VUV Transmission of PTFE for Xenon-based Particle Detectors

Liquid xenon (LXe) based detectors for rare event searches in particle and astroparticle physics are optimized for high xenon scintillation light collection and low background rate from detector materials. Polytetrafluoroethylene (PTFE, Teflon) is commonly used to encapsulate the active LXe volume due to its high reflectance for VUV LXe scintillation light with peak emission at 178 nm. Reflectance, transmission and number of background signals arising from PTFE depend on the thickness of the PTFE detector walls. In this work, we present VUV transmission measurements for PTFE of various thicknesses often considered in the design phase of LXe detectors. PTFE samples are measured in an apparatus previously used for reflectance measurements in LXe using collimated light at a wavelength of 178 nm. Measurements in vacuum as well as gaseous xenon yield a transmission coefficient of $\lambda = (0.89 \pm 0.05)\,\mathrm{mm}$ for light not reflected at the PTFE surface. The PTFE wall thickness of the XENONnT dark matter experiment was optimized by these measurements and selected as $\geq 3\,\mathrm{mm}$.

WIMPs are expected to produce a single scatter nuclear recoil signal [20], thus neutron-induced nuclear recoils are one of the most critical backgrounds [21]. PTFE is a fluoropolymer (C 2 F 4 ) n and contains two fluorine atoms per carbon atom. Fluorine only has one stable isotope, 19 F, which has a high neutron emission yield in (α, n) reactions [22]. In an experimental setting α-particles originate from the primordial 238 U and 232 Th decay chains present in trace amounts in any material [23]. In order to maximize sensitivity it is crucial to minimize the amount of background signals arising from PTFE by minimizing its amount in a given detector. On the other hand, PTFE serves important purposes in the detectors: apart from maximizing the light collection efficiency [1,24] in the TPC, it is also used as insulating structural material [25][26][27] and optically separates the active LXe TPC (in which light and charge signals are recorded) from the inactive LXe surrounding the TPC. Scintillation light produced in this outer volume must not enter the TPC, as it would lead to artefacts such as accidental coincidences and could reduce the ability to distinguish electron and nuclear recoils. Accordingly, a balance between background reduction and optical shielding has to be found when determining the required thickness of the PTFE walls. This is especially important in the context of third generation LXe dark matter detectors such as DARWIN [28,29].
Here, we report on measurements of the transmission of collimated VUV light at a wavelength of 178 nm through virgin grade PTFE in vacuum and gaseous xenon (GXe). Our results were used to optimize the thickness of the PTFE walls of the XENONnT TPC currently under commissioning [24,30]. Previous measurements indicated that the opacity increases with decreasing wavelength. However, no data was available below 400 nm [31].
The article is structured as follows: the experimental apparatus for the measurement is described in Chapter 2. The measurements and the data analysis are presented in Chapter 3. The results are discussed in Chapter 4.

Münster reflectance and transmission chamber
The apparatus to measure the transmission of VUV light around 178 nm through PTFE is installed at the University of Münster and shown in Figure 1. It was previously used for angle-resolved reflectance measurements off PTFE surfaces immersed in LXe [2][3][4][5] and for photomultiplier tube (PMT) tests [32].
VUV light is generated using a tunable deuterium lamp (McPherson Model 632) providing spectral lines in the range of 110 nm to 170 nm and continuous emission from 170 nm to 400 nm [33]. The wavelength of LXe scintillation light around 178 nm is selected with a vacuum monochromator (McPherson Model 218) consisting of a series of mirrors and a 1200 grooves/mm grating with a blaze wavelength of 150 nm. Its resolution is specified as 5.3 nm for a slit width of 2 mm and a reciprocal linear dispersion for the grating of 26.5 Å/mm [3]. Therefore, light in a range of 5.3 nm around 178 nm is emitted with equal intensity. Light at 178 nm has a mean free path of 3.7 cm in air at normal pressure [3], so the light source is evacuated using a turbo molecular pump (Pfeiffer TMU 071 P). The two main volumes of the instrument, VUV light source and main vacuum chamber, are decoupled by a UV-transmitting MgF 2 window with a transmission of 80% at 178 nm [34]. The light is focused by a UV lens and guided into the main vacuum chamber, accommodating the PTFE sample and light sensor, by a collimator with a circular aperture of 1 mm diameter. The main vacuum chamber, an O-ring sealed stainless steel cylinder of 65 cm diameter, is equipped with a cooled 1"-square PMT sensitive to LXe scintillation light (Hamamatsu R8520-406 PMT), the sample holder and a two-stage cold shield to reduce radiative heat influx. Scattering of light on the inner cold shield is reduced by using anodized and dyed black stainless steel. In order to allow measurements with LXe in the setup, PMT and cold shield are cooled by a cold head (Leybold RPK 1500 E), reaching 160 K and 80 K, respectively [3]. The sample holder is cooled by an Iwatani CryoMini (PD08) and kept at 193.15 K, further reducing heat influx to the system. For the measurements presented here, cooling with both cold heads is used to reduce the PMT noise. Light deflection and condensation of water on cold surfaces, such as the PMT or sample holder, is prevented by evacuation down to 10 −6 mbar [4] using a turbo molecular pump (Leybold TW 300). The pressure in both main volumes, and the temperatures of the laboratory, PMT, sample holder and cold shield are monitored during preparation and measurement phases using Pt100 sensors.
Light from the collimator is focused onto the PTFE sample and measured by the PMT. The PMT is mounted onto a copper support structure which can be rotated around the sample holder's center. The angular movement of the PMT has been automated using a step motor controlled by the data acquisition system. It is equipped with a 24-teeth gear wheel. A 96-teeth gear wheel is mounted on the PMT structure rotation feedthrough and connected by a rubber belt to the motor wheel. Thus, one step of the motor corresponds to 0.45°movement on the PMT rotation axis as illustrated in Figure 1. PMT signals are amplified by a factor 10 via a fast amplifier (CAEN N979) and counted using a combination of a Leading Edge Discriminator (CAEN N840) operated at a threshold of 70 mV and a custom digital counter unit. The adjustment of the PMT operation voltage and temperature dependencies are discussed in [3,35]. The 20.5 × 20.5 mm 2 active area of the PMT is reduced to a circle of 1.5 mm diameter using an aperture attached to the PMT. The distance of the PMT aperture opening and sample holder center is 4 cm. Therefore, the solid angle covered by the PMT with respect to the PTFE sample amounts to ∆Ω PMT = 1.1 mrad. Along the horizontal PMT rotation axis, the aperture covers ∆θ = 2.1°.
The sample holder, placed in the center of the main vacuum chamber, can be rotated by 360°a nd adjusted in height. PTFE samples of various thicknesses are stacked, with a vertical distance of 7 mm, and mounted on the holder allowing for consecutive measurements with comparable pressure, temperature and cleanliness conditions by adjusting the z-position of the sample holder, moving a specific sample into the VUV light beam. Samples can be measured either in vacuum or GXe -as presented in this work -or LXe. Measurements with xenon require to encapsulate the samples with a quartz tube as described in [3,4]. The tube (from Quarzglas Heinrich) with a wall thickness of 5 mm is made of UV-grade fused silica and can withstand several bar over-pressure. It has a nominal light transmission of~80% through the full tube at 178 nm, crossing four surfaces. The refraction index of the tube at VUV wavelength is n = 1.6, thus~5% of the light is reflected at each surface for photons with perpendicular incidence angle [4]. The quartz tube is mounted using threaded rods, blocking transmitted and emitted light for certain PMT angles, as depicted in Figure 1.
Samples of virgin grade PTFE were machined from the same raw material. Each PTFE sample is 7 mm high, 25 mm wide and up to 5 mm thick at the impact point of the collimated 178 nm light beam. The full width at half maximum (FWHM) of the light beam spot at the PTFE sample position is estimated to be about 1 mm [3]. A circular mill-machined recess of 6 mm diameter was machined around the impact point to leave PTFE walls of "0 mm" (hole), 0.80 mm, 1.00 mm, 1.40 mm, 2.15 mm and 3.55 mm thickness, measured with a systematic uncertainty of 0.05 mm. The PTFE surfaces were not treated and thus not optimized for reflectivity. All samples are stacked on top of each other and fixed by means of screws. Light will be transmitted and emitted from the flat PTFE surface facing the PMT, allowing for measurements of the transmission profile at a selected PMT height.

PTFE transmission measurements
In preparation for the actual transmission measurements, the alignment of PMT, sample holder and collimator is ensured with an optically visible light source from the inside of the monochromator by illuminating the PMT aperture through the "0 mm"-sample (hole). The focus point was checked with visible light and concluded to be in the center of the sample holder for the non-visible wavelength of 178 nm. Both, main vacuum chamber and monochromator volume, are then evacuated to sufficiently low pressures of below 3 × 10 −8 mbar and 7 × 10 −5 mbar, respectively, before the deuterium lamp is turned on. The IONIVAC pressure gauge in the main vacuum chamber is turned off during measurements as it generates light during operational mode at low pressures [4]. The PMT and cold shields are being cooled down while the deuterium lamp is reaching stable light emission at 178 nm. The PMT is moved out of the beam center to a position at 25°, see Figure 1. The PMT count rate during cooldown is monitored, revealing a combination of background and dark count rate of~100 Hz at room temperature and below~4 Hz at operation temperature with the selected data acquisition settings. The PTFE samples are passively cooled by the sample holder cold head which is kept at a constant temperature. Thus a constant temperature of the PTFE samples is assumed.
The alignment of the collimated light beam, PTFE sample and PMT aperture in both, relative angle θ and height z, is a critical part of the measurement preparations. The sample holder is lifted out of the beam path and the PMT is aligned with the center of the collimated beam by measuring the light signal along the PMT axis for various PMT heights in z. Measurements in 0.9°steps from θ = −20°to +20°for each 1 mm step in PMT height were taken similar to the profile shown in Figure 2. The center of the collimated 178 nm light beam was obtained with an uncertainty of ∆θ = 0.45°and ∆z = 0.5 mm. It exhibited a maximum count rate of~40 kHz at a distance of 4 cm to the focus point, in agreement with [4]. Next, the "0 mm"-PTFE sample (hole) is moved into the beam path and the sample holder is rotated such that the PTFE surface normal is pointing to θ = −10°. Therefore, light is not blocked by the threaded rods. The light beam is measured again through the quartz tube surrounding the PTFE samples, leading to a similar two-dimensional distribution as without quartz tube. The profile is shown in Figure 2 and has a maximum count rate of~31 kHz. Thus, the maximum count rate is decreased by the quartz tube to~78% as expected for a light beam passing 4 times through a quartz-vacuum transition. As shown in the bottom plot of Figure 2, a central one-dimensional slice through the two-dimensional intensity profile can be  described by a Gaussian distribution G(θ) with standard deviation σ = 1.5°and an amplitude of G 0 = 31 kHz: The standard deviation σ describes the effective broadening of the light beam measured at the position of the PMT given by the angular divergence and the diameter of the light beam at the focal point as well as the diameter of the aperture in front of the PMT. The light beam can be directed through a PTFE sample of defined thickness by exact vertical (z-axis) positioning of the sample holder. The distance between two consecutive PTFE sample centers is fixed to 7 mm by design. The precision for the vertical positioning of the sample holder is estimated as 0.5 mm. The effective thickness of the PTFE sample experienced by the collimated beam depends on the angle of the PTFE surface with respect to the beam. The angle of the sample holder is set to 90°relative to the collimator tube and the PTFE surface normal is pointing to θ = 0°. Measurements in vacuum and GXe were taken following the exact same procedure.
Several scans over the full range of PMT angular positions are taken for each PTFE sample, depicted as PMT axis in Figure 1. The PMT height is kept at the z-position of the beam maximum. Each scan covers a range from −145°to +170°, see Figure 3. Data points are taken every 1.8°c orresponding to 4 steps of the stepper motor. Each PMT position is measured for 120 s, while the counts of the first 4 s are excluded in the analysis to avoid a possible impact of small mechanical vibrations after the PMT movement. The PMT window and aperture opening face the PTFE sample center at any angle. Therefore, photons reach the photocathode at nearly 0°incident angle at the center of the PMT. Differences of the PMT output in dependence of the photon incident angle and position on the photocathode can thus be neglected. The absolute rate of photons detected is reduced compared to the number of photons transmitted for all measurements due to the PMT detection efficiency, defined by the combination of PMT quantum efficiency and collection efficiency. Up to 10 complete scans per PTFE sample are combined by calculating the total number of counts and measurement time to obtain the rate of transmitted photons detected by the PMT. As shown in Figure 3, for angles below −95°and above 115°photons are reflected off the PTFE surface and absorbed in the main chamber. No transmitted photons are detected at the PTFE sample sides at ±90°, as well as at the angular positions of the threaded rods (around −125°, −5°and 115°). The total count rate I(θ, d) at an angle θ is given by: and G 0 (θ) were confirmed by a toy Monte Carlo simulation. The background rate B(θ, a, b) consists of a temperature-dependent dark count rate and a constant stray light contribution. The dark count rate was measured to change linearly with time due to a slowly changing laboratory temperature. As the PMT angle θ is selected sequentially for each measurement of 120 s duration, this leads to a dark count rate model which depends linearly on θ. Not shown in Equation (3.2) is a correction of the angle θ for misalignments of the PTFE samples in the sample holder with respect to the PMT aperture. These misalignments in both directions within the scattering plane (top view in Figure 1) are about 2 mm in size and equal for all scans. They account for both, a shift of the PTFE sample surface with respect to the center of the sample holder and an offset of the PMT rotation around the sample holder center.
The measured count rates for all measurements shown in Figure 3 are simultaneously fitted by χ 2 -minimization using I(θ, d), sharing the parameters for G 0 (θ), the angles θ of the threaded rods and the misalignment correction. Background rate parameters a and b are determined for each PTFE sample measurement, ranging from 0.2 mHz/°to 0.7 mHz/°and 1.8 Hz to 4 Hz, respectively. Therefore, the maximum difference in the background rate B(θ, a, b) in a single measurement is 0.14 Hz. Signal amplitudes I 0 (d) are shown for each measurement in Figure 4.
Beer-Lambert's law is assumed to be applicable for modeling the amount of light I 0 (d) transmitted through a PTFE sample of thickness d defined as: where I 0 (0) represents the signal strength of light transmitted through an infinitesimally thin PTFE sample, which is expected to reflect and transmit but not absorb light, and λ is the transmission coefficient of the medium. Both parameters are obtained for the PTFE samples in vacuum and GXe separately by χ 2 -minimization to I 0, vacuum (0) = (5.7 ± 0.7) Hz, λ vacuum = (0.94 ± 0.07) mm,   The absolute light transmission for various thicknesses can be estimated by integrating the light beam profile G(θ) and rate of transmitted photons L(θ, d) over the full hemisphere assuming azimuthal symmetry. The solid angle of the detector ∆Ω PMT under which all rates have been measured needs to be taken into account to obtain the correct total rates. The total radiances G and L(d) are given by The radiance of the collimated light beam G hitting the PTFE samples and of the transmitted photons L(d) are both measured through the quartz tube under identical conditions. The total transmission probability T θ in =0 (d) for collimated light at perpendicular incidence can be defined as:

Results and Discussion
In this work we have shown that the experimental setup, previously used for reflectance measurements reported in [3][4][5], can also be used to measure the transmission of materials at the wavelength of LXe scintillation light. These measurements were taken using collimated light around 178 nm with 0°incidence angle and a PMT orientation equal to the direction of the transmitted photon for all accessible angles. The transmission coefficient of virgin grade PTFE is measured to λ = (0.89 ± 0.05) mm for samples at about −80°C in vacuum and GXe. This transmission coefficient is expected to depend on the surrounding medium, as well as the temperature of the PTFE samples due to thermal density change. It was found that the measurements for vacuum and GXe are in agreement, as expected from their similar refraction index. The measurements were taken using a collimated beam with a FWHM of about 1 mm illuminating the center of the 6 mm diameter mill-machined recess. The depth of the recess and thus the thickness d of the samples could only be measured over the whole recess diameter. PTFE components are used to encapsulate the inner active volume in LXe detectors from the outer non-active LXe volume, equipped with passive detector components like electric field shaping rings and cables. Scintillation light created outside the active volume, e.g., by α-decays, must be hindered from entering the inner active volume where it could contribute to accidental coincidence signals, or -if T θ in =0 would be too large -lead to leakage of background events into the signal region. Such light signals generated in the outer LXe region can be assumed as VUV scintillation photons isotropically emitted from a point-like source. Therefore, it is straightforward assuming the incident angle distribution to follow a sinusoidal distribution. Incident angles θ > 0°increase the effective path through an infinitely large PTFE wall by 1/cos(θ) reducing the absolute transmitted amount of photons.
According to Equations (3.3) and (3.9) the fraction of light which penetrates from an outside located isotropically emitting light source through a PTFE wall of thickness d is reduced by For a thickness of d = 3 mm and λ = 0.89 mm the integral results in 0.007, yielding: Since PTFE walls of LXe detectors are usually polished to further enhance the absolute reflectance, unlike the PTFE samples used in this work, the absolute reflectance can be assumed to be higher resulting in the product of total transmission L(0)/G and the absorption integral to be lower. In addition, PTFE immersed in LXe will further increases the absolute reflectance to well above 90% [37]. Both effects will reduce the absolute transmission T θ in isotropic (3 mm) further into the low 10 −4 range. Thus, the fraction of the primary α-decay energy (typically 4 − 6 MeV) penetrating into the active detector volume will be below the detector threshold (typically ∼1 keV). Most LXe detectors feature a complex detector design in the outer LXe volume severely reducing the light collection efficiency owed to absorption on other materials and PTFE wall components of partially increased thickness (e.g., for structural purposes). These effects of significantly reducing the amount of transmitted light are neglected in this conservative estimate. Also neglected is the fact that any accidental light signal transmitted through a PTFE wall of finite thickness needs to be paired with a suitable charge signal in dual-phase TPCs such that it can be detected and contribute to the background rate. The XENONnT direct dark matter experiment [24] was designed with the goal of having a PTFE wall as thin as possible for separating the inner active LXe volume from the outer non-active LXe volume in order to reduce the radioactivity originating from the PTFE wall itself while still optically separating the active from the inactive LXe volume. The measurements reported in this work have led to the decision to use PTFE walls of at least 3 mm thickness compared to the 5 mm used on previous XENON detectors [25,27].
Independent measurements of xenon scintillation light transmission through PTFE have been performed recently [38]. These measurements were taken with the PTFE samples being immersed in GXe at room temperature as well as at cryogenic temperatures in LXe. After combining both data sets the authors obtain an about two times smaller transmission coefficient λ, which significantly disagrees to the one presented here. Despite of this discrepancy, our conclusion that a 3 mm thick PTFE wall is sufficiently reducing the intensity of typical background light events from the outer passive volume at the XENONnT experiment is even more valid for the smaller transmission coefficients reported in [38]. The measurements of the work presented here benefit from wellknown and quantifiable systematic uncertainties, the fact that the collimated VUV light passes only once through the PTFE samples, and the ability for angle-resolved measurements. By using a well-motivated model, which is supported by the measured data, the angular information allows us to extrapolate the transmitted light signal into regions which were experimentally not directly accessible in order to ensure that no transmitted light is left unaccounted for in the analysis.