1. Second run with the Copper filter WA104 technical meeting – 23/03/2016

2. Summary first run WA104 Technical Meeting2 With the copper-based filter operated in liquid phase during filling and in gas phase during Argon recirculation (1 full volume/4-5 days), a purity higher than 10 ms (above sensitivity of our detector) was reached in a short time. The 50 l LAr-TPC was operated for several days at max purity, then O 2 was injected in steps for a total of ~16 g of O 2, over the whole run. Purity started getting worse after injection of ~1.5 g of O 2 but with a very slow rate of change: still 10 us electron lifetime after injection of 16 g of O 2, suggesting that the filter trapping/saturation mechanism is not completely understood. It is also possible that other impurities are present in the detector that are not trapped by the filter. Fitting the slow component  s of the LAr scintillation light, yielded a value around 0.8  s ( to be compared with  s > 1.2  s prior to oxygen injection). Before the end of test the amount of N 2 and H 2 O in the recirculating gas was measured with argon gas analyzer: 16.6 ppm and < 0.5 ppm, respectively. These are partially compatible with the measured slow light component and electron lifetime and tend to confirm the presence of additional impurities in the LAr-TPC not trapped by the molecular sieve and the copper.

3. Filter composition: - Molecular Sieve (Type 4A 1.6-2.5 mm) —> 2.46 kg - Copper (BASF Cu-0226 S 14x28 mesh or Q-5)—> 2.96 kg (  = 0.85 kg/dm 3 ) Copper fills 45 cm of the filter Adsorption capacity, as estimated at FNAL : 0.5 g O2 /kg in LAr, 5 g O2 /kg in GAr 3WA104 Technical Meeting

4. 4 FLIC Cryo-set up - reminder New filter with copper bed installed BEFORE recondenser

5. Overall purity plot WA104 Technical Meeting5

6. Further measurements WA104 Technical Meeting6 A new run was performed in late December, taking data for few days with the filter left as in previous test: no tracks could be collected during this run, which seems to confirm that the Copper filter was saturated; the measurement of the slow LAr light component was compatible with no contamination by other impurities, as it had been in the previous run, before starting Oxygen injection. In January 2016 the filter was sent back to CERN cryolab and is was regenerated with the following prescriptions: Argon Flow at high Temperature (> 200°) to activate molecular sieve, followed by Ar(98%)+H 2 (  2%) flow for Copper activation. Water content was continuously recorded to get an estimate of the trapped impurities level both in Mol. Sieve and Copper. Far less than 15 g of water was actually extracted during the purification phase. A new run with the LAr-TPC followed. A new bottle of Ar60+O 2 55 (1% molar), with much lower content of impurities other than Oxygen, was used for filter saturation (O 2 55 -> H 2 0 ≤ 0.5 ppm-mol; CO,CO 2,H 2 ≤ 0.1 ppm-mol).

7. Second run A new run was performed in the past February, after the filter was regenerated by the Cryolab group at CERN. The whole system was kept identical to previous run (cryo plant, chamber), as well as the procedure: After filling, initial life-time  30  s; we recirculated until overcoming 3 ms (about one week), then started Oxygen injection. The only difference with respect to the past run is related to the much purer gas bottle used. We forced fast recirculation, one volume in 2 days, by increasing heat input to the chamber, to fast achieve high purity. Ar-O 2 mixture was then injected at 0.265 g O2 /h, for a total of 17.5 g O2. WA104 Technical Meeting7

8. 8 Purity during O 2 injection phase

9. Comments WA104 Technical Meeting9 This time we observed purity increase always compatible with the recirculation time, for O 2 injection up to 13.5 g. Then, an abrupt decrease of the purity follows, with “vanishing tracks”. This happens in a few hours, and it is confirmed by the change in the slow component of the LAr scintillation light, cut in half after the purity drop. This behaviour is well compatible with Oxysorb filters and what we expected from a correctly activated cartridge. As well, the amount of Oxygen needed to saturate the Copper filter is around 15 g, compatible with the previous run and data shared by American colleagues. Such results strengthen the hypothesis that the odd outcome of the past run may be due to low grade of the Oxygen used, implying a large amount of unknown impurities injected along with O 2.

10. Saturation Region WA104 Technical Meeting10 Purity almost goes to “zero” in around 6 hours. At 0.265 g/h, this means (roughly) more than 1.5 g of O 2 -> 10% of the full capacity. This translates in 10% length over which the filter looses 100% efficiency (interface) -> over the length of the filter this means (rough estimate) 4.5 cm. Oxysorb interface is smaller (purity decrease sharper), probably due to larger size of allumina pellets in the copper filter. d d = f(mean free path) = f’(pellet size) Full portion

11. Vanishing track example WA104 Technical Meeting11 COLLECTION INDUCTION WIRES * * = 280 t-sample, or 112  s Run 3024, ev. 2200 Around 2-3 hours after purity starts dropping.

12. Summary and next steps WA104 Technical Meeting12 A new run has been performed with the copper filter and the ICARUS 50l chamber: we verified saturation in gas is achieved with  15 g of O 2. High-purity oxygen was injected and the filter behaviour reflected common knowledge: 100% efficiency up to saturation, followed by an abrupt decrease in purity (6 h at 0.265 g O2 /h). A new test will be put in place, to characterise purification in liquid phase. Simpler circuit, no use of the 50l TPC. Immersed LAr pump and filter, connected to the Ar-O 2 bottle and to long exhaust at the output (to gasify the liquid and avoid air back-diffusion). In series to the copper filter, we will place a small, glass-transparent cartridge of Oxysorb, which will change colour as soon as it start operating: this will tell us exactly when the copper filter is full. A scheme of the full system is in development.