Maxilight (Hidex, Turku, Finland) based on di-isopropyl naphthalene (DIN) was used as the scintillation cocktail to extract radon gas from samples. Maxilight is a water-immiscible scintillation cocktail, so only radon gas and its daughter nuclides are included in the scintillation cocktail [10]. The ²²⁶Ra radioactive standard source (SRM 4966A) from the National Institute of Standard and Technology was used to prepare the ²²²Rn calibration source. Plastic vials (Wheaton, Millville, NJ), glass vials (Wheaton, Millville, NJ), and Teflon vials (Flon Chemical, Osaka, Japan) were used as LSC vials, each with a capacity of 20 mL. For the ²²²Rn analysis, the Quantulus 1220 (Perkin Elmer, Waltham, MA) and 300SL (Hidex, Turku, Finland) were used. Like the Quantulus 1220, the 300SL is equipped with a cooling function and an alpha/beta separation function.

To prepare the ²²²Rn calibration source, the ²²⁶Ra radioactive standard source of 1 Bq was added to a LSC vial and mixed with deionized water until the final volume reached 10 mL. After mixing with 10 mL of the Maxilight scintillation cocktail, we waited one month until 226Ra reached radioactive equilibrium with ²²²Rn. The ²²²Rn calibration source was used to determine the counting efficiency and the optimal conditions for alpha/beta separation. Equal 10-mL samples of deionized water and Maxilight were put into an LSC vial and used as a background sample. A 1:1 ratio between the sample and scintillation cocktail showed a high counting efficiency, low background count rate, and low minimum detectable activity (MDA).

Sampling is the most important factor when analyzing radon in groundwater. Since radon is dissolved in groundwater in gas form, radon instantly escapes from the sample when the sample is exposed to the atmosphere. If researchers are not attentive during the sampling process, the results could underestimate the actual concentration of radon. After installing a tube that supplies groundwater at the base of a 1-L or larger container, approximately 3 to 4 times as much groundwater as the container volume was pumped through. Using a pipette, 10 mL was extracted from the bottom-most part of the container possible.

The Quantulus 1220 and 300SL are very different in terms of hardware. The Quantulus 1220 has a width of 101 cm, a depth of 92 cm, a height of 156 cm, and a weight of approximately 1 ton. In contrast, the 300SL has a width of 52 cm, a depth of 63 cm, a height of 68 cm, and a weight of 130 kg, making it possible to use on top of a desk. Its compact weight and size make the 300SL easily transportable, while the Quantulus 1220 is utilized as a fixed detector. The Quantulus 1220 is heavy due to the 750 kg of lead installed in it to shield against external radiation. Less than 10% of the amount of lead in the Quantulus 1220 is installed in the 300SL. This results in a difference in the background count rate. The Quantulus 1220 has a beta background count rate of 3.0±0.2 cpm (mean ± standard deviation) while the value is more than 10 times greater for the 300SL, at 43.7±2.8 cpm (based on the use of a plastic vial; Ultima Gold AB 10 mL).

As shown in Figure 1, the Quantulus 1220 uses two photon multiplier tubes (PMTs) as detectors, while the 300SL uses three PMTs. The Quantulus 1220 uses a guard detector and an anticoincidence system to reduce the effect of external radiation. For the 300SL, the operator can decide whether to operate a guard detector before analysis, and can reduce the background count rate by 20% if the guard detector is used. For alpha nuclide analysis, the counting efficiency of the 300SL, which operates three PMTs, shows a higher efficiency within the 10% range than the Quantulus 1220, which uses two PMTs (Table 1). This is because the photoelectrons that are not detected in a PMT composed of 180-degree intervals can be detected in a PMT composed of 120-degree intervals (Figure 1B and 1E). One advantage of LSCs is that they contain an autosampler, which is a useful function for routine analysis. With the use of a 20-mL LSC vial, the Quantulus 1220 can simultaneously analyze 60 samples, and the 300SL can simultaneously analyze 40 samples.

²²⁶Ra decays while generating ²²²Rn, ²¹⁸Po, and ²¹⁴Po alpha nuclides. Of these products, ²¹⁸Po and ²¹⁴Po have short half-lives; they reach radioactive equilibrium with the mother nuclide ²²²Rn after approximately 3 hours. When considering the near 100% alpha emitter counting efficiency for LSCs, the ²²²Rn counting efficiency (taking into account ²²²Rn, ²¹⁸Po, and ²¹⁴Po) can theoretically reach 300% in this study, which used the radon evaporation method [10]. This is because ²²²Rn, ²¹⁸Po, and ²¹⁴Po form a radioactive equilibrium in the scintillation cocktail in which only radon was extracted. Since ²²⁶Ra exists in parts of an aqueous solution, not a scintillation cocktail, its effect on the alpha count rate is so small that it can be disregarded. The alpha count rate is equal to or less than 0.05 cpm (n=5) when 5 Bq of 226Ra is placed into a 20-mL plastic vial, 10 mL of it is filled with 0.1M HNO₃, and the combination is then analyzed using the Quantulus 1220. Based on the LSC analytical results of the ²²²Rn source, the counting efficiency (ɛ) was determined by Equation 1.

The LSCs used in this study have an alpha/beta separation function. The Quantulus 1220 separates alpha nuclides and beta nuclides based on pulse shape analysis (PSA), while the 300SL separates them based on the pulse length index (PLI). Since the shapes of electrons generated by photons emitted from alpha and beta particles are different, alpha particles trigger a pulse characterized by delayed phosphorescence and beta particles trigger a pulse involving prompt fluorescence. The criterion for distinguishing between alpha and beta nuclides after measuring the pulse decay time or pulse length that occurs at different times is referred to as PSA or the PLI. ²⁴¹Am and ³⁶Cl sources can be used to determine alpha/beta separation conditions [5]. It is important to assess beta spillover (when alpha rays are misidentified and measured as beta rays) and to assess alpha spillover, the opposite phenomenon. Simultaneous alpha/beta analysis must take alpha spillover and beta spillover into account in order to measure accurate levels of alpha and beta emitters. In such cases, the optimal PSA level for separating alpha/beta nuclides is determined when the alpha spillover and beta spillover are at a minimum—in other words, the optimal PSA level is chosen at the point where the alpha count rate of ²⁴¹Am reaches its maximum and where, at the same time, the beta count rate of ³⁶Cl reaches its maximum [5]. The optimal PSA level of the Quantulus 1220 used in this study was determined based on the figure of merit (FOM) value of the ²²²Rn calibration source according to the PSA level. Since the ²²²Rn calibration source includes alpha daughter nuclides (²¹⁸Po, ²¹⁴Po) and beta daughter nuclides (²¹⁴Pb, ²¹⁴Bi), the count rate of alpha emitters reflects alpha spillover and beta spillover. Therefore, the higher the count rate in the alpha ray domain, the lower the alpha spillover and beta spillover. The FOM value increases with a lower background count rate and a higher counting efficiency (Equation 2). The ²²²Rn source and background sample were measured within the PSA 30 range and the 130 range, respectively. A PSA value of 80, with the maximum FOM value, was identified as the optimal alpha/beta analytical condition [10].

The 300SL has a simpler process for determining alpha/beta separation conditions than the Quantulus 1220. After measuring the ²²²Rn calibration source, the criterion for alpha/beta separation (PLI) is determined by using software. As shown in Figure 2, the analytical results of the ²²²Rn calibration source can be shown as alpha rays and beta rays in 3-dimensional graph form based on the PLI criterion. Compared to the prompt fluorescence that originates from beta particles, the length of the pulse due to the delayed phosphorescence resulting from alpha particles that react with the scintillation cocktail is longer. The 300SL can visually separate alpha emitters and beta emitters that include pulse length information, and can perform an alpha/beta separation based on a single measurement. In this study, the analysis was carried out based on a PLI of 12, and the counting efficiency and background count rate were determined accordingly.

The region of interest (ROI) is another important analytical factor. Since the background count rate and counting efficiency are closely related to the MDA, it is desirable to set an interval with a low background count rate and a high counting efficiency as the ROI. Figure 3 presents the alpha spectrum of the ²²²Rn calibration source generated using the Quantulus 1220 and 300SL. Table 1 presents the background count rate and counting efficiency of the Quantulus 1220 and 300SL in the 700–950 channel condition and the total channel condition. Compared to the analytical results of the total channel, the counting efficiency in the ROI condition was similar, at 230%-245%, but the background count rate decreased from 0.20 cpm to 0.01 cpm in the Quantulus 1220. Compared to the total channel set-up, when a ROI is set, the MDA can be lowered threefold for the same condition (sample volume and measuring time) using the Quantulus 1220.

In a previous study, plastic vials and glass vials were used to analyze groundwater with a high concentration of radon. When a toluene series scintillation cocktail was used, radon gas leakage was observed after the sample was stored in a vial. To avoid this phenomenon, Teflon vials were recommended [5]. This is because the toluene series scintillation cocktail permeated into the plastic vial and was volatilized outside of the container, causing some radon gas to leak as well. When a Maxilight scintillation cocktail was used, no radon gas leakage was observed from either plastic vials or glass vials. In each vial, 10 mL of groundwater was mixed with 10 mL of the Maxilight scintillation cocktail. The first analytical result used the LSC measurement as C₀ and the count rate measured after a certain amount of time as C_t (Equation 3). Based on C₀ (without considering the counting efficiency), the calculated count rate is shown in Figure 4, which presents the measured count rate using the LSC and the ²²²Rn decay rate. The measured value was strongly correlated with the calculated value. This means that no other reduction factors were present in addition to radiochemical radon decay in plastic or glass vials. When a DIN-series scintillation cocktail is used, it is not necessary to take radon leakage into account, even if expensive Teflon vials are not used.

Using the Quantulus 1220, the counting efficiency results of the ²²²Rn calibration source prepared in a glass vial, a plastic vial, and Teflon vial were 239%±3%, 235%±2%, and 236%±3%, respectively. The counting efficiency difference was not large among the vial types. However, the background count rate results of the glass vial and plastic vial were 0.01 cpm and 0.03 cpm, respectively, showing a relatively low background count rate for the glass vial.

The main variables that determine the MDA are sample quantity, counting time, background count rate, and counting efficiency. The counting efficiency values of Quantulus 1220 and 300 SL are similar, within a range of 10%. Under the ROI condition, the background count rate of Quantulus 1220 is 10 times lower than that of 300SL. When the sample volume is 10 mL and the counting time is 60 minutes, the MDAs of the Quantulus 1220 and 300SL are 0.08 Bq·L⁻¹ and 0.20 Bq·L⁻¹, respectively. While the MDA of 300SL is relatively high, it corresponds to 0.1% of 148 Bq·L⁻¹, the concentration of radon in drinking water suggested by EPA regulations. The MDA was calculated by the following equation [11].

To analyze actual radon in groundwater, a glass vial containing 10 mL of the Maxilight scintillation cocktail was prepared. As described in section 2.3, 10 mL of groundwater was extracted on site, put into a vial, and intensely mixed. To avoid sample exposure to the atmosphere, the pipette was placed at the bottom of the vial containing the scintillation cocktail to inject the sample. Analysis was conducted using a LSC after approximately 3 h, when ²²²Rn reached radioactive equilibrium with ²¹⁸Po and ²¹⁴Po. Table 2 presents the results of the analyses conducted with the Quantulus 1220 and 300SL. The counting efficiency and background count rate suitable for each piece of equipment were used, and the radon concentration was calculated using Equation 5.

The relative error was calculated based on the results of the Quantulus 1220, and the difference between the radon concentrations of the two LSCs matched within 10%.