Projected sensitivity of the SuperCDMS SNOLAB experiment

SuperCDMS SNOLAB will be a next-generation experiment aimed at directly detecting low-mass particles (with masses $\le 10\mathrm{GeV}/{\mathrm{c}}^2$) that may constitute dark matter by using cryogenic detectors of two types (HV and iZIP) and two target materials (germanium and silicon). The experiment is being designed with an initial sensitivity to nuclear recoil cross sections $\sim 1\times {10}^{-43}{\mathrm{cm}}^2$ for a dark matter particle mass of $1\mathrm{GeV}/{\mathrm{c}}^2$, and with capacity to continue exploration to both smaller masses and better sensitivities. The phonon sensitivity of the HV detectors will be sufficient to detect nuclear recoils from sub-GeV dark matter. A detailed calibration of the detector response to low-energy recoils will be needed to optimize running conditions of the HV detectors and to interpret their data for dark matter searches. Low-activity shielding, and the depth of SNOLAB, will reduce most backgrounds, but cosmogenically produced ${}^3\mathrm{H}$ and naturally occurring ${}^{32}\mathrm{Si}$ will be present in the detectors at some level. Even if these backgrounds are 10 times higher than expected, the science reach of the HV detectors would be over 3 orders of magnitude beyond current results for a dark matter mass of $1\mathrm{GeV}/{\mathrm{c}}^2$. The iZIP detectors are relatively insensitive to variations in detector response and backgrounds, and will provide better sensitivity for dark matter particles with masses $\gtrsim 5\mathrm{GeV}/{\mathrm{c}}^2$. The mix of detector types (HV and iZIP), and targets (germanium and silicon), planned for the experiment, as well as flexibility in how the detectors are operated, will allow us to maximize the low-mass reach, and understand the backgrounds that the experiment will encounter. Upgrades to the experiment, perhaps with a variety of ultra-low-background cryogenic detectors, will extend dark matter sensitivity down to the “neutrino floor,” where coherent scatters of solar neutrinos become a limiting background.

Figure 1

Channel layout for the HV (top) and iZIP (bottom) detectors. The HV detector has six phonon channels on each side, arranged as an inner “core,” surrounded by three wedge shaped channels and two outer rings designed to reject events near the edge. Each channel contains hundreds of lithographically defined superconducting sensors. The wedge channels on the bottom surface are rotated by 60° with respect to those on the top. The interleaved Z-sensitive Ionization Phonon (iZIP) detector also has six phonon channels on each side, arranged as an inner core, surrounded by four wedge-shaped channels and one outer ring. An “outer” ionization channel shares the same area and is interleaved with the outermost phonon ring, and an “inner” ionization channel is interleaved with the remaining phonon channels. The wedge channels on the bottom surface are rotated by 45° with respect to those on the top.

Figure 2

A schematic of the experiment shield and cryostat layers. The assembly rests on top of a seismic platform to provide isolation from major seismic events. The outer water tanks provide protection from cavern neutrons. A gamma shield protects from external gamma rays and the inner polyethylene layers serve to absorb radiogenic neutrons emitted from the cryostat and gamma shield.

Figure 3

Detailed simulation geometry of the SuperCDMS SNOLAB tower mechanical support and instrumentation for one of six detectors in a tower.

Figure 4

Raw background spectra of single-scatter interactions in a Si (left) and Ge (right) detector obtained from the Monte Carlo simulation. The spectra are broken down into components and shown as functions of recoil energy (ER or NR depending on the interaction). ${}^3\mathrm{H}$ (pink) and ${}^{32}\mathrm{Si}$ (purple) are the largest individual contributors to the backgrounds in the Ge and Si detectors, respectively. The remaining components are Compton scatters from gamma rays (red), surface betas (green), surface ${}^{206}\mathrm{Pb}$ recoils (orange), neutrons (blue) and coherent elastic neutrino-nucleus scattering (cyan). Note that the neutron spectrum (blue) has some spurious structure from the limited simulation statistics in the cavern component of the neutron background (cf. Table 5).

Figure 5

Ionization yield in Si. The energy range and yield behavior of the numbered regions are: (I) $0<E_{nr}<E_{y_0}$ (0.04 keV), no ionization production; (II) $0.04<E_{nr}<0.675\mathrm{keV}$, ionization yield described by power-law function; (III) $0.675<E_{nr}<15\mathrm{keV}$, ionization yield described by an empirical fit to DAMIC data [45, 46]; and (IV) $E_{nr}>15\mathrm{keV}$, ionization yield described by Lindhard theory.

Figure 6

Background spectra, before (left) and after (right) analysis cuts in Si (top) and Ge (bottom) HV detectors, shown as a function of nuclear recoil energy (keVnr). Thick black lines represent the total background rates. Electron recoils from Compton gamma rays, ${}^3\mathrm{H}$, and ${}^{32}\mathrm{Si}$ are grouped together (red). The Ge activation lines (grey) are shown convolved with a 10 eV rms resolution. The remaining components are surface betas (green), surface ${}^{206}\mathrm{Pb}$ recoils (orange), neutrons (blue), and $\mathrm{CE}\nu \mathrm{NS}$ (cyan). The large number of apparent Ge activation lines in the before cuts (left) spectrum is due to the discrete variation in the detector’s ionization response as a function of position ($\eta$) in the simplified detector response model used in this analysis; namely a given recoil energy can be reconstructed to multiple nuclear recoil energy values, according to the discrete parametrization in Table 6. After application of the radial fiducial-volume cut, the number of apparent lines is reduced because only the detector region with $\eta =1$ remains (minimizing the shortcomings of the simplified model). The value of $\eta$ for an actual detector is expected to be continuous, resulting in a less pronounced and more smeared out reconstruction of the activation peaks in the pre-cut spectrum compared to the model used here.

Figure 7

Background spectra, before (left) and after (right) analysis cuts in Si (top) and Ge (bottom) iZIP detectors, shown as a function of nuclear recoil energy (keVnr) to allow direct comparison of the various backgrounds to the natural dark matter interaction energy scale. Thick black lines represent the total background rates. Electron recoils from Compton gamma rays, ${}^3\mathrm{H}$ and ${}^{32}\mathrm{Si}$ are grouped together (red). The Ge activation lines (grey) are shown convolved with a 10 eV rms resolution. The remaining components are surface betas (green), surface ${}^{206}\mathrm{Pb}$ recoils (orange), neutrons (blue) and $\mathrm{CE}\nu \mathrm{NS}$ (cyan).

Figure 8

Projected exclusion sensitivity for the SuperCDMS SNOLAB direct detection dark matter experiment. The vertical axis is the spin-independent WIMP-nucleon cross section under standard halo assumptions [47], and the horizontal axis is the WIMP mass, where WIMP is used to mean any low-mass particle dark matter candidate. The blue dashed curves represent the expected sensitivities for the Si HV and iZIP detectors and the red dashed curves the expected sensitivities of the Ge HV and iZIP detectors. These sensitivity limits are determined using the optimum interval method [48, 49], which does not incorporate any knowledge of the specific disposition and source of background events observed during the experimental operation. The solid lines are the current experimental exclusion limits in the low-mass region, from the CRESST-II [50], SuperCDMS [4, 5] and LUX [51] experiments. The dotted orange line is the dark matter discovery limit from Ref. [52], which represents the cross section at which the interaction rate from dark matter particles becomes comparable to the solar neutrino coherent elastic scattering rate.

Figure 9

Si (left) and Ge (right) HV detector sensitivities for different background assumptions. The black dashed lines are the sensitivities for the nominal assumptions (same as Fig. 8). The green dashed lines correspond to varying the ${}^3\mathrm{H}$ contamination from zero (lower curve) to 3 times the nominal exposure (upper curve) while keeping all other backgrounds at their nominal values. For Si, the blue dashed lines correspond to varying the ${}^{32}\mathrm{Si}$ contamination from zero (lower curve) to 10 times the nominal value (upper curve), while keeping the ${}^3\mathrm{H}$ at its nominal value. The purple dashed line is the expected sensitivity if both the ${}^3\mathrm{H}$ and ${}^{32}\mathrm{Si}$ contamination levels are zero. The solid lines are the same experimental sensitivities presented in Fig. 8.

Figure 10

Si (left) and Ge (right) HV detector sensitivity studies pertaining to yield and analysis thresholds. The black dashed lines are the sensitivities for the nominal assumptions (same as Fig. 8). The orange dashed curves correspond to operating the detectors with zero voltage bias. For Si, the gray dashed curve shows the sensitivity using the standard Lindhard ionization yield model with an energy cutoff of $E_{y_0}=40\mathrm{eV}$. Solid lines are the same experimental sensitivities presented in Fig. 8.