Eur. Phys. J.

Plus (2012) 127: 114

DOI 10.1140/epjp/i2012-12114-y
THE EUROPEAN
PHYSICAL JOURNAL PLUS
Regular Article

The world deep underground laboratories

A. Bettinia,b
University of Padua “G. Galilei”, Physics Department and INFN, Via Marzolo 8, 35131 Padova, Italy
Laboratorio Subterráneo de Canfranc, Canfranc (Huesca), Spain

Received: 6 July 2012

Published online: 20 September 2012 – 
c Società Italiana di Fisica / Springer-Verlag 2012

Abstract. This paper is an introduction to a series of coordinated articles of an EPJ Plus Focus Point on
underground physics laboratories, written by the directors of the larger ones and by the coordinators of the
principal new projects. The paper is largely based on the text of my lecture Perspectives of underground
physics, given at the Enrico Fermi Varenna International School, Course CLXXXII (2011), Neutrino physics
and astrophysics, reproduced here by permission of the Italian Physical Society. Underground laboratories
provide the low radioactive background environment necessary to explore the highest energy scales that
cannot be reached with accelerators, by searching for extremely rare phenomena. Experiments range from
the direct search of the dark-matter particles that constitute the largest fraction of matter in the Universe,
to the exploration of the properties of the neutrinos, the most elusive of the known particles and which
might be particle and antiparticle at the same time, to the investigation on why our universe contains only
matter and almost no antimatter, and much more.

1 Why go underground?
The Focus Point on “Underground Physics Laboratories” is limited to the deep underground laboratories dedicated to
fundamental physics and astrophysics and other sciences that can profit of the underground location, such as under-
surface geodynamics and biology in extreme conditions. We shall not include laboratories at shallow depths, as those
used to shield experiments on neutrinos from power reactors or even from accelerators. Still different laboratories, at
shallow depths, are those dedicated to rock mechanics and engineering, radioactive waste storage and other technolog-
ical issues, such as the study of long-range thermal and mechanical effects of radioactive waste storage, the interference
with ground water, etc.
In the last half a century physicists have developed a theoretical description of the elementary building blocks of
matter and of the basic forces of Nature, the Standard Model. It is the most comprehensive theory ever built and has
been tested with high precision up to energies of a few hundred times the proton mass and down to distances 10000
smaller than the proton radius. A new collider, the LHC, has started to work now at still higher energy, discovering
the last missing element of the SM, the Higgs boson. But we know already that this, and any accelerator of the future,
will not be sufficient.
The reasons are the following: three of the four basic forces of Nature, strong, electromagnetic and weak, seem to
become equal at high energies. Unfortunately the energy scale of the unification is extremely high, so high that we
will never be able to reach it with an accelerator. It is some eleven orders of magnitude higher than the LHC energy.
Even higher is the Planck scale, the Big Bang energy, at which, presumably, also the fourth force, gravitation, becomes
unified. We need another way. Phenomena characterised by a high-energy scale do, in fact, happen naturally even at
lower, every-day energies. But the higher is their intrinsic energy scale the more rarely they happen.
The deep underground physics laboratories are dedicated to the search for these natural, but extremely rare,
nuclear and subnuclear phenomena, requiring very low radioactive background environment. The background is due
to cosmic rays and to decays of radioactive nuclei present, in traces, in all materials.
The following underground laboratories will be considered in this Focus Point: the Baksan Neutrino Observatory
(BNO) in the Russian Federation, the Canfranc Laboratory (LSC) in Spain, the Kamioka Observatory in Japan, the
Modane Laboratory (LSM) in France, the Gran Sasso Laboratory (LNGS) in Italy, the SNOLab in Canada and the
a
e-mail: [email protected]
b
e-mail: [email protected]
Page 2 of 11 Eur. Phys. J. Plus (2012) 127: 114

Fig. 1. Muon flux as a function of the “equivalent vertical depth” (see text).

Sanford Underground Resarch Facility (SURF) in the USA. The main projects for future facilities to be described are:
the India Based Neutrino Observatory (INO) in India, the China JinPing Underground Laboratory (CJPL) and the
Agua Negra (ANDES) Laboratory in South America.
In this paper I shall introduce the main science items that are common to these facilities. In the last sections I
shall briefly mention the facilities not considered in the articles on the above-quoted laboratories.

2 Backgrounds

The muon flux decreases with the thickness of the rock overburden roughly, but not exactly, exponentially. Under
a flat surface the flux is 10−3 m−2 s−1 at a depth of 2.3 km w.e. (1 km water equivalent is about 300 m of rock),
10−4 m−2 s−1 at 3.7 km w.e. and 10−5 m−2 s−1 at 5.3 km w.e. Under a mountain, like at Kamioka or Gran Sasso, the
angular dependence of the muon flux depends on the shape of the surface. It should be measured, or at least calculated,
to provide the input to the background simulations needed by the most delicate experiments. An “equivalent vertical
depth” can be defined as the vertical depth under a flat surface that would provide the same total muon flux. Clearly,
this is usually substantially smaller than the physical maximum vertical depth. It is shown in fig. 1.
Atmospheric muons are not only a background, they are also an important object of study; such a study led to
the discovery of neutrino oscillations. Moreover, muons are precious for calibrating the tracking detectors.
The local environment is a source of background, in particular neutrons and gamma rays, due to the presence of
the ubiquitous radioactive nuclides and to the secondaries of the muon interactions with the rocks and other materials.
These backgrounds can be reduced by many orders of magnitudes by shielding, i.e. by surrounding the detectors with
radio-clean materials absorbing the radiation.
Gamma rays originate from radioactive decays in the rocks (and in the concrete used for lining). Above shallow
depths of a few 100 m the gamma flux is independent of depth, but dependent on the local geology. The flux is a few
orders of magnitude lower than on surface, with typical values of the order of 104 m−2 s−1 . The energy spectrum falls
rapidly with increasing energy up to about 3 MeV.
Neutrons originate mainly from (α, n) and fission processes (U and Th) in the rocks and in the concrete used for
lining. The energies range from thermal to several MeV. Consequently these neutrons are not difficult to shield. The
neutron flux is substantially independent of depth (if > 100 m or so). It depends on the local geology. Typical values
of the n flux are from a few to several 10−2 m−2 s−1 . The largest fraction of the neutron flux originates within several
decimetres from the surfaces of the experimental halls. Consequently, the radioactivity of the concrete used for lining
may dominate the flux and should be carefully controlled.
Muons interactions in the rocks produce dangerous neutrons at a depth-dependent rate. Fluxes are typically 3–4
orders of magnitude smaller than the main n flux. The energies are large, up to several GeV, demanding thicker
shields.
Even more dangerous are the neutrons produced in the shields, in the detector and in the materials around it. In the
case of fast reactions, namely if the neutron immediately follows the muon, the background can be reduced by detecting
the muon and by anticoincidence. Neutrons from metastable nuclides can be eliminated by delayed coincidence and
Eur. Phys. J. Plus (2012) 127: 114 Page 3 of 11

working at larger depth. Their effect depends on the details of the experiment and is generally more severe for high-Z
materials.
More in general, the sensitivity to each background depends on the experiment. Consider, for example, the neutrons
always produced by the muons in a liquid scintillator detector. Once moderated in some 250 μs they almost always
end up by being captured by a proton with the emission of a 2.2 MeV gamma. Experiments searching for neutrino-less
double-beta (0ν2β) decay with the signal region at or below this energy need to take particular care of this problem.
Radon is a radioactive gas always present in the atmosphere, being continuously produced by the decay of 226 Ra
present in the rocks. An important source of Rn is ground water. Rn activity, which is typically 50–100 Bq/m3 in open
air, may be larger by one or two orders of magnitude in closed underground cavities (and in some sites on the surface,
too). The Rn activity has strong variation in time, both periodic (daily, seasonal, yearly) and aperiodic, depending on
the atmospheric conditions and other reasons. In the laboratory it is reduced by ventilation. The equilibrium activity
depends on the Rn content of the input air, on the Rn emanation rate and on the ventilation speed. The input air
duct, which may be a few km long, should be made of SS or similar materials to avoid Rn collection in the duct itself.
Air can be filtered with special antiradon factories underground. Such a facility at LSM produces air with 10 mBq/m3 .
All experiments need to be shielded. Indeed, the shield thickness determines the physical size of the most sensitive
experiments. For example, in several of the present-generation 0ν2β and dark-matter search experiments a major
shield component is a water tank with a typical diameter of 10–15 m, requiring a hall of at least 20 m in diameter
and height. This is much larger than the detectors themselves that are typically one metre across. Notice that the
maximum safe diameter tends to decrease while the costs tend to increase with depth.
The ultimate background sources, as already mentioned and as will be discussed later, are in the detector materials
themselves. The main steps to control them are the following.
A background model is developed by writing proper simulation computer codes, including all the detector compo-
nents, their materials and the assumed radioactive contaminants (at the state of the art). The model will foresee the
total background index (BI) in the energy region of the expected signal, namely the number of background counts
per unit energy interval (keV), sensitive mass (kg) and time (usually in days for dark matter, years for 0ν2β) and the
background budget, namely the tolerable contributions to the BI of each detector component.
The BI limits the sensitivity. Namely, to claim for a signal the experiment should have a counting rate significantly
larger than the background. However, background models that are reliable enough to design an experiment are not
such to safely claim a discovery. In practice, the BI should be as close to zero as possible. In general, a positive
claim based on having observed more events than those predicted by the background model cannot be considered
reliable. Consequently, increasing the sensitive mass increases the sensitivity only marginally if not accompanied by a
corresponding decrease in the BI. This usually requires years of R&D.
The background model must be checked by a number of specific measurements.
Finally, all the detector components must be screened in order to guarantee that their contributions to the back-
ground budget is inside the specifications. Several underground laboratories operate a screening facility with several
Ge detectors and other counters.
Notice that the internal backgrounds are the principal contributors to the background budget for the present- and
next-generation 0ν2β and dark-matter experiments (the principal ones) already at 1000–1500 m depths. Consequently
there is no reason to go deeper than that, at least up to the ton scale. Large proton decay and next-generation neutrino
detectors on accelerator beams need even less coverage, about 1000 m.

3 A bit of history
The first experiments underground were back in the 1960s. They were performed in very deep cavities in mines. In
1965 the first “natural” neutrinos, produced by cosmic-rays interactions in the atmosphere (atmospheric neutrinos),
were discovered, almost at the same time, by two groups working in the Kolar Gold Mine in South India [1] at a
depth of 2700 m and in the East Rand Property Gold Mine in South Africa [2] at a depth of 3200 m. A cavity in
the Homestake Mine in S. Dakota in the USA was the site of the fundamental experiment by Ray Davis [3], who
first detected in 1968 neutrinos coming from the Sun. Even more important was the comparison of the observed flux
with the accurate calculations by John Bahcall [4]. The observed flux was only about one third of the expected one.
As we have gradually learned with other underground experiments since then, this was the first evidence for “neutrino
oscillations”, a process that can happen only if neutrinos are different from those of the Standard Model.
A hall in a mine is not however a laboratory. A real laboratory must provide services underground and on surface.
The electric power must be of high quality and, for a fraction of it, continuous operation must be guaranteed. A quantity
of energy equal to the input one must be taken to the surface per unit time; this may require dedicated water cooling
circuit, reaching the surface at a few kilometres distance. Proper quality air (dust free, with low Rn level, stable in
temperature and humidity, etc.) must be provided at a sufficient rate. The same must be done for the water, taking
into account that different experiments may require different levels of radio-purity. Safe and easy-to-use access tunnels
or shafts must be guaranteed.
Page 4 of 11 Eur. Phys. J. Plus (2012) 127: 114

Safety is always an issue, both for the laboratory users and for those of the nearby different structures (mine,
freeway, aqueduct, etc.). Let us consider in particular the importance of fire prevention and seismic rules. Any possible
environmental impact, both on the surface and underground, should be evaluated and avoided. Of particular impor-
tance is the possible interference with the aquifer, usually present inside a mountain, both during the construction
and the operational phases. A careful training of the users on the matter is necessary.
In a few words, the scopes and the needs of a laboratory are completely different from those of a mine.
A laboratory needs facilities on the surface: mechanical shop, electronic, chemical, etc. laboratories, storage rooms,
offices, meeting and conference rooms, sleeping rooms, IT services, etc.
The first full-fledged underground laboratory is the Baksan Neutrino Observatory (BNO). In 1966, under the action
of M. Markov, head of the Physics Division, the Academy of Sciences of the USSR obtained a Decree of the Soviet
Government for the construction of the underground and surface facilities (the “Neutrino village”). Scientific activity
started under the leadership of G. Zatzepin and A. Chudakov. The underground laboratory, including a horizontal
access tunnel, was excavated and built under the mount Andyrchi in the Caucasus.
In 1979 a double tunnel was under construction for the freeway between Rome and the Adriatic Sea under the Gran
Sasso Mountain. A. Zichichi, at that time President of the INFN, saw the unique opportunity of building a world-class
underground laboratory with a broad spectrum of potential scientific programme, including a future neutrino beam
from CERN and a surface detectors array on the top of the mountain. In that year, after checking the geological and
radio cleanness properties of the rocks, he proposed to the Italian Parliament to build such a large laboratory. In 1982
the Parliament approved and funded the construction, which was completed in 1987, at a very low cost.
In 1983 M. Koshiba established the Kamioka Underground Observatory, in a modern working mine with horizontal
access, to host the KamiokaNDE water Cherenkov detector. Later on its bigger successor was built, SuperKamiokande,
which in 1998 discovered neutrino oscillations in the muon-neutrinos from the atmosphere, complementing the Davis
and Bahcall discovery.
Several other sites were built after those, of different sizes, different level of support, different degree of opening to
the international community and different rigor in the selection of the experiments.

4 Neutrino physics
4.1 Oscillations and MSW effect

In the Standard Model neutrinos are assumed to have no mass and to be stable. Experiments in underground labora-
tories have shown that both assumptions are wrong. Neutrinos have been observed to change flavour in two different
ways. The first phenomenon is neutrino oscillations, discovered by SuperKamiokande [5] in 1998. Cosmic-rays colli-
sions in the atmosphere produce electron neutrinos and muon neutrinos. SuperKamiokande has shown that the flux
of νμ ’s reaching the detector from very long distances is about 1/2 of the expected value. They convert into, oscillate
into, another neutrino species. The second phenomenon is the flavour conversion in matter, the Mikehev-Smirnov-
Wolfenstein [6–9] (MSW) effect. It was discovered, by a series of complementary experiments measuring the νe flux
from the Sun as well as the one of the other flavours, via neutral currents.
Neutrino oscillation is a purely kinematical phenomenon. It happens both in vacuum and in matter. It has been
discovered as the disappearance of the νμ ’s indirectly produced by cosmic rays in the atmosphere. Typical neutrino
energies (E) are from sub-GeV to multi-GeV, while the flight lengths (L) are up to a few thousands kilometres.
The probability to observe the flavour state νβ in a beam initially pure να in its first approximation is proportional
L(km)
to oscillating terms of the type Pαβ = A(θ12 , θ23 , θ13 ) sin2 1.27(m2i − m2j )(eV2 ) E(GeV) . The oscillation amplitudes A
are functions of the three mixing angles, θij ∈ [0, π/2]. The oscillation frequencies are proportional to the differences
between the squares of the masses of the eigenstates. We call Δm2 the square mass difference of the “atmospheric”
oscillation. The phenomenon (with a good approximation) is independent of the sign of m2i − m2j and, also, on
sgn(π/2 − θij ). Oscillations have been studied later with νμ beams from accelerators and ν̄e from reactors.
The MSW effect is a dynamical phenomenon, due to the νe e interaction in the matter. The interaction energy
depends on neutrino energy and matter density, and has opposite sign for neutrinos and antineutrinos. While crossing
a variable density medium νe ’s may reach a critical density layer where a resonant transition to a coherent superposition
of the other flavours may happen (level crossing). The phenomenon is the dominant process in the Sun, for above
MeV neutrino energies. We call δm2 the corresponding square mass difference. δm2 happens to be much smaller than
Δm2 . The matter effect does depend both on the sgn(δm2 ) and sgn(π/2 − θij ). In a supernova, due to its much larger
density, two level crossings are possible, at δm2 and at Δm2 .
The accumulated evidence tells us that the flavour states are not the mass eigenstates. We call the latter ν1 , ν2 and
3
ν3 , and m1 , m2 and m3 their masses. Flavour states are linear combinations of the eigenstates, νl = i=1 Ul,i νi , where
l = e, μ, τ and U is the mixing matrix. This can be written as the product of three rotation matrices including a phase
Eur. Phys. J. Plus (2012) 127: 114 Page 5 of 11

Fig. 2. Neutrino mass spectrum. Left: “normal” hierarchy. Right: “inverse” hierarchy.

factor, as in the case of quarks, and a fourth diagonal matrix with two more phases. The latter can be absorbed in the
wave functions only if neutrinos and antineutrinos are different (Dirac) particles. However, a fundamental unknown
property of neutrinos is whether or not they are their own antiparticles, as foreseen by E. Majorana in 1937 [10].
With cij = cos θij , and sij = sin θij we have
⎛ ⎞⎛ ⎞⎛ ⎞⎛ ⎞
1 0 0 c13 0 s13 eiδ c12 −s12 0 1 0 0
⎜ ⎟ ⎜ ⎟
U = ⎝0 c23 s23 ⎠ ⎝ 0 1 0 ⎠ ⎝s12 c12 0⎠ ⎝0 eiα 0 ⎠ .
0 −s23 c23 −s13 e−iδ 0 c13 0 0 1 0 0 eiβ

In total there are nine quantities to measure, three masses, three mixing angles and the three phases. The latter, if = 0
and = π, give CP -violation effects in the lepton sector. Majorana phases are, even if present, irrelevant in oscillation
and matter conversion phenomena; they appear only in 0ν2β decay.
Our present knowledge of neutrino mixing and masses comes from a number of different experiments, all in under-
ground laboratories. νμ disappearance experiments both with atmospheric neutrinos (SuperKamiokande and MACRO)
and accelerator neutrinos (K2K in Japan and MINOS in the USA) have measured |Δm2 | and θ23 ; disappearance of so-
lar νe , evidence for νμ +ντ appearance (SNO in Canada) and disappearance on long base line of reactor ν̄e (KamLAND)
have measured δm2 and θ12 . Moreover, ντ appearance (OPERA at LNGS) and νe appearance (T2K at Kamioka) are
reaching the three-standard-deviation sensitivity.
Figure 2 shows schematically the neutrino mass spectrum. It consists of a doublet (m1 and m2 ) and a singlet (m3 ).
Global fits of available data have been done, giving similar results. The fit by Fogli et al. [11] gives the one-standard-
deviation intervals,
    
Δm2  ≡ m23 − m22 + m21 /2 = 2260–2470 meV2 δm2 ≡ m22 − m21 = 73.2–78.0 meV2 .

We do not know whether m3 is larger or smaller than m1,2 nor the absolute value of the masses. An indirect upper
limit on each of the masses is given by cosmology, approximately mi < 200 meV.
The longer period oscillation, the one at Δm2 , is mainly νμ → ντ but has a minority component of probability
2
proportional to θ13 of νμ → νe . The latter can be measured either in a reactor ν̄e disappearance at short (about 2 km)
distance (Daya Bay, Reno and Double CHOOZ) or on νμ → νe on an accelerator beam (T2K). Up to this year, only
an upper limit of θ13 was known. A major breakthrough came in spring 2012, when two ν̄e disappearance experiments
from power reactors, Daya Bay [12] in China and RENO [13] in Korea, reported two fully compatible measurements,
with similar uncertainties.
The above-mentioned fit by Fogli et al. [11] gives the following ±1 σ intervals:

sin2 θ12 = 0.291–0.323 sin2 θ23 = 0.37–0.43 sin2 θ13 = 0.021–0.028.

Notice that the relative uncertainties on the three angles are similar: 5.4%, 13% and 13%, respectively.
Knowing that θ13 has a relatively high value, close to the previously known upper limit, the programme to measure
the sign of Δm2 and the CP phase, can be now defined.

4.2 Solar neutrinos

Let us now have a closer look at the solar neutrinos. Neutrinos are produced by fusion reactions in the core of the
star as pure νe by charged-current (CC) weak interactions. They will then move in matter and interact, again via CC,
with the electrons of the medium (one can show that the interactions with protons and neutrons are not relevant for
Page 6 of 11 Eur. Phys. J. Plus (2012) 127: 114

√
Fig. 3. Mass eigenvalues vs. A = 2 2GF ne Eν (ne is the electron density, Eν is the neutrino energy).

the phenomenon under study). Due to that interaction the neutrino Hamiltonian in the medium differs from that in
vacuum and its eigenstates are not the vacuum eigenstates.
Figure 3 shows the two eigenvalues (m̃i ) relevant for the Sun as functions of the density, ranging from zero density,
hence in vacuum, to the density in the core. In the Sun both oscillations, like in a vacuum, and the MSW effect can
take place. The former correspond to the kinetic term in the Hamiltonian, the latter to the dynamic one due to the
interaction with the electrons of the medium. The ratio between the two is
δm2 /(2Eν ) δm2
β= √ = ,
2GF ne A

where ne is the electron density.

The upper mass eigenstate ν̃2 is at high densities practically identical to the flavour state νe . The higher mass
eigenvalue gradually decreases with decreasing density. In its route to the exit the neutrino state follows the upper
curve and sooner or later it may encounter a layer in which the density satisfies the level crossing conditions, which
is 1/β = cos 2θ12 . The condition depends on the neutrino energy and can be satisfied, at the solar densities, above
about one MeV. At the crossing, the particle follows the upper curve, provided that the density variation is not too
fast. This condition, adiabaticity, is satisfied if θ12 > 1–2◦ , which is indeed the case.

4.3 The nature of neutrinos

In the SM neutrinos are assumed to be Dirac particles. In 1937, E. Majorana proposed [10] a new theory for neutral
spinors alternative to the “simple extension of the Dirac equations to neutral particles”, “even though it is perhaps
not yet possible to ask experiments to decide”. After 75 years we do not have the answer to this fundamental question,
but we have the means to actively ask the experiments a decision. A recent review, by J.J. Gomez Cadenas et al. [14]
has been recently published.
Majorana neutrino and antineutrino are two states of the same, massive, particle: neutrino is the state of the
negative helicity, antineutrino the state of the positive one. If this is the case, an extremely rare weak decay can
happen, the 0ν2β decay. Nuclides stable under β-decay decay via a second-order weak process, the two-neutrino
double-beta decay (2ν2β) N (A, Z) → N (A, Z + 2) + 2e− + 2ν̄e , which is a very rare, but standard process. It has been
observed in 10 nuclides with lifetimes of the order of 1019 –1021 y.
If it is a Majorana particle, the antineutrino produced at a nucleon can be absorbed by another one as neutrino:
the neutrino-antineutrino propagator mediates the 0ν2β process N (A, Z) → N (A, Z + 2) + 2e− . The observation of
this process would prove the Majorana nature ofthe electron neutrino. The decay probability is proportional to the
3
square of the so-called “Majorana mass” Mee = | i=1 Uei 2
mi |. To be precise, the Majorana mass is a 3 ×3 matrix Mlk ,
where l and k are the lepton flavours; Mee is its electron-electron element, the only one to be, in practice, accessible to
measurements. Notice that the phases of the complex Uei mixing matrix elements can induce cancellations. Figure 4
shows Mee as a function of the smallest neutrino mass, for the two possible hierarchies, calculated by Bilenki and
Giunti [15]. One notices that, unfortunately, Mee can be as small as zero. Consequently, if Mee = 0 neutrinos are
Majorana particles, but if Mee = 0 they may be Majorana or Dirac.
The measured, or limited, quantity is the half-life. It is related to the Majorana mass by the equation
 2 2
−1
T1/2 = GZ (Qββ ) M 0ν  Mee ,
Eur. Phys. J. Plus (2012) 127: 114 Page 7 of 11

Fig. 4. “Majorana mass” as a function of the lightest neutrino mass [13].

where GZ is the phase space volume, depending on the nucleus, which is not difficult to calculate (however the effect
of the atomic cloud on the beta electrons must be considered) and M 0ν is the “nuclear matrix element”. The latter are
very difficult to evaluate and have been uncertain by large factors. The situation is improving and recent calculations
agree within a factor of two [16–18]. Further work should reduce the uncertainties to 20–30%.
The majority of the experiments are based on the “calorimetric” approach, in which the source coincides with the
detector. The energy deposited by the two electrons is accurately measured. Its distribution is a continuous spectrum
from 2ν2β, which is certainly present, and a line after its end point if 0ν2β exists. Backgrounds can contribute both with
continuous spectra and with monochromatic lines, which may be close, sometimes very close, to Qββ . The sensitivity
of an experiment to the half-life T1/2 in the presence of a background index BI is proportional to the quality factor
 1/2
Mββ × T
FT1/2 ∝ ,
A × BI × ΔE

where Mββ is the sensitive mass of the double-beta active isotope (which is a fraction, say fββ of the total mass M ),
T the exposure time and ΔE, the energy resolution. Hence, the sensitivity to Mee scales as the fourth root of the
sensitive mass in the presence of background. This implies that the background must be reduced substantially to zero
in the region of interest (ROI), namely an energy interval of the order of the energy resolution (FWHM) around the
known position of the signal. In these conditions the sensitivity to Mee scales, in a first approximation, as the square
root of the exposure. Notice also that a reduction by a factor, say, 2 of the background index is equivalent to doubling
the isotope mass and similarly for the energy resolution.
As already mentioned, the detectors need to be shielded from the ambient radiations underground. Very small
radioactive traces in the inner parts of the shields, in the detector components, including read-out electronics, cables,
supports, etc., contribute to the background budget. Particularly dangerous are the surfaces.
Ultimately, the background from the much more frequent 2ν2β decay will remain. It is controlled only by the
5 2ν
S me Qββ T1/2
energy resolution. The contribution of the tail of its energy distribution in the signal region scales as B ; (ΔE)6 T1/2
0ν .

This expression shows the importance of the energy resolution.

To fix the orders of magnitude, we can say that for Mee = 100 meV, approximately the same number of events
per unit detector mass, 10−2 counts/(kg y), are expected for three technologies of the present generation experiments
based on natural TeO2 , 86% enriched Ge and 90% enriched Xe. The corresponding background-free condition is
BI × ΔE < 10−2 ct/(kg y).
Tagging of the final state by identifying the daughter atom, as foreseen by the EXO programme, or by pattern
recognition via tracking, as foreseen by NEXT, will be a powerful tool to discriminate the background.

5 Direct search for dark matter

The direct searches for dark matter look for the WIMPs (Weakly Interacting Massive Particles) elastic scattering on
nuclei, detecting the energy deposited by the recoiling nucleus.
If dark-matter particles are neutralinos (χ), their interaction cross-sections can be calculated, in principle, on the
basis of SUSY models. However, the unknown parameters of the theory are so many that the results span over many
Page 8 of 11 Eur. Phys. J. Plus (2012) 127: 114

orders of magnitudes in uncertainty. Two basic possible couplings can be distinguished: a) “spin dependent” (SD), an
axial vector coupling to the nucleons spin in which only unpaired nucleons couple; b) “spin independent” (SI), a scalar
interaction with the mass of the nucleons, proportional to A2 . This makes the search in the SI sector more sensitive
and we will concentrate on that. The WIMPs flux is uncertain too, depending on the assumed model of the Galactic
halo model.
The results of the calculations and of the experiments are usually represented in the plane of the cross-section,
reduced to a single nucleon, σp vs. the WIMP mass mχ . This corresponds to projecting on a plane the multidimensional
parameter space (local χ density, SI and SD cross-section ratio, ratio of SD coupling to neutrons and to protons, local
velocity, etc.).
There is no firm theoretical upper limit on mχ , but one can take reasonably 1000 GeV; the lower limit is about
6 GeV from cosmology. The theoretical expectations for the cross-sections have even more uncertainties, ranging from
10−5 to 10−12 pb. The corresponding rates depend on the target nucleus and on the energy threshold. Typically, they
may range from 1 to 10−7 counts/(kg keV y). The search for super-symmetric particles has just started at the LHC.
In the case of a discovery, the WIMP parameter space might be better defined. The “ultimate” detector should have
a sensitive mass of the order of 10 t with a background rate smaller than a few counts per year in a 10 keV window at,
say, 30 keV. Today experiments (not looking for modulation) have exposures of the order of 10–100 kg d, thresholds at
several keV and few background events.
Experimentally, the target coincides with the detector, in which one tries to detect the nucleus elastically hit by
the WIMP. The energy is deposited in three forms: as ionisation charge, scintillation light or phonons.
The experimental challenges are the extremely low signal rate, small energy deposits (several keV) and a signal
spectrum decreasing with increasing energy, similarly to backgrounds. The detectors must be operated in an under-
ground laboratory, but this is not enough. Specific R&D is needed for achieving the highest radio-purity and for the
systematic screening of all the set-up components, including the shields and the detector itself.
We can distinguish three basic backgrounds: 1) electromagnetic, due to γ’s, which is the dominant one; 2) radioactive
contaminants of the surfaces and their surroundings, which release only part of the energy in the sensitive volume,
simulating WIMP signal; 3) neutrons, which give nuclear recoils indistinguishable from WIMPs.
Backgrounds can be modelled, but the accuracy and the reliability of the models is limited. Hence, evidence for
dark matter cannot be claimed solely by observing a discrepancy with the background model. Consequently, the
experiments develop, in a blind analysis, selection criteria to define a “background-free” region in the experimental
parameter space. Then within the assumed model, for each mass mχ , the maximum possible signal rate allowed by
the data is calculated. The result is reported as an “exclusion plot” in the σp vs. mχ plane.
A second method, used so far only by the DAMA Collaboration, searches for a specific positive signature, namely
the annual modulation of the signal rate expected for the following reason. The solar system moves in the galaxy
with a speed of about 230 km/s. The velocity of our detector in the halo is modulated, with one year period, by the
speed of the Earth around the Sun of about 30 km/s. Correspondingly, the signal rate is expected to be maximum in
June, when the two velocities are in the same direction, minimum in December when they are opposite. Indeed the
DAMA (100 kg) and the DAMA/LIBRA (250 kg) experiments, with a set of hyper-pure NaI detectors have observed
the annual modulation with all the expected characteristics. This evidence has not been independently confirmed yet.
The ANAIS programme at LSC is presently in an advanced stage of development of hyper-pure NaI detectors.

6 Nuclear astrophysics
The majority of the nuclear reactions in the stars and in other astrophysical environments proceed at very low energies.
Consequently, their cross-sections are extremely small. An accelerator facility underground helps in protecting the
experiment from the cosmic-rays–originated backgrounds. To fix the orders of magnitudes, in some cases the interaction
rate may be as small as one per week.
Such a facility, LUNA, exists at LNGS since two decades with presently a 400 kV accelerator; the next stage, at a
few MeV, is under study. At these higher energies the accelerator and its experiments must be located in a dedicated
room, separated from those of the low background experiments, due to the neutron background they produce. The
development of a similar facility is under design also at LSC.

7 Other research chapters

7.1 Supernova neutrinos

The observation of the neutrino burst from a supernova collapse, the measurement of the energy spectra of the different
flavours and their evolution in time would provide extremely relevant information both on neutrino physics and on
astrophysics. This is an important part of the scientific programmes of several underground laboratories.
Eur. Phys. J. Plus (2012) 127: 114 Page 9 of 11

7.2 Geo-neutrinos
Antineutrinos produced in the Earth crust by radioactive decays of 238 U and 232 Th have been observed both by
Borexino at LNGS and KamLAND at Kamioka. They provide a novel tool to geology, a sector that will presumably
expand.

7.3 Gravity waves

The next generation of gravitational waves interferometers will go underground. The background to be suppressed is,
in this case, is the environmental noise due both to atmospheric phenomena and human activity.

7.4 Geodynamics
Smaller interferometers, with lengths of the order of 100 m, are used at LNGS, BNO and at LSC to monitor and
measure with nanometric sensitivity geodynamic phenomena at low frequencies, down to DC. Such are the slow fault
slippages (observed for the first time at LNGS), the Earth tides, the evolution of the aquifer (both seasonal and due
to seismic events), etc. The programme can be coupled with accelerometers and seismometers network of smaller or
larger aperture, integrated with counterparts on the surface.

7.5 Life under extreme conditions

Very special micro-organisms may be living deeply in the rock. How deep underground does life extend?

8 Other underground facilities

The articles in this Focus Point do not cover all the underground research sites. Other, usually smaller, underground
facilities exist in a number of Countries, run by a university or by a research institution and hosting one or a few
experiments. I will summarise here their characteristics, following and updating my review “The world underground
science facilities. A compendium” [19].

8.1 CUPP. Centre for Underground Physics in Pyhäsalmi (Finland)

http://cupp.oulu.fi/.
The centre is hosted in a working mine. Several cavities, dismissed by the mine, are available at different depths
down to 980 m, for a total area of more than 1000 m2 . Presently, the mine works at a depth between 1000 m and
1400 m. The access is both via a shaft and an inclined tunnel. The EMMA experiment has been installed. Small lab
and office space is available in a surface building. A guest house is also available.

8.2 Solotvina Underground Laboratory (Ukraine)

http://lpd.kinr.kiev.ua/LPD SUL.htm.
The Laboratory was constructed in 1984 under the leadership of Yuri Georgievich Zdesenko by the Lepton Physics
Department of the Institute for Nuclear Research (Ukrainian National Academy of Sciences) in a salt mine. The
laboratory space is divided into a Main Hall: 30 × 20 × 8(h) m3 and four chambers 6 × 6 × 3(h) m3 . The total area is
about 1000 m2 . Access is vertical by the mine cage, taking into account the timetable of the mine.
The lab is 430 m deep in salt. The muon flux is 1.7 × 10−2 m−2 s−1 . The n flux is < 2.7 × 10−2 m−2 s−1 . The radon
concentration in air is 33 Bq m−3 . The principal research lines are on 0ν2β decay.

8.3 Y2L. YangYang Laboratory (Korea)

http://dmrc.snu.ac.kr.
The Dark Matter Research Centre (DMRC) of Seoul National University operates the centre. The available area is
100 m2 underground and 100 m2 space for office space, computing and detector test facilities are available on surface.
Access is horizontal by car. The lab utilizes the space in the tunnel of the host YangYang Pumped Storage Power
Plant. The rock overburden is 700 m with a μ flux of 2.7 × 10−3 m−2 s−1 . The neutron flux is 8 × 10−3 m−2 s−1 for
1.5 MeV < En < 6.0 MeV. The radon activity is 40–80 Bq/m3 . The underground space is mostly occupied by the
Korea Invisible Mass Search (KIMS) experiment, currently taking data for a WIMP search with 100 kg CsI(Tl) crystal
detectors. Other activities include R&D for 0ν2β decay and background measurements with a HPGe counter.
Page 10 of 11 Eur. Phys. J. Plus (2012) 127: 114

8.4 Oto Cosmo Observatory (Japan)

http://wwwkm.phys.sci.osaka-u.ac.jp/info/syoukai/oto-e.html.
The laboratory has been developed by H. Ejri of the Osaka University and collaborators. Its area of consists of
Lab. 2 (50 m2 ) hosting ELEGANT V and MOON on dark-matter search with NaI and 0ν2β decay of 100 Mo. Lab. 1
(33 m2 ) hosts ELEGANT VI on 0ν2β decay of 48 Ca and dark-matter search with CaF2 . The rock coverage is 470 m,
with a μ flux of 4 × 10−3 m−2 s−1 . The neutron flux is 0.4 m−2 s−1 . The access is horizontal through a non-used railway
tunnel, which also provides the non-forced ventilation.

8.5 SUL. Soudan Underground Laboratory (USA)

http://www.soudan.umn.edu/.
The pretty large infrastructures are built in a former mine, now Soudan Underground Mine State Park. The
“Soudan” lab, which is 72×14×14(h) m3 , about 1000 m2 area, hosted CDMS2, a leading dark-matter search experiment,
which is now completed. The “MINOS” lab is 82×16×14(h) m3 , about 1300 m2 area, and hosts the MINOS experiment
on the NUMI neutrino beam from Fermilab.
Access is vertical via a two compartment slightly angled shaft. Diameters in excess of 1 m and lengths in excess of
10 m pose a problem. Access outside normal operating hours is possible. Normal laboratory safety requirements are in
place. The historic Park offers mine tours during the summer months to the public, and winter tours to school groups.
Some tours utilize a visitor’s gallery available in the MINOS laboratory, but do not interfere with either operation or
installation of equipment.
The overburden is 700 m of Soudan rock (2 km w.e.). The μ flux is 2 × 10−3 m−2 s−1 . The neutron interaction rates
are approximately 10 kg−1 d−1 (from U/TH, low energy) or 0.01 kg−1 d−1 (muon generated in the rock).
The Soudan mine has natural ventilation —about 550 m3 /h for the level of the laboratories. Half of this is diverted
to ventilate the MINOS and Soudan spaces. This results in a complete air change every 110. The radon concentration
is seasonal, varying from 300 Bq/m3 in the winter to 700 Bq/m3 in the summer.
The major facility on the surface is a receiving building with a 7.5 t crane and two office areas, kitchen and sanitary
facilities. It occupies approximately 650 m2 .

8.6 The DOE Waste Isolation Pilot Plant (WIPP)

http://www.wipp.energy.gov/.
The infrastructure hosts the EXO experiment, a leading experiment in the search for 0ν2β decay with an enriched
136
Xe liquid Time Projection Chamber. The overburden is 650 m, partially of rock partially of salt.

9 General considerations
Underground laboratories differ in many important aspects.
Only one laboratory, BNO, has an especially built horizontal access tunnel.
Several laboratories, LNGS, LSM, LSC and the ANDES project have been built, or are foreseen, near to a freeway
tunnel. The main excavations can be done only in phase with the tunnel construction, being it no more possible when
the tunnel is open to the traffic. This window of opportunity gives the chance of substantially reducing the construction
and operational costs. The advantages to “drive in” to the experiments are obvious. A mine with horizontal access,
without any time restriction, as in the case of Kamioka, gives the same advantages as well as a service tunnel for a
hydroelectric power plant as for the INO and CJPL projects. The presence of a permanent horizontal infrastructure
in the massif gives also the opportunity in the design stage of the underground laboratory in the area to accurately
study the rock quality both from the engineering point of view, by an adequate number of boreholes, and from the
radio purity aspect, by measuring the radioactivity of the ambient and of a number of samples. A proper location for
an optimized design is then defined.
Other laboratories, like SNOLab, have been built in a working mine with which they share the vertical access. This
situation is not ideal, for different reasons: the maximum size of the objects that can be introduced is limited by the
shaft diameter (as mentioned for Soudan, for example) the site(s) of the experimental halls may need to be far from
the shaft, increasing construction and operational costs; the access must be planned in order not to interfere with the
work of the mine. On the other hand, there are also advantages, as, for example, the opportunity to take profit of the
safety regulations and management of the mine.
In one case, SURF, the choice was to build a new laboratory in a former mine.
Eur. Phys. J. Plus (2012) 127: 114 Page 11 of 11

Some considerations on the investments costs may be useful. The surface infrastructures are an essential component
but their costs is only a small fraction of the total, ranging from 3 M¤ for a medium-size laboratory to some 20 M¤
for a large one.
The cost of the underground facility is obviously an increasing function of the excavated volume, above the fixed
costs to be covered at the start. The rock consolidation costs scale with the surface area. Other factors contribute. I
can give the following examples that I know.
LNGS. The project was approved by the Italian Parliament with two bills to the Italian public road department
(ANAS). The total appropriation was 77.159 MLit spent between 1983 and 1987. The project was complete, including
excavations for about 190 000 m3 , underground services (power ducts and two air pipes running in the freeway tunnel),
the procurement of a vast surface for the campus (half of that is presently used), a service building, which is small
but was sufficient for the first phase of the activity and related structures, etc. Considering the official inflation index
(factor 2.37) in Italy over the period 1985–2010 and converting to Euro one gets 96 M¤ (2011).
In 1990 the Italian Parliament approved the funding of two more halls and of an independent-access tunnel, for a
total of 90 Glit. However, the changed political climate in Italy delayed the project. In the early 2000, the political
climate was changed back to positive and I, as the lab director, coordinated the executive project for the access tunnel
(necessary, I think, for safety reasons). The tunnel had to be about 5 km long, of circular cross-section with 6.5 m
diameter. New service infrastructures including fresh air input and air (and smoke) extraction, new electrical power
input, infrastructure for traffic regulations, etc., were included. The cost was evaluated in 2002 in 60 M¤. Correction
for inflation (a factor 1.23) gives 70 M¤ 2011. Tunnels (and drifts) have large volumes and consequently are expensive.
The operation cost is around 15 M¤ per year.
The ULISSE project of LSM foresees a new Gran Sasso style experimental hall to host a large 0ν2β decay experiment
and a dark-matter next-generation experiment, with access and related services, for a volume of 17000 m3 and an
estimated cost of 15 M¤.
SURF project. As already mentioned, this project aims to build a scientific top-level laboratory in a former mine.
The Homestake Facility has benefited from an extensive geotechnical evaluation program. It has undergone extensive
rehabilitation and restorations and is now a fully functional research facility. This facility will be uncompromised by
competition with mining, transportation or other uses. The dedicated facility at SURF permits extensive expansion,
adaptation, and evolution as experimental requires evolve, similarly to SNOLab and Kamioka Observatory. The esti-
mates and designs for additional underground excavations including massive 65 m diameter cavities have been advanced
to mature baseline estimates. These estimates have been independently reviewed and formal risk-based contingencies
established.
A different chapter is that of the large next-generation laboratories for long base-line neutrinos, proton decay and
other physics, being developed in Japan, Europe and in the USA. The European Commission funded the LAGUNA
design study included preliminary projects, by different Companies, in seven different European sites. The design
studies include all the underground infrastructures, access and services needed for three types of detectors: three water
Cherenkov detectors, 200–250 kt each, a liquid scintillator detector of 50 kt and a LAr detector from 10 kt to possibly
100 kt. Evaluated costs range, depending on the detector type and on the site from less than 100 M¤ to less than
250 M¤.

References
1. C.V. Achar et al., Phys. Lett. 18, 196 (1965).
2. F. Reines et al., Phys. Rev. Lett. 15, 429 (1965).
3. R. Davis et al., Phys. Rev. Lett. 12, 302 (1964).
4. J.N. Bahcall et al., Astrophys. J. 137, 334 (1963).
5. Y. Fukuda et al., Phys. Rev. Lett. 81, 1562 (1998).
6. L. Wolfenstein, Phys. Rev. D 17, 2369 (1978).
7. L. Wolfenstein, Phys. Rev. D 20, 2634 (1979).
8. S.P. Mikheyev, A.Yu. Smirnov, Yad. Fiz. 42, 1441 (1985).
9. S.P. Mikheyev, A.Yu. Smirnov, Nuovo Cimento C 9, 17 (1986).
10. E. Majorana, Nuovo Cimento 14, 171 (1937).
11. G.L. Fogli et al., hep-ph/1205.5254 (2012).
12. F.P. An et al., Phys. Rev. Lett. 108, 171803 (2012).
13. J.K. Ahn et al., Phys. Rev. Lett. 108, 191802 (2012).
14. J.J. Gómez Cadenas et al., Riv. Nuovo Cimento 34, 1 (2012).
15. S.M. Bilenky, C. Giunti, arXiv:1203.5250 [hep-ph] (2012).
16. V.A. Rodin, A. Faessler, F. Simkovic, P. Vogel, Nucl. Phys. A 766, 107 (2006) Nucl. Phys. A 793, 213(E) (2007).
17. E. Caurier, J. Menendez, F. Nowacki, A. Poves, Phys. Rev. Lett. 100, 052503 (2008).
18. J. Barea, F. Iachello, Phys. Rev. C 79, 044301 (2009).
19. A. Bettini, The world underground science facilities. A compendium, arXiv:0712.1051.