1. The cryogenic neutron EDM experiment at ILL Technical challenges and solutions James Karamath University of Sussex

2. 2 In this talk… (n)EDM motivation Measurement principles and sensitivity Brief (recent) nEDM history The Cryo-EDM experiment  Overview of apparatus  Summary of my DPhil work Summary/conclusions James KaramathUniversity of Sussex27/02/2016 22:33:11

3. 3 (n)EDMs – so? I P- and T-violating CPV in SM not fully understood e.g. insufficient CPV for baryon asymmetry Strong CP problem  θ CP < 10 -10 rad. Axions? James KaramathUniversity of Sussex27/02/2016 22:33:11 n n  p   × S + - d S - + d

4. 4 (n)EDMs – so? II Estimated EDMs model dependent  SM d n ~ 10 -31 ecm  Other models typically 10 5-6 times greater e.g. SUSY:  CP < 10 -2 quark electric dipole moments: qq  gaugino squark

5. 5 nEDM measurement principle B0B0 E = + h/2 = - h/2 h (0) = -2μ.B h (  )= 2(-μ.B+d n.E) h (  )= 2(-μ.B-d n.E) B0B0 B0B0 E d n defined +ve ↑↑ - ↑↓ = Δ = 4d n.E / h Ramsey NMR performed on stored Ultra Cold Neutrons (UCN)

6. 6 Ramsey’s method of separated fields Start with spin polarised neutrons in uniform B-field (B z ) Apply oscillating B-field pulse (B xy ) perpendicular to B z. Precession axis rotates down to xy-plane Apply large E-field and allow to precess freely for ~300s Apply 2 nd, phase coherent with the first, oscillating B xy. Neutron precession axis rotates down to –z axis.

7. 7 Ramsey’s method of separated fields However if an EDM is present a phase difference builds up during the free precession If 180 out of phase second pulse returns spin back to +z axis.

8. 8 Ramsey’s method of separated fields (2n-1)π out of phase Experimental runs taken at approx π/2 off resonance. Here dN/dν is a maximum.

9. 9 nEDM statistical limit Fundamental statistical limit α = visibility [polarisation product] E = E-field strength T = NMR coherence time N = total # counted James KaramathUniversity of Sussex27/02/2016 22:33:11

10. 10 nEDM systematic limit Main concern: changes in B-field accidentally correlated with E-field changes give false d n signal h( ν ↑↑ – ν ↑↓ ) = 2|μ n |(B ↑↑ –B ↑↓ ) – 4d n E True nEDM signal False signal due to varying B 

11. 11 nEDM experiments: history Co-magnetometer era Cryogenic UCN era RT stored UCN eraBeam era ΔB ≈ v x E / c 2 limited

12. 12 RT nEDM experiment at ILL Create UCN, can then be guided & stored Polarise UCN UCN admitted into cell with E and B- fields and stored… Mercury polarised by Hg lamp and added to cell N S Storage cell Magnet & polarizing foil / analysing foil UCN Approx scale 1 m B E Magnetic field coil High voltage lead James KaramathUniversity of Sussex27/02/2016 22:33:11 Magnetic shielding

13. 13 RT nEDM experiment at ILL Ramsey NMR performed Released from cell Neutrons spin analysed (# f n of precession) Mercury spin analysed. Repeat: E=↓or 0, B=↓ N S Magnetic shielding Storage cell UCN detector Approx scale 1 m Magnetic field coil B High voltage lead E Magnet & polarizing foil / analysing foil James KaramathUniversity of Sussex27/02/2016 22:33:11

14. 14 Systematics I Mercury fills cell uniformly, UCN sag under gravity, lower by ~3 mm. Thus don’t sample EXACTLY the same B- field. Axial (z) gradients → problems… Magnetometer problems Hg n z James KaramathUniversity of Sussex27/02/2016 22:33:11

15. 15 Systematics II Two conspiring effects  v x E: motional particle in electric field experiences B-field: ΔB ≈ v x E / c 2  Axial field gradient dB/dz creates radial B-field (since .B=0) proportional to r, B r  r Let’s look at motion of a mercury atom across the storage cell Geometric Phase Effect (GPE)

16. 16 Systematics III Geometric Phase Effect (GPE) dB/dz → B  r B  v x E Scales with E like EDM!!! Scales with dB/dz (GPE Hg ~ 40GPE n ) Resultant i.e. B 0 field into page has gradient Shifts resonance of particle Using Mercury introduces error E and B 0 into page Rotating B field

17. 17 Final result Room temperature experiment gave the result; d n = (+0.6  1.5(stat)  0.8(syst)) x 10 -26 ) ecm i.e.|d n | < 3.0 x 10 -26 ecm (90% CL). New cryogenic experiment will eventually be x100 more sensitive… www.neutronedm.org

18. 18 The Cryogenic nEDM experiment Reminder: RTCryo N /day6x10 6 ~6x10 8 T /s~130~260  0.75~0.9 E /kV/cm~12~25 (B 0 /μT15) ~10 -28 ecm * * with new beamline x20 x5* x2 x1.2 x2

19. 19 Improved production of UCN (↑N) I Crosses at 0.89 nm for free (cold) n. Neutron loses all energy by phonon emission → UCN. Reverse suppressed by Boltzmann factor, He-II is at 0.5K, no 12K phonons. Dispersion curves for He-II and free neutrons James KaramathUniversity of Sussex27/02/2016 22:33:11

20. 20 Improved production of UCN (↑N) II Idea by Pendlebury and Golub in 1970’s, experimentally verified in 2002 (detected in He-II) for cold neutron beam at ILL (~1 UCN/cm 3 /sec). Also better guides – smoother & better neutron holding surfaces, Be / BeO / DLC coated → more neutrons guided/stored. Allows longer T too. James KaramathUniversity of Sussex27/02/2016 22:33:11

21. 21 Polarisation and detection (α) I Polarisation by Si-Fe multi-layer polarizer, 95±6% initial polarisation. Can lose polarisation in 2 ways:  “Wall losses” magnetic impurities in walls, generally not aligned with neutron spin  Gradients in B-field, if not smooth and steady have similar effect James KaramathUniversity of Sussex27/02/2016 22:33:11

22. 22 Polarisation and detection (α) II Detector: solid state, works in 0.5K He-II. n ( 6 Li, α) 3 H reaction - alpha or triton detected Thin, polarised Fe layer - spin analysis James KaramathUniversity of Sussex27/02/2016 22:33:11

23. 23 Magnetic field issues I Target – need ~ 100 fT stability (NMR) Need ~ 1 nT/m spatial homogeneity (GPE) Perturbations ~ 0.1 μT (cranes!) Need (axial) shielding factor ~ 10 6 MMu-metal shielding~ 50 SSuperconducting shielding~ 8x10 5 AActive shielding (feedback coils)~ 15 Shielding factors

24. 24 Magnetic field issues II CRYOGENIC nEDM! Utilise superconducting shield and B 0 solenoid. MMajor part of fluctuations across whole chamber (common mode variations) MMagnetometer (zero E-field) cell(s) see same VVery stable B 0 (t) current Holding field x5 to reduce GPE of the neutrons by factor of 25 (GPE n  1/B 0 2 ) Extra benefits James KaramathUniversity of Sussex27/02/2016 22:33:11

25. 25 Magnetic field issues III ~fT sensitivity 12 pickup loops will sit behind grounded electrodes. Will show temporal stability of B-field at this level. Additional sensitivity from zero-field cell(s) SQUIDS

26. 26 James KaramathUniversity of Sussex27/02/2016 22:33:11 Now have a 400 kV supply to connect to HV electrode. Will sit in 3bar SF 6. For 160 kV use N 2 :CO 2 first. Improving the E-field (↑E) I: The HV

27. 27 Improving the E-field (↑E) II: HV line 1 50 kV ~1 GOhm resistors Superfluid containment vessel (SCV) HV electrode Ground electrodes 400 kV bipolar stack N.B. Diamond-like-carbon coated titanium electrodes BeO spacers

28. 28 Improving the E-field (↑E) II: HV line 2 Spellman +130 kV Spellman -130 kV Thick walled PTFE tube and thin- walled SS tube HV “cryo-cable”. Standard 150 kV cable HV connection

29. 29 The dielectric strength of LHe Has been tested in the past, mostly at 4.2 K (760 torr), at small electrode gaps (sub- mm) and with small electrodes. Superfluid data is limited and generally at low voltages (sub-40 kV, often sub-20 kV). Usually the breakdown strength as a function of gap is studied. We’d like to know the strength as the pressure/temperature falls – esp. in the superfluid state. James KaramathUniversity of Sussex27/02/2016 22:33:11

30. 30 The dielectric strength of LHe II Past literature He-I data 4.2 < T(K) < 2.2 Nope – put in the final versions from thesis

31. 31 The dielectric strength of LHe III Past literature He-II data 2.2 < T(K) < 1.4

32. 32 The dielectric strength of LHe IV Test electrodes submerged in He-II in bath cryostat. Studying V bd and I leak as function of d, T, dielectric spacers, purity… up to 130 kV. Also electrode damage. E ±HV cryostat He-II (T, purity…) gap (d, V, spacers) Sussex HV tests

33. 33 The dielectric strength of LHe IV Sussex HV results

34. 34 The dielectric strength of LHe V

35. 35 The dielectric strength of LHe VI Statistics

36. 36 The dielectric strength of LHe VII Size effects: Weak but important dependence on electrode area or stressed fluid volume may decrease dielectric strength. Leakage currents never found to be >0.1 nA (sensitivity limited) even immediately below breakdown. ~0.3 mm craters in electrodes when breakdown occurs at >80 kV. Bad news for DLC coated electrodes.

37. 37 The dielectric strength of LHe VI Breakdown strength reduced by insulating BeO spacers by a factor of ~1.4. Due to surface tracking along the BeO.

38. 38 The dielectric strength of LHe summary At 0.7 cm gap the breakdown field strength was approx 80 ± 10 kV/cm. i.e. ~50 kV/cm for 1 in 1000 chance of breakdown. May have to half this if size effects indeed exist. What controls breakdown – pressure or temperature?! May hold key to improving V bd.

39. 39 And so, the CryoEDM experiment I n guide tubes + spin analyser E ~ 25kV/cm E = 0kV/cm Spin flipper coil (measure other spin)

40. 40 And so, the CryoEDM experiment II HV electrode Ground electrodes HV in z Carbon fibre support BeO spacers

41. 41 And so, the CryoEDM experiment III HV electrode Ground electrodes G10 Superfluid containment vessel HV in z Neutrons in/out Guides not shown 250l He-II 0.5K * * * BeO spacers/guides

42. 42 And so, the CryoEDM experiment IV 1m Dynamic shielding coils Magnetic (mu- metal) shields Superconducting shield and solenoid The shielded region

43. 43 Schedule / Future Finish construction THIS YEAR Start data taking THIS YEAR First results ~2009 Upgrade neutron guide to ↑N ~2009 ? James KaramathUniversity of Sussex27/02/2016 22:33:11

44. 44 Summary (n)EDMs help study T-violation and are constraining new physics. Final RT result: |d n | < 3.0 x 10 -26 ecm. Aim to push well into 10 -28 ecm. Further work needed to understand dielectric properties of He-II. Only 20 kV/cm? (Paper in preparation.) Can pressure/purity/electrode material make a difference?

45. 45 Done! Thanks for listening