Application of the Deconvolution Method to Alpha-Ray Energy Spectrum to Improve the Performance of a Radiation Airborne Particulate Monitor

Makoto Sasano; Masateru Hayashi; Yusuke Yanagawa; Masatoshi Kai; Yasushi Nakano; Yoshitsugu Osawa; Shunsuke Kurosawa; Yuki Morishita

doi:10.14407/jrpr.2023.00444

Sasano, Hayashi, Yanagawa, Kai, Nakano, Osawa, Kurosawa, and Morishita: Application of the Deconvolution Method to Alpha-Ray Energy Spectrum to Improve the Performance of a Radiation Airborne Particulate Monitor

Original Research

Journal of Radiation Protection and Research 2025; : jrpr.2023.00444.

Published online: February 10, 2025

DOI: https://doi.org/10.14407/jrpr.2023.00444

Application of the Deconvolution Method to Alpha-Ray Energy Spectrum to Improve the Performance of a Radiation Airborne Particulate Monitor

Makoto Sasano¹

, Masateru Hayashi¹, Yusuke Yanagawa², Masatoshi Kai², Yasushi Nakano², Yoshitsugu Osawa², Shunsuke Kurosawa^3,⁴, Yuki Morishita⁵

¹Mitsubishi Electric Corporation, Amagasaki, Japan

²Mitsubishi Electric Plant Engineering Corporation, Kobe, Japan

³Institute for Materials Research, Tohoku University, Sendai, Japan

⁴Institute of Lase Engineering, Osaka University, Suita, Japan

⁵Japan Atomic Energy Agency, Tokaimura, Japan

Corresponding author: Makoto Sasano, Mitsubishi Electric Corporation, 8-1-1 Tsukaguchi-Honmachi, Amagasaki City, Hyogo 661-8661, Japan E-mail: Sasano.Makoto@db.MitsubishiElectric.co.jp

Received August 3, 2023 Revised December 20, 2023 Accepted February 16, 2024

This is an open-access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract

Background

At nuclear facilities and decommissioning sites, monitoring radioactivity concentrations in airborne particulates is crucial to prevent worker exposure. To avoid internal exposure, alpha-decay radionuclides must be detected. When monitoring alpha-decay radionuclides in airborne dust, we want to measure the concentration of only artificial radionuclides (e.g., ²³⁸Pu, ²³⁹Pu, ²⁴⁰Pu, ²³⁵U, ²³⁸U, ²⁴¹Am, and ²⁴⁴Cm). The radioactivity concentration must be measured separately for artificial (4.3–5.8 MeV) and natural radionuclides (²¹²Bi, ²¹⁴Po, and ²¹²Po at 6.0, 7.7, and 8.8 MeV, respectively).

Materials and Methods

We created response functions for various alpha-ray energies using a radiation simulation toolkit. Utilizing these response functions, we deconvolved the alpha-ray energy spectra measured while collecting dust on filter paper. To ensure the precision of the response function, we prepared a model including the distance between the filter and the detector and the structure of the light shield in detail.

Results and Discussion

The deconvolved spectra had three clear peaks at 6.0, 7.7, and 8.8 MeV. These energies were consistent with those of ²¹²Bi, ²¹⁴Po, and ²¹²Po. The deconvolved energy spectra showed that only a few measurements (4.0–5.8 MeV) were included in the energy range due to artificial radionuclides. From these measurements, the decision threshold for artificial radionuclides was determined to be about 1.5×10⁻⁷ Bq/cm³.

Conclusion

Our findings demonstrated that we could measure artificial and natural radionuclides separately with the deconvolved alpha-ray energy spectra. Specifically, we were able to monitor artificial radionuclides down to low radioactivity concentrations in 10-minute measurements.

Keywords: Airborne Particulate Monitor, Alpha-Ray Energy Spectrum, Deconvolution

Introduction

Nuclear power plants and facilities that handle nuclear fuel materials or radioactive sources must monitor airborne radioactivity levels to prevent public exposure. When airborne radioactive materials are inhaled, they significantly impact internal radiation exposure. Alpha particles are particularly harmful inside the body and can cause significant damage. Therefore, emitted radionuclides must be detected as soon as possible at decommissioning sites and other locations where alpha-emitting radionuclides may be dispersed. In this respect, after the accident at the Fukushima Daiichi Nuclear Power Plant in 2011, ²⁴¹Am, ²³⁹Pu, ²⁴⁰Pu, and ²⁴¹Pu derived from fuel rods were detected [1]. In the future, when fuel debris is removed in the process of decommissioning, alpha-ray emitting nuclides may leak into surrounding areas. Therefore, technology is required to detect artificial alpha-ray emitting nuclides with high precision.

To detect artificial alpha-ray emitting nuclides (e.g., Pu and Am) quickly, the influence of naturally occurring alpha-ray emitting nuclides must be removed. Polonium-214 (²¹⁴Po), ²¹²Po, and ²¹²Bi are well-known naturally occurring alpha-ray emitters with short half-lives [2]. As each of these nuclides emits alpha rays at different energies, measuring them separately on the energy spectrum is desirable. When performing alpha-ray energy spectrum analysis, measuring samples in a vacuum is common practice to avoid the effects of energy loss in the air [3]. However, it is difficult for on-site monitors to measure samples in vacuum environments. Airborne particulate monitors typically collect dust on filter papers and measure them in the air [4]. The energy spectrum of alpha-rays obtained under these conditions has a peak shape with a tail at the low-energy side due to energy loss. In this shape, the peaks overlap each other, making it difficult to separate the overlapping nuclides. To produce a device capable of constant monitoring for extended periods, the detector must be prevented from malfunctioning or becoming contaminated. To achieve this, a structure that protects the detector is necessary, and a sheet must be provided to protect the detector while allowing alpha rays to pass through. Within this sheet, the alpha rays lose energy, resulting in a more significant change in the shape of the energy spectrum.

Silver-activated zinc sulfide (ZnS(Ag)) scintillators are widely used as alpha-ray detectors. These scintillators are very thin, helping reduce the gamma-ray and beta-ray background. However, they have the disadvantage of poor energy resolution [5]. In some cases, semiconductor detectors are used, and a collimator structure has been proposed to improve the energy resolution [6]. However, this method also has the disadvantage of reduced detection efficiency, and the time to detection becomes longer. Thus, improving the analysis and detection performance of alpha-rays simply by devising the type and structure of the detector is not easy.

This study provides a fast and precise way to measure airborne radioactivity to prevent worker exposure. Specifically, we examine a method of applying deconvolution to alpha-ray energy spectra using a silicon semiconductor detector. This method estimates the source spectrum from the measured spectrum and is used to estimate energy spectra in neutron and gamma rays [7, 8]. It is also applied to the energy spectrum of alpha-rays measured in a vacuum [3]. To estimate the source spectrum, a response function linking the measurement and the estimation is essential. To create this response function, the detector’s geometric structure and the protective sheet’s effect must be considered. We utilize radiation simulations to create a response function to reproduce our measurements. In this study, we perform deconvolution processing using the response function on alpha-ray energy spectra measured while collecting dust in the air. The analytical performance is then evaluated from the deconvolved spectra.

Materials and Methods

1. Setup for Alpha-Ray Spectroscopy

Our aim here is to obtain the energy spectra of alpha particles. Several detectors can acquire energy spectra. Scintillator detectors perform poorly when separating alpha and beta radiation because of the smaller emission in alpha detection. Semiconductor detectors are more suitable for obtaining spectra with better energy resolution. In particular, semiconductor detectors with large sensitive regions are effective for simultaneously obtaining the concentration of beta-emitting nuclides. In this verification, we used a Si-PIN photodiode from Hamamatsu Photonics. This photodiode has a maximum sensible thickness of 300 μm, making it highly efficient for detecting both alpha and beta radiation. To prevent alpha rays from being blocked by the package, we used one that had not been treated with protective epoxy. An aluminum mylar approximately 5 μm thick was placed in front of the detector for protection.

The processing of the detector signal had the same configuration as in general radiation measurements. An analog circuit of a preamplifier circuit and an amplifier circuit were installed immediately after the detector. A bias voltage of 70 V, which was the voltage for complete depletion, was also supplied to the detector. The output from the analog circuit was converted to an 8,192-channel histogram by the multichannel analyzer circuit. The histogram information was sent to the central processing unit (CPU) every second and converted into an energy spectrum in the CPU before deconvolution processing was performed.

To install the system on-site and perform constant measurements, dust must be collected on filter paper while performing the measurement. Fig. 1 shows a photograph of the prototype dust monitor. In the lower part of the photo, dust was collected on filter paper through air flow using a pump. A Si-PIN photodiode at the amplifier box detected radiation from the filter paper. The volume of air taken in from outside was set at about 45 L/min. We used a membrane-type filter paper with a thickness of 120 μm and a pore size of 1.0 μm. This filter advantageously has a high collection rate and low dust penetration into the filter.

The experiment was conducted in our laboratory room. The test started at 12:00 AM on September 15, 2022, and measurements were taken continuously for 80 hours. The dust collection pump was operated for the first 24 hours of the test, after which we stopped the pump and evaluated the decay of the collected radionuclides. Each measurement was 10 minutes long and was done continuously for 80 hours. The performance evaluation was based on the results of the deconvolution processing of the 10-minute data.

We performed the evaluation by checking sources before performing the collection. When performing source calibration, a plane source of the same size as that being collected was used. Americium-241 (²⁴¹Am) and ³⁶Cl were utilized to calibrate the alpha and beta efficiency, respectively. The radioactivities of the sources were 964 Bq for ²⁴¹Am and 1.17 kBq for ³⁶Cl, with uncertainties of 5% and 6%, respectively. These uncertainties were extended uncertainties with a coverage factor k=2 and a level of confidence of approximately 95%.

2. Preparation of the Response Matrix

To calculate the inverse problem, the response of the detector, including the effect of alpha-ray energy loss from the filter to the detector, was required. In the case of alpha-rays, energy attenuation in the air and the aluminum mylar had a significant effect. The dust monitor utilized in this study collected dust in a circular area. Therefore, because it had a broadened source geometry rather than a point source, we needed to consider the various paths that radiation could take to reach the detector from the filter paper. In this case, deriving the energy transfer at the detector analytically was difficult. We therefore attempted to create a response function using Monte Carlo simulation.

We used Geant4 [9–11] for the radiation simulation, which has previously been utilized for alpha-ray detection [12] and is suitable for calculating the reaction process of the present alpha radiation. Meanwhile, we used Fritiof parton, Bertini, and Precompound models (FTFP_BERT) as physical models [11]. Radiation from the collected dust was assumed to be emitted uniformly from the surface of the filter paper. The effect of dust penetration into the interior of the filter paper was not considered. In the simulations, the alpha-ray energy was varied by 0.05 MeV to generate 200 responses with source energies ranging from 0 to 10 MeV. We generated 1,000,000 alpha rays in each energy condition. This number of occurrences was enough to obtain sufficient statistics to reproduce the response function. The energy range was set to 0.05 MeV to perform the deconvolution process quickly.

The respective energy spectra of the generated alpha rays for their initial energy are shown in Fig. 2. Above 1.0 MeV, a peak was detected at an energy about 1.0 MeV lower than the generated energy because of the influence of air and aluminum plates. In addition, the spectral shape of the spectrum tailed off to the low-energy side: for an alpha ray of 5.5 MeV, which was close to the alpha-ray energy of ²⁴¹Am, the peak energy was 4.3 MeV with a full width at half maximum of 0.7 MeV.

In the deconvolution described below, these responses were utilized to recover the energy information of the alpha radiation emitted from the filter paper.

3. Algorithm of the Deconvolution Process

The alpha energy spectrum acquired by the detector includes the detector’s response, as described above. Consequently, the acquired spectrum loses initial information about the radiation emitted from the filter. This information is estimated through deconvolution.

Various algorithms are used to estimate the energy spectrum through deconvolution. Because the dust monitor that we developed is constantly operated in the field, it is desirable to use an algorithm that can derive a stable solution. However, the simplest inverse estimation involves the appearance of negative solutions. Thus, instead, we utilized deconvolution through successive approximation [3]. This algorithm used the spectrum obtained from the measurement as the initial value, thus preventing the divergence of the solution. The following two equations, Equations (1) and (2), were used for the estimation. During the calculation, the energy spectrum was treated as a vector.

(1)

Mn+1=R·Sn

(2)

Sn+1=(Mn+1Mn)·Sn

M, S, and R represent the estimated measurement spectrum, the estimated source spectrum, and the response function, respectively. The index n represents the number of repetitions of the operation. Here, M⁰ and S⁰, at the beginning of the operation, use the acquired energy spectrum. When the estimated measurement spectrum converges to Mⁿ⁺¹–Mⁿ, Sⁿ⁺¹ is almost the same as Sⁿ from Equation (2), and the source spectrum to be estimated also converges.

The parameter in this algorithm is the number of iterations. This number must be set to a value sufficient for the convergence of the spectrum to be estimated. Preliminary testing under various conditions confirmed that the solution converged at 1,000 iterations or more, so we used 1,000 iterations in the subsequent deconvolution process.

Results and Discussion

1. Spectra with Checking Source

As stated above, we used a ²⁴¹Am source for energy calibration. The energy spectrum obtained at this calibration is shown in Fig. 3. The original ²⁴¹Am alpha-ray energy is shown in Table 1 [13]. In the present measurement system, the peak was at about 1.2 MeV lower energy, and the tail was further drawn to the lower energy side. The ³⁶Cl irradiated spectrum was on the lower energy side in Fig. 3, and the count rate conversion factor to activity for the beta ray was 0.22 counts per second (cps)/Bq. The maximum energy of ³⁶Cl was 0.7 MeV, consistent with the beta-ray maximum energy of 0.71 MeV.

Fig. 4 shows the results of the comparison of the measured spectrum with the response function prepared with the simulation. In the low-energy part of the measured spectrum, there is a large discrepancy due to the chance coincidences between gamma rays and alpha rays from the source. The simulated and measured results agree within 20% above 1.0 MeV and reproduce within 10% near the peak energy.

We tried deconvolution with the obtained measured spectra to validate the deconvolution algorithm. The results are shown in Fig. 3, where the deconvolved result appears around 5.5 MeV. The alpha intensity obtained from the deconvolution result is 938 Bq, which agrees with the source 964 Bq within 5% of the source uncertainty. This result demonstrates that the deconvolution algorithm worked correctly.

2. Performance of Deconvolved Spectra with Airborne Dust Sample

We obtained spectra while the pump was operating and collected dust to investigate its performance with airborne dust. As described earlier, the measurements were conducted continuously for 80 hours. The air pump was run for just 1 day from the start. After that, measurements were made with the pump stopped.

The measured radioactivity and the time variation of the radioactivity concentration are shown in Fig. 5, along with the results of natural alpha-ray nuclides (²¹²Bi, ²¹²Po, and ²¹⁴Po with 6.0, 7.7, and 8.8 MeV, respectively) and artificial alpha-ray nuclides (²³⁸Pu, ²³⁹Pu, ²⁴⁰Pu, ²³⁵U, ²³⁸U, ²⁴¹Am, and ²⁴⁴Cm, 4.0–5.8 MeV) and beta-rays (0.1–1.5 MeV) derived from the deconvolution results. Here, we can see a rapid increase in ²¹⁴Po and a gradual increase in ²¹²Bi and ²¹²Po after the start of the measurements. The increase stopped once during dust collection but then increased and decreased. After the pump was stopped, the radioactivity decreased continuously, with ²¹⁴Po decreasing faster and ²¹²Bi and ²¹²Po decreasing at approximately the same time constant. A similar trend was found for beta radiation.

The raw and deconvolved spectra for the six conditions of 1, 4, 20, 24, 32, and 40 hours from the start are shown in Fig. 6. Although the radioactivity of ²¹²Bi, ²¹²Po, and ²¹⁴Po changed each time, the deconvoluted spectra showed that they could be separated into three peaks in the deconvolved spectra. In addition, the deconvolved spectra sometimes contained contamination in the energy region corresponding to the alpha-ray energy from the artificial nuclides. The decision threshold of the artificial nuclides with the deconvolved spectra was 1.5×10⁻⁷ Bq/cm³. This value was lower than the control concentration specified by Japanese law. Therefore, the performance was enough to detect artificial alpha-ray emitting nuclides to prevent public exposure. To achieve a better performance, we required a measurement time longer than 10 minutes.

The peak energy in the alpha-ray spectrum became higher when the humidity was high. The simulation evaluated the peak energy shift by the humidity. If the humidity changed from 0% to 100% at 50 °C, the peak energy increased by about 30 keV. In the deconvolved spectrum, the peak changed only one bin to higher energy. From this evaluation, the effect of humidity was negligible. The membrane filter paper collected 99% of dust on its surface [14]; thus, the effect of dust penetrating the filter paper was small. If the alpha-ray attenuation increased due to dust or humidity on the filter paper, the peak energy in the deconvolved spectrum was lower. At this time, the instrument should automatically replace the filter paper.

3. Decay of Natural Alpha-Ray Emitting Nuclides

Natural alpha-ray emitting nuclides on the filter decayed in a short time. The alpha rays detected in this alpha spectrum measurement were emitted by the following decay processes, as:

(3)

P212b→10.64 hrB212i→60.55 minP212o→0.299 μsP208b

(4)

P212b→10.64 hrB212i→60.55 minT208l→3.053 minP208b

(5)

P214b→26.8 minB214i→19.9 minP214o→164.3 μsP210b.

The processes in Equations (3) and (4) reached radiative equilibrium and decay in 10.64 hours, which was the starting point of the decay. The process in Equation (5) was also almost in radiative equilibrium, so this process decayed with a half-life of 26.8 minutes. To verify these results, we focused on the time variation after we stopped the air pump. The results are shown in Fig. 7. We fitted this time variation with an exponential function and derived the half-life shown in Table 2. The decay time from our experiment was 10% longer than that of a previous study on ²¹⁴Po. This was because our measurement time was 10 minutes and had a similar time scale to the ²¹⁴Po decay time. In the case of ²¹²Bi and ²¹²Po, our results were consistent with the known decay time within error. These findings demonstrate that changes in the short half-life can also be measured in real time.

4. Correlation Between Alpha- and Beta-Ray Activities

Both alpha and beta radiations measured in the energy spectrum were due to the processes in Equations (3)–(5). Therefore, their respective radiation activities should correlate. The results of plotting the alpha-ray radioactivity of ²¹²Bi, ²¹²Po, and ²¹⁴Po, which were derived from the deconvolved spectrum, and the beta-ray radioactivity derived from the count rate in the beta-ray region (0.1–1.5 MeV) are shown in Fig. 8. We confirmed that the alpha- and beta-ray radioactivity have a linear relationship. The result of fitting with a straight line is

(6)

β=2.39×α+1.06,

where α and β represent alpha and beta radioactivity, respectively. The offset of beta radiation was presumably due to background effects from cosmic rays and environmental gamma rays. By using the relationship in Equation (6), the effect of natural radionuclides in the beta radiation region can be removed. This method can thus improve the detection performance of beta-rays. To remove the effects of cosmic rays and environmental gamma rays, a compensating detector must be provided and the effects must be subtracted. Such research is outside the scope of the current paper.

Conclusion

We created a response matrix with a radiation simulation toolkit to deconvolve the measured alpha-ray energy spectra. Experiments and measurements with airborne collection confirmed that we could measure artificial and natural radionuclides separately with the deconvolved alpha-ray energy spectra. We also demonstrated that artificial radionuclides could be monitored down to low radioactivity concentrations in a 10-minute measurement.

Notes

Conflict of Interest

Makoto Sasano and Masateru Hayashi are employees of Mitsubishi Electric corporation. Yusuke Yanagawa, Masatoshi Kai, Yasushi Nakano, and Yoshitsugu Osawa are employees of Mitsubishi Electric Plant Engineering corporation. This study is funded by these companies, and the research design, data collection, analysis, or manuscript writing may be influenced by these companies.

Acknowledgements

This work was supported by the Japan Atomic Energy Agency Nuclear Energy S&T and Human Resource Development Project through concentrating wisdom Grant Number JPJA18P18071964.

Ethical Statement

This article does not contain any studies with human participants or animals performed by any of the authors.

Author Contribution

Conceptualization: Sasano M. Methodology: Sasano M, Hayashi M. Data curation: Sasano M. Formal analysis: Sasano M. Supervision: Sasano M. Funding acquisition: Sasano M, Hayashi M, Kurosawa S. Project administration: Sasano M. Investigation: Sasano M. Visualization: Sasano M. Resources: Sasano M. Software: Sasano M. Validation: Sasano M, Yanagawa Y. Writing - original draft: Sasano M. Writing - review & editing: Sasano M, Hayashi M, Yanagawa Y, Kai M, Nakano Y, Osawa Y, Kurosawa S, Morishita Y.

References

1. Schneider S, Walther C, Bister S, Schauer V, Christl M, Synal HA, et al. Plutonium release from Fukushima Daiichi fosters the need for more detailed investigations. Sci Rep. 2013;3:2988.

2. Pollanen R, Siiskonen T. Unfolding alpha-particle energy spectrum from a membrane air filter containing radon progeny. Radiat Meas. 2014;70:15-20.

3. Vajda N, Pollanen R, Martin P, Kim CK. Alpha spectrometry. In: L’Annunziata MF. Handbook of radioactivity analysis. 3rd ed. Academic Press. 2012;494-575.

4. Sung SH, Kim HR. Optimization of airborne alpha beta detection system modeling using MCNP simulation. Nucl Eng Technol. 2020;52(4):841-845.

5. Morishita Y, Yamamoto S, Izaki K, Kaneko JH, Toui K, Tsubota Y, et al. Performance comparison of scintillators for alpha particle detectors. Nucl Instrum Methods Phys Res A. 2014;764:383-386.

6. Kim MJ, Sung SH, Kim HR. Spectral resolution evaluation by MCNP simulation for airborne alpha detection system with a collimator. Nucl Eng Technol. 2021;53(4):1311-1317.

7. Reginatto M. Overview of spectral unfolding techniques and uncertainty estimation. Radiat Meas. 2010;45(10):1323-1329.

8. Li F, Gu Z, Ge L, Li H, Tang X, Lang X, et al. Review of recent gamma spectrum unfolding algorithms and their application. Results Phys. 2019;13:10211.

9. Agostinelli S, Allison J, Amako K, Apostolakis J, Araujo H, Arce P, et al. Geant4—A simulation toolkit. Nucl Instrum Methods Phys Res A. 2003;506(3):250-303.

10. Allison J, Amako K, Apostolakis J, Araujo H, Arce Dubois P, Asai M, et al. Geant4 developments and applications. IEEE Trans Nucl Sci. 2006;53(1):270-278.

11. Allison J, Amako K, Apostolakis J, Arce P, Asai M, Aso T, et al. Recent developments in Geant4. Nucl Instrum Methods Phys Res A. 2016;835:186-225.

12. Moslehi A, Baradaran S, Taheri M. GEANT4 modeling of alpha particles detection efficiency for polycarbonate SSNTD. Nucl Instrum Methods Phys Res A. 2022;1033:166714.

13. Nudat3.0. NuDat3.0 [Internet]. National Nuclear Data Center; 2024 [cited 2024 Jun 14]. Available from: https://www.nndc.bnl.gov/nudat3/

14. Takeshi I, Kenzo F, Shinji T, Ryuhei K. Characteristics of major filters used for ²²²Rn progeny measurements. Radiat Meas. 1998;29(2):161-164.

Fig. 1

(A) Photograph of the experimental setup. (B) Schematic diagram of the measurement setup. MCA, multichannel analyzer; CPU, central processing unit; Si-PIN, silicon-PIN photodiode.

Fig. 2

Response energy spectra. Gray, black, red, blue, green, yellow, cyan, and magenta colors denote the spectra of source energies 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, and 10.0 MeV, respectively.

Fig. 3

Raw energy spectra of ²⁴¹Am and ³⁶Cl checking source (left y-axis) and deconvolved ²⁴¹Am spectra (right y-axis). Dotted lines denote raw energy spectra. Red color bars denote the deconvolved spectrum. cps, counts per second.

Fig. 4

Comparison of the experiment and simulation with a ²⁴¹Am source. Solid and dotted lines denote experimental and simulation results, respectively. The deviation is below the spectra. cps, counts per second.

Fig. 5

Radioactivity (left y-axis) and density (right y-axis) trend graph of our 80-hour measurement. There are six colored lines: red (²¹²Bi), blue (²¹⁴Po), yellow (²¹²Po), magenta (²¹²Bi+²¹⁴Po+²¹²Po), cyan (beta ray), and black (artificial). The radioactivity and density of the artificial nuclides are magnified by 10. Arrows with (a–f) denote 1, 4, 20, 24, 32, and 40 hours from the beginning, respectively.

Fig. 6

Count rate (black dotted line, left y-axis) and deconvolved radioactivity (red bar, right y-axis) energy spectra. Symbols from (A–F) are the same as (a–f) in Fig. 5. cps, counts per second.

Fig. 7

Radioactivity trend graph after the air pump was stopped. Solid and dotted lines denote experimental data and fitted curves, respectively. Red, blue, and yellow lines denote the ²¹²Bi, ²¹⁴Po, and ²¹²Po trends, respectively.

Fig. 8

Plot between alpha- and beta-ray radioactivity.

Table 1

Main Alpha-Ray Energies of ²⁴¹Am, ²¹²Bi, ²¹²Po, and ²¹⁴Po [13]

Derived quantity	Alpha-ray energy (MeV)	Intensity (%)
²⁴¹Am	5.544	0.37
	5.512	0.225
	5.486	84.8
	5.469	0.02
	5.443	13.1
	5.417	0.01
	5.388	1.66
	5.322	0.015

²¹²Bi	6.090	9.7
	6.051	25.1
	5.768	0.61
	5.626	0.06

²¹²Po	8.785	100

²¹⁴Po	7.687	99.9895
	6.903	0.0105
	6.610	0.00005

Table 2

Experimental and Expected Decay Time

Nuclide	Experimental	Expected
²¹⁴Po	29.6±0.3 min	26.8 min
²¹²Po	10.8±0.2 hr	10.64 hr
²¹²Bi	10.8±0.2 hr	10.64 hr

Values are presented as mean±standard deviation.