Development of 2.2 μm cavity ring-down spectrometer for tritiated water analysis

A rapid and simple tritium analysis method is required for tracer application and the quantitative evaluation of radioactive waste. In this study, we focused on cavity ring-down spectroscopy (CRDS), which is an ultra-sensitive laser absorption spectroscopy, and developed a spectrometer for tritium analysis. A current modulation-assisted acoustic optical modulator switching method was developed in the prototype setup containing a 2.2 μm diode laser for accessing the 2ν1 absorption band of tritiated water vapor. The benefit of this switching method was investigated using the Allan deviation and compared to conventional acoustic optical modulator-only and current-only switching methods. Using the prototype setup with the proposed switching method, CRDS of stable H2O vapor was demonstrated. The detection limit for liquid tritium water analysis was estimated to be 2 × 101 kBq/10 μl for ten-minute measurements.


Introduction
Tritium ( 3 H or T) is a radioisotope of hydrogen and a betadecaying nuclide with a half-life of 12.32 years. Tritium analysis is required for tracer applications in biological and environmental studies, quantitative evaluations of radioactive waste from nuclear power plants and nuclear fusion experiments. [1][2][3] Although liquid scintillation counting is conventionally used for routine analysis of tritium, complicated chemical pretreatment such as quenching suppression and radioactive waste disposal of scintillation cocktails after every measurement are unavoidable. Especially for biological tracer applications and monitoring in exhaust and effluent, a rapid tritium analysis system is desired with simple pretreatment and without additional radioactive waste disposal. 4) Recently, a novel analytical method based on laser absorption spectroscopy has been developed for radioactive carbon-14, one of the radionuclides that emit low-energy β particles. [5][6][7][8][9] In this method, highly sensitive cavity-enhanced laser absorption spectroscopy, called cavity ring-down spectroscopy (CRDS), 10,11) is utilized, and quantitative and rapid determination of 14 C was demonstrated for biomedical applications and nuclear engineering. For tritium, optical spectroscopic data were obtained by several different techniques, such as laser absorption spectroscopy, Fourier transform infrared spectroscopy, and CRDS. [12][13][14][15][16][17][18] However, a quantitative analytical system for tritium based on laser spectroscopy has not yet been realized. Therefore, we are developing a cavity ring-down spectrometer for tritium analysis. In CRDS using a continuous wave laser (CW-CRDS), the incident laser beam to the optical cavity should be quickly blocked or tuned to off-resonance by an optical switch, however, if the extinction ratio and response time of the switch is not sufficient, it will cause a degradation of the measurement sensitivity. In this paper, the prototype spectrometer using a 2.2 μm diode laser including an optical switching method was developed. Since a tritiated water usage in the laboratory is now under preparation, a first demonstration of the absorption spectroscopy using our prototype system was conducted using stable water vapor.

2ν 1 Band absorption of tritiated water at 2.2 μm and interference from absorption of other gases
For tritium analysis by CRDS, the sample must be easily converted to a target molecular gas. Water vapor was selected as the target molecule in CRDS in this study because it can be easily converted through oxidation reactions by combustion. Although the fundamental absorption of the tritiated water (HTO) molecule exists in the 4 μm region, we adopted the overtone absorption 2ν 1 -band of tritiated water HTO at 2.2 μm from the viewpoint of both rather a high absorption strength and commercial availability of a laser source and optical elements. As there are many molecular absorptions in the region near the HTO absorption lines, interference due to these absorptions should be considered. Since HTO is difficult to separate from other water molecules such as H 2 O and HDO, the HTO absorption line must be distinguished spectroscopically from these isotopologues. In addition, concentrations of contaminant molecules that may be included in the sample gas must be reduced to a partial pressure below which HTO absorption is not disturbed as well.
Based on the HITRAN database 19) and HTO spectroscopic data, 20) a line-by-line calculation was conducted for the absorption spectra of HTO, H 2 O, HDO, and other molecules CO 2 , N 2 O, O 3 , and CH 4 having strong absorption lines in this wavelength. Figure 1 shows the calculated absorption spectra of HTO and other molecules at 2.2 μm with a gas temperature of 297 K and total pressure of 1 kPa. The spectra were calculated for 100% partial pressure of each molecule.
The HTO line shown here is a transition of the 2ν 1 band at a center wavenumber of 4596.485 cm −1 . Spectroscopic parameters of this transition are shown in Table I

Experimental setup
Our experimental setup based on CW-CRDS is shown in Fig. 2. In CW-CRDS, a laser beam is injected into a highfinesse optical cavity. When the laser wavelength and the cavity length satisfy the resonance condition, the light is coupled into the cavity. After switching off the incident laser beam, the intensity of transmitted light from the cavity starts an exponential decay, called a "ring-down" signal. The number density of the molecule N that absorbs photons at the laser wavelength can be derived from the decay rate of the where σ is the photo absorption cross-section and α is the absorbance in the spectrometer. CRDS is well-known as an ultra-sensitive laser spectroscopic technique because the high number of reflections/round-trips in the cavity allows for an effective optical path length of up to several kilometers and the decay rate measurement is less affected by intensity fluctuations of the laser source.
In this setup, a continuous wave distributed feedback (DFB) diode laser (Nanoplus Nanosystems and Technologies GmbH) with an output power of 4 mW was used as a probe laser because it has a sufficiently narrow bandwidth of around 1 MHz to distinguish the absorption line of HTO from those of other molecules. After passing through an optical isolator (Thorlabs, I2300C4), the probe laser was injected into an optical cavity. The bow-tie type cavity with high reflectance mirrors (Layertec, 140 983, reflectivity > 99.99%) was adopted to control background fluctuation caused by etalon effects. The one-round cavity length was 844 mm, corresponding to a free spectral range of 355 MHz. The cavity length was swept by a piezoelectric actuator attached to one of the cavity mirrors. Transmitted light from the cavity was detected by an InGaAs photodetector (Thorlabs, DET10D/M) with a current amplifier (Femto, DLPCA-200). A comparator was used to acquire the timing when the optical output in the detector exceeded a given threshold to trigger an optical switch for the incident laser beam. As the optical switch, optical beam steering by an Acoustic Optical Modulator (AOM, isomet, M1208-G80-4) and quick frequency shifting by laser diode current modulation were used. For the AOM operation and the current modulation, two trigger outputs were generated by a reconfigurable digital I/O module (National Instrument, PXIe-7856R) from the output of the comparator. After switching off the incident beam, the output from the current amplifier, i.e. the ring-down signal, was recorded by a digitizer (National Instrument, PXIE-5592).

Results and discussion
In CW-CRDS, the ring-down signals are obtained by quickly switching off the incident laser beam when the transmitted light intensity from the optical cavity reaches the threshold. In this study, we investigated the optical switching method. A typical ring-down signal is shown in Fig. 3. The AOM is typically used for high-speed optical beam steering in CW-CRDS to acquire the ring-down signal. However, dual-pass operation of the AOM switching was required due to the insufficient extinction ratio in single-pass operation in this setup. Since the output power of the diode laser was limited, large transmission loss by double pass in the AOM should be avoided.
Previous studies have reported switching using semiconductor optical amplifiers (SOA), which have shown that highspeed switching is possible with little output degradation (output can be slightly increased). [21][22][23][24] However, it is difficult to use this device in the 2.2 μm wavelength region because SOAs outside of the telecommunication wavelength range are not easily available. Although rapid sweeping of the cavity length was also investigated as another switching method without additional devices, it was difficult to reduce the fluctuation of the ring-down signal due to the large instability of the cavity. 25) On the other hand, a   The Japan Society of Applied Physics by IOP Publishing Ltd semiconductor diode laser can generally accept high bandwidth modulation of diode current. Therefore, a ring-down signal can be obtained by quick frequency shifting of the diode laser from the resonance condition of the optical cavity. This technique does not require an additional optical switch such as an AOM and provides a simpler experimental setup, however, the stability of the laser source often decreases by modulation of diode current and consequent temperature changes. In addition, the incident beam is not completely blocked, so unintended resonances may occur, and sensitivity is suppressed. Therefore, beam steering via AOM combined with frequency shifting by current modulation (named current modulation-assisted AOM switching) was developed as optical switching for CRDS at 2.2 μm. Figure 3 shows the time evolution of the transmitted light signal intensity, modulation voltage for laser diode current modulation, the trigger output from the comparator, and the AOM operation switching voltage. Whereas there was a delay of about 3 μs from the TTL trigger to the start of AOM operation, the laser current was able to modulate quickly by the laser current controller with the bandwidth of 2 MHz. The laser current modulation immediately after the trigger from the comparator may affect the laser light stored in the cavity, resulting the ring-down rate fluctuation. To investigate the issues, the start timing of the laser current modulation was controlled and delayed from the trigger. In the current modulation switching, the trigger output was used for current modulation after passing through an attenuator, a buffer amplifier, and a summing amplifier in order to adjust the modulation depth. In the case of AOM switching, the trigger output was applied to the analog modulation input of the RF driver of the AOM. The current modulation-assisted AOM switching was a switching method that combines the above two methods. Figure 4 shows typical ring-down signals and the residuals from fitting with an exponential function obtained by (a) the current modulation-assisted AOM switching and (b) only current switching. In the case of current switching, even after the ring-down signal was generated, undesired fluctuations sometimes occurred, likely due to accidental resonance of the laser with side-modes of the cavity. Thus, the decay rate evaluated by the fitting had a large fluctuation. In contrast, in the case of the current modulation-assisted AOM switching, such unwanted resonance was suppressed with beam shifting by the AOM. Therefore, a smaller fluctuation of the evaluated decay rate was achieved. Although the depth of current modulation in the current modulation-assisted AOM switching was varied in terms of frequency-shift equivalent to 5 GHz, 47 GHz, and 87 GHz, there was no difference in the fluctuation of the evaluated decay rate. Figure 5 shows the dependence of Allan deviation of absorbance estimated from the ring-down rate in the setup on the signal acquisition time (Allan plot). The Allan deviation is defined as follows: 26)   The Japan Society of Applied Physics by IOP Publishing Ltd from the TTL trigger to the start of AOM operation, the laser current modulation immediately after the trigger may affect the laser light stored in the cavity, resulting in the ring-down rate fluctuation. Therefore, we evaluated the Allan deviation by delaying the current modulation relative to the trigger. Even in case the current modulation was delayed compared to the switching time of the AOM by 1 μs or 2 μs, there was no difference in the Allan plot. Based on the results of the Allan plots, an uncertainty of the ring-down rate of 2 × 10 −9 cm −1 can be achieved by the current modulationassisted AOM switching for 50 s measurement (acquisition rate of approximately 10 2 signals s -1 ), while it is 3 × 10 −8 cm −1 for the AOM switching and 6 × 10 −9 cm −1 for the current switching. A deviation of 7 × 10 −10 cm −1 can be extrapolated assuming stable operation of the system for 600 s measurement.
Using the experimental setup with the current modulationassisted AOM switching, we first measured the CRDS spectrum of stable H 2 O vapor because vapor including HTO must be handled in radiation controlled area and further preparation is required. Water vapor was generated by bubbling water with nitrogen gas, which has no absorption in the 2.2 μm region and was introduced into the CRDS cavity. The laser wavelength was swept by a slow current ramp to obtain a CRDS spectrum. The laser wavelength was calibrated by several peaks of CO 2 absorption in the scanned wavelength region. Figure 6 shows the CRDS spectrum of stable H 2 O vapor at 293 K. The decay rate for the empty gas cell β 0 was evaluated under vacuum. The solid blue line is the result of fitting with the Voigt function using the HITRAN database, and the red dashed line is the center wavenumber of the selected absorption line of HTO in Sect. 2. The results show that the total pressure of the sample gas introduced into the gas cell was 0.29 bar and the partial pressure of water vapor was 0.35%.
In the current setup, background fluctuation did not reach 2 × 10 1 s −1 level corresponding to 7 × 10 −10 cm −1 because of the insufficient acquisition time and still remaining etalon effect. If this background fluctuation is achieved, it would be possible to analyze water liquid samples containing tritium equivalent to 2 × 10 1 kBq/10 μl (i.e. 2 × 10 −6 mol l −1 ). The current sensitivity is two orders of magnitude large for the requirement in biological tracer applications such as excretion examination in a drug study and even further improvement is necessary for monitoring exhaust and effluent at nuclear facilities. In our previous study, frequency stabilization of the DFB diode laser based on optical feedback has been developed for highly sensitive 14 C detection with CW-CRDS and a single detection limit of 1 × 10 −9 cm −1 Hz −1/2 and a reachable detection limit in 10 s measurement (acquisition rate of approximately 55 signals s −1 ) of 5 × 10 −11 cm −1 was demonstrated. 9,27) If systematic fluctuations can be controlled for 600 s with the acquisition rate of 10 2 signals s −1 , the required sensitivity can be achieved. In addition, since absorption line intensities in the 4 μm region are around one order of magnitude larger than those in the 2 μm region, further improvement in sensitivity could be expected by transitioning to that wavelength region. From the results of Fig. 1, to measure water liquid samples containing tritium mentioned above, the contamination levels of N 2 O, O 3 , and CH 4 should be suppressed to less than 10 −6 , 10 −4 , and 10 −4 of their partial pressures, respectively.

Conclusion
We developed a cavity ring-down spectrometer for tritium analysis. In the setup, the absorption line of 2ν 1 band of HTO water vapor with a center wavenumber of 4596.485 cm −1 (2.2 μm wavelength region) was set as the target absorption in CRDS. To achieve high sensitivity in the 2.2 μm CRDS system, a current modulation-assisted AOM switching method was developed. Using the spectrometer with this switching method, the CRDS spectrum of stable H 2 O vapor was obtained. Based on the estimation of background fluctuations in CRDS measurement, the sensitivity of 2 × 10 1 kBq/10 μl for tritium water liquid samples could be achieved. The currently evaluated sensitivity is not sufficient for tracer applications in a biological study, and a further improvement in the sensitivity will be expected with the implementation of frequency stabilization of the laser source and the construction of a setup using the 4 μm laser source. A sample introduction system for the spectrometer is under development and in the future, we intend to obtain HTO spectra with the CRDS system and demonstrate a quantitative analysis of tritium.