Dark matter search results from the PICO-60 ${\mathrm{CF}}_3\mathrm{I}$ bubble chamber

New data are reported from the operation of the PICO-60 dark matter detector, a bubble chamber filled with 36.8 kg of ${\mathrm{CF}}_3\mathrm{I}$ and located in the SNOLAB underground laboratory. PICO-60 is the largest bubble chamber to search for dark matter to date. With an analyzed exposure of 92.8 livedays, PICO-60 exhibits the same excellent background rejection observed in smaller bubble chambers. Alpha decays in PICO-60 exhibit frequency-dependent acoustic calorimetry, similar but not identical to that reported recently in a ${\mathrm{C}}_3{\mathrm{F}}_8$ bubble chamber. PICO-60 also observes a large population of unknown background events, exhibiting acoustic, spatial, and timing behaviors inconsistent with those expected from a dark matter signal. These behaviors allow for analysis cuts to remove all background events while retaining 48.2% of the exposure. Stringent limits on weakly interacting massive particles interacting via spin-dependent proton and spin-independent processes are set, and most interpretations of the DAMA/LIBRA modulation signal as dark matter interacting with iodine nuclei are ruled out.

Figure 1

A schematic of the PICO-60 bubble chamber.

Figure 2

Images of a multiple scattering neutron event from the two PICO-60 cameras. Reflection of the LED rings used for illumination are clearly visible on the front and back of the jar. The two vertical strings of acoustic sensors are visible running up the sides of the jar.

Figure 3

Total livetime in the dark matter search data vs Seitz threshold. Because of the temperature variations and the many pressure set points, the data sample a continuum of Seitz thresholds between 7 and 20 keV. There are a total of 92.8 livedays in the dark matter search data.

Figure 4

The best fit iodine (black), fluorine (red), and carbon (magenta) efficiency curves for $E_T=13.6\mathrm{keV}$ data are shown by the solid lines, and the light blue band shows the calculated Seitz threshold with the experimental and theoretical uncertainties (the solid curves are the same in both the top and bottom panels). In the top panel, the dashed lines show the curves used to determine sensitivity for a 20 GeV SD WIMP, corresponding to the set of curves with the least sensitivity to 20 GeV SD WIMP scattering consistent with the calibration data at $1\sigma$, while the dashed lines in the bottom panel show the curves used to determine sensitivity for a 20 GeV SI WIMP. The onset of nucleation for fluorine and carbon recoils occurs at energies greater than twice the Seitz threshold, while the response to iodine is much closer to the Seitz model.

Figure 5

AP distributions for neutron calibration (black) and WIMP search data (red) for all WIMP search data. The top figure shows ${\mathrm{AP}}_{\mathrm{low}}$ for frequency bands between 7 and 63 kHz and the bottom figure shows ${\mathrm{AP}}_{\mathrm{high}}$ for frequencies between 63 and 110 kHz. Events with ${\mathrm{AP}}_{\text{low}}>2.9$ are identified as alpha-decay events and shaded in both histograms. The rate of observed alpha decays is consistent between WIMP search data and calibration runs.

Figure 6

The decays of ${}^{222}\mathrm{Rn}$ and its daughters ${}^{218}\mathrm{Po}$ and ${}^{214}\mathrm{Po}$ produce alphas with energies 5.48, 6.0, and 7.68 MeV, respectively, with the half-lives shown.

Figure 7

The mean AP as a function of frequency bin for the first, second, and third decays of 82 triplets of consecutive alpha events whose timing is consistent with the fast radon decay chain. The data are normalized in each frequency bin to the neutron calibration data; i.e., the mean AP for neutron calibration data would appear flat at a value of 1.

Figure 8

The mean AP as a function of frequency bin in ${\mathrm{C}}_3{\mathrm{F}}_8$ [14] for the first, second, and third decays in 18 triplets of consecutive alpha events whose timing is consistent with the fast radon decay chain.

Figure 9

Event rate of the nonalpha background events (black) and alpha events (red) as a function of the length of time the chamber was in an expanded state. The rate is calculated for intervals of expansion time indicated by the horizontal error bars; the rates measured in neighboring bins are uncorrelated. A dark matter signal would be flat; by contrast, the background events cluster at early expansion times. Although a fraction of alpha decays do have timing correlations relevant on these scales (the ${}^{218}\mathrm{Po}$ decays), the total alpha distribution is dominated by the uncorrelated decays, nearly flat in expansion time, and can be viewed as a rough proxy for a dark matter signal. We include the alpha distribution here to show that systematic effects cannot account for the distribution of the background events.

Figure 10

Two-dimensional histogram of bubble location (${\mathrm{R}}^2/{\mathrm{R}}_{\mathrm{jar}}$ vs Z). The left-hand plot shows all alpha events while the right-hand plot shows the background events. A dark matter signal would be isotropic in these units. As a proxy for a dark matter signal, the alphas are more uniformly distributed in the jar than the background events, which are concentrated along the walls and near the interface.

Figure 11

A two-dimensional histogram of ${\mathrm{AP}}_{\mathrm{high}}$ and expansion time after applying the optimum fiducial cuts, divided into bins of equal exposure to dark matter (i.e., a dark matter signal would appear uniform in the histogram). All the background events populate the left and top of the histogram. The optimum cuts are represented by the red rectangle.

Figure 12

The 90% C.L. limit on the SI WIMP-nucleon cross section from PICO-60 is plotted in blue, along with limits from COUPP (light blue), LUX (black), XENON100 (orange), DarkSide-50 (green), and the reanalysis of CDMS-II (magenta) [10, 41, 42, 43, 44].

Figure 13

The 90% C.L. limit on the SD WIMP-proton cross section from PICO-60 is plotted in blue, along with limits from PICO-2L (red), COUPP (light blue region), PICASSO (dark blue), SIMPLE (green), XENON100 (orange), IceCube (dashed and solid pink), SuperK (dashed and solid black) and CMS (dashed orange) [10, 12, 13, 45, 46, 47, 48, 49]. For the IceCube and SuperK results, the dashed lines assume annihilation to $W$ pairs while the solid lines assume annihilation to $b$ quarks. Comparable limits assuming these and other annihilation channels are set by the ANTARES, Baikal and Baksan neutrino telescopes [50, 51, 52]. The CMS limit is from a monojet search and assumes an effective field theory, valid only for a heavy mediator [53, 54]. Comparable limits are set by ATLAS [55, 56]. The purple region represents parameter space of the CMSSM model of [57].

Figure 14

$C_n(x,\mu )$ for $n=0$, 1, 2, 3, 4 for simulations with $\mu =8$. For each $n$, the maximum value of $x$ for the restricted data set is indicated by an “x.” Over all $n$, the maximum of $C_n(x,\mu )$ for the restricted data set is 0.978 for $n=0$. For $\mu =8$, ${\overline{C}}_{\max }=0.946$, indicated by the horizontal black line. For both $n=0$ and $n=1$ the maximum of $C_n(x,\mu )$ exceeds ${\overline{C}}_{\max }$, thus excluding $\mu =8$ as too large at greater than the 90% C.L.

Figure 15

Maximum of $C_n(x,\mu )$ for the restricted data set for $n=0$, 1, 2, 3, 4, compared to ${\overline{C}}_{\max }$. Over the range of $\mu$ shown $C_{\max }$ is always taken from the $n=0$ curve. All WIMP couplings corresponding to $\mu \ge 5.4$, where $C_{\max }>{\overline{C}}_{\max }$, are excluded at the 90% C.L.