SPring-8, the large synchrotron radiation facility

Release Date

27 Feb, 2024

27 February 2024
Japan Synchrotron Radiation Research Institute (JASRI)
RIKEN

In the planned SPring-8-II facility, the intensity of high-energy synchrotron radiation X-rays in the 50–100 keV range is expected to increase by more than 100 times compared to the current SPring-8. To accurately measure X-ray diffraction and scattering using such high-intensity, high-energy X-rays, photon-counting two-dimensional detectors, which allow the setting of energy thresholds for X-rays, are considered suitable. This is because these detectors can effectively eliminate background noise—such as fluorescence from the sample and Compton scattering—that degrades data accuracy in high-energy X-ray measurements by applying energy thresholds.
However, even state-of-the-art photon-counting two-dimensional detectors experience photon pile-up at high count rates, leading to missed photon events. Therefore, these detectors are typically used with count-loss correction. Although it was qualitatively known that the accuracy (linearity) of corrected data varies with the bunch mode, a quantitative evaluation had not been conducted. As such, a quantitative assessment of the relationship between bunch modes and data accuracy was needed.

SPring-8 currently offers eight types of bunch modes (Modes A to H) for synchrotron radiation X-rays. Figure 1 illustrates the electron storage patterns in the storage ring for each mode. Since each bunch mode features a different time interval between X-ray pulses reaching the sample, their time structures vary. For example, in Mode A, as shown in the timing chart in Figure 2(a), X-ray pulses of several tens of picoseconds arrive every ~23.6 nanoseconds (ns). In Mode B, as shown in Figure 2(b), four consecutive pulses are emitted every 2 ns, followed by a 51.1 ns gap, and the pattern repeats. These timing structures, determined by the bunch mode, are critical for experiments that utilize the temporal characteristics of synchrotron X-ray pulses.

In this study, the researchers simulated the response of a photon-counting detector to the eight bunch modes (A to H) at SPring-8 using the Monte Carlo method. They determined the upper limit of reliable X-ray intensity for each mode.

Figure 3 shows an example of simulation results for detector response. In this case, the dead time of the detector is assumed to be 120 ns, and its response follows a paralyzable model. The horizontal axis represents the input X-ray intensity, while the vertical axis shows the detector’s output. The results reveal that the detector’s response varies significantly depending on the bunch mode. (An ideal detector would produce a straight line with a slope of 1, as shown by the red line.)
In this study, the upper limit of reliable X-ray intensity was defined as the maximum intensity at which the post-correction error remains below 1%, and this was termed the effective maximum count rate. In experiments requiring 1% accuracy, this defines the maximum X-ray intensity the detector can effectively handle. The study quantitatively revealed how this effective maximum count rate varies across different bunch modes.

Table 1 lists, and Figure 4 illustrates, the effective maximum count rates for the eight bunch modes and for an ideal perfect multi-bunch mode in which electrons are evenly distributed throughout the entire ring.

For a detector with a 120 ns dead time, in Mode A, accurate data with ≤1% error can be obtained if the input X-ray intensity is less than 0.916 Mcps/pixel (million counts per second per pixel). In contrast, in Mode F, the effective maximum count rate was only 0.012 Mcps/pixel, about 1/76 of that in Mode A. Thus, if 1% accuracy is required in high-intensity experiments, the X-ray intensity may need to be reduced in Mode F. Even if the detector itself can tolerate high count rates, it may not be effectively usable in Mode F. This study clarified how the acceptable maximum X-ray intensity depends on the bunch mode and the detector’s dead time, which will help ensure the acquisition of accurate data under strong X-ray conditions.

Furthermore, by reevaluating the bunch mode time structure—such as using a perfect multi-bunch mode—with the detector’s dead time in mind, it is possible to obtain reliable data even at higher intensities. This is because the bunch mode time structure has a significant impact on the effective maximum count rate of photon-counting detectors. Therefore, when determining the bunch structure for SPring-8-II, it is essential to consider not only the efficiency of time-resolved experiments but also the operational efficiency of the many photon-counting detectors in use across the beamlines.

The researchers have also developed a web application that allows users to easily simulate the count-loss characteristics of photon-counting detectors for any given bunch mode. The application is opened for public.

At SPring-8-II, where light source performance will be dramatically improved, ultra-precise measurements that currently take a full day are expected to be completed within several tens of minutes. This will enable new experimental approaches, such as large-scale measurements of over 1,000 samples or rapid temperature-change measurements. In such high-speed experiments, accuracy at high count rates is critical. This study provides fundamental insights for operating photon-counting two-dimensional detectors with high accuracy in high-energy synchrotron X-ray experiments. Based on these findings, further advancements are anticipated in research using high-intensity, high-energy X-rays at SPring-8-II.

Fig.1 Filling patterns of electrons in eight different bunch-modes at SPring-8

Fig. 2 Timing charts of synchrotron X-ray pulses in SPring-8 bunch-modes, A-mode (a) and B-mode (b), showing up to 137 nanoseconds.

Fig. 3 Response curves of a photon-counting detector obtained by simulation. The dead time is 120 nanoseconds. The detector response is assumed to follow the paralyzable model. (Multi) represents the complete multi-bunch mode (not currently operated at SPring-8).

Tab. 1 Effective maximum counting rates for SPring-8 bunch modes calculated for three different dead times (120 nanoseconds, 0.5 microseconds, and 3 microseconds). (Multi) is the complete multi-bunch mode in which electrons are equally accumulated all around the storage ring (not currently operated at SPring-8).

Fig. 4 Graphical representation of Table 1, showing the effective maximum count rates for SPring-8 bunch modes, calculated for three different dead times (120 nanoseconds, 0.5 microseconds, and 3 microseconds). (Multi) is the complete multi-bunch mode in which electrons are equally accumulated all around the storage ring (not currently operated at SPring-8).