Radio Detection of UHE Neutrinos in the Ice at South Pole

1. Radio Detection of UHE Neutrinos in the Ice at South Pole The ARA Experience and New Ideas Kael HANSON Wisconsin IceCube Particle Astrophysics Center and Department of Physics University of Wisconsin – Madison ARENA Workshop – June 7-10, 2016 Groningen, Netherlands

2. ARA Collaboration The University of Chicago Weizmann Institute of Science University College London ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

3. Credit Where Credit is Due • Ph.D. thesis work of Dr. Thomas Meures (ULB – now WIPAC) • http://arxiv.org/abs/1507.08991 • Phys. Rev. D 93 082003 (2016) • Other parallel efforts in ARA will be described later in this session by Ming Yuan Lu and Dr. Carl G. Pfender; • End-to-end calibration of ARA antennas / analog FE: Dr. K. Mase • See also ideas for phased array techniques in ARA – Dr. S. Wissel’s talk on Friday; • Retreat from presentation of far future ideas – these will be covered by Dr. Krijn de Vries: • Poster on cosmic ray detection with sub-surface RF arrays; • Talk later today on RADAR detection of neutrinos ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

4. Outline of Talk • Super brief background • UHE neutrinos from GZK interactions of cosmic ray protons • Askaryan radiation in the ice • ARA description (static since 2013); • Detail on first search using 10 months exposure during 2013 with 2 of 3 ARA stations in operation at the time; • Brief update on current operational status of ARA; • Discussion of 2017/2018 plans to deploy more stations. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

5. UHE Neutrinos – How and Why? The GZK mechanism A resonant interaction between UHECR and the CMB: • GZK neutrinos “guaranteed” to be there but could nevertheless remain out of reach of even planned facilities. • Neutrinos remain the only known particle capable of imaging the UHE sky. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

6. Detection of UHE Neutrinos in Ice • Via the Askaryan effect: • An excess negative charge (~20%) built up in neutrino induced cascades through: • Compton scattering • Other ionizing effects •  Moving current, emits electromagnetic radiation •  Coherent for radio wavelength ν • The advantages of RF detection schemes: •  RF attenuation length in ice ~ 1 km • Observe big detector volume with few sensors • Very cost effective means to instruments 100’s of Gton of target. ~15cm 56 deg ~15 m

7. The ARA Detector Drilling, Deployment, and Detector Hardware ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

8. The Askaryan Radio Array 3 of 37 planned currently deployed plus ARA TestBedused to evaluate the EMI env at Pole and ended up producing good scientific results (CGP talk today). • One station: • Measurement system: • 4 holes, 20m spacing • Deployed at depth of 180 m • 16 antennas, 150MHz – 850MHz(8 horizontally polarized., 8 vertically pol.) • Calibration system: 4 pulsing antennas • Each station is an autonomous detector! During 2013 season ARA-1 was not operational – hence this analysis only covers ARA-2 and 3 data. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

9. Shallow Ice Drilling • ARA 2012-13 Hot Water Drill delivered 200 m, 15 cm dry holes for deployment of ARA2, ARA3 instrumentation: • 0.6 – 1.0 m/min drill speed  5h drill time to 200 meters; • Improvement over previous (tough) season include closed loop water system  100% of heat generation went down hole instead of melting snow to make hot water. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

10. Shallow Ice Drilling ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

11. Instrument Deployment ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

12. The ARA Signal Chain and DAQ Each string has 2 Vpol and 2 Hpol antennas with local LNA. Long-distance (200 m) RF signal transport via fiber-optic translator. Triggering via tunnel diode power estimators + fast pulse discriminators with tunable threshold. Digitization performed by IRS2 (G. Varner, U. Hawaii). 8-ch SCA with 32k/ch analog buffer depth, divided into 512 64-sample blocks each randomly accessible. Analog sampling rates up to 4 GSPS possible – ARA configured to 3.2 GSPS (20 ns per 64 sample block). In principle ZeroDT capture for trigger rates ~ kHz. Early version of ASIC has some noise problems which prevented operation in this mode. Low power consumption ~ 20 mW/ch. • Surface electronics retrievable in shallow vault. Connected to ICL by Cu power line and optical fiber with media converters for networking. • Commercial COMExpress SBC on custom logic backplane (OSU). • Power consumption approx. 100 W per station. 180 m Sample jitter of 100’s ps and and severe non-linear amplitude response requires careful calibration. Time resolution of ~95 ps achieved on pulses from nearby calibration pulsers. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

13. Data Analysis Steps • Begin with 150 million triggers in 10 months data • Wait until next year because ARA only gets 1 GB/day of satellite transfer (1% of IceCube) so tapes, now disks, have to be physically transported to Madison and placed online. • Expectation: • 0.2 GZK/BZ neutrinos • 1000 impulsive RF events – non-thermal • Rest are thermal noise triggers • Step #1 – eliminate thermal noise • Step #2 – reconstruct emission vertex ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

14. Thermal Noise Filter • Fast and powerful noise rejection: from 1E18 – 1E19 92% signal retention and 99.9% background rejection. • Algorithm initially conceived to run in firmware but was applied in software offline. • Plane wave approximation – antennas with similar relative geometries will have similar speeds: • Histogram speeds in 5 different relative geom categories • Plane waves will exhibit peaks • Thermal noise evenly distributed ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

15. Vertex Reconstruction – Matrix Method Point emitter in ice produces spherically expanding wave (ray bending optics ignored) An expanding spherical front: May be linearized by considering pairs of observations and subtracting: ν Determine time difference by cross-correlation Linearization requires 5 observations to constrain solution (viz. 4 if you are willing to solve NL equations). Accommodates overconstrained system of linear equations  LLS or SVD fast matrix techniques exist for solution. Residual turns out to be good quality parameter Cross-correlation vs delay – make quality cut on max amplitude to reject weak pulses. Cal pulser waveforms shifted by time delay determined by x-corralgo. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

16. Vertex Reconstruction Performance Simulated Signal In-situ Calibration Pulsers ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

17. Final Cuts Only three cuts used to get to neutrino level: Thermal noise QP > 0.6 Log10 (residual) < -4 Angular cuts to remove surface noise / pulsers After evaluating cuts on 10% burn sample and obtaining OK to proceed with full 10 m evaluation, looked in the box and found no events which passed all cuts  neutrino limit Data ARA2 Simulated Signal ARA3 ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

18. Analysis Effective Area & UHE Neutrino Limit ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

19. LC-130 Cross-Check Instead of using drones ARA profited from opportunity to record transmissions of LC-130 departing from NPX. Compare angular reconstruction of ARA-2 alone with parallax reconstruction using combined signals in ARA-2 and ARA-3 stations. This required first aligning clocks which are normally not synchronized. Ground track of LC-130 Comparison of single station vs. multiple station reconstruction. The green band is error in parallax reconstruction. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

20. Current ARA Status • Fall 2015: replace hi-speed USB link between ARA logic PCB (ATRI) and SBC developed  failure of this link was mechanism which prevented ARA-1 from data taking in 2013. Other stations eventually manifested. • Firmware-only fix – data path is now over (higher bandwidth) PCIe link overcoming existing trigger rate bottleneck and improving the deadtime (already at percent level). • Deployed firmware Dec 2015: all 3 ARA stations taking data. • Improvements in operational procedures 2014+ much better at exploiting live time of detector. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

21. Future ARA Plans • Early 2016 - US NSF released funding on pending award to supplement the ARA array with at least two more stations. • 2017 / 2018 polar season will feature deployment of two, possibly three, additional ARA stations. • Possible changes/improvements for ARA in upcoming seasons: • Replace ARA Hot Water Drill with RAM drill – fast (20 holes per day) but 100 m max depth, smaller diameter holes, 10 cm; • Optimize station geometry – currently we have no distance to vertex so energy reconstruction is out window. Could be improved dramatically by enlarging station string spacing … preliminary analyses indicate no significant loss of trigger efficiency; • New digitizers – plug-in replacement for IRS2 being developed by M. Orbach (Weizmann Inst) based on PSI’s DRS4. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

22. You too can share in the ARA experience at a somewhat local establishment. Thank you

23. Backups ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

24. Systematic Uncertanties ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

25. Seasonal Variation of UHE Neutrinos EMI levels as a function of time (i.e. station open vs closed) are clearly visible. Most events which pass thermal noise and reconstruction quality cuts come from Austral Summer when activity levels are high (planes, snowmobiles, hand-held radios, …) Note however that angular cut still performs well – no events from ice volume during this period. Calibration pulsers spontaneously fired during winter season – these event are also easily rejcted by vertex angular cuts. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

26. Deep Pulser Reconstruction ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

27. Do we need to digitize at 3.2 GSPS? Shown at right (solid blue) is average energy vs frequency taken from UHE neutrino simulation for Vpol antennas. Due to nature of coherent Askaryan radiation, unless you are near the Cherenkov angle, signal energy tends to lie in the low-100’s of MHz. We could probably live with ~ 1 GHz sampler. ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

28. IRS2 – Ice Radio Sampler ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

29. IRS2 Calibration ARENA 2016 - Groningen, NL | Kael HANSON UW Madison

30. Gain Calibration ARENA 2016 - Groningen, NL | Kael HANSON UW Madison