HydroX: Hydrogen in Liquid Xenon TPCs

Maria Elena Monzani

Kavli Institute for Particle Astrophysics and Cosmology

Snowmass CF1 Meeting August 21, 2020 Dark Matter rates: low-mass vs high-mass

To enhance low-mass sensitivity, we want: 1. lighter target for better kinematic matching to DM mass 2. lower the threshold as much as possible

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 2 Hydrogen doping in Liquid Xenon

1. Dissolve H2 in LXe 2. Look for recoiling proton

Advantages: established low Advantages: kinematic matching background, plus Xe shielding. to light DM, lower threshold.

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 3 what are the expected S1 and S2 yields for He/Ne recoils in LXe? As described above, electrons deposit their energy into electronic excitations (electronic stopping) while xenon recoils in LXe deposit their energy into both electronic excitations and elastic collisions with nuclei (nuclear stopping). All the electronic energy is eventually collected as signal, but some of the energy given to nuclear recoils is lost as heat. Calculating the final electronic energy deposition from a xenon recoil is more complicated than simply taking the amount given directly from the primary recoil to Signalelectronic enhancement excitations, in H2 as+Xe secondary nuclei from the nuclear collisions in turn partition their energy into electronic and nuclear stopping. Lindhard theory [40] gives an approximation for the “Lindhard

WIMP recoil off Xe: mfactor”,Xe = mXe → or energy the total lost to electronic heat (Lindhard energy) deposition from nuclear recoils relative to electronic recoils of • at low energies, <20%the sameobservable…and energy. Figurefalling! 7 shows a plot of the Lindhard factor vs. energy for xenon, and the signal produced by low energy xenon recoils is less than 20% that produced by ER of the same energy.

Observable energy fraction after heat loss

0.25 Recoil Lindhard SRIM Xenon 0.02 0.02 0.2 Neon 0.20 0.09

Lindhard factor 0.15 Helium 0.68 0.69

0.1 0 1 10 10 Recoil energy [keV] Table 1: Estimated fraction of energy given to electronic stopping for nuclear recoils (not ac- Figure 7: Fraction of energy going into ob- counting for secondary cascades) from Xe, He, servable signal (Lindhard factor) vs. recoil and Ne recoils in LXe, calculated using Lindhard M.E. Monzani – HydroX energyCF1 Meeting, for August xenon 21, 2020 recoils in LXe. 4 theory [41] or the SRIM simulation package [42].

Because helium and neon are so much lighter than xenon, they will not lose as much energy in elastic collisions with xenon atoms, leaving more energy for electronic excitation and a corre- spondingly larger signal. Simple approximations for the Lindhard factor do not exist for nuclei moving through fluids composed of a di↵erent element, but one can estimate the raw stopping powers (before accounting for the secondary cascades) using either Lindhard theory [41] or the SRIM simulation package [42]. Table 1 shows the predicted amount of energy going directly from the primary recoil into electronic stopping from 5 keV Xe, He, and Ne recoils in LXe calculated via both methods. Neon and especially helium have a much larger fraction of energy deposited directly to electrons, i.e. directly into signal, without accounting for the secondary cascades that can only increase these fractions. It should be noted that the e↵ect of the cascades will be reduced for neon and helium because they will not eciently transfer energy to the predominantly xenon atoms around them, leading to more sub-ionization energy depositions. Even so, one can expect larger signals (both charge and light) from helium and neon recoils in LXe than from xenon recoils, and a correspondingly lower energy threshold. The second key question is how will that increased signal be partitioned into S1 and S2; what happens to the S2/S1 ratio that is so important for rejecting electron recoil backgrounds in LZ? Given that the ratio is determined by track structures, and recoiling electrons will still be interacting with xenon atoms, the S2/S1 ratio for electrons should be unchanged. As it is not fully understood what drives the partitioning between S1 and S2 for xenon recoils in LXe, the most that can be said here is that He/Ne recoils will likely lie below the electron band shown in Fig. 4. As one example, in scintillating CaWO4 crystals operated by the CRESST dark matter experiment, oxygen recoils produce a light/heat ratio that lies between the electron and tungsten recoil bands [43]. One can

9 Signal enhancement in H2+Xe

WIMP recoil off Xe: mXe = mXe → energy lost to heat (Lindhard) • at low energies, <20% observable…and falling!

WIMP recoil off H2: mp ≪ mXe → all electronic excitations • ~100% of energy is observable

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 5 Xenon shielding in a large detector

• Retain (self-) shielding properties of (a large mass of) LXe • Suppress external backgrounds like in a large-scale TPC • Plus: vetoes, water tank, intensive radio-cleanliness of LZ A kg-scale DM detector “embedded” in a ton-scale one!

103

102

101

0 Liquid H 10 2 Liquid xenon Interaction length [cm]

10-1 10-1 100 101 Gamma energy [MeV]

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 6 Feasibility Questions and Assumptions

Can we get H2 in LXe? And how much? Is H2 safe for PMTs?

• Originally considered He as dopant • 5% H2 fraction seems achievable • He can diffuse through PMT glass • Assume: 1% mol fraction, 1 kg in LZ • Larger H2 molecule slows diffusion • Can run safely for ~106 days

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 7 More Feasibility Questions, R&D

Circulation and cryogenics questions: • establish solubility of H2 in Xe • operations: Xe purification removes H2 Effect of doping on signal generation: • measure charge (and light?) losses • estimate effect on discrimination • calibrate ultra-low energy recoils on H2

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 8 HydroX Sensitivity Estimates

Sensitivity assumptions: • Yields from SRIM + LZ modeling

• 1 kg of H2 in LZ (1% mol fraction) • S2/S1 for NR recoils on proton is ER-like (i.e. no discrimination) • 500 live-days exposure

SI

SD

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 9 HydroX “Collaboration” and Timeline

Collaboration is under construction! • Current team: SLAC, LBL, FNAL, UCSB, Northwestern, Penn State, University of Michigan, SURF, Imperial College (UK) • ECA funding at UCSB, LDRD at SLAC, “Xenon Futures” (STFC) at Imperial • Growing interest from LZ collaboration • Please reach out if you are interested!

Hypothetical HydroX timeline: • 2020-21: R&D at FNAL/SLAC/UCSB/IC • 2022-23: cryogenics demonstration • 2025-26: possible upgrade to LZ? • 2027-30: HydroX science run

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 10 Conclusions and Outlook

H-doping gives new capabilities to existing and upcoming LXe TPCs (LZ, XENON, PandaX, ...) • Favorable kinematic matching to low-mass DM, plus signal enhancement • Xe shielding, not available in a conventional low-mass experiment • Probe additional 100 MeV – 5 GeV DM mass range, with SI and SD sensitivity

Even more interesting for a G3 experiment:

• Planning for H2 doping from scratch would simplify and optimize detector design • If SiPM readout, He is back on the table (potentially easier cryogenics/purification) • Global mass reach: 100 MeV - 10 TeV!

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 11 Thank you!

Slides https://and smu.zoom.usplots credits:/j/93594661518?pwd= UmlGdG1ZeDlpL1RxUy9FSFQwTmhqZz0 • H. Lippincott/A.9 Monte (UCSB) • A. Fan/M.E. Monzani/T. Shutt (SLAC)

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 12 Signal enhancement in H2+Xe

Nuclear recoil off Xe: mXe = mXe → energy lost to heat (Lindhard) • at low energies, <20% observable…and falling!

Nuclear recoil off H2: mp ≪ mXe → all electronic excitations • ~100% of energy is observable

M.E. Monzani – HydroX CF1 Meeting, August 21, 2020 13