Searching for Dark Matter with COHERENT

Searching for Dark Matter with COHERENT

Dan Pershey (Duke University)

for the COHERENT Collaboration

The Magnificent CEvNS, Nov 10 2019

D. Pershey Searching for Dark Matter with COHERENT

Dark Matter at the Spallation Neutron Source

❑Huge number of proton-Hg collisions may produce portal particles (𝑉) that mediate interactions between SM and hidden sector particles (𝜒)

❑Such a particle makes an attractive dark matter candidate consistent with thermal freeze-out for masses below the Lee-Weinberg bound

❑Portal particles would be produced mainly through π0/η0 → 𝑉γ and the hidden sector particles may interact within our detectors in Neutrino Alley

Vector portal:

Baryonic portal:

2

deNiverville et al., Phys Rev D92 095005 (2015)

NeutrinoDark Matter


Need for Accelerator-Based Searches

❑Direct detection experiments can search for light dark matter through 𝜒 − 𝑒 scattering

❑Can test thermal flux cross section assumingDM is a scalar

❑But… cross sections for different spin scenarios heavily suppressed at slow speeds

3

Plots from arXiv: 1707.04591

(MeV)

❑Any DM produced at accelerators would be relativistic – the interaction strengths for a thermal relic depend weakly on spin

❑And, relevant interaction strengths are within two orders-of-magnitude of current experimental bounds


DM Scattering in COHERENT Detectors

❑Detectors with a low-enough threshold to observe CEvNS would also be sensitive to coherent 𝜒 − 𝐴 scattering

❑The cross section is similarly large and precisely calculable so that a dark matter detector sensitive to CEvNS may be competitive with much larger experiments

❑Would expect a harder recoil spectrum than CEvNS as 𝑉 → χχ′ happens in flight

4

CEvNS:at low 𝑄2, a neutrino may interact with an atom whose nucleons recoil in-phase, increasing the cross section

𝜒 𝜒

𝑉

γ×

Dark Matter:In these dark matter models, an analogous channel exists with a similarly enhanced cross section


DM Potential for COHERENT Detectors❑CsI:

Fully capable of detecting dark matter, analysis of existing data underway

❑NaI:Very high background rates and low 𝑁 make this challenging

❑Ge:Low mass, and sensitivity to low recoil energies less helpful for observing the relatively hard recoil spectrum

❑LAr:Scalability and relatively low backgrounds make argon an ideal detection medium for detection of coherent DM scattering

5







This talk focuses on calculating the sensitivity of a 610 kg (fiducial) liquid argon detector in Neutrino Alley, LAr-1t

6





Large-scale, low-threshold cryogenic crystals would be sensitive to low-mass DM



This talk focuses on calculating the sensitivity of a 610 kg (fiducial) liquid argon detector in Neutrino Alley, LAr-1t

7


Predicting the LAr-1t Selected SampleBuilding off our experience with CENNS-10, we have data-driven estimates for backgrounds, CEvNS, and DM signal

❑Steady-state background• Dominated by 39Ar decay and radioactivity from an off-gas pipe running

through Neutrino Alley; rates measured in CENNS-10

• Detector is assumed to run with underground argon, depleting 39Ar concentration by a factor of 100

❑Beam neutron background• Simulation tuned to CENNS-10 data

• With simulation of additional to be placed in Neutrino Alley

❑CEvNS• Detection efficiency tuned to CENNS-10 calibration data

• Uncertainties on SM cross section and detector response included in analysis

❑DM signal• Generated with BdNMC convolved with detector response used for CEvNS

8

deNiverville et al., Phys Rev D95 035006 (2017)


Expected Sample After 3 Years

❑CEvNS very clearly visible after three years and is the largest background to a dark matter search

❑Energy and time are both important handles for separating DM and CEvNS• DM particles are relativistic, scatter within the prompt window (0 < t < 1 μs)

• Delayed window (1 < t < 6 μs) gives in-situ measurement of backgrounds

9

Vector portal:𝑚χ = 15 MeV

𝑚𝑉 = 45 MeVε = 8.77×10-5

α′ = 0.5Steady-state Bkg 2790

Beam Neutrons 971

CEvNS 5469

DM 1501

Prompt events after 3 years


Constraining CEvNS with Timing Info❑Proposed dark matter would be relativistic → the prompt time window is the

analysis region of interest

❑But … systematic uncertainties stifle our ability to robustly identify signal

❑Underestimating the neutrino flux by 10% would give a ≈550 event excess in the ROI that could be mis-interpreted as a dark matter signal

10

Prompt

Prediction 5469

Fake Data 6016 (+10%)

Bkg-Subtracted Counts

Assumed no DM


Constraining CEvNS with Timing Info❑But, an uncertainty on the neutrino flux would affect delayed events too

❑Observed data in the delayed timing sideband constrains systematic uncertainties relevant for understanding the CEvNS background in the ROI

11

Prompt Delayed

Prediction 5469 7450

Fake Data 6016 (+10%) 8195 (+10%)


Assumed no DM

Delayed sideband gives unique information for experiments in π-DAR beams!


Constraining CEvNS with Timing Info

12

Prompt eventsignored in this fit

❑To understand the power of the constraint, imagine fitting just the delayed sideband to this fake-data with the flux normalization shifted at +10%

❑Without analyzing events in the ROI, we’ve determined we must increase the CEvNS background prediction by 9.3%

Prompt Delayed


Fake Data 6016 (+10%) 8195 (+10%)

Fit 5980 (+9.3%) 8146 (+9.3%)


Base simulation underestimates the prompt CEvNS bkg by 10%, but a sideband reduces the effect to 0.7%

Assumed no DM


Error Budget for Identifying DM Scatters

❑Without this delayed sideband constraint, a dark matter search would be limited by systematic uncertainties

❑Allows for a more detailed understanding of the distinctive recoil energy spectrum expected for DM scatters

❑After COHERENT dark matter program, the analysis will be dominated by statistical errors → future paths for DM searches with CEvNS detectors

13

Leptophobic portal:𝑚χ = 5 MeV

𝑚𝑉 = 50 MeVα𝐵= 1.4e-7


Projected COHERENT Sensitivity

14

Vector Portal

Baryonic Portal

With 3 years of LAr-1t data, we expect to improve on current constraints for 𝑚χ < 80 MeV

Will probe thermal target for scalar DM for α′ < 0.1

Expect significant improvement for 𝑚χ < 500 MeV apart from

energies near 𝑚π

Most leading limits from beam dump experiments set with χ–𝑒scattering


Aside: Importance of Understanding Beam T0❑Using available timing information in all our detectors, we currently can

only determine the timing of the CEvNS pulse to about 200 ns

❑Fitting the sideband to fake-data, generated with a proton arrival time 200 ns earlier than expected, leads to a DM-like excess

❑Currently working with SNS on robust method to precisely measure the proton arrival time so this will not be significant uncertainty in the future

15

Prompt Delayed


Fake Data 6030 (+10%) 6801 (-8.7%)

Fit 5188 (-5.1%) 6834 (-8.3%)


Long-term Strategies for ImprovementLAr-1t will test currently unexplored parameter space, but CEvNS detectors have further potential

16


Leptophobic portal:𝑚χ = 5 MeV

𝑚𝑉 = 50 MeVα𝐵= 1.4e-7

Long-term Strategies for Improvement

17

❑After 610 kg x 3 yrs, the search is limited by statistical uncertainty• An order-of-magnitude increase in mass would further improve limits

❑Argon is scalable, and exploratory studies of a 10-t detector have begun

LAr-1t will test currently unexplored parameter space, but CEvNS detectors have further potential



18

❑The neutrino flux in a π-DAR beam is isotropic, but the flux of any dark matter would be boosted

❑LAr-1t will be 138 off-axis while an on-axis detector would roughly 6x the dark matter flux with the same CEvNS background

Off-axis Angle

Highest DM Flux

LAr-1t




❑We will also have an improved understanding of our neutrino flux

❑Though not the dominate error, the D20 detector win reduce the uncertainty to a few percent, which will improve DM constraints

19

See talk from J. Newby tomorrow morning



Long-term Sensitivity

❑ If we were to manage a 10 t on-axis detector with improved understanding of the flux, we could place very ambitious DM constraints

❑ In the most conservative scenarios, we will probe parameters for scalar DM that would explain the cosmologically observed dark matter concentration

❑Scalar DM excluded for all α′ < 1 for 5 < 𝑚χ < 100

20

Assuming a scalar DM candidate


Probing Non-Scalar Dark Matter

21

A 10 t on-axis detector wouldpush constraints on scalar and Majorana DM past the perturbative limit

Can test reasonable parameter space for pseudo-Dirac DM

The COHERENT liquid argon detector will test DM consistent with cosmological flux assuming α′ < 0.1

D. Pershey Searching for Dark Matter with COHERENT 22

Thank you from the COHERENT collaboration!

DM sensitivity paper soon to besubmitted to arXiv!

Documents

Searching for Dark Matter with COHERENT