NA62 Liquid Krypton Purity Monitor

A system for determining the purity of liquid krypton employed in the NA62 rare kaon decay experiment at CERN was developed based on the use of a time projection chamber. The attenuation of drifting ionization electrons from absorption of 511 keV gamma rays in liquid krypton was measured to estimate the purity. The setup was tested with krypton purified from commercial sources.


I. INTRODUCTION
The liquid krypton (LKr) ionization calorimeter (LKRC) plays a key role in the photon veto and particle identification systems for rare kaon decay measurements done at the CERN experiment NA62 [1].Detailed descriptions of the LKRC can be found in [2,3].The active volume is longitudinally segmented by 127 cm-long beryllium-copper ribbons which form an octagonal grid of 13248 2 × 2 cm 2 individual towers facing the beam direction.The cylindrical cryo-vessel contains about 9 m 3 of LKr.Boil-off gas passes through filters before being re-condensed in the LKRC by an argon cooler, ensuring continuous recirculation.
Due to the transitory occurrence of small leaks in the LKRC system during the period 2016-17, it became necessary to replenish the Kr supply using commercially obtained gas which is typically available with purity at the ppm level.However, for operation in the LKRC purification of the Kr gas at the ppb level is required and measurement of the electron lifetime is desirable before it is transferred to the LKRC.
The principal electro-negative impurities in such systems are usually oxygen and water.During the development and commissioning phase of the LKRC, the maximum electron drift time was measured under normal operating conditions (with 3 kV applied to the anodes) to be 3 µs [4].The electron lifetime was found be >300 µs [3] which is sufficient for NA62 operations.
To determine the ionization electron lifetime in the LKRC and in LKr purified from commercially sourced Kr, a new purity monitoring system was developed.The system is based on the use of a small (32 cm 3 ) time projection chamber (TPC) installed in a cryogenic vessel containing about 2 l of LKr.The device allows measurement of the attenuation of ionization charge as a function of the drift time and applied electric field, and, therefore, can be used to determine the electron lifetime in LKr.
This report describes the cryogenic and purity measurement systems, and the purity measurements made with gas sampled from the LKRC and with newly purified Kr.The following sections cover the cryogenic system (II), the detection apparatus (III), and the measurements performed (IV).

II. PURIFICATION AND CRYOGENIC SYSTEMS
The cryogenic setup and the procedures used to purify and condense LKr are described in [5].Briefly, filtered Kr gas was condensed inside the LKr vessel with a heat exchanger (HEX) linked to a cryocooler.The filters used to purify commercially supplied Kr gas were the same as in the LKRC recirculation system: two Sertronic type N gas purifiers1 in series.In the filters, the feed gas passes through a special catalyst which traps oxygen, while water and carbon dioxide are adsorbed on a molecular sieve bed.The principal impurities removed are O 2 , CO, CO 2 , and H 2 O.A single stage coaxial pulse tube cryocooler was used as a cooling source to liquefy Kr.The cooling power of the cryocooler (with a 3 kW compressor) at 120 K is approximately 55 W. The temperature of the cold head was controlled at 119 K during condensation and then at 119.8 K during stable operation.Figure 1 shows an assembly drawing of the vacuum insulated LKr vessel with the TPC, HEX, and the radiation entrance window.The inner LKr vessel cryostat is covered in 20 layers of multi-layer insulation.Prior to measurements described in Sec.IV, the LKr vessel was baked at 60 C for 3 weeks by running a current through a temporarily applied heating tape 2 and pumping vacuum to a level p < 2 × 10 −7 mbar.The TPC detector system, previously used in the measurements on liquid xenon [6], was modified for the present measurements of electron lifetime in LKr.As discussed below, avalanche photodiodes (APD) were replaced by Multi-Pixel Photon Counters (MPPC) to provide triggering for the ionization drift time measurement and preamplifiers were installed on the anodes.As shown in Figure 2, the TPC features ground potential anodes, a shielding grid (25 µm dia.wires, spaced 3 mm apart) separated from the anodes by 3 mm, a field cage, and a negatively biased cathode plane with a 3 x 3 x 3.6 cm3 drift volume.Charge was collected on a central 1 cm dia.electrode (A1) or on a 3 x 3 cm 2 outer electrode (A2).An electric drift field up to 1.5 kV/cm could be applied between the cathode and the shielding grid separated by 3.6 cm; the positive electric field between the anodes and grid was maintained at twice the drift field value.The field cage consisted of nine wires with a spacing of 3 mm strung around the four walls of the chamber; the voltage was distributed by 100 MΩ resistors.Taking into account the energy needed to produce an electron-ion pair W = 20.5 ± 1.5 eV [8], and assuming charge collection efficiency 80 % of the saturation value at 0.8 kV/cm, and 80 % grid transparency, an A1 anode signal of about 2.5 fC was expected for a 511 keV photon undergoing absorption by photo-electron emission.The anode signals were read out with adjacent N-Channel JFET (BF862) amplifiers situated in the LKr (see Figure 5) and connected by coaxial cables to room temperature post-amplifiers located outside the cryostat.
The readout system is outlined in Figure 6.The signal from the NaI(Tl) detector was split to allow high and low level discriminators used to select the 511 keV photo-peak events.The signals from the NaI(Tl) detector, MPPCs, and anodes were read out using a CAEN FIG. 5 -Schematic of the cold preamps (see text).DT5725 14-bit 500 MS/s FADC.The FADC was connected through a USB port to a laptop computer running the MIDAS [9] data acquisition system.

IV. MEASUREMENTS
Initial measurements were performed at the CERN Cryolab and, after qualification, the apparatus was transported and reassembled adjacent to the LKRC in the NA62 experimental area.Prior to measurements, the LKr vessel and connecting piping were baked-out as mentioned in Sec.II and purged to remove any residual gas; then, the vacuum insulation space between the outer and inner regions was evacuated.
Measurements were made with boil-off gas from the LKRC with no additional filtering.Figure 7 shows the charge distribution on the TPC anode A1 triggered by coincidence of the MPPC and NaI(Tl) signals for the full 16 µs drift time window at a drift field of 0.83 kV/cm.A Gaussian fit to the 511 keV photoelectron peak gave a resolution σ = 7.3 %.
The average light signal varies in amplitude due to the solid angle acceptance of the MPPCs which is maximum at the center of the TPC.After selecting the 511 keV charge peak, the observed energy resolution based on the light signal at the central 1 µs drift time slice was σ = 13 %.Since the light and charge signals are anti-correlated in LKr [10], the charge signal increases and the light signal decreases with increasing electric field.Combining the signals as discussed in [11] resulted in improved resolution of σ = 4.3 % as indicated in Figure 8.The correlation angle in the charge vs. light plot was 18 • .
To estimate the electron lifetime, the drift time was segmented into 16 1 µs regions and the amplitudes of the 511 keV peaks were determined by fitting the peak regions with a Gaussian in the presence of functions representing the Compton edge and background.Figure 9 shows the amplitude of the 511 keV photopeak as a function of drift time for data taken with Kr gas taken directly from the LKRC.The attenuation was estimated to be ∆A/A(0) = (−6.6 ± 5.8) × 10 −4 /cm corresponding to a lower limit on the electron lifetime τ e > 2.7 ms (90 % c.l. [12]) using the observed drift velocity v d = 2.3 mm/µs at electric field E = 0.83 kV/cm.Using the average of the atomic electron attachment cross sections available for LAr and LXe compiled by Doke [13] at this electric field, we estimated an equivalent O 2 (or H 2 O) contamination of < 0.1 ppb for the system measuring gas from LKRC.
In other measurements, commercially obtained Kr gas 4 was passed once through the Figure 10 shows the amplitude of the 511 keV photoelectron peak as a function of drift time for data taken at electric field E = 0.83 kV/cm with condensed Kr from commercially supplied Kr gas passed through the filters; the limit obtained on the electron lifetime was τ e > 1.9 ms (90 % c.l.) and the equivalent O 2 (or H 2 O) contamination was estimated to be ppm, O 2 ≤ 0.5 ppm, HC ≤ 0.5 ppm, N 2 ≤ 2 ppm, Ar ≤ 1 ppm, CF 4 ≤ 1 ppm, Xe ≤ 1 ppm, CO+CO 2 ≤ 1 ppm, H 2 ≤ 1 ppm. 5The detection threshold was estimated to be 100 keV to be compared with the 85 Kr beta decay endpoint energy of 687 keV.< 0.2 ppb.Since the attenuation is reduced further at higher fields, it is clear that these levels of impurity contamination have a negligible effect on the operation of the LKRC at its nominal field of E = 3 kV/cm.

V. CONCLUSION
A system for measuring the ionization electron lifetime, and consequently, the purity of LKr was developed.The attenuation of drifting ionization was measured in a small-time projection chamber triggered by scintillation light detected by MPPCs in coincidence with signals from a NaI(Tl) crystal observing back-to-back 511 keV annihilation photons from a 22 Na source.Electron lifetimes 2 ms were observed in gas samples from the NA62 LKr calorimeter and from commercially supplied Kr gas bottles after filtering, more than satisfying the requirements for operation of the NA62 experiment.with the setup and Dirk Mergelkuhl for assistance with alignment.We are also greatful to the CERN Geodetic Metrology Group and the CERN Vacuum Group for their assistance.This work was supported by TRIUMF and NSERC (Canada) grant SAPPJ-2018-0017.

FIG. 2 -
FIG. 2 -Schematic of the TPC.Left: End view showing the central (A1) and outer (A2) anodes and locations of the MPPC light sensors.Right: Side view showing the drift region with anodes, grid, cathode, and field cage wires; 511 keV gamma rays from annihilation of positrons emitted by a 22 Na source enter through the cathode from the left.

FIG. 6 -
FIG.6-Overview of the readout and data acquisition systems.The amplified MPPC and NaI(Tl) signals are passed through NIM-based logic units used to construct a trigger signal and fed to the ADC for digitization.

FIG. 7 -FIG. 8 -
FIG.7-Energy distribution from the charge measurement for all drift times at electric field E = 0.83 kV/cm; a Gaussian fit to the 511 keV photoelectron peak is shown (see text).

FIG. 9 -
FIG.9-Charge signal photo-peak amplitude (keV) vs. drift time (channels) in 1 µs slices at electric field E = 0.83 kV/cm for condensed Kr gas from the LKRC.Each channel is 4 ns.

FIG. 10 -
FIG.10-Charge signal photo-peak amplitude (keV) vs. drift time (channels) in 1 µs slices at electric field E = 0.83 kV/cm for condensed filtered Kr gas from bottles.Each channel is 4 ns.