were applied at 7 keV and 20 keV. Under this condition, the counting of a photon follows the condition: photons with energies below 7 keV were completely rejected, while pixels registering energies above 7 keV were counted only if the energies of their four neighboring pixels also exceeded 20 keV.
The pure Ag signal was also measured using CSM (red stars in Figure 6-d). A comparison of the background, Ag signal, and mixed signals is presented in Figure 6.
Figure 6 – CSM measurements of (a) Mo background, (b) pure Ag signal, (c) combined signal (Ag + Mo) and (d) comparison of the three configurations presenting signal over background (S/B) values.
In Figure 7-d, a direct comparison is shown between the pure signal (red stars), the signal obtained with SPM (green), and the signal obtained with CSM (purple) when both signals are mixed. The results clearly demonstrate the superior signal-to- background ratio provided by CSM.
Figure 7 – Comparison between SPM and CSM for isolating the Ag signal from a Mo background. The pure Ag signal is presented by the red stars, alongside the mixed signals processed with SPM (green line) and CSM (purple line). CSM clearly demonstrates superior separation capability, particularly when the signal and background energies are close.
This example demonstrates the improvement of energy resolution when using CSM. This is particularly important when dual-threshold measurements are performed, as CSM allows for a more precise separation of high-energy photons.
At high energies, using SPM in dual-threshold mode can lead to an overcounting of photons when the subtraction of signals is performed. Thus, a high-energy photon that has its energy spread among several pixels can be counted as several low-energy photons.
3. Countcorrection:
At high energies (see below for more details about each sensor recommendation), using the CSM correct the miscounting of photons caused by charge sharing among pixels.
How to implement the CSM in your measurement?
The CSM is available for 12- and 24-bit mode for high-Z sensor detectors. To apply CSM when acquiring data:
Before acquiring an image.
- Turn the charge summing mode selection on
- Set the lower threshold. The lower threshold is related to the local maximum
and can be modified to account 1⁄4 of the beam energy to avoid overcounting. Here you may consider possible noise/ fluorescence of your sample and set the minimum threshold above this value. The standard values used to suppress noise is 7 keV.
- Set the upper threshold: The Upper threshold is related to the sum of the neighboring pixels. To limit charge sharing losses, we recommend operating with threshold about 5 keV below the expected event energy. If a tighter energy cutoff is required, the threshold can be increased, but this will result in an overall loss of counts. Standard recommended values are above 15 keV.
- Here is an example at X-Spectrum GUI:
Figure 8 – X-Spectrum GUI showing CSM parameters.
When to NOT use CSM?
Although CSM can improve spatial resolution and correct the number of counts, care must be taken when using this feature. Below we exemplify some situations in which the use of CSM is not recommended:
1. Low minimum threshold:
When CSM is applied, the threshold must be set higher than in SPM. This is due to the noise contributions from the four neighboring pixels, which are also summed when CSM is activated, increasing the overall noise level. As a result, CSM is not suitable for applications that require low-energy threshold detection.
2. High Flux:
When using CSM, the linearity between the detected signal and the photon flux degrades sooner than inSPM. This is particularly relevant in high flux experiments, where CSM loses linearity approximately ten times earlier than SPM. Therefore, CSM is not recommended for fluxes exceeding 107 photons/mm2/s, as signal distortion and non-linear responses can significantly impact data accuracy under such conditions.
References for further reading
KOENIG, Thomas, et al. Charge summing in spectroscopic x-ray detectors with high- Z sensors. IEEE Transactions on Nuclear Science, 2013, 60.6: 4713-4718.
PENNICARD, D., et al. Simulations of charge summing and threshold dispersion effects in Medipix3. Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, 2011, 636.1: 74- 81.