Colorado Underground Research Institute (CURIE)

• CURIE is a shallow-underground, low-background facility offering accessible horizontal entry for detector R&D, calibration, and environmental studies.
• It achieves an effective shielding of 0.415 km.w.e. with a ~700× muon flux reduction relative to sea level, verified by precise direct measurements.
• The institute serves as a benchmark for simulation frameworks (Daemonflux, MUTE, PROPOSAL, and GEANT4) to validate muon-induced secondary backgrounds in complex topographies.

The Colorado Underground Research Institute (CURIE) is a shallow-underground low-background laboratory being developed inside the Edgar Experimental Mine (EEM) in Idaho Springs, Colorado, and operated by the Colorado School of Mines. Its defining technical characteristic is an effective shielding of approximately $0.415 \pm 0.027\ \mathrm{km.w.e.}$, corresponding to a measured underground muon flux of order $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$ and a reduction of the sea-level muon flux by roughly $700\times$. Within the current literature, CURIE is presented as an accessible, horizontally entered underground environment for detector R&D, calibration, environmental characterization, and selected low-background physics experiments that can tolerate moderate residual cosmogenic rates, while also serving as a test case for shallow-site muon and muon-induced background modeling (Keblbeck et al., 2024, Keblbeck et al., 7 Oct 2025).

1. Site, access, and physical configuration

CURIE is located in the EEM near Idaho Springs, Colorado, with the EEM entrance near longitude $-105.52$ and latitude $39.74$, at an altitude of $2401\ \mathrm{m}$ above sea level. The mine has two main tunnels, the Army tunnel and the Miami tunnel, and the Miami tunnel provides the direct, shortest horizontal access to the CURIE sites. Sites 0, 1, and 2 are roughly 400 horizontal meters into the Miami tunnel and are connected internally to the Army tunnel. The vertical height to the surface is approximately 200 m for all sites (Keblbeck et al., 2024).

A central operational feature of CURIE is horizontal tunnel access. The published description emphasizes that this enables straightforward logistics, transport, and installation compared with vertical shafts typical of some deep laboratories. Tunnel openings vary from roughly $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$ at the narrowest to $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$ at the widest, while the Army tunnel has a narrowest access of approximately $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$. This geometry is directly relevant to experimental handling, cryogenic delivery, shielding deployment, and staged installation (Keblbeck et al., 2024).

The mine is excavated in gneiss, described as analogous to granite but foliated. The reported consequence is excellent rock stability and safety, allowing flexible siting. EEM infrastructure includes 110 V single-phase and 440 V three-phase power, extensive ventilation, compressed air, and water. Site 2 is specifically described as tailored for a cryogenic low-background facility aimed at developing high-resolution, low-threshold detectors and enabling controlled studies of ionizing radiation, electromagnetic disturbances, and vibrations on superconducting circuits, including qubits and cryo-CMOS (Keblbeck et al., 2024).

By 2025, two research spaces had become the focus of detailed secondary-background simulations: the Subatomic Particle Hideout (SPH), approximately $$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$, brought online in 2024, and Cryolab I (CLI), approximately $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$0, under development. The literature characterizes these spaces as representative cavern environments within a non-flat mountain overburden where local geometry and rock composition materially affect underground spectra and source terms (Keblbeck et al., 7 Oct 2025).

2. Direct muon-flux measurements and overburden characterization

The first quantitative characterization of CURIE’s shielding environment was based on underground muon-flux measurements at Site 0 and Site 1. The detector system used two stacked plastic scintillators with a lead sheet between them, operated in coincidence to suppress gamma backgrounds. The top scintillator $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$1 measured $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$2, the bottom scintillator $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$3 $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$4, and the lead sheet $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$5, with each dimension measured to $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$6. Each scintillator was coupled to a PMT operated at 800 V bias. The coincidence time window was approximately $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$7, and the discriminator threshold on the larger panel was approximately 2 MeV (Keblbeck et al., 2024).

The effective acceptance was determined with GEANT4 geometric Monte Carlo using geantinos and angular weighting from MUTE’s underground intensity map, yielding

$10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$8

Efficiency calibration used single-mode operation of the top panel and the relation $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$9, giving $700\times$0 at Site 0 and $700\times$1 at Site 1. Live times were $700\times$2 at Site 0 and $700\times$3 at Site 1 (Keblbeck et al., 2024).

Muon and gamma contributions were separated by converting channel number to deposited energy using simulated minimum-ionizing muon energy loss in plastic, $700\times$4, and fitting the gamma background with

$700\times$5

where the constant term preserves the muon pedestal below the cutoff and only the exponential term is subtracted. The adjusted total muon counts were $700\times$6 for Site 0 and $700\times$7 for Site 1. Fluxes were then computed as

$700\times$8

giving measured underground muon fluxes of $700\times$9 at Site 0 and $-105.52$0 at Site 1 (Keblbeck et al., 2024).

These measurements correspond to an approximately $-105.52$1 reduction relative to a sea-level reference flux of about $-105.52$2. The same work reports an average equivalent vertical depth of $-105.52$3, placing CURIE in the shallow-underground regime. MUTE predicts nearly identical underground muon energy spectra across sites, with mean underground muon energy $-105.52$4 (Keblbeck et al., 2024).

A notable feature of the measurements is strong angular dependence induced by mountain geometry. At Site 0, a directional comparison between a South-facing orientation $-105.52$5 and a Northwest-facing orientation $-105.52$6 found a measured flux reduction of $-105.52$7, while GEANT4 plus MUTE predicted $-105.52$8. This is direct evidence that topographic anisotropy is experimentally resolvable at CURIE and cannot be collapsed to a single scalar depth without some loss of information (Keblbeck et al., 2024).

3. Computational pipeline and flat-depth normalization

CURIE has also served as a validation platform for a shallow-site simulation chain combining Daemonflux, MUTE, PROPOSAL, and GEANT4. In the muon-flux study, Daemonflux v0.8.1 generated surface muon fluxes using MCEq cascade equations calibrated to the Global Spline Fit and Data-Driven Model inclusive hadronic data with SIBYLL-2.3c; MUTE v2.0.1 propagated muons through rock using PROPOSAL; and GEANT4 handled detector acceptance and effective-area calculations. The overburden was constructed from USGS DEM data over approximately $-105.52$9 around EEM, with DEM accuracy of roughly 1 m in $39.74$0–$39.74$1 and $39.74$2 vertically (Keblbeck et al., 2024).

Slant depth was converted to water-equivalent depth using

$39.74$3

with $39.74$4 at $39.74$5. Density models included standard rock $39.74$6, a weighted average $39.74$7, and quadrant-dependent averages $39.74$8, $39.74$9, $2401\ \mathrm{m}$0, and $2401\ \mathrm{m}$1. Because differences among $2401\ \mathrm{m}$2 and $2401\ \mathrm{m}$3 were below 1% and the difference from standard rock below 3%, the final simulations adopted $2401\ \mathrm{m}$4 and assigned a 3% systematic to inhomogeneities, water saturation, and voids (Keblbeck et al., 2024).

Air gaps in mountain trajectories were corrected using a path-length algorithm that scans repeated $2401\ \mathrm{m}$5 values along $2401\ \mathrm{m}$6 columns over the full $2401\ \mathrm{m}$7 steradian. At $2401\ \mathrm{m}$8 angular sampling, the correction increased the average flux by 1–2%, and a $2401\ \mathrm{m}$9 correction was adopted. With full-scale simulations of approximately 30 million propagated muons, the predicted total underground muon fluxes were $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$0 for Site 0, $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$1 for Site 1, and $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$2 for Site 2, in agreement with measured values at roughly the $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$3 level (Keblbeck et al., 2024).

A distinct contribution of the 2024 work is a new depth–intensity relationship for converting non-flat mountain overburden to equivalent flat depth: $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$4 with $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$5, $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$6, and $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$7. Using this relation, the measured equivalent depths were $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$8 for Site 0 and $1.8\ \mathrm{m} \times 1.8\ \mathrm{m}$9 for Site 1; the simulated values were $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$0, $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$1, and $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$2 for Sites 0, 1, and 2, respectively. For the measured fluxes, Mitrica et al. would yield $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$3–$4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$4, while Mei–Hime would yield $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$5–$4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$6, so the new relation is presented as better aligned with world data at shallow depths (Keblbeck et al., 2024).

The same study identifies this as the first successful validation of the Daemonflux+MUTE+PROPOSAL framework against measurements at depths below $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$7. In practice, this establishes CURIE not only as an experimental site but also as a benchmark environment for shallow-overburden transport and normalization methods (Keblbeck et al., 2024).

4. Muon-induced secondary backgrounds

A subsequent end-to-end study modeled muon-induced secondaries for SPH and CLI with a two-stage mute–GEANT4 framework. Stage 1 used mute v2.0.1 to propagate atmospheric muons through the mountain profile and produce angle-dependent underground differential muon fluxes $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$8. Stage 2 used Geant4 v11-02-01 to transport these muons through a simplified, site-specific rock-and-cavern geometry and to simulate production and transport of electromagnetic, hadronic, and neutrino secondaries. The laboratory was embedded in a homogeneous hemispherical rock volume with a minimum 4 m thickness from hemisphere boundary to cavern wall to ensure shower equilibrium, and secondaries were tracked to their first entry through the rock–cavern boundary (Keblbeck et al., 7 Oct 2025).

The rock models were built from SEM analyses of local samples. For SPH, the density was $4.5\ \mathrm{m} \times 4.5\ \mathrm{m}$9 with mass fractions, in percent, of Si 36.22, O 49.84, Al 5.89, K 1.65, Na 3.47, Mg 1.08, Fe 1.24, and Other 0.61; $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$0, $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$1, and $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$2. For CLI, the density was $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$3 with Si 36.01, O 49.71, Al 5.47, K 2.40, Na 2.42, Mg 1.84, Fe 1.17, and Other 0.98; $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$4, $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$5, and $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$6. The paper explicitly links these quantities to ionization, spallation, pair production, and bremsstrahlung rates (Keblbeck et al., 7 Oct 2025).

Rather than sampling muons from a single mean underground energy, the simulations used the full angle-dependent $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$7. For SPH, $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$8, but replacing the angle-dependent distribution by a single mean energy elevated secondary fluxes by about 7.6% for neutrons. The authors therefore treat angle-dependent sampling as necessary for accurate background prediction in complex topography (Keblbeck et al., 7 Oct 2025).

Research space	Muon-induced neutron flux at rock–cavern boundary	Muon-induced $2.4\ \mathrm{m} \times 2.4\ \mathrm{m}$9-ray flux at rock–cavern boundary
SPH	$$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$0	$$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$1
CLI	$$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$2	$$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$3

For context, the primary muon fluxes at the boundary were $$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$4 for SPH and $$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$5 for CLI. Neutron thresholds were also reported. At SPH, the neutron fluxes above 1 MeV, 10 MeV, 100 MeV, and 1000 MeV were $$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$6, $$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$7, $$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$8, and $$5.2\ \mathrm{m} \times 3.0\ \mathrm{m} \times 3.0\ \mathrm{m}$$9, respectively; at CLI the corresponding values were $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$00, $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$01, $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$02, and $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$03 (Keblbeck et al., 7 Oct 2025).

The predicted neutron spectrum has four components: a thermal peak around $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$04, an epithermal/fast region, evaporation neutrons roughly $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$05–$10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$06, and a spallation component peaking near $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$07 with a high-energy tail to multiple GeV. An absorption-like feature just above $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$08 is described as plausibly linked to resonances in $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$09 and $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$10 in the local rock. The study also reports that the electromagnetic component dominates the total muon-induced flux at CURIE depth, while at approximately GeV energies the proton flux approaches and then exceeds the neutron flux (Keblbeck et al., 7 Oct 2025).

Uncertainties are explicitly systematic-dominated: $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$11 from correlated rock density and muon-energy effects, $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$12 from the Sternheimer parameter approximation, $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$13 from live-time normalization, $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$14 from laboratory wall-area uncertainty, and $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$15 from Geant4 physics-list choice. Muon bundles are neglected on the basis of shallow-depth measurements indicating single-muon fractions $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$16 in the 100–300 m.w.e. range (Keblbeck et al., 7 Oct 2025).

5. Experimental scope and infrastructure requirements

The published CURIE studies consistently frame the facility as suitable for detector R&D, low-threshold detector development, calibration, and environmental studies on superconducting circuits where moderate muon rates are acceptable. The literature also states that experiments using passive shielding and/or active muon veto systems can mitigate residual muon-induced backgrounds such as spallation neutrons. Horizontal access, stable rock, and existing infrastructure are presented as expanding the range of deployable instruments relative to limited U.S. shallow facilities (Keblbeck et al., 2024, Keblbeck et al., 7 Oct 2025).

The Snowmass 2021 topical report on underground facilities for the Cosmic Frontier does not explicitly mention CURIE, Henderson Mine, Edgar Mine, or any Colorado-based underground facility proposals or sites. It does, however, articulate facility requirements for next-generation underground dark-matter programs that are directly applicable to the concept of a Colorado-based laboratory (Cooley et al., 2022).

For noble-liquid experiments, the report identifies next-generation detector scales of 50–300 tons and states that the detector must be in an experiment hall that is large both in floor space and ceiling height, with room for external active veto and shielding. It further specifies significant continuous cooling power, underground liquid-nitrogen capacity extrapolating from XENON1T’s $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$17 tank to a next-generation scale of order $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$18, and on-site cryogenic distillation columns. The cited examples include a 5.5 m purification column releasing some 10s of kW of heat into the cavern and a 3.8 m radon distillation column requiring a few kW of cooling power. The report also calls for large underground low-radon clean rooms with staging areas, cleanliness better than ISO-6, radon concentration better than $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$19, and substantial vertical space and cranes for heavy lifts (Cooley et al., 2022).

For cryogenic bolometers, the report states that a vertical space of at least 4 m is highly beneficial for $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$20 dilution refrigeration, that environmental vibrations should be below $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$21 at all frequencies, and that superconducting electronics such as SQUIDs require quiet electromagnetic conditions and minimized DC magnetic fields. It also emphasizes the value of underground R&D space for developmental and R&D-scale efforts because long-duration events and pile-up complicate above-ground characterization (Cooley et al., 2022).

For other technologies, including skipper CCDs, point-contact Ge detectors, low-pressure gas directional detectors such as CYGNUS, scintillator experiments, and superheated water detectors, the report states that present facility needs are relatively compact in footprint and would generally be met if the environmental controls required by noble-liquid and cryogenic-bolometer experiments are available. This suggests that a Colorado laboratory with the environmental conditions measured and modeled at CURIE could support compact early-stage directional or low-threshold programs while remaining especially relevant to R&D-scale deployments (Cooley et al., 2022).

6. Role, limitations, and broader context

CURIE occupies an intermediate position between surface or minimally shielded installations and deep underground laboratories. The measured reduction in muon flux, approximately $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$22, exceeds the values cited for PNNL at approximately 30 m.w.e. ($10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$23) and Fermilab shallow sites at approximately 225–300 m.w.e. ($10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$24–$10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$25). At the same time, CURIE remains a shallow-underground facility with residual muon intensities of order $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$26, not a deep laboratory in the SNOLAB or SURF sense (Keblbeck et al., 2024).

Several limitations follow directly from this depth regime. Directional muon-flux variations due to mountain topography require orientation-aware shielding and veto strategies. Muon-induced neutrons and spallation products in rock and shielding require continued characterization and mitigation. The 2025 simulation further shows that local geology and overburden geometry alter yields and energy spectra, so simple depth-based extrapolation is insufficient for high-fidelity background budgeting (Keblbeck et al., 2024, Keblbeck et al., 7 Oct 2025).

A common misconception would be to treat all $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$27-class sites as interchangeable once total muon flux is matched. The CURIE studies argue against that simplification: they report measurable azimuthal anisotropy, site-specific rock-composition effects, and a 5–8% bias in secondary fluxes when angle-dependent muon energies are replaced by a single $10^{-1}\ \mathrm{m^{-2}\ s^{-1}}$28. The Snowmass report also cautions, in a different context, that depth requirements for future low-threshold cryogenic-bolometer experiments remain difficult to assess until the relevant R&D matures (Keblbeck et al., 2024, Keblbeck et al., 7 Oct 2025, Cooley et al., 2022).

Within the U.S. underground ecosystem, the Snowmass report concludes that current host laboratories are mostly full and that there is a clear need for expanded underground space tailored to needs beyond the late 2020s. A plausible implication is that a facility with CURIE’s measured accessibility, infrastructure, and environmental characterization could augment national capacity for shallow- to intermediate-depth low-background work, especially for technology development, environmental studies, and experiments that can combine passive shielding with active vetoing. In that sense, CURIE is best understood not as a substitute for deep laboratories, but as a quantified and increasingly well-characterized platform for the segment of low-background research that benefits from substantial muon attenuation without the logistical constraints of deep-lab access (Cooley et al., 2022, Keblbeck et al., 2024).