- Email: [email protected]
High-accuracy measurement of the emission spectrum of liquid xenon in the vacuum ultraviolet region
Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
High-accuracy measurement of the emission spectrum of liquid xenon in the vacuum ultraviolet region Keiko Fujii a, Yuya Endo a, Yui Torigoe a, Shogo Nakamura a, Tomiyoshi Haruyama b, Katsuyu Kasami b, Satoshi Mihara b,c, Kiwamu Saito b,c, Shinichi Sasaki b,c, Hiroko Tawara b a
Faculty of Engineering, Yokohama National University, Yokohama, Kanagawa 240-8501, Japan High Energy Accelerator Research Organization (KEK), Tsukuba, Ibaraki 305-0801, Japan c The Graduate School of Advanced Studies, Hayama, Kanagawa 240-0193, Japan b
art ic l e i nf o
a b s t r a c t
Article history: Received 1 February 2015 Received in revised form 20 May 2015 Accepted 27 May 2015 Available online 5 June 2015
The emission spectrum of cryogenic liquid xenon in the vacuum ultraviolet region was measured by irradiating liquid xenon with gamma-rays from a radioactive source. To achieve a high signal-to-noise ratio, we employed coincident photon counting. Additionally, the charge of the photo-sensor signals was measured to estimate the number of detected photons accurately. In addition, proper corrections were incorporated for the wavelength; response functions of the apparatus obtained using a low-pressure mercury lamp, and photon detection efficiencies of the optical system were considered. The obtained emission spectrum is found to be in the shape of a Gaussian function, with the center at 57,199 734 (stat.) 7 33 (syst.) cm 1 (174.8 70.1 (stat.) 70.1 (syst.) nm) and the full width at half maximum of 3328772 (stat.) 7 65 (syst.) cm 1 (10.27 0.2 (stat.) 7 0.2 (sys.) nm). These results are the most accurate values obtained in terms of the data acquisition method and the calibration for the experimental system and provide valuable information regarding the high-precision instruments that employ a liquid-xenon scintillator. & 2015 Elsevier B.V. All rights reserved.
Keywords: Scintillator Liquid xenon Photon counting Emission spectrum
1. Introduction Liquid xenon is an excellent scintillation material because of a number of advantages such as its large photon yield and fast response. This material is used primarily in astrophysics and particle physics studies. A currently on-going underground experiment [1] employs a ton-scale detector containing ultrapure liquid xenon as the scintillator with the aim of directly detecting the weakly interacting massive particle (WIMP), which is supposed to constitute dark matter. An accelerator-based experiment [2] searches for rare muon decays using a ton-scale liquid-xenon scintillator as an efficient gamma-ray detector to observe new signs of physics beyond the Standard Model. Time-projection chambers using liquid xenon as a drift medium are now under development [3] with the aim of furthering the use of positron emission tomography for medical applications. The principal part of the emission spectrum of liquid xenon is in the vacuum ultraviolet (VUV) region, as is also the case for gaseous xenon. The emission mechanism [4] is considered to be a common radiative de-excitation process of the excimer Xen2. This excimer is produced from the primary ionized or excited xenon atoms through interactions with ambient ground-state xenon.
E-mail address: [email protected] (K. Fujii). http://dx.doi.org/10.1016/j.nima.2015.05.065 0168-9002/& 2015 Elsevier B.V. All rights reserved.
However, different values for the emission wavelength have been reported and used in previous studies. Jortner et al. reported [5] that the emission spectrum of liquid xenon is centered at 56,180 cm 1 with the full width at half maximum (FWHM) of 4500 cm 1. These values can be readily converted into wavelengths of 178.0 nm and 14.3 nm, respectively, with the central value being used widely. However, a textbook [6] listed slightly shorter values, such as 174 nm for the central wavelength. This small difference is not negligible when planning or performing accurate experiments. Contrary to the case in the visible region, the same amount of difference in the wavelength affects various optical parameters significantly in the VUV region. The Rayleigh scattering length is a representative example. We can calculate the scattering length following the discussion by Seidel et al. [7] using recent values of the refractive index of liquid xenon [8]. The results show that a difference of only 4 nm in the wavelengths can cause a change of 20% or more in the Rayleigh scattering length of approximately 40–50 cm for the VUV scintillation light of liquid xenon. Furthermore, for other practical applications of liquid-xenon scintillators, knowledge of the correct emission spectrum is indispensable, notably in determining the accurate response of a detector. The correct wavelength dependence of photon transmittance of any devices must be considered. Optical processes occurring inside or at the boundary of the liquid xenon, such as
294
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
scattering, reflection, and refraction, are usually wavelength dependent. The purpose of this study was to determine the emission spectrum of liquid xenon accurately. We developed a sophisticated optical system suitable for use with cryogenic liquid xenon and performed spectroscopic intensity measurements for the observed scintillation phenomenon with high sensitivity. The optical system was calibrated using a standard light source, and the data were corrected carefully using proper spectroscopic procedures.
2. Experimental setup 2.1. Optical system We constructed an optical system on the basis of the coincidentphoton-counting technique [9] to measure the monochromated weak emission of liquid xenon. Fig. 1 shows the side view of the optical system. The optical components were placed under vacuum condition to allow for VUV light conduction and to thermally insulate the liquid-xenon container effectively. Liquid xenon was cooled cryogenically in a small cylindrical container placed within a vacuum chamber and excited by gamma-rays (1.17 and 1.33 MeV) from a standard checking 60Co source (1.8 MBq), which was placed outside the surface of the vacuum chamber. The container was made of stainless steel and was equipped with MgF2 viewports at both ends. The inner dimensions of the container were the following: diameter of 16 mm and length of 56.6 mm. The scintillation light emitted through one end of the container was detected directly by a VUV-sensitive photomultiplier (PMT) with a quartz window (Hamamatsu Photonics, a custom product based on R7600, referred to as PMT1). The voltage applied to PMT1 was þ800 V. The scintillation light exiting the other end of the container was first monochromated by a Seya-Namioka-type vacuum monochromator (Acton, VM-502-S). The primary specifications of the monochromator were as follows: wavelength accuracy of 0.1 nm, focal length of 0.2 m. and F-number of 4.5. By considering the Fnumber, the container size and its position were set to minimize the amount of light reflected by the inner surface of the container; therefore, all light entered directly into the monochromator. The wavelength setting and scan rate of the monochromator were controlled using a Linux-based software program implemented in a data acquisition system. The monochromated light was then measured with another VUV-sensitive, high-gain PMT with a MgF2 window (Hamamatsu Photonics, R6836PX, referred to as PMT2). The voltage applied to PMT2 was þ1450 V. 2.2. Data acquisition system The intensity of the monochromated scintillation light was measured using a scaler and a charge-sensitive analog-to-digital converter (ADC) (HOSHIN ELECTRONICS CO., LTD, C009) module in
parallel under the control of a computer automated measurement and control (CAMAC) system. To measure the faint signal efficiently even in the intrinsic dark noise of the PMTs, the coincidences and accidental coincidences were counted to deduce the net signal counts. A schematic diagram of the data acquisition system is shown in Fig. 2. Each analog pulse output from PMT1 was amplified 7.1 times and fed into a constant-fraction discriminator (CFD) to issue the nuclear instrumentation module (NIM) pulse for the scintillation timing. The output analog pulse of PMT2, which detected photons of monochromated light, was first duplicated by a linear fan-in/fan-out module (LeCroy, 428F). One pulse was fed into the analog input of the ADC to evaluate the signal charge in each pulse. Here, the charge integration time was selected to be 250–300 ns (depending on the run) to cover most of the scintillation. The other analog pulse was fed into the CFD to output a NIM pulse used to count the numbers of coincidences and accidental coincidences in combination with the NIM pulse of PMT1. The coincidence condition was that the PMT2 signal should appear within 240–300 ns (depending on the run) after the PMT1 signal. By contrast, the accidental coincidence condition was that a 1 μs delay was added to the PMT2 signal for the coincidence condition; this condition did not result in a correlation. In addition, these NIM pulses were counted separately as single rates of every PMT. The NIM pulses were also used for other purposes in the electronics. The gate pulse was generated for the ADC using the NIM pulse of PMT1. Another NIM pulse, with a width of 500 μs, was used to inhibit the input of the ADC during the ADC data conversion, which lasted for 200 μs. 2.3. Xenon-handling system A diagram of the xenon-handling system is shown in Fig. 3. After evacuating the entire gas line to a residual pressure on the order of 10 5 Pa, pure xenon gas was introduced into the line through a purifier containing 5000 activated getter pills (SAES Getters, St707 Pill/4-2/50). While maintaining the target pressure at approximately 110 kPa, the xenon gas was cooled and condensed in the container continuously using a pulse-tube refrigerator (Iwatani Corp., PDC08Y); additional xenon gas was supplied continuously. The pressure of the xenon gas was measured with a digital manometer (Yokogawa Electric Corp., MT110). The refrigeration power was compensated for with a heater attached on the cold head to maintain a target temperature; a platinum resistance thermometer and a digital indicating controller (CHINO Corp., DB1000) were employed for this purpose. The fluid level of the liquid xenon was monitored using three platinum resistance thermometers (LakeShore, PT-111) set vertically at different points in the container at an interval of approximately 8 mm from the bottom. The temperatures from the three thermometers were read with a data acquisition/switch unit
Fig. 1. The schematic side view of the optical system under vacuum. A 60Co source is located on the side surface of the chamber and irradiates liquid xenon. The scintillation light was emitted through two ends of the liquid xenon container; one end was detected directly by a PMT, and the other was detected by another PMT after passing through the monochromator.
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
295
Fig. 2. Data acquisition system for coincident photon counting.
Fig. 3. Overview of the xenon handling system. The xenon gas in a buffer cylinder was introduced into the liquid xenon container through the purifier and cooled and condensed using a refrigerator.
Table 1 Summary of measurement conditions. Slit width (mm)/height (mm) Thresholds of CFD (mV)
run#1 0.5/16 run#2 0.8/4
PMT1
PMT2
50 29
30 30
Scanning range of wavelength (nm) Scanning interval Gate width Live time per one wavelength of wavelength (nm) for ADC (ns) (s)
160.0–191.2 160.0–189.9
(Hewlett-Packard, 34970A). A sensor above the fluid surface measured higher temperature than that of the liquid because of the heat generated by the sensor itself and a worse heat conduction in the gas than in the liquid.
3. Measurements and data analysis 3.1. Measurement conditions During the measurements, 11 ml of liquid xenon was maintained in the cylindrical container. The steady state of the liquid xenon was confirmed visually through the viewports throughout the test runs. We confirmed the stability of the state of the xenon; xenon gas and liquid phases are in equilibrium without showing any bubble formation.
0.8 1.3
300 250
ADC
Scaler
1280 780
1500 1200
For the measurement runs, two different slits were used for the monochromator: one slit set a width of 0.5 mm and a height of 16 mm (hereafter referred to as run#1), and the other slit set a width of 0.8 mm and a height of 4 mm (hereafter referred to as run#2). The physical conditions were 110 kPa and 168 K for run#1 and 117 kPa and 169 K for run#2. These narrow slits ensured that the intensity of the monochromated light was low (at the level of single-photon counting); this was confirmed in advance by the charge distribution measured by the ADC. For the monochromated light signal from PMT2, the threshold level of the CFD was set to 30 mV, corresponding to 0.2 photoelectrons. For the amplified signal from PMT1, the threshold level of the CFD was 50 mV for run#1 and 29 mV for run#2. These conditions are summarized in Table 1 with the other parameters, including the scanning wavelengths, the gate widths for the ADC, and the live times of the data acquisition for each wavelength data set.
296
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
3.2. Correction of wavelength To obtain the correct emission spectrum, wavelength calibrations were performed using an atomic emission line (184.90 nm) of a standard low-pressure mercury lamp. We measured the identical emission line with slits of varying widths and heights and determined the other measurement conditions corresponding to the apparatus used. On the basis of the assumption that the intrinsic width of the emission line is smaller than the resolution of the experimental apparatus used, the observed peak widths and positions were safely used to estimate the characteristics of the experimental system. In Fig. 4, the peak wavelengths of the line observed using the experimental system are plotted as functions of the slit width. As shown in the figure, the peaks shifted to larger wavelengths with an increase in the slit width, ultimately asymptotically reaching constant values. The degree of shift in the peaks was larger for the slit with a larger height. These results are fully consistent with those expected for normal Seya-Namioka-type monochromators such as VM-502-S, which accounts for coma-type aberration and astigmatism. To quantify the shift, we examined the relationship between the peak shift and slit width for specific slit heights on the basis of a previous work [10] on the characteristics of gratings of the same type. The results of a detailed analysis and numerical evaluation of the skewed image of the entrance slit suggested that the peak shift can be accurately expressed as a function of the slit width using a following simple formula: f ðxÞ ¼ Að1 e
Bx
ÞþC
ð1Þ 1
for a slit height of where A and B are 0.035 nm and 69.0 mm 4 mm, and 0.56 nm and 2.95 mm 1 for a slit height of 16 mm, respectively. Parameter C is the unknown common systematic shift in the wavelength and is independent of the properties of the slit of the monochromator. We found the value of C to be 0.11 nm by fitting the functions to all of the data simultaneously. This value is consistent with the accuracy of the monochromator. The fitting functions are also shown in Fig. 4. From these analyses, we concluded that the wavelength shift was þ0.54 nm for run#1 and þ0.15 nm for run#2; these values were then used in the following analysis. 3.3. Background subtraction and corrections of light intensity To obtain the emission spectrum, the light intensity was calculated from the number of events as a function of the wavelength. In the case of the ADC data analysis, the charge distributions obtained at below or above the apparent emission peak (approximately 160 nm
Fig. 4. Peak wavelengths measured for the 184.90 nm line from a mercury lamp with slits of 16 mm (round plot) and 4 mm (square plot) height. The curves are the best-fit functions determined on the simulation (cross plot) basis in the previous study [10] of the gratings of a similar monochromator.
or 190 nm) were first subtracted from the signal charge distributions to remove the background events. The resultant charge distributions were then examined closely, and the number of events was calculated for events above the relevant threshold. This threshold was defined as lying between the pedestal peak and the higher tail of the single photoelectron (p.e.) distribution, including the low-charge dynode pulse events [11] because of the single electron emission at the dynodes. The thresholds set for the charge distributions were low enough and corresponded to 0.04 p.e. for run#1 and 0.06 p.e. for run#2. For the scaler data analysis, the number of accidental coincidences was simply subtracted from the number of coincidences to subtract the background. A comparison of the emission spectra obtained from the ADC and the scaler shows that they were proportional. No significant differences between the two were found, except in the absolute number of entries. Although the spectrum from the scaler had a greater number of entries resulting in smaller statistical errors, we only used the ADC data because the criterion for signal selection was set on the basis of the signal charge distribution. The background-subtracted spectra were corrected on the basis of the detection efficiency of scintillation photons using efficiencies of various optical components of the system. The corrected intensity, N(λ), was obtained from the number of events, C(λ), for each wavelength, λ, using the following equation: NðλÞ ¼
CðλÞ TðλÞ UGEðλÞ UMRðλÞ UQ EðλÞ
ð2Þ
where T(λ) is the transmittance of the MgF2 window of the container facing the monochromator, GE(λ) is the grating efficiency, MR(λ) is the reflectance of the mirror in the monochromator, and QE(λ) is the quantum efficiency of PMT2. These characteristic data were taken from the respective specification sheets and are summarized in Fig. 5. As for, T(λ), the refractive index of liquid xenon was also considered to evaluate the reflection of light by the inner surface of the MgF2 window. Fig. 6 shows the corrected spectra of the liquid xenon scintillation light for run#1 and run#2 as functions of the wave number. As seen in the figure, the spectra exhibited good overlapping when one normalizes the heights of the peaks. The shape could be approximated with a Gaussian function. The best-fit Gaussian curves are also shown in the figure. 3.4. Deconvolution The spectra of liquid xenon scintillation light shown in Fig. 6 are still convoluted with the apparatus function. Therefore, the
Fig. 5. Efficiencies of the optical components involved in the system: transmittance of the MgF2 window, reflectance of the mirror, efficiency of the grating, and quantum efficiency of PMT2.
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
297
32787167 (stat.)765 (syst.) cm 1 for run#2. The systematic errors in the FWHM values were estimated from the uncertainty of the threshold set for the charge distributions of the ADC data. Because these two results for the different slit dimensions and physical conditions were consistent with each other, we calculated the weighted average (using errors) of the results: the center is 57,199 734 (stat.) 733 (syst.) cm 1 (174.8 70.1 (stat.) 70.1 (syst.) nm), and the FWHM is 3328772 (stat.) 765 (syst.) cm 1 (10.2 70.2 (stat.) 70.2 (sys.) nm). 4. Summary and discussion
Fig. 6. Emission spectra of liquid xenon obtained from the ADC data for run#1 (round plot) and run#2 (square plot). The wavelength and the intensity corrections are applied. The curves are the best-fitting functions for run#1 (solid curve) and run#2 (dashed curve), respectively, as determined assuming a Gaussian function.
Fig. 7. Spectral resolution of the system obtained from the FWHM of an atomic line of mercury at slit widths of 16 mm (round plot and dashed line) and 4 mm (square plot and solid line).
peaks can be considered to be slightly broadened, depending on the slit dimensions. We performed a deconvolution of the spectra to extract the intrinsic emission spectrum of the liquid xenon scintillation light using the known apparatus function. In practice, because the convoluted spectra and the apparatus function were both well-approximated by Gaussian functions, the intrinsic spectrum could also be expected to be a Gaussian and centered at the identical position with a smaller FWHM. Furthermore, the FWHM could be readily determined using the following formula: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi W 2o W 2r ð3Þ where Wo and Wr are the FWHMs of the observed emission peak of liquid xenon and the apparatus function, respectively. The FWHMs of the apparatus function are shown in Fig. 7. As seen from the figure, the FWHM depended on the slit width and was 600713 (stat.)739 (syst.) cm 1 for run#1 and 864714 (stat.) 765 (syst.) cm 1 for run#2. For both cases, the FWHM was smaller than that of the unfolded emission peak of liquid xenon scintillation light. Therefore, we concluded that the deconvolution did not change the shape of the emission spectrum significantly. After the deconvolution procedure and although the contribution was small, the center and FWHM of the emission peak were evaluated to be 57,218737 (stat.)739 (syst.) cm 1 and 3339779 (stat.)739 (syst.) cm 1 for run#1 and 57,116779 (stat.)765 (syst.) cm 1 and
The emission spectrum of cryogenic liquid xenon in the VUV region was investigated by irradiating the liquid xenon with gamma rays from a radioactive source. Although the monochromated light of scintillation was weak, a high signal-to-noise ratio could be achieved using the constrained photon counting with relevant threshold corresponding to the occurrence of scintillation. To obtain the accurate spectrum, we carefully performed wavelength correction on the basis of measurements using a low-pressure mercury lamp, in addition to proper corrections of optical efficiencies. We could obtain the correct emission spectrum of liquid xenon. The obtained emission spectrum is found to be in the shape of a Gaussian function centered at 57,1997 34 (stat.) 733 (syst.) cm 1 (174.8 70.1 (stat.) 70.1 (syst.) nm) with an FWHM of 33287 72 (stat.) 765 (syst.) cm 1 (10.2 70.2 (stat.) 70.2 (sys.) nm). Compared to the values reported by Jortner et al. [5], the center is approximately 3 nm shorter and the FWHM is 4 nm smaller. However, the center is approximately 1 nm longer than that reported in a textbook [6]. The causes of these differences are not clear; however, we believe that the results of the present study are the most accurate ever obtained and therefore provide valuable information regarding instruments that use a liquid-xenon scintillator. Furthermore, the technique used in the present study should be applicable for other wavelength regions and for other cryogenic liquid rare gases. Acknowledgments We thank Ikuko Murayama, Takanori Fujita, Shuhei Oyama, Mao Yoshida, Ryo Hamanishi, and Yuki Katada for their assistance with the experiments. We also thank Professors Masatoshi Tanaka and Takao Sekiya, both at Yokohama National University, for fruitful discussions. This work was supported by JSPS KAKENHI Grant no. 22540307 from the Japan Society for the Promotion of Science, Japan, and partially carried out by the joint research program of the Institute for Cosmic Ray Research, University of Tokyo. References [1] K. Abe, K. Hieda, K. Hiraide, et al., Physics Letters B 719 (2013) 78. [2] J. Adam, X. Bai, A.M. Baldini, et al., Physical Review Letters 110 (2013) 201801. [3] A. Miceli, P. Amaudruz, F. Benard, et al., Journal of Physics: Conference Series 312 (2011). [4] E. Aprile, T. Doke, Reviews of Modern Physics 82 (2010) 2053. [5] J. Jortner, L. Meyer, S.A. Rice, et al., The Journal of Chemical Physics 42 (1965) 4250. [6] C. Grupen and B. Shwartz, Particle Detectors, 2nd ed., (2008) 127. [7] G.M. Seidel, R.E. Lanou, W. Yao, Nuclear Instruments and Methods in Physics Research A 489 (2002) 189. [8] S. Nakamura, Y. Ozaki, K. Saito et al., Proceedings of the Workshop on Ionization and Scintillation Counters and Their Uses (unpublished, 2007). [9] J.E. McMillan, C.J. Martoff, Measurement Science and Technology 17 (2006) 2362. [10] Tatsuo Harada, Toshiaki Kita, Applied Optics 19 (1980) 3987. [11] W Becker, Advanced Time-Correlated Single Photon Counting Techniques (2005), Springer-Verlag Berlin Heidelberg.
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
High-accuracy measurement of the emission spectrum of liquid xenon in the vacuum ultraviolet region Keiko Fujii a, Yuya Endo a, Yui Torigoe a, Shogo Nakamura a, Tomiyoshi Haruyama b, Katsuyu Kasami b, Satoshi Mihara b,c, Kiwamu Saito b,c, Shinichi Sasaki b,c, Hiroko Tawara b a
Faculty of Engineering, Yokohama National University, Yokohama, Kanagawa 240-8501, Japan High Energy Accelerator Research Organization (KEK), Tsukuba, Ibaraki 305-0801, Japan c The Graduate School of Advanced Studies, Hayama, Kanagawa 240-0193, Japan b
art ic l e i nf o
a b s t r a c t
Article history: Received 1 February 2015 Received in revised form 20 May 2015 Accepted 27 May 2015 Available online 5 June 2015
The emission spectrum of cryogenic liquid xenon in the vacuum ultraviolet region was measured by irradiating liquid xenon with gamma-rays from a radioactive source. To achieve a high signal-to-noise ratio, we employed coincident photon counting. Additionally, the charge of the photo-sensor signals was measured to estimate the number of detected photons accurately. In addition, proper corrections were incorporated for the wavelength; response functions of the apparatus obtained using a low-pressure mercury lamp, and photon detection efficiencies of the optical system were considered. The obtained emission spectrum is found to be in the shape of a Gaussian function, with the center at 57,199 734 (stat.) 7 33 (syst.) cm 1 (174.8 70.1 (stat.) 70.1 (syst.) nm) and the full width at half maximum of 3328772 (stat.) 7 65 (syst.) cm 1 (10.27 0.2 (stat.) 7 0.2 (sys.) nm). These results are the most accurate values obtained in terms of the data acquisition method and the calibration for the experimental system and provide valuable information regarding the high-precision instruments that employ a liquid-xenon scintillator. & 2015 Elsevier B.V. All rights reserved.
Keywords: Scintillator Liquid xenon Photon counting Emission spectrum
1. Introduction Liquid xenon is an excellent scintillation material because of a number of advantages such as its large photon yield and fast response. This material is used primarily in astrophysics and particle physics studies. A currently on-going underground experiment [1] employs a ton-scale detector containing ultrapure liquid xenon as the scintillator with the aim of directly detecting the weakly interacting massive particle (WIMP), which is supposed to constitute dark matter. An accelerator-based experiment [2] searches for rare muon decays using a ton-scale liquid-xenon scintillator as an efficient gamma-ray detector to observe new signs of physics beyond the Standard Model. Time-projection chambers using liquid xenon as a drift medium are now under development [3] with the aim of furthering the use of positron emission tomography for medical applications. The principal part of the emission spectrum of liquid xenon is in the vacuum ultraviolet (VUV) region, as is also the case for gaseous xenon. The emission mechanism [4] is considered to be a common radiative de-excitation process of the excimer Xen2. This excimer is produced from the primary ionized or excited xenon atoms through interactions with ambient ground-state xenon.
E-mail address: [email protected] (K. Fujii). http://dx.doi.org/10.1016/j.nima.2015.05.065 0168-9002/& 2015 Elsevier B.V. All rights reserved.
However, different values for the emission wavelength have been reported and used in previous studies. Jortner et al. reported [5] that the emission spectrum of liquid xenon is centered at 56,180 cm 1 with the full width at half maximum (FWHM) of 4500 cm 1. These values can be readily converted into wavelengths of 178.0 nm and 14.3 nm, respectively, with the central value being used widely. However, a textbook [6] listed slightly shorter values, such as 174 nm for the central wavelength. This small difference is not negligible when planning or performing accurate experiments. Contrary to the case in the visible region, the same amount of difference in the wavelength affects various optical parameters significantly in the VUV region. The Rayleigh scattering length is a representative example. We can calculate the scattering length following the discussion by Seidel et al. [7] using recent values of the refractive index of liquid xenon [8]. The results show that a difference of only 4 nm in the wavelengths can cause a change of 20% or more in the Rayleigh scattering length of approximately 40–50 cm for the VUV scintillation light of liquid xenon. Furthermore, for other practical applications of liquid-xenon scintillators, knowledge of the correct emission spectrum is indispensable, notably in determining the accurate response of a detector. The correct wavelength dependence of photon transmittance of any devices must be considered. Optical processes occurring inside or at the boundary of the liquid xenon, such as
294
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
scattering, reflection, and refraction, are usually wavelength dependent. The purpose of this study was to determine the emission spectrum of liquid xenon accurately. We developed a sophisticated optical system suitable for use with cryogenic liquid xenon and performed spectroscopic intensity measurements for the observed scintillation phenomenon with high sensitivity. The optical system was calibrated using a standard light source, and the data were corrected carefully using proper spectroscopic procedures.
2. Experimental setup 2.1. Optical system We constructed an optical system on the basis of the coincidentphoton-counting technique [9] to measure the monochromated weak emission of liquid xenon. Fig. 1 shows the side view of the optical system. The optical components were placed under vacuum condition to allow for VUV light conduction and to thermally insulate the liquid-xenon container effectively. Liquid xenon was cooled cryogenically in a small cylindrical container placed within a vacuum chamber and excited by gamma-rays (1.17 and 1.33 MeV) from a standard checking 60Co source (1.8 MBq), which was placed outside the surface of the vacuum chamber. The container was made of stainless steel and was equipped with MgF2 viewports at both ends. The inner dimensions of the container were the following: diameter of 16 mm and length of 56.6 mm. The scintillation light emitted through one end of the container was detected directly by a VUV-sensitive photomultiplier (PMT) with a quartz window (Hamamatsu Photonics, a custom product based on R7600, referred to as PMT1). The voltage applied to PMT1 was þ800 V. The scintillation light exiting the other end of the container was first monochromated by a Seya-Namioka-type vacuum monochromator (Acton, VM-502-S). The primary specifications of the monochromator were as follows: wavelength accuracy of 0.1 nm, focal length of 0.2 m. and F-number of 4.5. By considering the Fnumber, the container size and its position were set to minimize the amount of light reflected by the inner surface of the container; therefore, all light entered directly into the monochromator. The wavelength setting and scan rate of the monochromator were controlled using a Linux-based software program implemented in a data acquisition system. The monochromated light was then measured with another VUV-sensitive, high-gain PMT with a MgF2 window (Hamamatsu Photonics, R6836PX, referred to as PMT2). The voltage applied to PMT2 was þ1450 V. 2.2. Data acquisition system The intensity of the monochromated scintillation light was measured using a scaler and a charge-sensitive analog-to-digital converter (ADC) (HOSHIN ELECTRONICS CO., LTD, C009) module in
parallel under the control of a computer automated measurement and control (CAMAC) system. To measure the faint signal efficiently even in the intrinsic dark noise of the PMTs, the coincidences and accidental coincidences were counted to deduce the net signal counts. A schematic diagram of the data acquisition system is shown in Fig. 2. Each analog pulse output from PMT1 was amplified 7.1 times and fed into a constant-fraction discriminator (CFD) to issue the nuclear instrumentation module (NIM) pulse for the scintillation timing. The output analog pulse of PMT2, which detected photons of monochromated light, was first duplicated by a linear fan-in/fan-out module (LeCroy, 428F). One pulse was fed into the analog input of the ADC to evaluate the signal charge in each pulse. Here, the charge integration time was selected to be 250–300 ns (depending on the run) to cover most of the scintillation. The other analog pulse was fed into the CFD to output a NIM pulse used to count the numbers of coincidences and accidental coincidences in combination with the NIM pulse of PMT1. The coincidence condition was that the PMT2 signal should appear within 240–300 ns (depending on the run) after the PMT1 signal. By contrast, the accidental coincidence condition was that a 1 μs delay was added to the PMT2 signal for the coincidence condition; this condition did not result in a correlation. In addition, these NIM pulses were counted separately as single rates of every PMT. The NIM pulses were also used for other purposes in the electronics. The gate pulse was generated for the ADC using the NIM pulse of PMT1. Another NIM pulse, with a width of 500 μs, was used to inhibit the input of the ADC during the ADC data conversion, which lasted for 200 μs. 2.3. Xenon-handling system A diagram of the xenon-handling system is shown in Fig. 3. After evacuating the entire gas line to a residual pressure on the order of 10 5 Pa, pure xenon gas was introduced into the line through a purifier containing 5000 activated getter pills (SAES Getters, St707 Pill/4-2/50). While maintaining the target pressure at approximately 110 kPa, the xenon gas was cooled and condensed in the container continuously using a pulse-tube refrigerator (Iwatani Corp., PDC08Y); additional xenon gas was supplied continuously. The pressure of the xenon gas was measured with a digital manometer (Yokogawa Electric Corp., MT110). The refrigeration power was compensated for with a heater attached on the cold head to maintain a target temperature; a platinum resistance thermometer and a digital indicating controller (CHINO Corp., DB1000) were employed for this purpose. The fluid level of the liquid xenon was monitored using three platinum resistance thermometers (LakeShore, PT-111) set vertically at different points in the container at an interval of approximately 8 mm from the bottom. The temperatures from the three thermometers were read with a data acquisition/switch unit
Fig. 1. The schematic side view of the optical system under vacuum. A 60Co source is located on the side surface of the chamber and irradiates liquid xenon. The scintillation light was emitted through two ends of the liquid xenon container; one end was detected directly by a PMT, and the other was detected by another PMT after passing through the monochromator.
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
295
Fig. 2. Data acquisition system for coincident photon counting.
Fig. 3. Overview of the xenon handling system. The xenon gas in a buffer cylinder was introduced into the liquid xenon container through the purifier and cooled and condensed using a refrigerator.
Table 1 Summary of measurement conditions. Slit width (mm)/height (mm) Thresholds of CFD (mV)
run#1 0.5/16 run#2 0.8/4
PMT1
PMT2
50 29
30 30
Scanning range of wavelength (nm) Scanning interval Gate width Live time per one wavelength of wavelength (nm) for ADC (ns) (s)
160.0–191.2 160.0–189.9
(Hewlett-Packard, 34970A). A sensor above the fluid surface measured higher temperature than that of the liquid because of the heat generated by the sensor itself and a worse heat conduction in the gas than in the liquid.
3. Measurements and data analysis 3.1. Measurement conditions During the measurements, 11 ml of liquid xenon was maintained in the cylindrical container. The steady state of the liquid xenon was confirmed visually through the viewports throughout the test runs. We confirmed the stability of the state of the xenon; xenon gas and liquid phases are in equilibrium without showing any bubble formation.
0.8 1.3
300 250
ADC
Scaler
1280 780
1500 1200
For the measurement runs, two different slits were used for the monochromator: one slit set a width of 0.5 mm and a height of 16 mm (hereafter referred to as run#1), and the other slit set a width of 0.8 mm and a height of 4 mm (hereafter referred to as run#2). The physical conditions were 110 kPa and 168 K for run#1 and 117 kPa and 169 K for run#2. These narrow slits ensured that the intensity of the monochromated light was low (at the level of single-photon counting); this was confirmed in advance by the charge distribution measured by the ADC. For the monochromated light signal from PMT2, the threshold level of the CFD was set to 30 mV, corresponding to 0.2 photoelectrons. For the amplified signal from PMT1, the threshold level of the CFD was 50 mV for run#1 and 29 mV for run#2. These conditions are summarized in Table 1 with the other parameters, including the scanning wavelengths, the gate widths for the ADC, and the live times of the data acquisition for each wavelength data set.
296
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
3.2. Correction of wavelength To obtain the correct emission spectrum, wavelength calibrations were performed using an atomic emission line (184.90 nm) of a standard low-pressure mercury lamp. We measured the identical emission line with slits of varying widths and heights and determined the other measurement conditions corresponding to the apparatus used. On the basis of the assumption that the intrinsic width of the emission line is smaller than the resolution of the experimental apparatus used, the observed peak widths and positions were safely used to estimate the characteristics of the experimental system. In Fig. 4, the peak wavelengths of the line observed using the experimental system are plotted as functions of the slit width. As shown in the figure, the peaks shifted to larger wavelengths with an increase in the slit width, ultimately asymptotically reaching constant values. The degree of shift in the peaks was larger for the slit with a larger height. These results are fully consistent with those expected for normal Seya-Namioka-type monochromators such as VM-502-S, which accounts for coma-type aberration and astigmatism. To quantify the shift, we examined the relationship between the peak shift and slit width for specific slit heights on the basis of a previous work [10] on the characteristics of gratings of the same type. The results of a detailed analysis and numerical evaluation of the skewed image of the entrance slit suggested that the peak shift can be accurately expressed as a function of the slit width using a following simple formula: f ðxÞ ¼ Að1 e
Bx
ÞþC
ð1Þ 1
for a slit height of where A and B are 0.035 nm and 69.0 mm 4 mm, and 0.56 nm and 2.95 mm 1 for a slit height of 16 mm, respectively. Parameter C is the unknown common systematic shift in the wavelength and is independent of the properties of the slit of the monochromator. We found the value of C to be 0.11 nm by fitting the functions to all of the data simultaneously. This value is consistent with the accuracy of the monochromator. The fitting functions are also shown in Fig. 4. From these analyses, we concluded that the wavelength shift was þ0.54 nm for run#1 and þ0.15 nm for run#2; these values were then used in the following analysis. 3.3. Background subtraction and corrections of light intensity To obtain the emission spectrum, the light intensity was calculated from the number of events as a function of the wavelength. In the case of the ADC data analysis, the charge distributions obtained at below or above the apparent emission peak (approximately 160 nm
Fig. 4. Peak wavelengths measured for the 184.90 nm line from a mercury lamp with slits of 16 mm (round plot) and 4 mm (square plot) height. The curves are the best-fit functions determined on the simulation (cross plot) basis in the previous study [10] of the gratings of a similar monochromator.
or 190 nm) were first subtracted from the signal charge distributions to remove the background events. The resultant charge distributions were then examined closely, and the number of events was calculated for events above the relevant threshold. This threshold was defined as lying between the pedestal peak and the higher tail of the single photoelectron (p.e.) distribution, including the low-charge dynode pulse events [11] because of the single electron emission at the dynodes. The thresholds set for the charge distributions were low enough and corresponded to 0.04 p.e. for run#1 and 0.06 p.e. for run#2. For the scaler data analysis, the number of accidental coincidences was simply subtracted from the number of coincidences to subtract the background. A comparison of the emission spectra obtained from the ADC and the scaler shows that they were proportional. No significant differences between the two were found, except in the absolute number of entries. Although the spectrum from the scaler had a greater number of entries resulting in smaller statistical errors, we only used the ADC data because the criterion for signal selection was set on the basis of the signal charge distribution. The background-subtracted spectra were corrected on the basis of the detection efficiency of scintillation photons using efficiencies of various optical components of the system. The corrected intensity, N(λ), was obtained from the number of events, C(λ), for each wavelength, λ, using the following equation: NðλÞ ¼
CðλÞ TðλÞ UGEðλÞ UMRðλÞ UQ EðλÞ
ð2Þ
where T(λ) is the transmittance of the MgF2 window of the container facing the monochromator, GE(λ) is the grating efficiency, MR(λ) is the reflectance of the mirror in the monochromator, and QE(λ) is the quantum efficiency of PMT2. These characteristic data were taken from the respective specification sheets and are summarized in Fig. 5. As for, T(λ), the refractive index of liquid xenon was also considered to evaluate the reflection of light by the inner surface of the MgF2 window. Fig. 6 shows the corrected spectra of the liquid xenon scintillation light for run#1 and run#2 as functions of the wave number. As seen in the figure, the spectra exhibited good overlapping when one normalizes the heights of the peaks. The shape could be approximated with a Gaussian function. The best-fit Gaussian curves are also shown in the figure. 3.4. Deconvolution The spectra of liquid xenon scintillation light shown in Fig. 6 are still convoluted with the apparatus function. Therefore, the
Fig. 5. Efficiencies of the optical components involved in the system: transmittance of the MgF2 window, reflectance of the mirror, efficiency of the grating, and quantum efficiency of PMT2.
K. Fujii et al. / Nuclear Instruments and Methods in Physics Research A 795 (2015) 293–297
297
32787167 (stat.)765 (syst.) cm 1 for run#2. The systematic errors in the FWHM values were estimated from the uncertainty of the threshold set for the charge distributions of the ADC data. Because these two results for the different slit dimensions and physical conditions were consistent with each other, we calculated the weighted average (using errors) of the results: the center is 57,199 734 (stat.) 733 (syst.) cm 1 (174.8 70.1 (stat.) 70.1 (syst.) nm), and the FWHM is 3328772 (stat.) 765 (syst.) cm 1 (10.2 70.2 (stat.) 70.2 (sys.) nm). 4. Summary and discussion
Fig. 6. Emission spectra of liquid xenon obtained from the ADC data for run#1 (round plot) and run#2 (square plot). The wavelength and the intensity corrections are applied. The curves are the best-fitting functions for run#1 (solid curve) and run#2 (dashed curve), respectively, as determined assuming a Gaussian function.
Fig. 7. Spectral resolution of the system obtained from the FWHM of an atomic line of mercury at slit widths of 16 mm (round plot and dashed line) and 4 mm (square plot and solid line).
peaks can be considered to be slightly broadened, depending on the slit dimensions. We performed a deconvolution of the spectra to extract the intrinsic emission spectrum of the liquid xenon scintillation light using the known apparatus function. In practice, because the convoluted spectra and the apparatus function were both well-approximated by Gaussian functions, the intrinsic spectrum could also be expected to be a Gaussian and centered at the identical position with a smaller FWHM. Furthermore, the FWHM could be readily determined using the following formula: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi W 2o W 2r ð3Þ where Wo and Wr are the FWHMs of the observed emission peak of liquid xenon and the apparatus function, respectively. The FWHMs of the apparatus function are shown in Fig. 7. As seen from the figure, the FWHM depended on the slit width and was 600713 (stat.)739 (syst.) cm 1 for run#1 and 864714 (stat.) 765 (syst.) cm 1 for run#2. For both cases, the FWHM was smaller than that of the unfolded emission peak of liquid xenon scintillation light. Therefore, we concluded that the deconvolution did not change the shape of the emission spectrum significantly. After the deconvolution procedure and although the contribution was small, the center and FWHM of the emission peak were evaluated to be 57,218737 (stat.)739 (syst.) cm 1 and 3339779 (stat.)739 (syst.) cm 1 for run#1 and 57,116779 (stat.)765 (syst.) cm 1 and
The emission spectrum of cryogenic liquid xenon in the VUV region was investigated by irradiating the liquid xenon with gamma rays from a radioactive source. Although the monochromated light of scintillation was weak, a high signal-to-noise ratio could be achieved using the constrained photon counting with relevant threshold corresponding to the occurrence of scintillation. To obtain the accurate spectrum, we carefully performed wavelength correction on the basis of measurements using a low-pressure mercury lamp, in addition to proper corrections of optical efficiencies. We could obtain the correct emission spectrum of liquid xenon. The obtained emission spectrum is found to be in the shape of a Gaussian function centered at 57,1997 34 (stat.) 733 (syst.) cm 1 (174.8 70.1 (stat.) 70.1 (syst.) nm) with an FWHM of 33287 72 (stat.) 765 (syst.) cm 1 (10.2 70.2 (stat.) 70.2 (sys.) nm). Compared to the values reported by Jortner et al. [5], the center is approximately 3 nm shorter and the FWHM is 4 nm smaller. However, the center is approximately 1 nm longer than that reported in a textbook [6]. The causes of these differences are not clear; however, we believe that the results of the present study are the most accurate ever obtained and therefore provide valuable information regarding instruments that use a liquid-xenon scintillator. Furthermore, the technique used in the present study should be applicable for other wavelength regions and for other cryogenic liquid rare gases. Acknowledgments We thank Ikuko Murayama, Takanori Fujita, Shuhei Oyama, Mao Yoshida, Ryo Hamanishi, and Yuki Katada for their assistance with the experiments. We also thank Professors Masatoshi Tanaka and Takao Sekiya, both at Yokohama National University, for fruitful discussions. This work was supported by JSPS KAKENHI Grant no. 22540307 from the Japan Society for the Promotion of Science, Japan, and partially carried out by the joint research program of the Institute for Cosmic Ray Research, University of Tokyo. References [1] K. Abe, K. Hieda, K. Hiraide, et al., Physics Letters B 719 (2013) 78. [2] J. Adam, X. Bai, A.M. Baldini, et al., Physical Review Letters 110 (2013) 201801. [3] A. Miceli, P. Amaudruz, F. Benard, et al., Journal of Physics: Conference Series 312 (2011). [4] E. Aprile, T. Doke, Reviews of Modern Physics 82 (2010) 2053. [5] J. Jortner, L. Meyer, S.A. Rice, et al., The Journal of Chemical Physics 42 (1965) 4250. [6] C. Grupen and B. Shwartz, Particle Detectors, 2nd ed., (2008) 127. [7] G.M. Seidel, R.E. Lanou, W. Yao, Nuclear Instruments and Methods in Physics Research A 489 (2002) 189. [8] S. Nakamura, Y. Ozaki, K. Saito et al., Proceedings of the Workshop on Ionization and Scintillation Counters and Their Uses (unpublished, 2007). [9] J.E. McMillan, C.J. Martoff, Measurement Science and Technology 17 (2006) 2362. [10] Tatsuo Harada, Toshiaki Kita, Applied Optics 19 (1980) 3987. [11] W Becker, Advanced Time-Correlated Single Photon Counting Techniques (2005), Springer-Verlag Berlin Heidelberg.