LLNL This work was partially performed under the auspices of the US Department of Energy by the University of California, Lawrence Livermore National Laboratory,

LLNLThis work was partially performed under the auspices of the US Department of Energy by the University of

California, Lawrence Livermore National Laboratory, under contract No. W-7405-Eng-48.

Safeguards and Cooperative Monitoring of Reactors With

Antineutrino Detectors

Adam Bernstein, (P.I.)

Jan Batteux

Dennis Carr

Celeste Winant

Chris Hagmann

Norm Madden

John Estrada (P.I.)

Nathaniel Bowden

Jim Lund

C. Michael Greaves

N. Mascarhenas

Lawrence Livermore National Laboratory

Sandia National LaboratoriesCalifornia

Stanford University Giorgio Gratta, Yifang Wang

University of AlabamaAndreas Piepke

Oak Ridge National Laboratory Ron Ellis

Collaborators

LLNL

Project Timeline

1999-2000Late 2000

Research into 1 kT explosion detectionRecognize futility of this effort – publish paper

2000/2001 Research into reactor monitoring

2002 Begin installation at San Onofre

Oct. 2003 First data taking

Dec. 2003 IAEA interest / experts meeting

Now Operational for 100 days, 70 events/day

Feb 9th 2004 Refueling shutdown

Summer 2004

800-1000 events per day

LLNL

Properties of Antineutrinos and Antineutrino Detectors

• Rates near reactors are high 1 ton detector, 24 m from reactor core Not untypical core thermal power = 3.46 GW 3925 events/day/ton (100% efficient detector)

• Rate and spectrum are sensitive to the isotopic composition of the core

• Cost and complexity can be made comparable to that of a few high-end Germanium detectors

LLNL

Monitoring Reactors with Antineutrino Detectors

A. ~1 cubic meter antineutrino detectors placed a few tens of meters from the reactor core

B. Compare measured and predicted total daily or weekly antineutrino rates (or spectrum) to search for anomalous changes in the total fission rate

C. Identify changes in fissile content based on changes in antineutrino rate (“the burnup effect”)

A. Measured in previous experimentsB. Rovno quotes 540 kg +- 1% fissile content from shape analysis

LLNL

What Good Is That ?

1. Detecting unauthorized production of plutonium outside of declarations

2. Measuring enrichment of freshly loaded fuel and burn-up or plutonium content of spent fuel destined for reprocessing or storage shipper-receiver difference

3. Checking progress of plutonium disposition, and ensuring burnup is appropriate to core type

4. Monitoring core conversion

• An integral, continuous, high statistics, non-intrusive, unattended measurement

LLNL

Fission Rates Vary with Time and Isotope

U-238

Pu-241

Input fuel enrichment can be changed in PWRs increased plutonium production even at constant power

Easy to alter for CANDU (online refueling)

LLNL

Detected Antineutrino Rates Vary With Isotope

LLNL

The Antineutrino Rate Tracks Inventory Changes

• The total antineutrino rate changes with the relative U/Pu content of the core About 250 kg of Pu is generated during the cycle

• Rate calculation based on a detailed reactor simulation shows an antineutrino rate change of about 10% through a 500 day equilibrium reactor cycle This “burnup effect” seen and corrected for in

past experiments Modern detectors reach 3% precision

The change in antineutrino rate directly tracks the fissile inventory even at constant power

LLNL

A 1 Cubic Meter Detector, 10 Meters From PWR Core

• fuel rods with 20 kg Pu replaced with fresh rods (0 kg Pu)

• assumed 3% systematic error

• 50% detection efficiency

• A standard statistical test can identify the switch with

> 90% confidence with one month’s data

The systematic shift in inventory is reflected by the antineutrino count rate over time

Days

Counts per day

LLNL

Inputs Needed to Predict/Extract an Absolute Antineutrino Measurement

1. Core model with the input parameters: Secondary calorimetric power Pressure Flow rates Boron concentration Inlet temperature total model error 1% (power

dominates) *

2. Antineutrino energy density error = 3 % 3. Null result from near-reactor oscillation

experiments4. Well understood antineutrino detector

(* from “Estimation of Expected Neutrino Signal at Palo Verde”, Lester Miller, Stanford University,unpublished note)

LLNL

• 3.46 GWt reactor

• Antineutrino detector in “tendon gallery” with 1017 / s per m2

• Installation/testing begun May 2002

The Site, Detector, Signal, Backgrounds

data taking began in late September 2003

LLNL

The Underground Experimental Site

20 meter concrete/rock overburden

24 meters from core

LLNL

Cutaway Diagram of the LLNL/Sandia Antineutrino Detector

Gd-doped

Currently operational:

2 cells instrumented with

4 pmts;

0.32 tonnes of Gd-scintillator;

quasi-hermetic muon veto

hermetic water shield

LLNL

• The antineutrino interacts with a proton producing…

– A 1-7 MeV positron

– A few keV neutron

– mean time interval 28 sec

• Both final state particles deposit energy in a scintillating detector over 10s or 100s of microsecond time intervals (depending on the medium)

• Both energy depositions and the time interval are measured

Detection of Antineutrinos

LLNL

Backgrounds

1. Muon generated neutrons create “correlated” events

fast neutron proton knock-on thermalization and capture within time window

energies and time correlation can mimic the antineutrino

2. Neutron and gamma “singles” can fall within the time coincidence window defining the antineutrino event

gammas from surroundings neutrons from S.F., activated nuclei from surroundings muon-induced neutrons Background rate ~ a few dozen per day

LLNL

Four Variables Define the Antineutrino Signal

Variable Eff.

T > 100

(sec)

95% the time between a muon veto and a cube signal

10 < Tcube < 100

(sec)

70% the time between the two energy depositions mean = 28 sec

3 < Eprompt < 10

(MeV)

62% (analytic formula)

The prompt, positron-like signal (including annihilation gammas)1.022 < E1 < 7 MeV, peak at ~3 MeV

4 < Edel < 12

MeV

68% (MC) the delayed, neutron-like signal from Gd gamma cascade

“geometry cut” ~80% Events with large asymmetries in PMT energy distribution within a cell

1280 events per day 68 events per day (now) over 800 (with simple upgrades)

100% efficiency 5% efficiency 30% efficiency and 2x volume

LLNL

Event Candidates Since the Last Muon

antineutrino-like backgrounds(spallation and capture) more likely to occur near a muon

T > 100 sec

Cut on time since last muon:

LLNL

2. Interevent time

Interevent time distribution well fit by e-t/ 28 sec

(Capture time set by 0.1% Gdconcentration in scintillator)

10 < Tcube < 100 sec

Cut on interevent time:

LLNL

Conclusions

1) That antineutrinos can track burnup and plutonium inventory is firmly established by prior experiments and shortly confirmed by us

2) Detector deployment essential for demonstrating practical utility and potential

3) Main challenges: • Controlling detector systematic effects (spectrum

error, fiducial error, event containment…)• Shrinking footprint (coherent scatter, better IVB

detector)• Transforming a delicate physics instrument into a

robust cooperative monitoring deviceMust compare to existing safeguards methods and demonstrate that the benefit is worth the cost of deployment

LLNL

Literature on Applied Antineutrino Physics

Reactor Monitoring Bernstein, A., Wang, Y., Gratta, G., West, T.,

Nuclear Reactor Safeguards and Monitoring with Antineutrino Detectors, J. Of Appl. Phys V.91, Num. 7, p 4672, April 2002

Klimov, Yu. et al. The remote measurement of power and energy release using a neutrino method, Inst. Obshch. Yad. Fiz, Russia, At. Energ., (1994) 76(2) 130-5, CODEN:AENGAB; ISSN: 0004-7163

Detection of Antineutrinos for Non-Proliferation M.M. Nieto, A. C. Hayes, C. M. Teeter, W. B. Wilson,W. D. Stanbro, arXiv:nucl-th/0309018 v1 9 Sep 2003

Conclusion: power and isotopic measurement at 10-100 m is feasible

Explosion DetectionA. Bernstein, T. West V. Gupta, An Assessment of Antineutrino Detection as a Tool for

Monitoring Nuclear Explosions, Science & Global Security, Volume 9 pp 235, April 2001

Conclusion: 1 kT remote detection is not feasible