Mechanical structure

The core of the MBM is a grid of multi-clad square 1-mm-wide SCSF-3HF scintillating fibers produced by Kuraray [32, 33] coupled at one end to an S13360-1350PE SiPM from Hamamatsu [34], forming a square beam window arranged along the X and Y axes, where the orthogonal axis is perpendicular to the beamline direction. The effective photosensitive area of the selected SiPM is 1.3 mm× 1.3 mm, which matches the cross-sectional area of the scintillating fiber, and the spectral responses of the scintillating fiber and SiPM used are well matched, which can minimize the loss of photons at the transmission interface. The use of square fibers makes the detector response independent of the position of the muon trajectory inside the fiber and minimizes the dead space. The fibers were cut and polished with a heated tungsten steel cutter [35], and silicone grease was coated on the contact end to further improve photon transport efficiency via the fiber–SiPM interface. The fibers were divided into two perpendicular layers spaced 0.8 mm apart, which can cover a beam window of 30× 30 cm². Each layer was made of 128 fibers with a length of 500 mm and 1.3-mm spacing to measure the beam profile in the two orthogonal directions (X and Y).

Scintillating fiber has a total reflection cladding structure and can achieve extremely high light collection efficiency and effectively improve the detection efficiency of the detector. When charged particles go through a scintillating fiber, energy loss occurs through ionization transfer in the scintillating fiber. Ionization energy loss can be estimated using the Bethe—-Bloch formula [36]. For organic scintillating materials with densities close to 1 g/cm³, the corresponding minimum ionization energy loss dE/dx is ~200 keV/mm [37], which is acceptable for a high-intensity muon beam with an energy of $\mathcal{O}(10)$ MeV. Simultaneously, the mesh structure can further reduce the average energy loss of the muon beam such that such a detector structure can achieve quasi-noninvasive effects. Using this quasi-noninvasive design and a pure geometric calculation, we observe that 60% of the muons passing through the beam window will be recorded, whereas 16% of the muons will trigger both layers and generate coincident signals. The signals will provide the hit map of the muons and allow characterization of the beam profile.

All the scintillating fibers and their corresponding modules were fixed on an aluminum (Al) plate, and the grid of the scintillating fibers was additionally shielded in a stainless-steel box as a light shield. The beam window on the stainless-steel box was cut off and covered with two layers of Al foil with a thickness of 50 μm. In addition to the grid of fibers shielded in the stainless-steel box, we installed an electronic system on the Al plate, including two electronic modules and one time logic unit (TLU). The two electronic modules were fixed on the two sides of the detector with the SiPMs coupled to the ends of the fibers. The TLU module was fixed at the corner of the Al plate for easy communication with the electronics on both sides. All electronic boards were covered by a resin shell with a metal layer on the surface, which can prevent dust and reduce occasional single-particle flip damage caused by beam-—particle scattering to electronic components. The whole MBM scheme is illustrated in Fig. 1.

Fig. 1Full-screen View

(Color online) Global setup of the MBM structure. The top left shows the brief grid of scintillating fibers, which in fact covers a beam window of 30 × 30 cm²

Electronic readout

The electronic readout of the MBM includes two main parts: two electronic modules to read out the signals and upload data to the upper computer and one TLU to synchronize the clock time of the two electronic modules. Each electronic module contains three parts: a SiPM carrier, a frontend electronics board (FEB), and a data acquisition (DAQ) board. The SiPM carrier contained 128 SiPMs arranged in a 16 × 8 array and was connected to the FEB with an edge connector. The SiPM transforms photons to photoelectrons and generates a current signal, which is then integrated using a charge-sensitive preamplifier. A pulse signal with a fixed time length and voltage of 3.3 V was produced and processed in the FEB. A power supply module was installed on the SiPM carrier to ensure stable power to all 128 channels. The power supply module was equipped with a built-in temperature compensation system that could adjust the voltage according to the external temperature, thereby improving the stability of the SiPM in different environments.

The FEB board comprises two application-specific integrated circuit (ASIC) chips [38] and a field-programmable gate array (FPGA). The ASIC chip has 64 channels, each of which includes a charge-sensitive preamplifier, a CR-—RC shaping circuit, a screener, and other modules. The ASIC chip, which is connected to the TLU in real time through HDMI, amplifies the signals and transmits them to the FPGA. The TLU sends a clock and sampling start signal to the FEB, combining the signal with the timestamp generated by the timer inside the FEB to achieve time synchronization of SiPMs. The FPGA will pack the data and send them to the DAQ board, which transfers the data to the switch, and the data are stored in the computer (or MIDAS Bank) for further processing. We also developed graph interface software that can receive data from the MBM and draw histograms for trigger rates in X and Y directions to monitor the MBM running status in real time.

Owing to the limit of the ASIC chip, the electronics system can only record the time information triggered by the signal while the charge information remains blank. Fortunately, counting and recording the signals that pass through the threshold are sufficient to measure the beam profile and time structure. However, because SiPM characteristics such as the dark count rate and quantum efficiency can quite different, the response of the 256 channels of the MBM would also be diverse, which would degrade detector performance. In this case, we carefully designed the electronic readout configuration such that it was possible to modify the threshold of each channel and adjust all channels in uniform responses to the signals. For each channel, the dark count rate was maintained at a low level of ~1 Hz by calibrating the threshold value. Therefore, we expect a high signal-to-noise ratio in the beam-monitoring run.

Validation of detector response

After assembling the MBM, it is necessary to validate the working status of the detector before installing it on the beamline. Therefore, we continuously recorded cosmic-ray data for ~3 h. With this test, we are able to validate two questions: 1. We checked whether the detector can run properly in the whole period; 2. by comparing the collected data with our rough estimates, we checked whether the detector can record the charged particles (cosmic muons) with high efficiency.

For the first purpose, we monitored the working status of the detector in the long run to identify stable detector data without any breaks, as shown by the trigger rate for this period in Fig. 2a.

Fig. 2Full-screen View

(Color online) Tests with cosmic-ray muons in the local laboratory. (a) Hit rate of cosmic-ray muons recorded by the MBM detector. (b) Hit map caused by cosmic-ray muons

Second, we analyzed the recorded data and compared them with our expectations. The cosmic-ray muon flux at sea level is ~1 min^-1 cm^-2 [39]; therefore, we must expect ~3 Hz in the MBM. In the actual configuration, the dark count rate was reduced to 1 Hz for each channel by setting an appropriate threshold. Therefore, the trigger rate of all channels in the MBM remains high and at the level of $\mathcal{O}(100)$ Hz, which is approximately two orders of magnitude higher than the naive expectation from cosmic-ray events. Fortunately, in this case, we can switch to coincident measurement and collect cosmic-ray muon events by restricting the X and Y layers triggered within 20 ns, which is 3σ of the fiber decay time. Based on a coincidence time window of 20 ns, the accidental coincidence count rate caused by dark noise in two orthogonal scintillating fibers is 6.6 × 10^-4 Hz, which is much lower than the muon count rate of 2.8 Hz based on Monte Carlo simulation results. Therefore, the dark noise frequency is acceptable.

Finally, satisfactory results are obtained, as shown in Fig. 2. In Fig. 2a, we observe that the coincidence-triggered rate is ~3 Hz, which is consistent with our expected results. In addition, the rate of triggered events was stable in the plot, which indicates the stability of the detector in a relatively long run. In addition, the detector response was almost uniform, indicating good performance for all channels in the MBM (see Fig. 2b). During detector operation, the power was maintained at ~60 W, although it fluctuated slightly during different runs. Based on the above observations, we conclude that the MBM can operate stably for a long time and has the required functionalities to achieve physics objectives.

Proton beam test at the CSNS

To understand the detector response to a high-intensity beam, we conducted a quick beam test using the Associated Proton beam Experiment Platform (APEP) at the CSNS on January 9, 2023 [40]. The detector was installed on the platform on a movable bracket that could move it on the X–Y plane with accurate positioning. Prior to the beam tests, a laser collimator was used to align the beam window to match the center of the beam.

In the APEP setup, the extracted proton beam has a repetition rate of 25 Hz with a 400-μs beam pulse length for each bunch. A suite of degraders was placed downstream of the proton beam window, which allowed us to adjust the proton energy in the range of 10–80 MeV. A collimation system was installed on the beamline to adjust the beam spot size and intensity. Using this system, we can tune the uniform proton beam spot sizes at both the vacuum and air test points from 10 mm × 10 mm to 50 mm × 50 mm. However, the scattering effect of atmospheric molecules on the beam spot in a nonvacuum experimental environment results in an actual beam spot size that is significantly larger than the original size inside the vacuum beamline. For example, the setup of the collimation system of 10 mm × 10 mm would result in a beam spot size of ~70 mm × 70 mm at the air test point [40].

We have two purposes for the beam test. The first objective is to validate the ability of the MBM detector to record the time structure of the beam. Considering the 25-Hz repetition rate of the beam, we selected data within 280 ms, including eight pulses. The time structure is shown in Fig. 3a, and we observe a clear time structure with a regular peak every 40 ms, which is consistent with the 25-Hz beam repetition rate. The other purpose is to validate the ability of the detector to measure the profile of the beam. Therefore, we must draw a two-dimensional plot by selecting the X–Y coincidence events.

Fig. 3Full-screen View

(Color online) Beam test results with a high-intensity 80-MeV proton beam at CSNS. (a) Timing structure of the proton beam at CSNS. (b) Beam spot during CSNS beam tests

However, during the beam test, we found that radiation seriously affected some scintillating fibers operating in the 80-MeV proton beam for a long time, in which almost all channels got triggered continuously at an ultrahigh rate, especially for the previously scanned points. In this case, the true proton events became seriously polluted with the radiation background. Therefore, stricter cuts had to be made to obtain a clean triggered signal:

• The event should be triggered within 10 ns on the X and Y layers.

• The trigger time of the event should be in the range of the beam’s spill time.

Using these cuts, we obtained the results shown in Fig. 3b. In the plot, we can observe a clear beam spot at the top center, which is the exact location of the beam arrangement. The size of the beam spot was ~70 mm × 70 mm, which is consistent with the simulation results provided by the CSNS [40]. Based on these tests, we can confirm that the MBM detector has a good response to the beam particles and can record the timing structure and profile characteristics of the beam quite well.

COMET phase-α commissioning

To measure the characteristics of proton beams and π and μ production with less ambiguity, which is essential for COMET experiments, the collaborating group proposed a low-intensity beam run called phase-α commissioning. An essential goal of this run was to monitor and measure the properties of muon beams.

The phase-α detector system comprised four subdetectors: the MBM, the straw tube tracker, the range counter (RC), and the proton beam monitor. As the first detector at the backend of the muon beam, the MBM was used to monitor the timing structure and position information of the muon beam and to provide a reference for the downstream detector and beamline quality. The straw tube tracker was installed after the MBM and its task was to measure the position and direction of the injected particles [41]. Because the straw rube tracker is part of the detection system in COMET phase I, phase-α commissioning was used to validate its performance in the beam environment, especially for its electronics system.

At the end of this system is the RC detector, which consists of a series of several plastic scintillators coupled with photomultiplier tubes, aiming to measure the deposited energy and the hit time of the muon events. In addition, the electronic system of the RC detector allows it to record approximately a 10 μs photomultiplier tube waveform; thus, it can measure the decay time of muons decayed in orbit events. The proton beam monitor was installed on a proton beamline and provided real-time proton-beam monitoring. Furthermore, a dedicated beam-masking system was installed upstream of the transport solenoid entrance to study the optics and beam dynamics of the transport solenoid.

3.2.2

Experimental results

For phase-α commissioning, the goal of MBM included two aspects: 1. characterizing the time structure of the muon beam and comparing it with the bunched proton time structure from the J-PARC MR; 2. measuring the muon beam spatial profile to gain a deeper understanding of the beam production and transportation. Moreover, assessing the operational stability of the MBM in harsh environments is essential in the long run.

The timing structure of the muon beam in the COMET is one of the key questions to address. As previously mentioned, the COMET experiment requires a dedicated 8-GeV bunched proton beam that generates the secondary muon beam. In the design of the COMET experiment, the J-PARC MR synchrotron accelerates the protons cycle by cycle. There were nine beam buckets in the MR, with each bucket separated by 586 ns. The beam buckets were not filled completely to obtain the pulsed beam required for the experiment. The MR cycle consists of three bunches (type a) spaced by 1.17 μs and one bunch spaced by 1.76 μs (type b), which are used to effectively reduce the physical background caused by other beam particles. Owing to the low beam intensity in the phase-α commissioning, there are, on average, only 0.02 events recorded by the MBM in one bunch; thus, many events must be superimposed to obtain a plot with a visible structure. However, because of the limited timing resolution from the MR, we were unable to confirm the precise serial number of bunches for the triggered hit. Therefore, the time structure is a superposition of four possibilities (aaab, aaba, abaa, and baaa). Consequently, we expect a time structure with one large peak and six small peaks generated by the misalignment of the beam structure after superposition. The large peak should be spaced at 1.1 μs compared to the small peaks, whereas the small peaks should be separated by 586 ns in the enlarged time window, which is consistent with the MR beam structure. Accumulating data from ~800 runs enabled us to draw the timing structure of the muon beam, as shown in Fig. 4, which meets our expectations.

Fig. 4Full-screen View

Monitoring the timing structure of the muon beam during phase-α commissioning. (a) Overview of the timing structure. (b)Details of the peak fittings in a short time window

Another important concern is the muon-beam profile of the COMET. We obtained a two-dimensional profile of the muon beam, as shown in Fig 5. In the actual beam trial operation stage, the measured beam profile is not completely circular as a result of interference from many factors. One of the reasons for this is that the magnetic field in the transport solenoid was set to 1.5 T, which is half of that originally designed in the COMET phase I design. In addition, the detectors in the phase-α commissioning were installed out of the exit of the transport solenoid without any magnetic field and scattering in the air had an effect. Moreover, the beam contains several kinds of charged particles, including e^-, μ^-, and π^-, each of which has a different beam direction. Based on the above analysis, the beam was assumed to be widely spread. This results in different beam spots in the beam profile.

Fig. 5Full-screen View

(Color online) Two-dimensional muon beam profile during COMET phase-α commissioning

Phase-α commissioning started from March 3 to 13 with a 1.5-T magnetic field. During this period, the muon beam operated stably, and detailed measurements of the time and beam profile structures were conducted. The results are reflected by the stable beam spot, as shown in Fig. 6. To obtain more comprehensive beam profile features, we drew the barycenter of the beam profile and the counting rate of the detector, which also indicates the long-term stability of the detector during the entire phase-α commissioning.

Fig. 6Full-screen View

Long-term stability of the muon beam. (a) Change of center location during the beam time. (b) Change in hit rates during the beam time

After the successful run in the muon mode, we turned off the magnetic field for several hours and then inverted the direction of the magnetic field and measured the antimuon component in the beam (called the “Mu+ Run,’’) to study the beam component and the effect of the magnetic field. In addition, we moved the location of the beam-masking system several times during the Mu+ Run period to validate the beam optics and dynamics in the curved transport solenoid. Figure 6 shows the location of the beam center and the stable trigger rate most of the time. It should be noted that the beam jumps several times after Run 3257, which is caused by the swapping of the magnetic field polarization and the movement of the beam-masking system. In general, the COMET muon beam runs stably, and the MBM reflects the running status of the beam quite well.

To cross-check whether the detector can correctly identify the beam position, we also conducted a coincidence measurement with the RC by selecting the first hits in every event. Consequently, we obtained the hit maps shown in Fig. 7. We can observe a clear hot zone in the plots, which indicates the position of the RC.

Fig. 7Full-screen View

(Color online) MBM–RC coincidence hit maps for when the RC is (a) located in the bottom right and (b) moved to the top left

We moved the position of the RC to several different locations, allowing us to validate the MBM’s response to the beam location. We recorded the relative location of the RC during this period and drew the MBM–RC coincidence beam profile using the corresponding data. In Fig. 7, the beam spot is located at the bottom right (Fig. 7a and top left (Fig. 7b, which corresponds to RC movement from the bottom right to the top left. Based on these results, we verified the accuracy of the MBM response to the beams.