1. Far-field Monitoring of Rogue Nuclear Activity with an Array of Antineutrino Detectors Neutrino Sciences 2005 University of Hawaii, Manoa December 14-16, 2005 Eugene H. Guillian University of Hawaii, Manoa Neutrino Sciences 2005 University of Hawaii, Manoa December 14-16, 2005 Eugene H. Guillian University of Hawaii, Manoa

2. Rogue Nuclear Activity Fission Reactor Fission Bomb Detonation Purpose Produce Weapons-Grade Plutonium (93.8% 239 Pu) Make sure a given bomb design works (i.e. it explodes, not just fizzling) Expected Size < ≈100 MW th ≈ 1 kiloton TNT Typical commercial reactor ≈ 2500 MW th Little Boy ≈ 15 kiloton Fat Man ≈ 23 kiloton Hard to Detect! Small!

3. Far-Field Monitoring Uncooperative Regime Access within ~100 km not allowed Far Away Small event rate (1 / distance 2 ) Large Detector Shielding from cosmic rays Detector must be affordable These factors strongly constrain the detector & array specifications

4. Antineutrino Detector Module Specification Far-field monitoring Distance > ≈ 100 km Detector must be on the order of 10 6 m 3 100 m Water is the only economically feasible target H 2 O loaded with 0.2% GdCl 3 C.f. GADZOOKS! (Super-K with GdCl 3 ) J. F. Beacom & M. R, Vagins, Phys. Rev. Lett. 93, 171101 (2004)

5. Detection Mechanism Inverse Beta Decay Delayed Event ≈ 20µs n + Gd  Gd +  cascade E vis ≈ 3~8 MeV Prompt Event Cherenkov radiation

6. Neutrino Energy Spectrum GADZOOKS! Threshold E > 3.8 MeV KamLAND Threshold E > 3.4 MeV GADZOOKS! Efficiency 58% of entire spectrum (E > 1.8 MeV) 82% of KamLAND efficiency

7. A Very Basic Look at a Megaton Detector Module 100 m Photo-Sensor Requirement ≈ 120,000 units (10  Super-Kamiokande) Gadolinium 2000 metric tons Water Purification 200  Super-Kamiokande’s capacity ~$120 Million @ $1000 per unit ~$10 Million @ $3 / kg Cost? The cost of just one module looks to be easily about $500 Million!

8. Is a Megaton Module Outlandish? The linear dimensions are not that much larger than those of Super-Kamiokande Challenges Deep-Ocean environment Remote operations Mega-structure engineering

9. Cosmic Ray Background Like bullets! Occasionally they destroy atomic nuclei Unstable nucleiSometimes indistinguishable from antineutrinos!

10. Shielding from Cosmic Rays Super-Kamiokande Shielded by 1000 m of rock (equivalent to 2700 m of water) Mitsui Mining Co. property Super-Kamoikande (and similar experiments) would have cost too much if shielding had to be erected from scratch! For the megaton module array, we assume that cost of shielding on land is prohibitive. Ocean & Lake = Affordable Shielding

11. Array Configurations Global Monitoring RegimeRegional Monitoring Regime Want sensitivity to anywhere on EarthWant sensitivity to a well-defined region Can’t optimize module positioning Module positions can be optimized because of prior knowledge of likely locations Larger Modules Required 10 Megatons 1 year exposure Smaller Modules Will Do 1 Megatons 1 year exposure

12. Global Array 1 5º  5º Array Total of 1596 modules

13. Global Array 2 Equidistant Array Total of 623 modules Minimum nearest- neighbor distance ≈ 600 km

14. Global Array 3 Coast-hugging Array Total of 1482 modules Minimum nearest- neighbor distance ≈ 100 km Modules removed from coast line by ≈ 100 km

15. Regional Array North Korea Choose locations based on sensitivity map (red dots are candidate module positions) 250 MW th fission reactor deep inside of North Korea Background from commercial nuclear reactors

16. Rogue Activity Detection Strategy (1) Assume that no rogue activity is taking place (2) If this assumption is incorrect AND if the rogue activity is sufficiently large, there would be a discrepancy between observation & expectation (3) Use a statistical technique (minimum log-likelihood) to estimate the position & power of the rogue activity

17. Illustration of the Detection Strategy If no rogue activity takes place, module 1, 2, & 3 detects B 1, B 2, and B 3 events With rogue activity, module 1, 2, and 3 sees an extra S 1, S 2, and S 3 events The size of the excess goes as: Power / Distance 2

18. Seeing the Rogue Activity Above Random Fluctuations Observed Number of Events Background only Observed Number of Events Small Signal + Background Random Statistical Fluctuation Large Signal + Background

19. B = # background events S = # signal events Signal Strength = statistical uncertainty Signal Strength S S

20. Map of Signal Strength Rogue Activity 2000 MW th

21. Equidistant Detector Array Configuration 10 Megaton per module 1 year exposure

22. Detectors with Signal Strength > 3

23. Detectors with Signal Strength > 2

24. Detectors with Signal Strength > 1

25. Remarks on Rogue Activity Detection Rogue Activity: 1.Has sufficiently large power 2.Is sufficiently close to detector modules Cluster of nearby detector modules with significant excess Pin-Pointing Rogue activity location given roughly by the position of the cluster Cutting on Signal Strength Tight cut  low background noise, but loss of signal Loose cut  more signal, but more background noise Measuring Power Use log-likelihood to obtain the most likely power

26. P 99 : Benchmark for Array Performance Log-likelihood Function A statistical tool used for hypothesis testing Hypothesis No rogue activity is taking place Information Used in Log- likelihood Function Expected number of background events in each detector (from commercial nuclear reactors) Observed number of events in each detector The log-likelihood value is not defined a priori because of random fluctuations in the measurement Its distribution, however, is defined a priori

27. 99% of measurements give log-likelihood above the alarm threshold 1% of measurements sets off false alarm 1% False Positive

28. Rogue Reactor Exists  Hypothesis Incorrect  Log-likelihood function is biased to lower values Rogue Reactor Power is Weak  Large overlap between observed vs. expected distributions  Can’t reliably detect rogue activity Unacceptably Large Frequency of False Negatives

29. Definition of P 99 P 99 = Rogue reactor power which gives 1% chance of false negative

30. Global Array Performance For each array configuration, make a map of P 99 Procedure for making map: 1.Vary the rogue reactor position 2.At each location, determine P 99

31. P 99 Map: 5º  5º MW th

32. P 99 Map: Equidistant Scaled to 1596 Modules MW th

33. P 99 Map: Coast-hugging Scaled to 1596 Modules MW th

34. 5º  5º Equidistant Coast-Hugging P 99 Summary In Water< 100 MW th W/in several 100 km of coast Several 100 MW th Deep in continent Up to 2000 MW th

35. Regional Monitoring Example: A rogue reactor in North Korea Signal Background Signal Strength About the Plots Signal Rogue power = 250 MW th Detector mass = 1 Megaton Exposure = 1 year Background Commercial nuclear reactors 1 Megaton 1 year

36. Detector Locations 23 candidate locations based on map of sensitivity

37. Performance of Various Array Configurations Consider configurations with 2, 3, and 4 detector modules For each configuration, determine: P 99 Estimated area that contains rogue reactor

38. Two Modules 95% Confidence 99% Confidence P 99 = 250 MW th

39. Two Modules 95% Confidence 99% Confidence P 99 = 120 MW th

40. Three Modules 95% Confidence 99% Confidence P 99 = 626 MW th

41. Four Modules 95% Confidence 99% Confidence P 99 = 336 MW th

42. Four Modules 95% Confidence 99% Confidence P 99 = 502 MW th

43. What if a Georeactor Exists? The Georeactor Hypothesis: Unorthodox, but surprising things can happen…. If it does exist, its power is likely to be 1~10 TW th Total commercial nuclear activity ≈ 1 TW th If a terawatt-level georeactor does exist, the background level for rogue activity monitoring increases significantly!

44. log 10 Background No Georeactor log 10 Background 3 TW th Georeactor Ratio 3 TW th / No Georeactor

45. Squeezing More Information from the Data

46. Fission Bomb Monitoring

47. Conclusion Global Monitoring ≈ 1000  10 Megaton modules  10 Gigaton-year P 99 Water < ≈ 100 MW th Several 100 km from coast Several 100 MW th Deep in continent < ≈ 2000 MW th Regional Monitoring Several Megaton-year P 99 100 ~ several 100 MW th Location (95% Confidence Level) BestWorst < ~ 100 km Band stretching over several 100 km One module costs several hundred million dollars  multiply this by number of modules in array A terawatt-level georeactor increases the background level by a factor of several in most locations around the world A regional monitoring regime seems not-too-outlandish

48. Appendix

49. Antineutrino Detection Rate for H 2 O + GdCl 3 Detector Reactor Assume 100% detection efficiency for E > 1.8 MeV Fission Bomb Assume 100% detection efficiency for E > 1.8 MeV Integrated over 10 sec. burst time

50. Antineutrino Detection Rate for H 2 O + GdCl 3 Detectors Reactor Assume 100% detection efficiency for E > 1.8 MeV Fission Bomb Assume 100% detection efficiency for E > 1.8 MeV Integrated over 10 sec. burst time

51. Background Processes Antineutrinos from sources other than the rogue reactor Non-antineutrino background mimicking antineutrino events Commercial nuclear reactors Geo-neutrinos Georeactor (possibly) Cosmic rays Radioactivity in the detector Require E > 3.4 MeV Place detector at > 3 km depth under water Fiducial volume cut + radon free environment

52. Antineutrino Detection with a H 2 O + GdCl 3 Detector Inverse beta decay on target hydrogen nuclei e + p  n + e + Prompt Event Delayed Event E > 1.8 MeV E e ≈ E – 1.3 MeV Detector Threshold:  E e > 2.5 MeV E > 3.8 MeV Physics Threshold: ≈ 20 µs n + Gd  Gd* E cascade ≈ 3~8 MeV Gd +  cascade 90% neutron captured by Gd @ 0.2% concentration

53. Commercial Nuclear Reactors 433 reactors Total thermal power ≈ 1 TW

54. The effect of commercial nuclear reactors on the detection sensitivity for a rogue nuclear reactor Assume that a rogue reactor with P = 250 MW th is operating just north of Hawaii Top: Middle: Bottom: log 10 S log 10 B # events from rogue reactor # events from commercial reactors 3.5 7.0 1.5 Detector target mass = 10 megatons 1 year exposure Detectors allowed only in oceans & large lakes 100% detection efficiency S, B, and S/sqrt(S+B)

55. Possible Detector Locations 23 Locations based on S/sqrt(S+B) log 10 (S) log 10 (B) Map of S, B, and S/sqrt(S+B) for 1 megaton target exposed for 1 year

56. If a Geo-Reactor Exists… If it does exist, its power is expected to be 1 ~ 10 TW th, 3 TW th being the most favored value. The total power from all commercial reactors world-wide ≈ 1 TW th In most locations around the world, antineutrinos from a georeactor would outnumber those from commercial reactors 3 TW th Georeactor