Dark Matter Searches with the PICO Bubble Chambers Scott Fallows Tevpa 2017 – August 8

Home , PICO

Dark matter searches with the PICO bubble chambers Scott Fallows TeVPA 2017 – August 8 – Ohio State University

2 2 Overview

• Dark matter direct detection: spin-independent (SI) vs. spin-dependent (SD) WIMP-nucleon couplings

• Bubble chambers for direct detection: the PICO program

• The PICO-60 C3F8 detector; Run-1 WIMP-search results

• Next chamber: PICO-40L

Design/background goals, timeline, physics reach

• Future ton-scale chamber: PICO-500

3 Direct detection: channels

• Bubble chambers don’t use the typical direct detection channels

• They’re(exception: threshold scintillating LXedetectors BC – see slides from J. Zhang’s talk, TeVPA 2017, Mon 7 Aug, 5:15pm) • No direct measurement of recoil energy, however…

• Acoustic signals do enable both spectroscopy and discrimination (arguably ﬁts Zornoza, NIM. A 742 130-138 (2014) into the “heat” category)

4 Direct detection: channels

• Bubble chambers don’t use the typical direct detection channels

• They’re threshold detectors

• No direct measurement of recoil energy, however…

• Acoustic signals do enable both spectroscopy and discrimination (arguably ﬁts in the “heat” category)

5 Direct detection: SI vs SD

Two primary interactions considered by experimentalists:

• Spin-independent (SI): couples to all nucleons

• A2 enhancement for large nuclei (coherent scattering)

• Spin-dependent (SD): couples to the spin of the

nucleus, i.e. unpaired spin of one nucleon 129/131Xe, …) 73 , (19F, Ge • “SDp” and “SDn” are similar but not directly comparable

6 SPIN DEPENDENCE OF DARK MATTER SCATTERING PHYSICAL REVIEW D 78, 056007 (2008) IceCube muon rate IceCube signal significance

mSUGRA mSUGRA UED UED LHT LHT xSM xSM RHN Discovery RHN

Evidence µ σ N 10 1 2 10 1 2 10 10 − − log log 2 2 − − 3 3 − − 4 4 − − 0 200 400 600 800 1000 0 200 400 600 800 1000

MDM MDM

FIG. 11 (color online). The expected numbers of events and the statistical significance of DM signals for 3 and 5 years running of the IceCube neutrino telescope. Only the FP region in mSUGRA and a portion of the mUED parameter space give a significance >5 for 5 years of running. oughly scan the parameter spaces, we adopted the section. This is a consequence of the DM annihila- Bayesian method that is the foundation of the Markov tion occurring through the Z boson. To counter the chain Monte Carlo approach. The DM models can have small annihilation rate in the early universe (due to distinct phenomenological predictions. We showed that the the small neutralino mass in the model), the neu- DM model possibilities can be narrowed by measurements tralino pair is required to have a larger Z boson of both SI and SD elastic scattering. The direct signals for coupling, resulting in a large SD cross section. DM in recoil and neutrino telescope experiments are com- (iii) In the singlet extended SM, the DM candidate is plementary to LHC experiments in distinguishing the be- spin-0, which gives a vanishing SD cross section. yond the standard model physics scenarios [135]. TheDifferent SI cross section is generally small,Responses below Favor We summarize below the model predictions for the DM 10 8 pb, and occurs through Higgs boson ex- À cross sections; the posterior distributions are summarized change. If SD scattering of DM is observed, the in Fig.SI12. vs. SD (vs. nuclearclass physics/EFT of models with spin-0 DM would be imme-) (i) In mSUGRA, the FP region provides the largest SI diately excludedDifferent as being the sole origin of the DM Elements! • andSpin-independent SD cross sections. This searches is due to the have mixed receivedin the the Universe. most attention (iv) For stable Dirac neutrino DM, the Z boson domi- Higgsino nature of the2 lightest neutralino; the neu- tralinodue couplings to the to A the rate Higgs enhancement and Z bosons are large. (>15,000nates for inxenon) the calculation Protons of both the relic density and Neutrons The Bino nature of the lightest neutralino in the CA elastic scattering cross section and makes the SI and and AF regions causes these scenarios to have sub- SD cross sections tightly correlated and large. PICO-250L DOE G2 Proposal Mp Sp'' Sp' Fp'' Fp' 6Dp Mn Sn'' Sn' Fn'' Fn' Dn • stantiallyBut the smaller actual cross mechanism sections. is unknown,(v) In mUED, a sweet spot of SD O 10À pb exists for which the DM relic density is reproduced.ð Þ The environment(ii) called The for tadpole it (e.g., to nMSSM follow up a signal model in a haslarge largexenon detector). SD scattering, In addition, the use of F multiple mass targets will be needed to characterize the WIMP mass and velocity distribution in the event F and may3 be more complicated relic density is strongly dependent on the curvature1.0 1.0 of a discovery of[21][22] order. 10À pb, and a wide range of SI cross parameter and fixes its value. The KK quarks have SI vs. SD approximatelyNa the same mass as the inverse curva-0.8 Na 0.8 mSUGRA UED ture and the SD cross section is thus closely tied to LHT 2 0.6 0.6

− nMSSM RHN theSi relic density. On the other hand, the SI cross Si section is more dispersed due to the larger variation 0.4 0.4 4

− of the Higgs boson mass. (pb) Ge Ge SD (vi) In the LHT model, the SD interaction occurs0.2 0.2 σ 10

6 through T-odd quarks which have a small hyper- − log charge.I Therefore, the SD cross section in this0 I 0 ---640246 ---640246

8 model is typically very small. In contrast, the SI − scatteringXe proceeds through the Higgs and T-odd Xe quarks, giving experimentally accessible SI cross 10 − section values. −12 −11 −10 −9 −8 −7 −6 We provide a summary of the SI and SD cross sections log10σSI (pb)

Figure 2: Comparison of SD-proton vs. SI cross sections for a set of dark matter models, showing theby the boxSensitivity and whisker of different plots in Fig. p-coupling13. The boxes operators repre- complementary andSD necessary vs. SIreach cross of both channels section to explore predictions possible parameter for space (fromDM [17]p ). FIG. 12 (color online). The posterior distributions of SI À sents the coverageto of various the middle nuclear 50-percentile. targets We sum- DM p

and SDdifferentÀ for six models models. (Barger, PRD, 78 056007) marize the forecast for observing a signal in neutrino 7 (L. Fitzpatrick, INT Workshop, 2014) b)(Description(of(Experimental(Method;(Performance(Requirements(

The(Case(for(Bubble(Chambers(as(Dark(Matter(Detectors( Throughout this section, we refer to several existing bubble chambers. COUPP-4 is a 2-liter chamber that was filled with CF3I at Fermilab and then at SNOLAB, producing excellent physics results056007-13 [9][10]. COUPP-60 has been running with 37 kg of CF3I at SNOLAB since June 2013. PICO-2L is a 2-liter chamber that has replaced COUPP-4 at SNOLAB; commissioned in October 2013, it represents the first large-scale bubble chamber to be filled with C3F8 and also the first joint effort of the combined PICO collaboration.

The strengths of the bubble chamber technology for dark matter searches can be summarized as follows:

Electron)recoil)insensitivity) A principal strength of the bubble chamber technology for a dark matter search application is the extraordinary insensitivity (~10-10) to electron recoil backgrounds. The ability of the bubble chamber to attain this large background rejection factor while maintaining high efficiency for detecting nuclear recoils arises naturally from the physics of bubble nucleation in a superheated liquid, which requires a critical energy deposition within a small volume to nucleate the bubble. The bubble formation scale and energy threshold are classical thermodynamic properties determined by the temperature and pressure of the superheated fluid [23][24][25]. These can be adjusted to cleanly discriminate between the nuclear

5 SPIN DEPENDENCE OF DARK MATTER SCATTERING PHYSICAL REVIEW D 78, 056007 (2008) IceCube muon rate IceCube signal significance

mSUGRA mSUGRA UED UED LHT LHT xSM xSM RHN Discovery RHN

Evidence µ σ N 10 1 2 10 1 2 10 10 − − log log 2 2 − − 3 3 − − 4 4 − − 0 200 400 600 800 1000 0 200 400 600 800 1000

MDM MDM

FIG. 11 (color online). The expected numbers of events and the statistical significance of DM signals for 3 and 5 years running of the IceCube neutrino telescope. Only the FP region in mSUGRA and a portion of the mUED parameter space give a significance >5 for 5 years of running. oughly scan the parameter spaces, we adopted the section. This is a consequence of the DM annihila- Bayesian method that is the foundation of the Markov tion occurring through the Z boson. To counter the chain Monte Carlo approach. The DM models can have small annihilation rate in the early universe (due to distinct phenomenological predictions. We showed that the the small neutralino mass in the model), the neu- DM model possibilities can be narrowed by measurements tralino pair is required to have a larger Z boson of both SI and SD elastic scattering. The direct signals for coupling, resulting in a large SD cross section. DM in recoil and neutrino telescope experiments are com- (iii) In the singlet extended SM, the DM candidate is plementary to LHC experiments in distinguishing the be- spin-0, which gives a vanishing SD cross section. yond the standard model physics scenarios [135]. TheDifferent SI cross section is generally small,Responses below Favor We summarize below the model predictions for the DM 10 8 pb, and occurs through Higgs boson ex- À cross sections; the posterior distributions are summarized change. If SD scattering of DM is observed, the in Fig.SI12. vs. SD (vs. nuclearclass physics/EFT of models with spin-0 DM would be imme-) (i) In mSUGRA, the FP region provides the largest SI diately excludedDifferent as being the sole origin of the DM Elements! • andSpin-independent SD cross sections. This searches is due to the have mixed receivedin the the Universe. most attention (iv) For stable Dirac neutrino DM, the Z boson domi- Higgsino nature of the2 lightest neutralino; the neu- tralinodue couplings to the to A the rate Higgs enhancement and Z bosons are large. (>15,000nates for inxenon) the calculation Protons of both the relic density and Neutrons The Bino nature of the lightest neutralino in the CA elastic scattering cross section and makes the SI and and AF regions causes these scenarios to have sub- SD cross sections tightly correlated and large. PICO-250L DOE G2 Proposal Mp Sp'' Sp' Fp'' Fp' 6Dp Mn Sn'' Sn' Fn'' Fn' Dn • stantiallyBut the smaller actual cross mechanism sections. is unknown,(v) In mUED, a sweet spot of SD O 10À pb exists for which the DM relic density is reproduced.ð Þ The environment(ii) called The for tadpole it (e.g., to nMSSM follow up a signal model in a haslarge largexenon detector). SD scattering, In addition, the use of F multiple mass targets will be needed to characterize the WIMP mass and velocity distribution in the event F and may3 be more complicated relic density is strongly dependent on the curvature1.0 1.0 of a discovery of[21][22] order. 10À pb, and a wide range of SI cross parameter and fixes its value. The KK quarks have SI: scaling w/ mass SI vs. SD approximatelyNa the same mass as the inverse curva-0.8 Na 0.8 mSUGRA UED ture and the SD cross section is thus closely tied to LHT 2 0.6 0.6

− nMSSM RHN theSi relic density. On the other hand, the SI cross Si section is more dispersed due to the larger variation 0.4 0.4 4

− of the Higgs boson mass. (pb) Ge Ge SD (vi) In the LHT model, the SD interaction occurs0.2 0.2 σ 10

6 through T-odd quarks which have a small hyper- − log charge.I Therefore, the SD cross section in this0 I 0 ---640246 ---640246

The strengths of the bubble chamber technology for dark matter searches can be summarized as follows:

5 SPIN DEPENDENCE OF DARK MATTER SCATTERING PHYSICAL REVIEW D 78, 056007 (2008) IceCube muon rate IceCube signal significance

mSUGRA mSUGRA UED UED LHT LHT xSM xSM RHN Discovery RHN

Evidence µ σ N 10 1 2 10 1 2 10 10 − − log log 2 2 − − 3 3 − − 4 4 − − 0 200 400 600 800 1000 0 200 400 600 800 1000

MDM MDM

FIG. 11 (color online). The expected numbers of events and the statistical significance of DM signals for 3 and 5 years running of the IceCube neutrino telescope. Only the FP region in mSUGRA and a portion of the mUED parameter space give a significance >5 for 5 years of running. oughly scan the parameter spaces, we adopted the section. This is a consequence of the DM annihila- Bayesian method that is the foundation of the Markov tion occurring through the Z boson. To counter the chain Monte Carlo approach. The DM models can have small annihilation rate in the early universe (due to distinct phenomenological predictions. We showed that the the small neutralino mass in the model), the neu- DM model possibilities can be narrowed by measurements tralino pair is required to have a larger Z boson of both SI and SD elastic scattering. The direct signals for coupling, resulting in a large SD cross section. DM in recoil and neutrino telescope experiments are com- (iii) In the singlet extended SM, the DM candidate is plementary to LHC experiments in distinguishing the be- spin-0, which gives a vanishing SD cross section. yond the standard model physics scenarios [135]. TheDifferent SI cross section is generally small,Responses below Favor We summarize below the model predictions for the DM 10 8 pb, and occurs through Higgs boson ex- À cross sections; the posterior distributions are summarized change. If SD scattering of DM is observed, the in Fig.SI12. vs. SD (vs. nuclearclass physics/EFT of models with spin-0 DM would be imme-) (i) In mSUGRA, the FP region provides the largest SI diately excludedDifferent as being the sole origin of the DM Elements! • andSpin-independent SD cross sections. This searches is due to the have mixed receivedin the the Universe. most attention Fluorine’s (iv) For stable Dirac neutrino DM, the Z boson domi- Higgsino nature of the2 lightest neutralino; the neu- tralinodue couplings to the to A the rate Higgs enhancement and Z bosons are large. (>15,000nates for inxenon) the calculation Protons of both the relic densitygreat and for Neutrons The Bino nature of the lightest neutralino in the CA elastic scattering cross section and makesproton the SI and spin! and AF regions causes these scenarios to have sub- SD cross sections tightly correlated and large. PICO-250L DOE G2 Proposal Mp Sp'' Sp' Fp'' Fp' 6Dp (SDp) Mn Sn'' Sn' Fn'' Fn' Dn • stantiallyBut the smaller actual cross mechanism sections. is unknown,(v) In mUED, a sweet spot of SD O 10À pb exists for which the DM relic density is reproduced.ð Þ The environment(ii) called The for tadpole it (e.g., to nMSSM follow up a signal model in a haslarge largexenon detector). SD scattering, In addition, the use of F multiple mass targets will be needed to characterize the WIMP mass and velocity distribution in the event F and may3 be more complicated relic density is strongly dependent on the curvature1.0 1.0 of a discovery of[21][22] order. 10À pb, and a wide range of SI cross parameter and fixes its value. The KK quarks have SI: scaling w/ mass SI vs. SD approximatelyNa the same mass as the inverse curva-0.8 Na 0.8 mSUGRA UED ture and the SD cross section is thus closely tied to LHT 2 0.6 0.6

− nMSSM RHN theSi relic density. On the other hand, the SI cross Si section is more dispersed due to the larger variation 0.4 0.4 4

− of the Higgs boson mass. (pb) Ge Ge SD (vi) In the LHT model, the SD interaction occurs0.2 0.2 σ 10

6 through T-odd quarks which have a small hyper- − log charge.I Therefore, the SD cross section in this0 I 0 ---640246 ---640246

The strengths of the bubble chamber technology for dark matter searches can be summarized as follows:

5 SPIN DEPENDENCE OF DARK MATTER SCATTERING PHYSICAL REVIEW D 78, 056007 (2008) IceCube muon rate IceCube signal significance

mSUGRA mSUGRA UED UED LHT LHT xSM xSM RHN Discovery RHN

Evidence µ σ N 10 1 2 10 1 2 10 10 − − log log 2 2 − − 3 3 − − 4 4 − − 0 200 400 600 800 1000 0 200 400 600 800 1000

MDM MDM

FIG. 11 (color online). The expected numbers of events and the statistical significance of DM signals for 3 and 5 years running of the IceCube neutrino telescope. Only the FP region in mSUGRA and a portion of the mUED parameter space give a significance >5 for 5 years of running. oughly scan the parameter spaces, we adopted the section. This is a consequence of the DM annihila- Bayesian method that is the foundation of the Markov tion occurring through the Z boson. To counter the chain Monte Carlo approach. The DM models can have small annihilation rate in the early universe (due to distinct phenomenological predictions. We showed that the the small neutralino mass in the model), the neu- DM model possibilities can be narrowed by measurements tralino pair is required to have a larger Z boson of both SI and SD elastic scattering. The direct signals for coupling, resulting in a large SD cross section. DM in recoil and neutrino telescope experiments are com- (iii) In the singlet extended SM, the DM candidate is plementary to LHC experiments in distinguishing the be- spin-0, which gives a vanishing SD cross section. yond the standard model physics scenarios [135]. TheDifferent SI cross section is generally small,Responses below Favor We summarize below the model predictions for the DM 10 8 pb, and occurs through Higgs boson ex- À cross sections; the posterior distributions are summarized change. If SD scattering of DM is observed, the in Fig.SI12. vs. SD (vs. nuclearclass physics/EFT of models with spin-0 DM would be imme-) (i) In mSUGRA, the FP region provides the largest SI diately excludedDifferent as being the sole origin of the DM Elements! • andSpin-independent SD cross sections. This searches is due to the have mixed receivedin the the Universe. most attention Fluorine’s (iv) For stable Dirac neutrino DM, the Z boson domi- Higgsino nature of the2 lightest neutralino; the neu- tralinodue couplings to the to A the rate Higgs enhancement and Z bosons are large. (>15,000nates for inxenon) the calculation Protons of both the relic densitygreat and for Neutrons The Bino nature of the lightest neutralino in the CA elastic scattering cross section and makesproton the SI and spin! and AF regions causes these scenarios to have sub- SD cross sections tightly correlated and large. PICO-250L DOE G2 Proposal Mp Sp'' Sp' Fp'' Fp' 6Dp (SDp) Mn Sn'' Sn' Fn'' Fn' Dn • stantiallyBut the smaller actual cross mechanism sections. is unknown,(v) In mUED, a sweet spot of SD O 10À pb exists for which the DM relic density is reproduced.ð Þ The environment(ii) called The for tadpole it (e.g., to nMSSM follow up a signal model in a haslarge largexenon detector). SD scattering, In addition, the use of F multiple mass targets will be needed to characterize the WIMP mass and velocity distribution in the event F and may3 be more complicated relic density is strongly dependent on the curvature1.0 1.0 of a discovery of[21][22] order. 10À pb, and a wide range of SI cross parameter and fixes its value. The KK quarks have SI: scaling w/ mass SI vs. SD approximatelyNa the same mass as the inverse curva-0.8 Na 0.8 mSUGRA UED ture and the SD cross section is thus closely tied to LHT 2 0.6 0.6

− nMSSM RHN theSi relic density. On the other hand, the SI cross Si section is more dispersed due to the larger variation 0.4 0.4 4

− of the Higgs boson mass. (pb) Ge Ge SD (vi) In the LHT model, the SD interaction occurs0.2 0.2 σ 10

6 through T-odd quarks which have a small hyper- − log charge.I Therefore, the SD cross section in this0 I 0 ---640246 ---640246

8 model is typically very small. In contrast, the SI − scatteringXe proceeds through the Higgs and T-odd Xe quarks, giving experimentally accessibleIodine: SI nuclear cross 10 − section values. angular momentum −12 −11 −10 −9 −8 −7 −6 We provide a summary of the SI and SD cross sections log10σSI (pb)

The strengths of the bubble chamber technology for dark matter searches can be summarized as follows:

5 Why is SD interesting? SD WIMP-proton/-neutron combined plot • PICO-500’s SD-proton sensitivity is very similar to LZ’s SD-neutron sensitivity

• So: given typical SD/SI cross- section ratios, a ﬁrst discovery by PICO-500 is plausible

• The CEνNS ﬂoor is much lower for F than for Xe: we actually want to minimize SI sensitivity in order to maximize SD reach

8 Complementarity w/ xenon, colliders 5 *C. Amole et al., Phys. Rev. Lett. 118, 251301 (2017) 0.2 1000 LUX ] 90% CL 2 PICO-60 C F • A more model-independent Panda-X II 3 8 0.1 800 CMS DM+J/V PICO-60 C3F8 Axial-vector mediator, Dirac DM comparison: bubble chamber g =0.25 , g =1 600 q DM p

(ﬂuorine) results constrain effective a 0 400 PICO-2L PICO-60 CF3I proton coupling in a complementary -0.1 way to xenon TPC constraints on 50 GeV/c2 200

-0.2 WIMP mass [GeV/c effective neutron coupling -0.4 -0.2 0 0.2 0.4 5 0 0 500 1000 1500 2000 a 2

] n Mediator mass [GeV/c ] 2 -38 10 104 2 CMS DM+J/V m − m • Complementarity with LHC: FIG. 5.Axial-vector PICO-60 mediator, constraints Dirac (blue)DM on the effective spin- FIG. 6. Exclusion limits at 95% C.L. in the DM med g =0.25 , g =1 CMS DM+photon a dependent3 q WIMP-proton DM and WIMP-neutronPICO-60 couplings, C F p plane. PICO-60 constraints (blue) are compared against col- 10 2 38 limit from simplified collider and an,fora50GeV/c WIMP mass. Also shown are results lider constraints from CMS (red) [32] for an axial-vector me- from PANDAX-II (cyan) [34], LUX (yellow) [39], PICO-2L diator using the monojet and mono-V channels. production model for CMS, 2 -40 (purple)10 [9], and PICO-60 C3FI (red) [10]. 10 following recommendations of LHC Dark Matter Working Group 101 tance and useful discussion. We wish to acknowledge but to interpret LHC searches, one must assume a spe- the support of the National Sciences and Engineering WIMP mass [GeV/c ] cific0 model to generate the signal that is then looked for -42 10 1 2 3 Research Council of Canada (NSERC) and the Canada 10 in the data.10 This can make10 it difficult to compare10 LHC 2 Foundation for Innovation (CFI) for funding. We ac- PICO-60 C F results9 with directMediator detection mass experiments, [GeV/c ] as the latter 3 8 knowledge the support from National Science Foundation tend to be more general. The LHC Dark Matter Work- (NSF) (Grant 0919526, 1506337, 1242637 and 1205987). FIG. 6. Exclusion limits at 95% C.L. in the mDM − m ing Group (LHCDMWG) has made recommendations on medWe acknowledge that this work is supported by the U.S. -44 plane. PICO-60 constraints (thick blue) are compared against 10 a set of simplified models to be used in LHC searches Department of Energy (DOE) Office of Science, Office of

SI WIMP-nucleon cross section [cm 0 1 2 10 10 10 colliderand constraints the best way from to present CMS such for an results axial-vector [29–31]. mediatorFor a High Energy Physics (under award DE-SC-0012161), by 2 usinggiven the monojet/mono-V simplified model involving (red) [33] a mediator and mono-photon exchanged (or- WIMP mass [GeV/c ] the DOE Office of Science Graduate Student Research s ange)via [34] the channels.-channel, The there shaded are four regions free signify parameters: excluded the pa-(SCGSR) award, by DGAPA-UNAM through grant PA- m m rameterdark space matter for mass the chosenDM, the model. mediator A similarmass med analysis,the byPIIT No. IA100316 and CONACyT (Mexico) through FIG. 4. The 90% C.L. limit on the SI WIMP-nucleon cross- g ATLASuniversal can be mediator found in coupling [52]. to quarks q,andtheme- grant No. 252167, by the Department of Atomic Energy section from PICO-60 C3F8 plotted in thick blue, along g diator coupling to dark matter DM.TheLHCDMWG (DAE), the Government of India, under the Center of with limits from PICO-60 CF3I(thickred)[10],PICO-2L recommends that results of simplified model searches be (thick purple) [9], LUX (yellow) [44], PandaX-II (cyan) [45], AstroParticle Physics II project (CAPP-II) at SAHA In- directpresented comparison by plotting of the confidence sensitivity level limits of PICO as a function to that ofstitute of nuclear Physics (SINP), the Czech Ministry CRESST-II (magenta) [46], and CDMS-lite (black) [47]. of the two mass parameters m and m for a fixed CMS [33, 34] by applying our resultsDM to themed specific case ofof Education, Youth and Sports (Grant LM2015072) While we choose to highlight this result, LUX sets the set of couplings gq and gDM. Here, we follow the example strongest limits on WIMP masses greater than 6 GeV/c2.Ad- a simplified dark matter model involving an axial-vectorand the the Spanish Ministerio de Econom´ıay Competi- set by the LHCDMWG to make a direct comparison of tividad, Consolider MultiDark (Grant CSD2009-00064). ditional limits, not shown for clarity, are set by PICASSO [14], s-channelthe sensitivity mediator. of PICO Following to that Eq. of CMS 4.7-4.10 [32] by of applying Ref. [32], XENON100 [41], DarkSide-50 [48], SuperCDMS [49], CDMS- This work is partially supported by the Kavli Institute we findour an results expression to the specific for the case spin-dependent of a simplified crossdark mat- sectionfor Cosmological Physics at the University of Chicago II [50], and Edelweiss-III [51]. s as ater function model involving of those freean axial-vector parameters,-channel and we mediator. invert thisthrough NSF grant 1125897, and an endowment from the Following Eq. 4.7-4.10 of Ref. [31], we find an expres- expression to find mmed as a function of cross section.Kavli Foundation and its founder Fred Kavli. We also sion for the spin-dependent cross section as a function of For this comparison, we assume gq =0.25 and gDM =1.wish to acknowledge the support from Fermi National 0.2 those free parameters, and we invert this expression to Accelerator Laboratory under Contract No. De-AC02- 2 With thism simple translation onto the specified model, we 50 GeV/c find med as a function of cross section. For this com- 07CH11359, and Pacific Northwest National Laboratory, can plot our limits ong the same. mDMg − mmed plane, and 0.1 parison, we assume q =025 and DM = 1. With this which is operated by Battelle for the U.S. Department of the results are shown in Fig. 6. simple translation, we can plot our limits on the same Energy under Contract No. DE-AC05-76RL01830. p mDM − mmed plane, and the results are shown in Fig. 6. a 0 The PICO collaboration wishes to thank SNOLAB and its staTheff for PICO support collaboration through wishes underground to thank SNOLAB space, logisti-and -0.1 cal andits sta technicalff for support services. through SNOLAB underground operations space, logisti- are sup- cal and technical services. SNOLAB operations are supported by the Canada Foundation for Innovation and ∗ -0.2 ported by the Canada Foundation for Innovation and the [email protected] the Province of Ontario Ministry of Research and In- † -0.4 -0.2 0 0.2 0.4 Province of Ontario Ministry of Research and Innova- [email protected] novation, with underground access provided by Vale at [1] K.A. Olive et al. (Particle Data Group), Chinese Phys. a tion, with underground access provided by Vale at the 38 n the Creighton mine site. We are grateful to Kristian C ,090001(2014). Creighton mine site. We are grateful to Kristian Hahn [2] E. Komatsu et al.,Astroph.J.Suppl.180,330(2009) Hahn and Stanislava Sevova of Northwestern University FIG. 5. PICO-60 constraints (blue) on the effective spin- and Stanislava Sevova of Northwestern University and and refs. therein. andBjörn Björn Penning Penning of the of theUniversity University of Bristol of Bristol for their for assis- their [3] G. Jungman, M. Kamionkowski, and K. Griest, Phys. dependent WIMP-proton and WIMP-neutron couplings, ap 2 and an,fora50GeV/c WIMP mass. Parameter space out- assistance and useful discussion. We wish to acknowl- side of the band is excluded. Also shown are results from edge the support of the Natural Sciences and Engineering PANDAX-II (cyan) [35], LUX (yellow) [40], PICO-2L (pur- Research Council of Canada (NSERC) and the Canada ple) [9], and PICO-60 C3FI (red) [10]. Foundation for Innovation (CFI) for funding. We acknowledge the support from National Science Foundation (NSF) (Grant 0919526, 1506337, 1242637 and 1205987). detection constraints in the parameter space of Fig. 3. We acknowledge that this work is supported by the U.S. We choose instead to compare our limits with those of Department of Energy (DOE) Office of Science, Office of the LHC on the chosen model, as shown in Fig. 6. The High Energy Physics (under award DE-SC-0012161), by LHC Dark Matter Working Group has made recommen- the DOE Office of Science Graduate Student Research dations on a set of simplified models to be used in LHC (SCGSR) award, by DGAPA-UNAM through grant PA- searches and the best way to present such results [30– PIIT No. IA100316 and CONACyT (Mexico) through 32]. For a given simplified model involving a mediator grant No. 252167, by the Department of Atomic Energy exchanged via the s-channel, there are four free param- (DAE), the Government of India, under the Center of eters: the dark matter mass mDM, the mediator mass AstroParticle Physics II project (CAPP-II) at SAHA In- mmed, the universal mediator coupling to quarks gq,and stitute of nuclear Physics (SINP), the Czech Ministry the mediator coupling to dark matter gDM.Wemakea of Education, Youth and Sports (Grant LM2015072) Why bubble chambers?

19 • High density of F means great SDp sensitivity

• Intrinsic rejection of electron recoil backgrounds

• Low energy recoil sensitivity (< 5.5 keV)

• Large, monolithic (self-shielding) target mass – ton-scale next generation

• Multiple target nuclei: ability to test scattering rate dependence on atomic number, nuclear spin, etc.

• Disadvantages: no measurement of recoil energy; threshold calibrations may be difﬁcult; recompression dead-time requires very low overall event rate

10 Target: superheated ﬂuid Lower pressure in target liquid until it is in metastable superheated state Energy deposition nucleates small bubble that grows to visible size

Cameras watch for visible bubble and issue the primary trigger

compressed

expanded

superheated

11 (plots by Eric Dahl) Gamma rejection

Set temp. & pressure for sensitivity to nuclear recoils (α, n, nuclei, WIMPs), and insensitivity to electron recoils (γ/β) [protobubble immediately collapses] 12 Gamma rejection

-2 Gamma Rejection by Chamber

10 Various PICO Detectors PICO-0.1 10-3 U Chicago COUPP-1L C3F8 10-4 Queen's CF I COUPP-4 -5 3 PICO-2L 10 PICO-60 10-6

10-7

10-8

10-9 PICO-60

Probability of Nucleation 10-10 We can set the threshold 10-11 Aim for this level for 1-year exposure on dE/dx (ET/rc) 10-12 0 2 4 6 8 10 12 Threshold (keV)

(Dan Baxter, Conference on Science at SURF, May 14, 2017) 13 4 ] -1 0 1 2 3 4 2 10-37 60 Neutron 40 WIMP search 10-38 20 Counts

Acoustic discrimination0 -39 4 1 10 ] -1 0 1 2 3 4 2 10-37 • Acoustic discrimination against alphas discovered 60 by PICASSO 0.5 Neutron -40 40 WIMP search 10 (Aubin et al., New J. Phys.10:103017, 2008) 10-38

NN score 20 PICO-60 C F 0 Counts 3 8 • -1 0 1 2 3 4 -41 Alphas deposit their energy over tens of μm 0 -3910 10 SD WIMP-proton cross section [cm 1 log(AP) 101 102 103 • Nuclear recoils deposit energy over tens of nm WIMP mass [GeV/c2] 0.5 252 FIG. 2. Top: AP distributions for AmBe and Cf neu- 10-40 tron calibration data (black) and WIMP search data (red) at • In PICO, alphas are several times louder than recoils NN score FIG. 3.PICO-60 The 90% C F C.L. limit on the SD WIMP-proton cross 3.3 keV threshold. Bottom: AP and NN score for the same 3 8 0 section from PICO-60 C F plotted in thick blue, along -1 0 1 2 3 4 3 8 dataset. The acceptance region for nuclear recoil candidates, 10-41 SD WIMP-proton cross section [cm with limits from PICO-60 CF3I (thick red) [10], PICO-2L • For a WIMP-search run, the acoustic signals aredefined before WIMP search acousticlog(AP) data unmasking using 101 102 103 (thick purple) [9], PICASSO (green band) [14], SIMPLE (or- neutron calibration data, are displayed with dashed lines and WIMP mass [GeV/c2] blinded in order to set an unbiased cut on this 252 ange) [15], PandaX-II (cyan) [35], IceCube (dashed and dot- reveal noFIG. candidate 2. Top: events AP distributions in the WIMP for AmBe search and data.Cf Alphas neu- “acoustic parameter” (“AP”) tron calibration data (black) and WIMP search data (red) at ted pink) [36], and SuperK (dashed and dotted black) [37, 38]. 222 FIG. 3. The 90% C.L. limit on the SD WIMP-proton cross from the3.3 keVRn threshold. decay chain Bottom: can be AP identified and NN score by their for the time same sig- neutrons alphas sectionThe from indirect PICO-60 limits C F fromplotted IceCube in thick and blue, SuperK along assume anni- nature anddataset. populate The acceptance the two region peaks for in nuclear the WIMP recoil candidates, search data 3 8 Observable bubble ~mm withhilation limits from to ⌧ PICO-60leptons CF (dashed)I (thick red) and [10],b quarks PICO-2L (dotted). The defined before WIMP search acoustic data214 unmasking using 3 at high AP. Higher energy alphas from Po are producing (thick purple) [9], PICASSO (green band) [14], SIMPLE (or- neutron calibration data, are displayed with dashed lines and purple region represents parameter space of the constrained ~40 μmlarger acoustic signals. ange) [15], PandaX-II (cyan) [35], IceCube (dashed and dot- reveal no candidate events in the WIMP search data. Alphas minimal supersymmetric model of [39]. Additional limits, not ted pink) [36], and SuperK (dashed and dotted black) [37, 38]. from the 222Rn decay chain can be identified by their time sig- shown for clarity, are set by LUX [40] and XENON100 [41] The indirect limits from IceCube and SuperK assume anni- nature and populate the two peaks in the WIMP search data ~50 nm hilation(comparable to ⌧ leptons to (dashed) PandaX-II) and b andquarks by ANTARES (dotted). The [42, 43] (com- at high AP. Higher energy alphas from 214Po are producing purpleparable region to represents IceCube). parameter space of the constrained to be 0.25larger acoustic0.09 (0.96 signals. 0.34) single(multiple)-bubble minimal supersymmetric model of [39]. Additional limits, not events. PICO-60± was exposed± to a 1 mCi 133Ba source Daughter heavy nucleus Helium nucleus Multiple radiating bubbles shown for clarity, are set by LUX [40] and XENON100 [41] (~100 keV) (~5both MeV) before and after the WIMP search data, which, (comparable to PandaX-II) and by ANTARES [42, 43] (com- comparedto be against 0.25 a0.09 Geant4 (0.96 [21]0.34) Monte single(multiple)-bubble Carlo simulation, parableWIMP to IceCube). mass, are shown in Fig. 3 and 4. These limits, ± ± gives aevents. measured PICO-60 nucleation was exposed eciency to a 1 for mCi electron133Ba source recoil corresponding to an upper limit on the spin-dependent both before and after the WIMP search data,10 which, 41 2 events above 3.3 keV of (1.80 0.38) 10 . Combin- WIMP-proton cross section of 3.4 10 cm for a compared against a Geant4 [21]± Monte⇥ Carlo simulation, WIMP mass, are2 shown in Fig. 3 and 4. These⇥ limits, ing thisgives with a measured a Monte nucleation Carlo simulation eciency for of electron the external recoil corresponding30 GeV c toWIMP, an upper are limit currently on the spin-dependent the world-leading con- 10 41 2 gammaevents flux from above [16, 3.3 22], keV we of (1.80 predict0.38) 0.02610 0.007. Combin- events WIMP-protonstraints in cross the WIMP-proton section of 3.4 spin-dependent10 cm for a sector and ± ⇥ ± 2 ⇥ due to electroning this with recoils a Monte in the Carlo WIMP simulation search of exposure. the external The 30 GeVindicate c WIMP, an improved are currently sensitivity the world-leading to the dark con- matter signal backgroundgamma from flux from coherent [16, 22], scattering we predict 0.026 of 8B solar0.007 events neutri- straintsof a in factor the WIMP-proton of 17, compared spin-dependent to previously sector andreported PICO ± indicate an improved sensitivity to the dark matter signal nos is calculateddue to electron to recoils be 0.055 in the WIMP0.007 search events. exposure. The results. background from coherent± scattering of 8B solar neutri- of a factor of 17, compared to previously reported PICO We usenos isthe calculated same shapes to be 0.055 of the0.007 nucleation events. eciency results.A comparison of our proton-only SD limits with ± curves forWe fluorine use the and same carbon shapes nuclear of the nucleation recoils as e foundciency in Aneutron-only comparison of SD our limits proton-only set by SD other limits dark with matter search Ref. [8],curves rescaled for fluorine upwards and in carbon recoil nuclear energy recoils to asaccount found in for neutron-onlyexperiments SD limits is achieved set by other by dark setting matter constraints search on the the 2%Ref. di↵erence [8], rescaled in thermodynamic upwards in recoil threshold. energy to account We adopt for experimentse↵ective is spin-dependent achieved by setting WIMP-neutron constraints on the and WIMP- the 2% di↵erence in thermodynamic threshold. We adopt e↵ective spin-dependent WIMP-neutron and WIMP- proton couplings an and ap that are calculated according the standard halo parametrization [23], with the follow- proton couplings an and ap that are calculated according the standard halo parametrization2 [23],3 with the follow- 2 3 to the method proposed in Ref. [29]. The expectation ing parameters:ing parameters:⇢D=0.3⇢D=0.3 GeV GeV c ccmcm, ,vvescesc = 544 544 km/s, km/s, to the method proposed in Ref. [29]. The expectation values for the proton and neutron spins19 for the 19Fnu- vEarth =vEarth 232= km/s, 232 km/s, and andvo =vo 220= 220 km/s. km/s. We We use the the e↵ eec-↵ec- values for the proton and neutron spins for the Fnu- tive field theory treatment and nuclear form factors de- cleuscleus are taken are taken from Ref. from [24]. Ref. The [24]. allowed The region allowed in region in tive field theory treatment and nuclear form factors de- 2 the a a plane is shown for a 50 GeV c WIMP in 2 scribedscribed in Refs. in Refs. [24–27] [24–27] to determine to determine sensitivity sensitivity to to both both then anp ap plane is shown for a 50 GeV c WIMP in spin-dependent and spin-independent dark matter inter- Fig.Fig. 5. We 5. find We that find PICO-60 that PICO-60 C3F8 improves C F theimproves con- the con- spin-dependent and spin-independent dark matter inter- straints on a and a , in complementarity3 with8 other actions. For the SI case, we use the M response of Table straintsn on a pand a , in complementarity with other actions.1 For in Ref. the [24], SI case, and for we SD use interactions, the M response we use the of Table sum dark matter searchn experimentsp that are more sensitive dark matter search experiments that are more sensitive 1 in Ref.of the [24],⌃0 andand for⌃00 terms SD interactions, from the same we table. use the To im- sum to the WIMP-neutron coupling. of the ⌃plement0 and these⌃00 terms interactions from and the form same factors, table. we use To the im- Theto the LHC WIMP-neutron has significant sensitivity coupling. to dark matter, plementpublicly these available interactionsdmdd code and package form factors, [27, 28]. Thewe use calcu- the but toThe interpret LHC LHC has searches, significant one must sensitivity assume a to spe- dark matter, publiclylated available Poissondmdd uppercode limits package at the 90% [27, C.L. 28]. for The the spin- calcu- cificbut model to to interpret generate LHCthe signal searches, that is then one looked must for assume a spe- dependent WIMP-proton and spin-independent WIMP- in the data. Despite this subtlety, the convention has lated Poissonnucleon upper elastic scattering limits at cross-sections, the 90% C.L. as a for function the spin- of beencific to show model LHC to limits generate alongside the signal more general that is direct then looked for dependent WIMP-proton and spin-independent WIMP- in the data. Despite this subtlety, the convention has nucleon elastic scattering cross-sections, as a function of been to show LHC limits alongside more general direct Neutron background

• Single-scatter neutrons are indistinguishable from WIMPs in these detectors

• Can’t discriminate against them, so minimize them

• Two neutron sources for PICO-60:

• Cosmogenic: spallation in rock near detector by high energy cosmic

ray muons (veto present for C3F8 Run-1, saw no muons)

• Radiogenic: natural radioactivity in rock and detector apparatus (alpha-n and spontaneous ﬁssion)

• Total neutron background estimate for PICO-60 C3F8 Run-1: 0.25 ± 0.09 (0.96 ± 0.34) single- (multiple)-bubble events

15 Cameras as a “neutron veto” Multiply-scattering neutrons won’t be mistaken for WIMPs (3:1)

Four views of a neutron event from an AmBe source

16 Cameras as a “neutron veto” Multiply-scattering neutrons won’t be mistaken for WIMPs (3:1)

Four views26 bubbles of a neutron in the event small from 2L chamber! an AmBe source

16 Backgrounds checklist

• Gammas/betas:

• dE/dx threshold in superheated detectors affords “intrinsic” -11 rejection ~10 for typical PICO energy thresholds in C3F8

• Alpha decays:

• large acoustic signals allow discrimination at >99.4% (stats. limited)

• Neutrons:

• reject multiple scatters visually, veto detector-adjacent cosmogenics, minimize other sources (extensive material screening, shielding)

17 The PICO program • PICO – 2012 merger of the PICASSO and COUPP collaborations

• Small surface test chambers at Université de Montréal, Queen’s University, Northwestern, Drexel, NEIU (for threshold calibration, etc.)

• PICO-2L C3F8 (2014-17) C. Amole et al., PRL 114, 231302 (2015) PICO-2L COUPP-60 → PICO-60 C. Amole et al., PRD 93, 061101 (2016) C3F8 CF3I, C3F8

• PICO-60 CF3I (2013) C. Amole et al., PRD 93, 061101 (2016)

PICO-60 C3F8 (2016-17) PICO C. Amole et al., PRL 118, 251301 (2017)

• PICO-40L (2017-19)

• PICO-500 (~2018+) 18 The PICO-60 detector

• Deployed 2 km underground at SNOLAB

• C3F8 target: 52 kg total (45.7 ± 0.5 kg ﬁducial, 87.7%)

• Synthetic fused silica inner vessel, stainless steel pressure vessel, water tank, muon veto

• Bellows allow expansion to superheated state with typical per-event cycle of 800s, >80% live-fraction

• Four cameras monitor for bubble nucleation using LED illumination

• Eight piezoelectric acoustic sensors monitor sound of bubble nucleation

19 Event cycle

Live-time begins

26 Trigger

0 20 40 60 80

(plot by Dan Baxter) Expand to target pressure; begin counting live-time after 25s stability

Primary trigger: changes in image information content (bubble appearance)

Time-out trigger set to 2000s – regular cycling improves detector stability 20 PICO-60 Camera Trigger

Scott Fallows, Pitam Mitra December 15, 2016

Abstract [note: this is not a general abstract yet, discusses masked trigger only] The ﬁrst 5s ⇡ after initiating expansion to 30 psi have enough visible motion to trigger the cameras. With elevated rates during calibration, this 5s period without a live video trigger is long enough for the cameras to regularly miss bubbles, leading to reduced stability. By masking oﬀ the moving parts of the images, we should be able to keep the camera trigger live for the entire expansion.

1CameraTriggerOverview

Two cameras on each of two “cam-slave” servers monitor their view of the chamber, acquiring frames at 200 Hz. On each camera, every even-numbered frame is compared to the previous even-numbered frame to check for bubblesFast or other visualcamera events at 100 Hz.trigger The quantity used to define the trigger is the image entropy SI [1]oftheabsolutedifference −3 Diff. entropy | 20161106_1/9 | Cam 0 of the two frames under comparison,• Primary where trigger: “image entropy” ×10 16 Image Entropy Present trigger SI = Pi log Pi. (1) 2 14 i X 12 • Calculate absolute difference of Entropy of diff. frames typical single An image histogram is constructed from the difference-images with 16 intensity bins running successive frames, searching for 10 bubble event over the full 8-bit dynamic rangechanges from 0 in to information 255. The content values Pi are the probabilities for the intensity of any pixel in the image to fall into intensity bin i,sothehistogramisnormalizedby8 the number of entries (pixels). The• Images quantity initiallySI acquiredfor this diat ff200-image Hz is6 then calculated by running the sum in Eq. 1 over the 16 bins of the normalized image histogram. – increased to hardware maximum 4

The quantity SI is then compared340 Hz to for each low threshold camera’s run predeﬁned – fast threshold, tunedtrigger to be just high enough to avoid generating spurioustrigger ensures triggers stable on operations pixel noise: 2 frame

at very low pressures 0 30 35 40 45 50 55 60 65 70 Camera Diff-entropy threshold Frame Number (time) 0 0.00035 21 Figure 2: The rise in differenced-frame image entropy for a typical bubble, exceeding threshold 1 0.0003 on frame 48. Frame 50 indicates the time that the master trigger response from the main DAQ 2 0.0003 was received. 3 0.0003 3.1 Motivation The first 5.25 s after initiating expansion to 30 psi have enough visible motion to trigger ⇡ the cameras, so the video trigger is not live during this period. With elevated rates during If any diff-image’s SI exceeds that camera’s threshold,calibration, the cameraDAQ this 5.25 s period is long code enough sets for the the cameras video to regularly miss bubbles, leading to reduced stability. By masking off the moving parts of the images, we intend to keep the trigger TTL line True for that camera server (conversely,camera if trigger neither live for the camera’s entire expansion. threshold is ex- ceeded, it is kept False, or switched to False if the previous state was True). The main DAQ responds to this by issuing a master trigger, compressing4Maskdefinition the chamber and signaling that the cameraDAQ should stop image acquisition and begin savingMasks are defined images to prevent from triggering the on ring parts that buff moveer during and expansion, while still catching other outputs to disk. all bubbles with minimal delay, ideally 10 ms or less. 4.1 Mask generation The masks are generated in a multi-step process using OpenCV: 1 1. Absolute difference images from fully-compressed 190 psi state and fully-expanded 30 psi state.

2. Maximize dynamic range, etc.

3. Shift gamma

3 PICO-60 Run-1: blinded 2 • Following “pre-physics” background and calibration data, detector Candidates pre-acoustic cut erated at SNOLAB in Sudbury,performance Ontario, Canada. was assessed Here good 600 we report results from the firstenough run to of allow PICO-60 a blind analysis with C3F8, with an efficiency-corrected dark matter exposure 500 of 1167 kg-days, taken between• Acquired November acoustically 2016 and blinded Jan- 400 uary 2017. background data from 28 Nov 2016 The PICO Collaboration previouslyto 13 Jan reported2017 (no power the ob- outages, 300 remarkably stable running) servation of anomalous background events in dark matter 200 Z [mm] search data with the 2-liter PICO-2L C3F8 [8] and the 18- • Saw 106 bulk singles in WIMP-search 100 liter PICO-60 CF3I [10] bubble chambers.dataset: consistent Improvements with Rn decay rate in fluid handling and bubble chamberseen in unblinded operation pre-physics elimi- data 0 nated this anomalous background in a second run of the -100 PICO-2L detector [9]. A leading• Saw hypothesis 3 multiples for, so the given cause 3:1 multiples- 0 50 100 150 200 to-singles ratio from n calibration and of these background events is bubble nucleation due to R2/R [mm] surface tension effects introducedsimulation, by the expected contamination 0-3 bulk n singles jar of the active target with particulate matter and water 22 droplets [15]. The PICO-60 detector was recommissioned FIG. 1. Spatial distribution of bubble events in the WIMP following a rigorous cleaning procedure targeting partic- search data. Z is the reconstructed vertical position of the 2 ulate contamination. Every component was cleaned to bubble, and R /Rjar is the distance from the center axis MIL-STD-1246 Level 50 standard [16] prior to assembly, squared, normalized by the nominal jar radius (145 mm). The fiducial cut is represented by the dashed line. Red squares are and samples of the water buffer were taken using an in situ the 106 events in the fiducial bulk volume passing all cuts and filtration system during commissioning to monitor grey dots are all other single-bubble events. particulate injection. A final measurement after C3F8 distillation confirmed that the total assembly met MIL- STD-1246 Level 100, after which the inner volume was after a pressure-rise trigger. In the WIMP search dataset, closed. the chamber was expanded for 34.3 of the 44.6 days that The PICO-60 apparatus was described in Ref. [10], and the detector was operational. here we restrict ourselves to describing subsequent im- The WIMP search dataset was taken at 30.2 ± 0.3 psi provements and changes. A new seal design was deployed and 13.9 ± 0.1◦C, corresponding to a thermodynamic between the silica jar and the stainless steel bellows to threshold of 3.29 ± 0.09 keV, the calculation of which minimize particulate generation, replacing the gold wire is detailed in Ref. [10]. There is an additional 0.2 keV seal described in Ref. [10] with a PTFE gasket. The uncertainty in the threshold due to the thermodynamic C3F8 target does not require the addition of chemicals properties of C3F8 [18]. In situ nuclear and electronic re- to remove free ions, unlike CF3I. While the same water coil calibrations were performed by exposing the chamber tank is used, a new chiller system holds the temperature to AmBe and 252Cf neutron sources and a 133Ba gamma in the water tank uniform to approximately 0.1◦C. The source both before and after the WIMP search run. Pre- target volume was more than doubled, requiring a corre- physics background data were taken during commission- sponding increase from two to four cameras (in two ver- ing to measure the alpha backgrounds due to 222Rn chain tical columns). Eight piezoelectric acoustic transducers decays. For the WIMP search run, we performed a blind identical to those used in Ref. [9] were attached, evenly- analysis by masking the acoustic information that allows spaced around the outside of the silica jar, using a spring the discrimination between alpha decays and nuclear re- loaded HDPE ring. Five sensors failed during commis- coils. This information was processed only after cuts and sioning, leaving three operable sensors for the duration efficiencies for single bulk nuclear recoil candidates were of the experiment. set, using source calibrations and pre-physics background The chamber expansion cycle was similar to that em- data. ployed in the previous run [10]. First, the chamber is For the WIMP search dataset, periods of unstable op- expanded to a predetermined pressure, superheating the eration are removed, these being defined as times within C3F8 active liquid. Following a trigger, the hydraulic one hour of radioactive source transport near the detec- system initiates a fast compression, raising the pressure tor or in a 24-hour window following any significant in- above 150 psia in roughly 100 ms. The primary trigger terruption to operation. The first 25 s of every expansion uses the change in entropy between two consecutive cam- is discarded to remove transient effects. Of the 34.3 days era images [17] to detect the appearance of a gas bubble the detector was expanded, 30.0 live-days (87.4%) are in the chamber. A trigger is also sent if a rise in pressure considered in the WIMP search. is detected or when the chamber has been expanded for Bubble images are identified using the same entropy 2000 s. The chamber begins a new expansion after the algorithm as used for the optical trigger. The pixel co- chamber has been compressed for 100 s. A long compres- ordinates are then reconstructed into spatial coordinates sion of 600 s is imposed on every tenth compression or using ray propagation in a simulated optical geometry. 4 ] -1 0 1 2 3 4 2 10-37 60 Neutron 40 WIMP search 10-38 20 Counts 4 0 -39

PICO-60 Run-1: unblinding ] 10

1 2 -1 0 1 2 3 4 10-37 60 Neutron 0.5 -40 • 30 live-day run at 3.3 keV threshold, 40 WIMP search 10 10-38

NN score 0 candidates 20 PICO-60 C F

published in PRL*: a background-free Counts 0 3 8 -1 0 1 2 3 4 -41 1167 kg-day WIMP-search exposure 0 4 10-39 1 log(AP) 10SD WIMP-proton cross section [cm 101 102 103

] 2 -1 0 1 2 3 4 2 -37 WIMP mass [GeV/c ] • Factor of 17 improvement in upper FIG.10 2. Top: AP distributions for AmBe and 252Cf neu- 60 0.5 tron calibration data (black) and WIMP search data (red) at 10-40 limit on spin-dependentNeutron FIG. 3. The 90% C.L. limit on the SD WIMP-proton cross

3.3 keVNN score threshold. Bottom: APPICASSO and NN score for the same 40 WIMP search PICO-2L C section fromPICO-60 PICO-60 C F C3F8 plotted in thick blue, along WIMP-proton cross-section dataset.10-380 The acceptance region for nuclear recoil candidates, 3 8 -1 0 1 2 3 4 with-41 limits from PICO-60 CF3I (thick red) [10], PICO-2L 20 defined before WIMP search acoustic data unmasking using 10 SD WIMP-proton cross section [cm Counts (thick purple) [9], PICASSO (green band) [14], SIMPLE (or- neutron calibration data, are displayedlog(AP) with dashed lines and 101 102 103 3 I ange) [15], PandaX-II (cyan) [35], IceCube (dashed and dot- • Additional blinded exposure reveal no candidate eventsF in the WIMP search data. Alphas3 2 0 -39 8 252 ted pink) [36], and SuperKWIMP mass (dashed [GeV/c and dotted] black) [37, 38]. 1 fromFIG.10 the 2.222 Top:Rn decay AP distributions chain can be foridentified AmBe by and theirCf time neu- sig- acquired at lower thresholds tron calibration data (black) and WIMP searchPICO-60 data (red)CF at The indirect limits from IceCube and SuperK assume anni- nature and populate the two peaks in the WIMP search data FIG. 3. The 90% C.L. limit on the SD WIMP-proton cross 3.3 keV threshold. Bottom: AP and NN214 score for the same hilation to ⌧ leptons (dashed) and b quarks (dotted). The at high AP. Higher energy17x alphas from Po are producing section from PICO-60 C3F8 plotted in thick blue, along dataset. The acceptance region forSuperK nuclear recoil candidates, purple region represents parameter space of the constrained 0.5 larger-40 acoustic signals. with limits from PICO-60 CF I (thick red) [10], PICO-2L • defined10 before WIMP search acoustic data unmasking using minimal supersymmetric model3 of [39]. Additional limits, not Now decommissioning, as any (thick purple) [9], PICASSO (green band) [14], SIMPLE (or-

NN score neutron calibration data, are displayed with dashedIceCube lines and shown for clarity, are set by LUX [40] and XENON100 [41] PICO-60 C F ange) [15], PandaX-II (cyan) [35], IceCube (dashed and dot- additional exposure would be reveal no candidate events3 8 in the WIMP search data. Alphas (comparable to PandaX-II) and by ANTARES [42, 43] (com- 0 ted pink) [36], and SuperK (dashed and dotted black) [37, 38]. -1 0 1 2 3 4 from-41 the 222Rn decay chain can be identified by their time sig- parable to IceCube). expected to be background limited to10 be 0.25 0.09 (0.96 0.34) single(multiple)-bubble The indirect limits from IceCube and SuperK assume anni- natureSD WIMP-proton cross section [cm and populate1 the two peaks in the2 WIMP search data3 log(AP) ± 10 ± 10 133 10 hilation to ⌧ leptons (dashed) and b quarks (dotted). The events.at high PICO-60 AP. Higher was energy exposed alphas to from a 1214 mCiPo areBa producing source 2 purple region represents parameter space of the constrained bothlarger before acoustic and signals. afterWIMP the mass WIMP [GeV/c search] data, which, FIG. 2. Top: AP distributions for AmBe and 252Cf neu- minimal supersymmetric model of [39]. Additional limits, not 23compared*C. Amole against et a al. Geant4, Phys. Rev. [21] Lett. Monte 118 Carlo, 251301 simulation, (2017) WIMP mass, are shown in Fig. 3 and 4. These limits, tron calibration data (black) and WIMP search data (red) at shown for clarity, are set by LUX [40] and XENON100 [41] FIG.gives 3. a The measured 90% C.L. nucleation limit on e theciency SD WIMP-proton for electron recoil cross (comparablecorresponding to PandaX-II) to an upper and by limit ANTARES on the [42, spin-dependent 43] (com- 3.3 keV threshold. Bottom: AP and NN score for the same 41 2 section from PICO-60 C3F8 plotted in thick10 blue, along WIMP-proton cross section of 3.4 10 cm for a dataset. The acceptance region for nuclear recoil candidates, events above 3.3 keV of (1.80 0.38) 10 . Combin- parable to IceCube). withto limits be 0.25 from0.09 PICO-60 (0.96 CF0.34)I± (thick single(multiple)-bubble red)⇥ [10], PICO-2L 2 ⇥ defined before WIMP search acoustic data unmasking using ing this with± a Monte Carlo± 3 simulation of133 the external 30 GeV c WIMP, are currently the world-leading con- (thickevents. purple) PICO-60 [9], PICASSO was exposed (green to band) a 1 mCi [14], SIMPLEBa source (or- neutron calibration data, are displayed with dashed lines and gamma flux from [16, 22], we predict 0.026 0.007 events straints in the WIMP-proton spin-dependent sector and ange)both [15], before PandaX-II and after (cyan) the [35], WIMP IceCube search (dashed± data, and which, dot- reveal no candidate events in the WIMP search data. Alphas due to electron recoils in the WIMP search exposure. The indicate an improved sensitivity to the dark matter signal tedcompared pink) [36], against and SuperK a Geant4 (dashed [21] andMonte dotted Carlo black) simulation, [37, 38]. WIMP mass, are shown in Fig. 3 and 4. These limits, from the 222Rn decay chain can be identified by their time sig- 8 of a factor of 17, compared to previously reported PICO Thebackgroundgives indirect a measured limits from from coherent nucleation IceCube scattering e andciency SuperK of forB electron solarassume neutri- recoil anni- corresponding to an upper limit on the spin-dependent nature and populate the two peaks in the WIMP search data 10 results. 41 2 hilationnosevents is calculated to above⌧ leptons 3.3 to keV (dashed) be of 0.055 (1.80 and0.0070.38)b quarks events.10 (dotted).. Combin- The WIMP-proton cross section of 3.4 10 cm for a at high AP. Higher energy alphas from 214Po are producing ±± ⇥ 2 ⇥ purpleing thisregion with represents a Monte parameter Carlo simulation space of of the the constrained external 30 GeVA comparison c WIMP, are of currently our proton-only the world-leading SD limits con- with larger acoustic signals. We use the same shapes of the nucleation eciency minimalgamma supersymmetric flux from [16, 22], model we ofpredict [39]. 0.026Additional0.007 limits, events not straintsneutron-only in the WIMP-proton SD limits set spin-dependent by other dark mattersector and search curves for fluorine and carbon nuclear recoils± as found in shownRef.due [8], for to electron rescaled clarity, recoils are upwards set in by the in LUX WIMP recoil [40] energy search and XENON100 exposure. to account The [41] for indicateexperiments an improved is achieved sensitivity by to setting the dark constraints matter signal on the 8 of a factor of 17, compared to previously reported PICO (comparablethebackground 2% di↵erence to from PandaX-II) in coherent thermodynamic and scattering by ANTARES threshold. of B solar [42, We 43] neutri- adopt (com- e↵ective spin-dependent WIMP-neutron and WIMP- parablenos is to calculated IceCube). to be 0.055 0.007 events. results. to be 0.25 0.09 (0.96 0.34) single(multiple)-bubble the standard halo parametrization± [23], with the follow- proton couplings an and ap that are calculated according ± ± 133 2 3 A comparison of our proton-only SD limits with events. PICO-60 was exposed to a 1 mCi Ba source ing parameters:We use the same⇢D=0.3 shapes GeV of c thecm nucleation , vesc = 544 eciency km/s, to the method proposed in Ref. [29]. The expectation neutron-only SD limits set by other dark matter search19 both before and after the WIMP search data, which, v curves= for 232 fluorine km/s, and carbonv = 220 nuclear km/s. recoils We use as thefound e↵ inec- values for the proton and neutron spins for the Fnu- Earth o experiments is achieved by setting constraints on the compared against a Geant4 [21] Monte Carlo simulation, WIMPtiveRef. field [8], mass, theory rescaled are treatment shown upwards in in andFig. recoil nuclear 3 andenergy 4. form to These account factors limits, for de- cleus are taken from Ref. [24]. The allowed region in the 2% di↵erence in thermodynamic threshold. We adopt e↵ective spin-dependent WIMP-neutron and2 WIMP- gives a measured nucleation eciency for electron recoil correspondingscribed in Refs. to [24–27] an upper to determine limit on the sensitivity spin-dependent to both the an ap plane is shown for a 50 GeV c WIMP in the standard halo parametrization [23], with41 the follow-2 proton couplings an and ap that are calculated according 10 WIMP-proton cross section of 3.4 10 cm for a Fig. 5. We find that PICO-60 C3F8 improves the con- events above 3.3 keV of (1.80 0.38) 10 . Combin- spin-dependent and spin-independent2 3 dark matter inter- to the method proposed in Ref. [29]. The expectation ± ⇥ ing parameters:2 ⇢D=0.3 GeV c cm⇥ , vesc = 544 km/s, 30actions. GeV c ForWIMP, the SI case,are currently we use the theM world-leadingresponse of Table constraints on an and ap, in complementarity19 with other ing this with a Monte Carlo simulation of the external v = 232 km/s, and v = 220 km/s. We use the e↵ec- values for the proton and neutron spins for the Fnu- straints1 inEarth Ref. in the[24], WIMP-proton and for SDo interactions, spin-dependent we use sector the sum and dark matter search experiments that are more sensitive gamma flux from [16, 22], we predict 0.026 0.007 events tive field theory treatment and nuclear form factors de- cleus are taken from Ref. [24]. The allowed region in ± of the ⌃ and ⌃ terms from the same table. To im- to the WIMP-neutron coupling. 2 due to electron recoils in the WIMP search exposure. The indicatescribed an0 in improved Refs.00 [24–27] sensitivity to determine to the sensitivity dark matter to signalboth the an ap plane is shown for a 50 GeV c WIMP in 8 background from coherent scattering of B solar neutri- ofplement aspin-dependent factor these of 17, interactions compared and spin-independent to and previously form factors, dark reported matter we use inter- PICO the Fig.The 5. We LHC find has that significant PICO-60 sensitivity C3F8 improves to dark the con- matter, nos is calculated to be 0.055 0.007 events. results.publiclyactions. available For the SIdmdd case,code we use package the M [27,response 28]. The of Table calcu- straintsbut to on interpretan and LHCap, in searches, complementarity one must withassume other a spe- ± lated Poisson upper limits at the 90% C.L. for the spin- darkcific matter model search to generate experiments the signal that that are moreis then sensitive looked for We use the same shapes of the nucleation eciency A1 in comparison Ref. [24], and of for our SD proton-only interactions, SD we use limits the sum with dependent WIMP-proton and spin-independent WIMP- toin the the WIMP-neutron data. Despite coupling. this subtlety, the convention has curves for fluorine and carbon nuclear recoils as found in neutron-onlyof the ⌃0 and SD⌃ limits00 terms set from by other the same dark table. matter To search im- nucleonplement elastic these scattering interactions cross-sections, and form factors, as a we function use the of beenThe to LHC show has LHC significant limits sensitivity alongside more to dark general matter, direct Ref. [8], rescaled upwards in recoil energy to account for experiments is achieved by setting constraints on the e↵ectivepublicly spin-dependent available dmdd code WIMP-neutron package [27, 28]. and The WIMP- calcu- but to interpret LHC searches, one must assume a spe- the 2% di↵erence in thermodynamic threshold. We adopt lated Poisson upper limits at the 90% C.L. for the spin- cific model to generate the signal that is then looked for proton couplings a and a that are calculated according the standard halo parametrization [23], with the follow- dependent WIMP-protonn p and spin-independent WIMP- in the data. Despite this subtlety, the convention has 2 3 ing parameters: ⇢D=0.3 GeV c cm , vesc = 544 km/s, to the method proposed in Ref. [29]. The expectation nucleon elastic scattering cross-sections, as a function19 of been to show LHC limits alongside more general direct vEarth = 232 km/s, and vo = 220 km/s. We use the e↵ec- values for the proton and neutron spins for the Fnu- tive field theory treatment and nuclear form factors de- cleus are taken from Ref. [24]. The allowed region in 2 the a a plane is shown for a 50 GeV c WIMP in scribed in Refs. [24–27] to determine sensitivity to both n p spin-dependent and spin-independent dark matter inter- Fig. 5. We find that PICO-60 C3F8 improves the con- actions. For the SI case, we use the M response of Table straints on an and ap, in complementarity with other 1 in Ref. [24], and for SD interactions, we use the sum dark matter search experiments that are more sensitive of the ⌃0 and ⌃00 terms from the same table. To im- to the WIMP-neutron coupling. plement these interactions and form factors, we use the The LHC has significant sensitivity to dark matter, publicly available dmdd code package [27, 28]. The calcu- but to interpret LHC searches, one must assume a spe- lated Poisson upper limits at the 90% C.L. for the spin- cific model to generate the signal that is then looked for dependent WIMP-proton and spin-independent WIMP- in the data. Despite this subtlety, the convention has nucleon elastic scattering cross-sections, as a function of been to show LHC limits alongside more general direct 4 ] not apply acoustic or fiducial cuts, resulting in the larger 2 10-37 exposure shown in Table I. Instead, given 99.5 ± 0.1% efficiency to reconstruct at least one bubble in the bulk for a multiple-bubble event, every passing event is scanned 10-38 for multiplicity. This scan reveals 3 multiple-bubble events in the WIMP search dataset. Based on a detailed Monte Carlo simulation, the background from neutrons is 10-39 predicted to be 0.25 ± 0.09 (0.96 ± 0.34) single(multiple)- bubble events. PICO-60 was exposed to a 1 mCi 133Ba source both before and after the WIMP search data, 10-40 which, compared against a Geant4 [20] Monte Carlo sim- PICO-60 C F ulation, gives a measured nucleation efficiency for elec- 3 8 ± × −10 tron recoil events above 3.3 keV of (1.80 0.38) 10 . 10-41 Combining this with a Monte Carlo simulation of the ex- SD WIMP-proton cross section [cm 101 102 103 ternal gamma flux from [15, 21], we predict 0.026 ± 0.007 WIMP mass [GeV/c2] events due to electron recoils in the WIMP search exposure. The background from coherent scattering of 8B FIG. 3. The 90% C.L. limit on the SD WIMP-proton cross solar neutrinos is calculated to be 0.055 ± 0.007 events. section from PICO-60 C3F8 plotted in blue, along with lim- The unmasking of the acoustic data, performed after its from PICO-60 CF3I(red)[10],PICO-2L(purple)[9], completion of the WIMP search run, reveals that none of PICASSO (green) [14], SIMPLE (orange) [33], PandaX-II (cyan) [34], IceCube (dashed and dotted pink) [35], and Su- the 106 single bulk bubbles are consistent with the nu- perK (dashed and dotted black) [36, 37]. The indirect limits clear recoil hypothesis defined by AP and the NN score, from IceCube and SuperK assume annihilation to τ leptons as shown in Fig. 2. (dashed) and b quarks (dotted). The purple region represents We use the same procedure and calibration data de- parameter space of the constrained minimal supersymmetric scribed in Ref. [8] to evaluate nucleation efficiency curves model of [38]. Additional limits, not shown for clarity, are set for fluorine and carbon recoils. We adopt the standard by LUX [39] and XENON100 [40] (comparable to PandaX-II) −2 −3 and by ANTARES [41, 42] (comparable to IceCube). halo parametrization [22], with ρD=0.3 GeV c cm , vesc =544km/s,vEarth =232km/s,andvo =220km/s. We use the effective field theory treatment and nuclear Nearly competitive in SI at low mass form factors described in Refs. [23–26] to determine sensi- ] …additional reach available at lower threshold? 2 -38 tivity to both spin-dependent and spin-independent dark 10 matter interactions. For the SI case, we use the M response of Table 1 in Ref. [23], and for SD interactions, we use the sum of the Σ′ and Σ′′ terms from the -40 same table. To implement these interactions and form 10 CRESST-II factors, we use the publicly available dmdd code pack- CDMSlite age [26, 27]. The calculated limits at the 90% C.L. for the spin-dependent WIMP-proton and spin-independent PICO-2L -42 WIMP-nucleon elastic scattering cross-sections, with no 10 PICO-60 C F background subtraction, as a function of WIMP mass, 3 8 are shown in Fig. 3 and 4. These limits are currently LUX Panda-X II the world-leading constraints in the WIMP-proton spin- PICO-60 CF3I dependent sector and indicate an improved sensitivity to 10-44

SI WIMP-nucleon cross section [cm 0 1 2 the dark matter signal of a factor of 17, compared to 10 10 10 2 previously reported PICO results. WIMP mass [GeV/c ] 24 Constraints on the effective spin-dependent WIMP- FIG. 4. The 90% C.L. limit on the SI WIMP-nucleon neutron and WIMP-proton couplings an and ap are cal- cross-section from PICO-60 C3F8 plotted in blue, along culated according to the method proposed in Ref. [28]. with limits from PICO-60 CF3I(red)[10],PICO-2L(pur- The expectation values for the proton and neutron spins ple) [9], LUX (yellow) [43], PandaX-II (cyan) [44], CRESST- 19 for the F nucleus are taken from Ref. [23]. The allowed II (magenta) [45], and CDMS-lite (black) [46]. While we −2 region in the an − ap plane is shown for a 50 GeV c choose to highlight this result, LUX sets the strongest lim- 2 WIMP in Fig. 5. We find that PICO-60 C3F8 improves its on WIMP masses greater than 6 GeV/c .Additional the constraints on an and ap, in complementarity with limits, not shown for clarity, are set by PICASSO [14], other dark matter search experiments that are more sen- XENON100 [40], DarkSide-50 [47], SuperCDMS [48], CDMS- II [49], and Edelweiss-III [50]. sitive to the WIMP-neutron coupling. The LHC has significant sensitivity to dark matter, PICO-60 low threshold run

• Second physics run prompted by observation of far fewer recoil events than expected at lower thresholds

• Decided on a threshold of 2.4 keV, where backgrounds were projected to produce <5 events over a 30 live-day exposure, now acquired

• Analysis is wrapping up, results soon…

3.3 keV blind run (PRL)

2.4 keV blind run

25 Why end PICO-60?

• Only published 30 live days, with ~30 more on the way…

• After over a year commissioning, why acquire so little data?

• Short answer: 3 multiple-scatter neutron events in Run-1 meant expectation of 1 single-scatter neutron (which we didn’t see)

• That rate now appears to have been a slight upward ﬂuctuation, but the full (3.3, 2.4) keV dataset (~60 days) will almost certainly be background limited: very slow gains if we’d continued

• More pressing need: build the next chamber!

26 PICO-40L Goals

where the “L” again indicates, approximately, “demonstrator for next, bigger chamber” • Science: acquire one-year background-free exposure

• Order of magnitude improvement on PICO-60 limits

• Engineering: demonstrate background reduction and technology improvements for PICO-500

• Focus on (neutron) background reduction

• Conﬁrm “RSU” design used in prototype chambers

27 PICO-60 → PICO-40L

Pressure vessel

C3F8/freon (target) Cameras Hydraulic ﬂuid (mineral oil)

Thermal gradient → (no water buffer) Heating element Bellows Thermal insulation

28 PICO-40L detector design

• To be deployed 2 km underground at SNOLAB (“ladder labs” area)

• Target: ~40L C3F8, (proj. >90% ﬁducial) Superheated (15C) • Synthetic fused silica inner vessel and piston (no more “water piston”)

• Larger stainless steel pressure Normal (-25C) vessel, 20t water tank, muon veto – all minimize neutron backgrounds

29 PICO-40L detector design

• Inversion eliminates potential sources of background:

• water droplets

• surface tension effects Superheated (15C)

• particulates – would now fall out of active region into cold annulus

• No buffer: allows wider choice of Normal (-25C) target ﬂuid, wider range of operating temperatures; directly enables full target recirculation and puriﬁcation

30 Many upgraded systems

• Optics/DAQ: much better Basler cameras using new Sony IMX174 CMOS sensor

• running on newer USB3 Vision interface for more programming ﬂexibility

• better lenses (higher resolution, reduced barrel distortion, etc.)

• better stereoscopic viewing angles and camera mounts

Basler CMOS camera • better retroreﬂector and improved LED lighting rings with new IMX714 sensor

• Hydraulic system control: brought into alignment with new designs used on several test chambers; will enable continuous active recirculation/ﬁltration

• Piezo acoustic sensors: better physical coupling, and improved longevity in different hydraulic ﬂuid (mineral oil rather than propylene glycol)

31 PICO-40L timeline

• Pressure vessel arrived to SNOLAB surface 18 May 2017

• Clean surface commissioning ongoing presently

• Full detector assembly to be shipped underground to SNOLAB Dec 2017

• First data January 2018

• End of physics data in 2019

32 PICO-500

• Planned ton-scale detector • Intended to begin surface commissioningPICO%500(Physics(Case( as soon as late 2018 • Goal is to begin data-taking in 2019

(cf. PICO-40L)

PICO-500 33 Russell&Neilson&

8/25/16& Drexel&University& 1& PICO-500

10 -37

• Designed to have an additional 10 -38

] PICO-2L

order of magnitude sensitivity 2 -39 10 PICO-60 CF3I LUX beyond PICO-40L PICO-60 C3F8 10 -40

IceCube • Could run C3F8 and/or several 10 -41 PICO-40 (proj)

other targets (i.e. CF3I or 10 -42 hydrocarbons: C2H2F4, etc.) to -43 probe higher/lower mass or 10 -44 reduce a WIMP signal in a SD WIMP-proton cross section [cm 10

predictable way 10 -45 C3F8 neutrino bkgd

10 -46 10 1 10 2 10 3 WIMP mass [GeV/c 2 ]

34 PICO-500

10 -37-37

• Designed to have an additional 10 -38-38

] ] PICO-2L

order of magnitude sensitivity 2 2 -39-39 10 PICO-60 CF3I LUX beyond PICO-40L PICO-60 C3F8 10 -40-40

IceCubeIceCube • Could run C3F8 and/or several 10 -41-41 PICO-40 (proj)PICO-500 (proj)

other targets (i.e. CF3I or 10 -42-42 hydrocarbons: C2H2F4, etc.) to -43-43 probe higher/lower mass or 10 -44-44 reduce a WIMP signal in a SD WIMP-proton cross section [cm SD WIMP-proton cross section [cm 10

predictable way 10 -45-45 C3F8 neutrino bkgd

10 -46-46 10 1 10 2 10 3 WIMP mass [GeV/c 2 ]]

34 Summary and outlook

• PICO-40L commissioning now, data in early 2018

• One year background-free run – order of magnitude improvement on PICO-60 result

• Demonstrate background reduction advances enabling ton-scale PICO-500

• PICO-500 could begin data taking as early as 2019

• Sensitivity to additional order of magnitude in SDp beyond PICO-40L, covering signiﬁcant new well motivated parameter space

• Could check itself/signals from other detectors with a target change

• PICO detectors are relatively cheap and ﬂexible, with very quick turnaround time

• SD WIMP interactions are arguably just as promising as SI! Imagine a ﬁrst signal in 2019?

35 Extra slides SD WIMP-proton/-neutron combined sensitivity plot

37 Direct Detection Rates

• DM density component

• Unknown cross section – what we set upper limits on

2 5/3 • Nuclear form factor: F ∝ exp(-Q/Q0) where Q0 ~ (80 MeV)/A

• Velocity distribution of dark matter in the galactic halo

38 Comparing SD limits

SD WIMP-nucleus cross-section at q=0

WIMP-nucleon cross-sections (in limiting cases an,p=0)

from experiment

39 (a) (b)

Figure 2: A comparison of LHC results to the mDM–SI (a) and mDM–SD (b) planes. Unlike in the mass-mass plane, the limits are shown at 90% CL. The LHC contour in the SI (SD) plane is for a vector (axial-vector) mediator, Dirac DM and couplings gq =0.25 and gDM = 1. The LHC SI exclusion contour is compared with the LUX, CDMSLite and CRESST-II limits, which are the most constraining in the shown mass range. The SD exclusion contour constrains the DM-proton cross section and is compared with limits from the PICO experiments, the IceCube limit for the tt¯ annihilation channel and the Super-Kamiokande limit for the b¯b annihilation channel. The depicted LHC results are intended for illustration only and are not based on real data.

n,p n,p n,p n,p Here f =1 fq . The state-of-the-art values for fq are from [48] (for fu and TG q=u,d,s f n,p) and [49] (for f n,p) and read f n =0.019, f n =0.045 and f n =0.043. The values for d P s u d s the proton are slightly di↵erent, but in practice the di↵erence can be ignored. Substituting these values, we ﬁnd that numerically

3 f(g )=1.16 10 g , (4.5) q · q and therefore the size of a typical cross section is

2 4 2 43 2 gq gDM 125 GeV µn SI 6.9 10 cm . (4.6) ' ⇥ · 1 Mmed 1 GeV LHC comparison⇣ ⌘ ✓ ◆ ⇣ method⌘ 4.1.2 SD case: Axial-vector mediator For the axial-vector mediator, the scattering is SD and the corresponding cross section can be written as 2 2 2 3f (gq)gDMµn SD = 4 . (4.7) ⇡Mmed p,n In general f (gq)di↵ers for protons and neutrons and is given by

p,n (p,n) (p,n) (p,n) f (gq)=u gu + d gd + s gs , (4.8)

(p) (n) (p) (n) where u = =0.84, = u = 0.43 and = 0.09 are the values rec- d d s ommended by the Particle Data Group [50]. Other values are also used in the literature (see e.g. [51]) and di↵er by up to (5%). – 12 – O Under the assumption that the coupling gq is equal for all quarks, one ﬁnds

f(gq)=0.32gq , (4.9)

and thus

2 4 2 SD 42 2 gq gDM 1 TeV µn 2.4 10 cm . (4.10) ' ⇥ · 0.25 Mmed 1 GeV ⇣ ⌘ ✓ ◆ ⇣ ⌘ We emphasise that the same result is obtained both for the SD DM-proton scattering p n (arXiv:1603.04156) cross section SD and the SD DM-neutron scattering40 cross section SD. Using (4.10)itis therefore possible to map collider results on both parameter planes conventionally shown by DD experiments. Should only one plot be required, we recommend comparing the LHC p results to the DD bounds on SD, which is typically more dicult to constrain. In the future, it is desirable to consider not only the case gu = gd = gs, but also the case g = g = g , which is well-motivated from embedding the simpliﬁed model in the u d s SM gauge group and can be included without much additional e↵ort. For g = g = g u d s one obtains approximately f p(g )=1.36 g and f n(g )= 1.18 g , i.e. the DM-neutron q u q u cross section is slightly smaller than the DM-proton cross section.4

4.1.3 Neutrino observatories: IceCube and Super-Kamiokande The IceCube [53] and Super-Kamiokande [54] neutrino observatories are also able to constrain the SI and SD cross sections. When DM particles elastically scatter with elements in the Sun, they can lose enough energy to become gravitationally bound. Self-annihilation of the DM particles produces neutrinos (either directly or in showering) that can be searched for in a neutrino observatory. When the DM capture and annihilation rates are in equilib- rium, the neutrino ﬂux depends only on the initial capture rate, which is determined by the SI or SD cross section [55]. p The IceCube and Super-Kamiokande limits on SD are of particular interest as they can be stronger than the corresponding bounds from DD experiments. The former bounds are however more model dependent, since they depend on the particular DM annihilation channel. For annihilation only into light quarks, the limits are weaker than DD experiments.

For mb mt, the dominant annihilation channel is to tt¯ and the resulting constraints from IceCube are stronger than DD experiments. Both the Super-Kamiokande and IceCube limits can be shown together with other bounds on the SD DM-proton scattering cross section.

4 LHC searches are only sensitive to the relative sign between gu and gd if both types of quarks are present in a single process (e.g. ud¯ ud¯+¯ or uu¯ dd¯+¯). Such processes give a subleading e↵ect in mono-jet ! ! searches and are presently not included in the signal computation. As a result, the signal prediction for mono-jets turns out to be independent of the relative sign between the individual quark couplings [52].

– 13 – Spin-independent 41