Detection structure

Figure 4(a) presents a comprehensive structural diagram of the detector system, offering clarity on its fundamental configuration. The system operates as follows: The target sample is introduced directly into the detection chamber through an injection port, pump, and flowmeter. Once inside the detection chamber, the sample undergoes a measurement process of specific duration. The detector functions by transferring energy from beta rays within the aqueous medium to plastic scintillating fibers, leading to the emission of scintillation photons. These photons then travel through the plastic scintillation fiber array and water medium, ultimately reaching the photomultiplier tubes, where they are converted into electrical signals. These signals undergo further processing to derive the desired measurement outcomes. After the measurement process is complete, the sample is discharged from the detection chamber and guided through a pipeline and flowmeter to a waste liquid tank for proper disposal.

Fig. 4Full-screen View

Design of the system and detector: (a) Design of the system, (b) Design of the detector

Figure 4(b) presents a detailed illustration of the probe chamber’s structural arrangement. The plastic scintillation fiber array is positioned within the probe chamber using a grid system, ensuring that the PSFs’ ends are close to the probe window. This placement minimizes water sample retention in the central region of the chamber. An airtight connection between the photomultiplier tubes and the probe is established using optical silicone. Additionally, the tubes are precisely aligned at both ends, aligning them with the chamber’s geometric center. Efforts are made to direct the flow of water samples toward the lower end of the detection chamber to facilitate their exit from the upper end. Maintaining a predominantly horizontal flow is crucial to prevent bubble formation and potential interference within the chamber.

Signal analysis

The LeCroy-640Zi oscilloscope is an indispensable tool in our signal analysis process, especially after assembling the device under study. It utilizes continuous sampling over 1600 s to thoroughly analyze the detector’s signals.

Our experimental setup involves the decay of tritiated water, which emits low-energy β-rays that interact with the plastic scintillator. However, due to limited energy transfer, fewer scintillation photons are produced, reducing the probability of simultaneous detection by the photomultiplier tubes located at both ends of the scintillator. Additionally, the quantum efficiency of the PMTs is diminished in these circumstances, resulting in relatively small amplitude signals that can be easily overshadowed by the inherent dark current noise in the PMTs. To address this issue, our system employs two PMTs for coincidental measurements at both ends of the plastic scintillator. Assuming that the dark current is randomly generated and that the scintillation photons propagate simultaneously towards both ends, we expect the photons to coincidentally reach the photocathode of each PMT and be detected. This approach ensures accurate capture of genuine coincidence signals while effectively suppressing the influence of interspersed dark-current noise [32, 33].

However, simultaneous detection of the dark current generated by both PMTs by their respective photocathodes can lead to false coincidence events. Long-term experiments should be conducted to assess the effect of these false coincidences on measurement results. Based on the observed curve and the relationship between coincidence counting and dark current during our experiments, we adjusted the voltage levels of the PMTs to 1000 V and 1050 V, respectively. This adjustment takes into account the expected performance variability among PMTs of the same model. In long-term measurements using these PMTs operating at elevated voltages, the coincidence count originating from dark current remained consistently below 1 for 1600 s. Thus, the use of two PMTs in coincidence measurements and voltage adjustment effectively addresses the issue of dark current noise and false coincidence events.

The low energy levels of beta rays emitted during tritium decay, typically on the order of a few thousand electron volts, pose a significant challenge to the detection process, as the weak signal is prone to interference from ambient noise. To address this issue, signal processing is necessary to filter out unwanted noise and extract the tritium detection signal. It should be noted that the high-amplitude end of the detector output signal is primarily occupied by high-energy gamma rays and high-energy cosmic particles, which must be filtered out to avoid overwhelming the detector’s output signal. Conversely, the low-amplitude end contains a mixture of environmental noise and tritium detection signals, making it crucial to analyze and process these signals separately to avoid potential interference between them. Thus, appropriate signal processing techniques need to be employed to distinguish the tritium detection signal from other signals in the low-amplitude end of the output signal. This can be achieved through careful analysis of the signal spectrum and the application of appropriate filters, such as bandpass filters, to isolate the tritium detection signal. Overall, effective signal processing techniques are essential to ensure accurate and reliable detection of tritium, especially in the presence of ambient noise.

The acquired signal data undergoes processing and analysis using specialized software. The initial step in this process involves applying a filter algorithm to the signal to remove any unwanted noise that may have been captured during acquisition. This step is crucial for ensuring the accuracy and reliability of subsequent signal analysis. Following the application of the filter algorithm, a peak search algorithm is utilized to identify peak signal counts from each photomultiplier tube individually. This information is essential for identifying coincidence events and evaluating system performance. Finally, the coincidence algorithm is employed to optimize the delay time between the two signals and minimize background noise counts. This algorithm ensures that the signal meets the experiment’s requirements and remains unaffected by external factors that could compromise its accuracy. Optimizing the delay time reduces the likelihood of false coincidence events and improves measurement accuracy. Overall, the application of these algorithms significantly enhances the reliability and accuracy of our measurements. It establishes a robust framework for processing and analyzing signal data, enabling us to obtain precise and dependable results.

Figure 5 provides a comprehensive depiction of the various constituent elements comprising the detector system. Figure 5(a) represents the photomultiplier tube sleeve, which serves the dual purpose of securely holding and shielding the photomultiplier tube against interference noise. Figure 5(b) showcases the detection chamber, constructed from Teflon material, with dimensions of 90 mm in length and 22 mm in diameter, matching the size of the photomultiplier tube. The window component of the detection chamber is exemplified in Fig. 5(c), comprising a coated acrylic plate characterized by impressive light transmittance exceeding 95%. Moving on to Fig. 5(d), it represents an array consisting of 104 plastic scintillation fibers, measuring 100 mm in length and 1 mm in diameter. The PSF array is meticulously arranged using a hexagonal dense packing configuration, with a spacing of approximately 0.5 mm between each PSF. Lastly, Fig. 5(e) represents the photomultiplier itself, specifically utilizing the R3550 model manufactured by Hamamatsu, Japan.

Fig. 5Full-screen View

(Color online) Physical drawings of the components of the detector: (a) Photomultiplier tube sleeve, (b) the detection chamber, (c) Window of the detection chamber, (d) an array of PSFs, (e) a photomultiplier

Figure 6 illustrates a representative distribution of the signal amplitudes obtained from the measurements of the tritium water samples with an activity of 5000 BqL^-1. The figure clearly shows that this distribution corresponds to the energy spectrum of beta rays emitted during tritium decay. Due to the 1 millivolt amplitude interval used in the oscilloscope sampling, the tritium energy spectrum distribution could not be completely fitted. The analysis of this spectrum revealed distinct characteristics and allowed the identification of tritium signals amid background noise interference. In the high-energy region of the spectrum (above 0.05 V), which predominantly consists of signals with higher amplitudes, the presence of background noise interference is more pronounced. The tritium signals in this region are typically minimal and can be easily overshadowed by noise. Consequently, this high-energy region is commonly excluded from tritium signal analysis to ensure accurate and reliable measurements.

Fig. 6Full-screen View

Typical signal amplitude spectrum measured with the test equipment

However, special attention was paid to optimizing the signal processing algorithm for the low-energy region (below 0.005 V) to mitigate the effects of noise interference. The algorithm was fine-tuned to effectively minimize the noise levels in this region, enhancing the sensitivity to low-amplitude tritium signals. A visual examination of Fig. 6 demonstrates the effectiveness of this optimization, as it reveals minimal fluctuations or disturbances in the low-energy portion of the spectrum, which is often referred to as a weakened signal burr. By excluding the high-energy region with pronounced background noise interference and applying an optimized signal processing algorithm to the low-energy region, we can focus on accurately detecting and analyzing tritium signals within the spectrum. This approach ensured that our measurements maintained a high level of precision and reliability, enabling us to draw meaningful conclusions from the data.

Experimental result and minimum detectable activity

After completing the assembly, debugging, and signal algorithm optimization of the equipment, the minimum detectable activity of the equipment was calibrated. The specific parameters of the equipment are as follows: The test equipment employed in this study incorporated a detection array consisting of 104 plastic scintillation fibers, 100 mm in length and 1 mm in diameter. The PSFs are arranged in a hexagonal dense packing configuration, with a spacing of approximately 0.5 mm between adjacent PSFs. This arrangement ensures optimal coverage and sensitivity of the detection system. The detection chamber, in which the PSF array was housed, had a length of 90 millimeters and a diameter of 22 mm. This chamber can accommodate a water sample volume of 26.03 mL, allowing precise and accurate measurements. To ensure reliable and precise measurements, the experiments were conducted under strict light-avoidance conditions.

Initially, low-background distilled water was introduced into the experimental setup to assess and investigate the background count in the testing environment. The background count is influenced by three primary factors: environmental radiation and noise, supply voltage crosstalk, and the crosstalk count of the photomultiplier. In the absence of adequate lead shielding, the background noise count exhibited fluctuations ranging from 1.12 to 1.25 counts per second during prolonged measurement periods. Consequently, the experiment necessitated multiple measurements to mitigate the influence of background noise on equipment performance. By conducting repeated measurements, we aimed to minimize the impact of background noise and obtain more accurate and reliable experimental data.

Once we determined the experimental method involving multiple measurements, we settled on conducting ten measurements. To ensure the equipment’s sensitivity to tritium water with lower activity levels, we employed a descending gradient activity measurement method during the experimental process. This approach systematically evaluated the equipment’s capability to detect and quantify tritium water samples with decreasing levels of activity, helping us identify the minimum detectable activity and assess the equipment’s performance in effectively capturing and measuring tritium activity even at low levels. For calibration purposes, we used standard tritium water samples to determine the detection efficiency of the equipment. The specific activity of the tritium water samples used in the experiment was calibrated using a liquid scintillation counter (ALOKA LSC-LB7, Japan). Each signal measurement lasted for 1600 s, and the tritium water samples had specific activities of 935 BqL^-1, 4250 BqL^-1, and 6200 BqL^-1, respectively.

The results obtained from the equipment are depicted in Fig. 7 (a). This figure illustrates a strong positive correlation between the counting rate and the activity of the tritium water sample. As the activity of the tritium water sample increases, the counting rate of the equipment also increases accordingly. The detection efficiency is calculated as the ratio of the actual count of the detector to the theoretical total number of the sample. Based on this correlation, the detection efficiency of the device was calculated to be 1.6e-3. This detection efficiency is an order of magnitude higher than that of similar detectors used by other researchers, which is 3.11e-4 [24]. The calibration procedure and the determination of the detection efficiency using standard tritium water samples ensure accurate and reliable measurements throughout the study. With the established detection efficiency, subsequent experimental samples’ tritium activity levels can be quantified with a high degree of confidence.

Fig. 7Full-screen View

Experimental results and Evaluation of the minimum detectable activity (MDA): (a) Experimental results, (b) background count rate and MDA when the measurement time is 1600 s, (c) measurement time and MDA, the red symbols are the performance confirmed by the test device in this study

To measure low-level radioactivity effectively, it’s crucial to minimize the instrument’s Minimum Detectable Activity (MDA) as much as possible. MDA typically characterizes the instrument’s measurement threshold. The calculation formula for MDA is presented in Eq. (1) [34]. $\text{MDA}=\left(2.71+4.65\sqrt{N_{\text{b}}\times T}\right)/\left(\epsilon \times V\times T\right)$ (1) where N_b is the background count rate (cps), T is the measurement time (s), ε is the detection efficiency and V is the sample volume (L). During the experiment, the measurement duration was set to 1600 s, the volume of the water sample was precisely measured at 26.03 mL, and the instrument exhibited a detection efficiency of 1.6e-3. The background count rate registered at 1.25 counts per second. Based on these parameters, the minimum detectable activity of the device was calculated as 3165 BqL^-1.

Using the experimental results and calculated MDA, we can explore the device’s capabilities further. We can calculate the Minimum Detectable Activity (MDA) under varying background noise counts while keeping the measurement time fixed at 1600 s. Additionally, assuming a background noise count of 1.25 cps, we can determine the MDA for different measurement times. This analysis allows us to evaluate the device’s performance under different conditions.

To visually depict the MDA change, Fig. 7 (b) and (c) show the MDA curve with a red line. Specifically, Fig. 7(b) emphasizes the substantial enhancement in the device’s detection performance due to shielding measures. Additionally, Fig. 7(c) illustrates variations in the device’s performance at different measurement times during practical use. These fluctuations underscore the device’s stringent detection performance requirements, especially in short-term measurements where precision and effectiveness are paramount.

The absence of a lead shielding layer in the experimental setup resulted in a noticeable increase in background noise count. While this might have somewhat mitigated noise impact when measuring highly active tritium water samples, it’s important to note that for low-activity tritium water measurement, a lead shielding layer is crucial. Adding a lead shielding layer can significantly enhance detector performance, reducing minimum detectable activity and improving overall accuracy. Given this result, it’s highly advisable to include a lead shielding layer in future experiments. This shield acts as a protective barrier, effectively reducing external interferences and minimizing background noise count. Thus, the detection system will be better equipped to accurately measure and quantify low levels of radioactivity in tritium water samples. Incorporating a lead shielding layer not only enhances detector sensitivity and precision but also ensures reliable and reproducible results. Therefore, for more robust and accurate measurements of low-activity tritium water samples, it is essential to include a lead shielding layer in future experimental setups.

Analysis of experimental results and outlook

The device’s detection efficiency was calculated to be 1.6e-3, while the simulation result for the entire detection system is 4e-4. The actual equipment’s detection efficiency is approximately four times higher than that of the simulated equipment. This significant difference can be attributed to two main reasons. Firstly, inaccurate simulation results of low-energy β rays in the model led to a substantial systematic deviation in the results. Additionally, the actual PSF exhibits a certain arc, increasing the contact volume with the water body and resulting in higher detection efficiency compared to the simulation results. However, it’s important to note that simulation results are generally instructive. Currently, the existing device is in the initial stage of development. With a measurement time of 1600 s, it achieves a minimum detectable activity of 3165 BqL^-1. The water sample volume is 26.03 milliliters, and the sample injection process can be completed within 10 s using a peristaltic pump. Consequently, it can function as a warning measurement device for high-activity tritium water samples, suitable for alarm systems in nuclear power plants and other specialized facilities. The simulation results indicate that increasing the number of plastic scintillation fibers can positively impact both the detection efficiency and volume of the entire system. Additionally, extending the length of the PSF contributes to an increase in the detection volume. Reducing the diameter of the PSF allows for a higher density of PSFs within the same area, thereby enhancing the detection efficiency by augmenting the detection volume. To further improve detection performance, strict shielding methods can be implemented to effectively control the background noise count. Many detection devices have successfully achieved a background noise count between 0.01 and 1 count per second, or even lower. By employing rigorous shielding measures to eliminate interference from ambient noise counts and restrict the background noise count to below 0.1 cps, the equipment’s detection performance can be significantly enhanced. This reduction in background noise count also leads to a decrease in the minimum detectable activity of the device. Based on the above conditions and simulation results, it is predicted that with a measurement time set at 1600 s, the MDA of the device will be below 50 BqL^-1. Notably, China mandates that the concentration of tritium in water 1 km downstream of inland nuclear power plants should not exceed 100 BqL^-1, while the European Council Directive 2013/51/Euratom requires a maximum tritium level in water for human consumption to be lower than 100 BqL^-1. We are actively engaged in refining the equipment and conducting experiments to validate the accuracy of the inferred results. With such exceptional performance, the device can effectively monitor tritium content in both nuclear power plant discharge outlets and common drinking water sources, enabling early warning measurements.