Supernova Observation Via Neutrino-Nucleus Elastic Scattering in the CLEAN

Detector
∗
C. J. Horowitz
Nuclear Theory Center and Dept. of Physics, Indiana University, Bloomington, IN 47405

†
K. J. Coakley
National Institute of Standards and Technology, Boulder, CO 80305

D. N. McKinsey
Physics Department, Princeton University, Princeton, NJ 08544
(June 11, 2018)
Development of large mass detectors for low-energy neutrinos and dark matter may allow super-
arXiv:astro-ph/0302071v1 5 Feb 2003

nova detection via neutrino-nucleus elastic scattering. An elastic-scattering detector could observe
a few, or more, events per ton for a galactic supernova at 10 kpc (3.1 × 1020 m). This large yield, a
factor of at least 20 greater than that for existing light-water detectors, arises because of the very
large coherent cross section and the sensitivity to all flavors of neutrinos and antineutrinos. An
elastic scattering detector can provide important information on the flux and spectrum of νµ and ντ
from supernovae. We consider many detectors and a range of target materials from 4 He to 208 Pb.
Monte Carlo simulations of low-energy backgrounds are presented for the liquid-neon-based Cryo-
genic Low Energy Astrophysics with Noble gases (CLEAN) detector. The simulated background is
much smaller than the expected signal from a galactic supernova.

I. INTRODUCTION

Rich information on neutrino properties, oscillations, the supernova mechanism, and very dense matter is contained
in the neutrinos from core-collapse supernovae [1]. Existing detectors such as Super-Kamiokande [2] should accurately
measure the ν̄e component of the supernova signal. However the very interesting νµ , ντ , ν̄µ , and ν̄τ (collectively νx )
components may be detected without direct energy information and or in the presence of significant backgrounds from
other neutrino induced reactions. Therefore, additional νx detectors could be very useful.
Perhaps the “ultimate” supernova detector involves neutrino-nucleus elastic scattering [3,4]. The count rate in such
a detector could be very high because the coherent elastic cross section is large and all six neutrino components (νe ,ν̄e ,
and the four νx ) contribute to the signal. In particular, the detector is sensitive to νx , which are expected to have a
high energy and large cross section. Elastic scattering detectors can have yields of a few or more νx events per ton for
a supernova at 10 kpc (3.1 × 1020 m). This is an increase by a factor of 20 or more over existing light-water detector
yields of hundreds of ν̄e and tens of νx events per kiloton.
Furthermore, the energy of nuclear recoils provides direct information on the νx spectrum. Existing detectors
measure νx via neutral-current inelastic reactions on oxygen [5], deuterium [6], or carbon [7]. Here the observed
energy deposition does not depend on the neutrino energy as long as it is above threshold. Perhaps neutrino-proton
elastic scattering [8] can be detected in KamLAND [7]. This is similar to neutrino-nucleus elastic scattering but has
a smaller cross section.
Alternatively, it may be possible to detect νx using inelastic excitations of Pb. Proposals include using lead
perchlorate, as suggested by Elliott [9], OMNIS [10] and LAND [11]. Here some information on νx energies may
be obtained by measuring the ratio of single- to two-neutron knockout. However, the inelastic Pb cross sections are
somewhat uncertain. In contrast, neutrino-nucleus elastic cross sections can be calculated accurately with very little
theoretical uncertainty.
The νx spectrum depends on how neutrinos thermalize with matter in a supernova, and is somewhat uncertain.
Keil, Raffelt, and Janka have studied the effects of NN bremsstrahlung, pair annihilation, and nucleon recoil on the
νx spectrum [12]. These effects can be measured with an elastic-scattering detector.

∗
e-mail: horowitz@iucf.indiana.edu
†
Contributions of NIST staff to this work are not subject to copyright laws in the US.

1
 Obtaining direct information on νx energies may be very important because the difference in energies for νx compared
to νe or ν̄e is the primary lever arm for observing neutrino oscillations. For example, νx → νe oscillations could lead to
high-energy νe . However, deducing the oscillation probability may depend crucially on knowing how hot the νx were to
begin with. Neutrino-nucleus elastic scattering itself is “flavor blind”. Therefore, the signal should be independent of
neutrino oscillations (among active species). Thus elastic scattering may provide a baseline with which to characterize
the supernova source. Comparing this information to other flavor-dependent information and theoretical simulations
may provide the best evidence of oscillations.
The kinetic energy of the recoiling nuclei is low, typically below 100 keV. It is difficult to detect such low-energy
events in the presence of radioactive backgrounds. Furthermore, scintillation signals from the nuclear recoils may be
reduced by quenching because of the very high ionization density. However, recent progress in designing detectors for
low-energy solar neutrinos suggests that detection may be feasible. In general, backgrounds for solar neutrinos with
a low count-rate signal may be more severe than those for a supernova, where all of the events are concentrated in a
few-second interval.
Cryogenic Low Energy Astrophysics with Noble gases (CLEAN) is a proposed detector for low-energy solar neutrinos
based on scintillation in an ultrapure cryogenic liquid [13]. This will detect electrons from neutrino-electron scattering
at energies comparable to the recoil energy of nuclei from supernova neutrinos. In this paper, we discuss the utility
of CLEAN for supernovae detection via neutrino-nucleus elastic scattering.
There is considerable interest in the direct detection of weakly interacting massive particles (WIMPS). These are
expected to produce recoiling nuclei with a spectrum somewhat similar [14] to that of supernova neutrino-nucleus
elastic scattering. Again, backgrounds for WIMP detection may be larger than for supernovae because of the low
WIMP count rates. Present WIMP detectors, for example CDMS [15], have small target masses. However, future
detectors may be larger.
It is important to search for neutrinoless double-beta decay, as this can distinguish Dirac from Majorana neutrinos.
Existing 76 Ge experiments use multikilogram masses [16]. The next-generation experiments such as Majorana [17] or
Genius [18] may employ up to a ton of Ge. The need for good energy resolution, to tell neutrinoless from two-neutrino
double-beta decay, often aids in the detection of low-energy recoils. We calculate that the largest double-beta decay
experiments may soon be sensitive to galactic supernovae via elastic scattering.
Finally, micropattern gas detectors [19] may have a threshold low enough to detect nuclear recoils. This may allow
the observation of neutrino-nucleus elastic scattering using reactor antineutrinos.
Thus the technical requirements for detecting low-energy solar neutrinos, WIMPS, double-beta decay, and super-
novae via nuclear-elastic scattering may be similar. One detector or approach for low-threshold, low-background,
large mass measurements may have applications in multiple areas including supernova detection via neutrino-nucleus
scattering.
In section II, we fold elastic scattering cross sections with a model supernova neutrino spectrum to produce recoil
spectra and yields. We consider a range of noble-gas targets from 4 He to 132 Xe along with 12 C, 28 Si, 76 Ge, 114 Cd,
130
Te, and 208 Pb. We also discuss extrapolating yields to nearby isotopes. Section III focuses on the liquid-Ne-based
CLEAN detector, which appears to be very promising. A Monte Carlo simulation of backgrounds in CLEAN is
presented and compared to the expected supernova signal. We discuss the large signal-to-background ratio, choice of
fiducial volume, and possible detector thresholds. We conclude in section IV.

II. SUPERNOVA SIGNALS IN VARIOUS DETECTORS

The neutrino-nucleus elastic-scattering cross section dσ/dΩ is [20,3],

dσ G2 2 Q2w
= k (1 + cosθ) F (Q2 )2 , (1)
dΩ 4π 2 4
for a neutrino of energy k scattering at angle θ. The Fermi constant is G. This coherent cross section depends on the
square of the weak charge Qw

Qw = N − (1 − 4sin2 ΘW )Z (2)

of a nucleus with N neutrons and Z protons. The weak mixing angle is sin2 ΘW ≈ 0.231 [21]. We assume a spin-zero
target. Finally, the ground-state elastic form factor F (Q2 ) at momentum transfer Q,

Q2 = 2k 2 (1 − cosθ), (3)

2
is
1 sin(Qr)
Z
F (Q2 ) = d3 r [ρn (r) − (1 − 4sin2 ΘW )ρp (r)]. (4)
Qw Qr

Here ρn (r) is the neutron density and ρp (r) is the proton density. The form factor is normalized F (Q2 = 0) = 1. We
neglect a small correction from the single-nucleon form factors.
The inclusion of F (Q2 ) is crucial for heavier targets. However, we evaluate it at relatively small Q2 so the exact
form of the densities is not important. The proton density is often well constrained by measured charge densities.
For simplicity we use theoretical densities from simple relativistic-mean-field calculations using the successful NL3
effective interaction [22]. These calculations assume spherical ground states and do not include pairing corrections.
The use of other densities is not expected to change our results significantly.
We now consider a simple “standard model” for the supernova-neutrino spectra, see for example [23]. This model is
close to what others have used. A total energy of 3 × 1053 ergs (1 erg = 10−7 J) is assumed to be radiated in neutrinos
from a supernova at a distance d of 10 kpc (3.1 × 1020 m). For simplicity we use Boltzmann spectra at temperatures
of kB T = 3.5, 5 and 8 MeV for the νe , ν̄e and νx components respectively. Here kB is the Boltzmann constant, which
we set to one in the rest of the paper. The use of Fermi Dirac spectra (at zero chemical potential) should give similar
results. However, neutrino-nucleus elastic scattering is sensitive to the high-energy tails in the spectra. Therefore non
thermal spectra could modify our results somewhat and should be investigated in future work.
We assume equal partition in energy among the νe ν̄e and the four νx components. Therefore this standard
supernova radiates a total of Nνe = 3.0 × 1057 , Nν̄e = 2.1 × 1057 and Nνx = 5.2 × 1057 neutrinos. The time integral
of the neutrino flux at Earth φi (k) for a neutrino of energy k is
1 Ni 2 −k/Ti
φi (k) = k e , (5)
4πd2 2Ti3

for i = νe , ν̄e , or νx .
Microscopic simulations of supernovae suggest that equal partition of energy may be good only to ≈ 25 %. Further-
more, there is important uncertainty in the νx spectrum, with estimates of Tνx ranging from ≈ 6 to 8 MeV. It is an
important goal of elastic-scattering detectors to measure Tνx . Therefore, the predictions of our supernova spectrum
have significant uncertainties. Nevertheless, this simple model should provide order-of-magnitude estimates and may
allow easy comparisons to other calculations.
The yield of recoiling nuclei with energy E and mass M is
Z ∞ Z 1
Q2 dσ
Y (E) = 2πNt Σi=νe ,ν̄e ,νx dkφi (k) dcosθδ(E − ) , (6)
0 −1 2M dΩ

where Nt is the total number of target atoms. For our Boltzmann spectra this integral is simple:

G2 Q2w Nt

Y (E) = M F 2 (2M E) Σi=νe ,ν̄e ,νx Ni (ti + 1)e−ti , (7)
π 4 4πd2
with ti = (M E/(2Ti2 ))1/2 . For large recoil energy E, Y (E) is proportional to
M
−( )1/2 E 1/2
Y (E) → F 2 (2M E)e
2
2Tν
x . (8)

For light nuclei the high-energy tail continues to hundreds of keV and is produced by the scattering of very-high-energy
νx . However, for heavier nuclei the tail is sharply reduced by the nuclear form factor.
We consider first noble-liquid detectors from 4 He to 132 Xe and then a range of other detectors in order of increasing
mass number A from 12 C to 208 Pb. The yield Y (E) from Eq. (7) is shown in Figure 1 for detectors made of 4 He, 20 Ne,
40
Ar, 84 Kr, and 132 Xe. We don’t mean to imply that detectors would be feasible with all of these liquids. However,
we show these nuclei to illustrate how the yield depends on A for a broad range of A. The spectra in Figure 1 are
peaked at low recoil energy E. Increasing A raises the cross section because coherent scattering is proportional to N 2 .
Thus at low energies Y (E) increases significantly with A. However, as A increases the spectrum is strongly shifted
to lower energies by the form factor and the large target mass. The energy integral of Y (E), or total yield, is given
in Table I in events per ton of detector. Also listed are events above a threshold of 5, 10, 25, or 50 keV. Finally, the
average recoil energy of the nuclei is given. This average is influenced by a small number of events at high energies,
while the spectrum is peaked at low energies.

3
 The optimal choice of A may involve a trade-off between cross section, which favors high A, and recoil energy,
which favors small A. This choice may depend on the attainable threshold. Furthermore, the choice of target material
depends on a host of other practical considerations including the presence of possible backgrounds from radioactive
isotopes. Although 4 He has a relatively small cross section it has high recoil energies. We find a total yield in Table I
of 0.85 events per ton. Helium-based solar-neutrino detectors include HERON [24] and TPC [25] (or HELLAZ [26]).
Perhaps nuclei near A = 40 give a reasonable balance between cross section and recoil energy. However, Ar may
have backgrounds from radioactive 39 Ar and 42 Ar, while Kr may have backgrounds from 85 Kr. Xenon is being used in
several dark-matter, double-beta decay and solar-neutrino detectors including XMASS [27], XENON [28] and ZEPLIN
[29]. The total yield is very large, 31 events per ton. However, because 132 Xe is heavy, there is a strong premium
on obtaining a low threshold. Background from the 2ν double-beta decay of 136 Xe should not be a problem for a
supernova detector.
Above a threshold of 25 keV the yield is a relatively slow function of A. Therefore one may have considerable
freedom in the choice of target material. We consider the CLEAN liquid-Ne solar-neutrino detector [13] at some
length. The total yield of 3.99 events per ton is dominated by 3.08 νx events with only 0.38 νe and 0.53 ν̄e events.
One is very sensitive to the νx spectrum. For example, if the νx temperature is not the expected 8 MeV but instead
is near that for ν̄e Tx = 5 MeV, the spectrum in Figure 1 is greatly changed. This verifies that the recoil spectrum
contains direct information on the νx energies. A Monte Carlo simulation of backgrounds in CLEAN is presented in
section III.
We now consider a number of other targets. Detectors based on organic scintillator such as Borexino and KamLAND
[30,7] have a yield of 2.50 events per ton of 12 C, see Table I. With any carbon-based detector there will be backgrounds
from 14 C. Borexino should have a ratio 14 C/ 12 C of the order 10−18 [30]. At a concentration of 10−18 there will be
about 2 14 C decays per ton during the 10 seconds of a supernova neutrino burst. This is comparable to the number
of elastic recoils. However, the 14 C background should have a different spectrum and can be well measured at other
times. Therefore, this background may not prevent the use of carbon as a supernova detector, even if it does prevent
the observation of pp solar neutrinos. Quenching may be a serious problem for organic scintillator. The amount of
light produced from recoiling C ions may be much less than that for recoiling electrons [8].
An organic scintillator will also have events from neutrino-proton elastic scattering [8]. We estimate a yield of about
0.33 events per ton of CH2 . Because of the light proton mass these events will have a larger recoil energy. The proton
elastic-scattering cross section has a theoretical uncertainty of 10 to 20 % from possible strange quark contributions to
the nucleon’s axial current and spin [31]. It would be very useful to have better laboratory measurements of neutrino-
proton elastic scattering. In contrast, strange quarks are not expected to make significant contributions for supernova
neutrino-nucleus elastic scattering. Indeed, there is almost no theoretical uncertainty in the neutrino-nucleus elastic
cross sections.
Silicon detectors such as those described in [4] have a yield of 5.5 events per ton and the recoil spectrum is shown
in Figure 2, while 76 Ge detectors have a yield of 18.6 events per ton and an average recoil energy of 9.5 keV. The
double-beta decay experiment Majorana [17] is proposed to have a 500 kg mass, while Genius [18] has a proposed
mass of one ton. These detectors should have very low backgrounds and low thresholds. Therefore they should be
sensitive to a supernova at 10 kpc. It is remarkable that such small target masses can yield statistics for our own
galaxy comparable to those for the historic IMB and Kamiokande signals from SN1987A in the Large Magellanic
Cloud.
Finally, 114 Cd, 130 Te and 208 Pb yields are listed in Table I. The yield and spectrum for 130 Te is very close to 132 Xe
(see Figures 1 and 2) since they both have 78 neutrons. The heavy nucleus 208 Pb has a very large yield of 47.5 events
per ton. However, the average recoil energy is only 2.6 keV. Backgrounds and the need for a very low threshold may
make a Pb detector very difficult to build.
The nuclei in Figures 1 and 2 display a range of recoil spectra. For point nuclei there would be a single universal
spectral shape, with the recoil energy decreasing and the yield increasing with increasing A. However, the different
nuclear form factors modify the spectra for heavy nuclei.
Finally, we provide a simple formula to extrapolate the yields in Table I to nearby isotopes. If one ignores small
changes in the form factors of nearby nuclei, then the yield will be approximately proportional to the square of an
effective weak charge, Qef f ,

Q2ef f = δA,odd 3ga2 + Q2W , (9)

with QW from Eq. (2), and δA,odd = 1 for odd A nuclei and δA,odd = 0 for even A nuclei. This factor takes into
account the axial current of the last nucleon with ga = 1.26. Because this term adds in quadrature with Q2W , it makes

4
a very small contribution, except for very light systems such as the proton1 . For example, using Eq. (9) we find that
21
Ne and 22 Ne have cross sections respectively 1.29 and 1.48 times that of 20 Ne. Natural Ne is 0.3 % 21 Ne and 8.8 %
22
Ne so this will lead to a slight increase in yield over that for pure 20 Ne.

III. THE CLEAN DETECTOR

CLEAN, a detector concept based on liquid Ne, was originally proposed for the detection of low-energy solar
neutrinos. It will also have high sensitivity to weakly interacting massive particles (WIMPs). Liquid neon has a high
scintillation yield, has no long-lived radioactive isotopes, and can be easily purified using cold traps. In addition, neon
is inexpensive, dense, and transparent to its own scintillation light, making it practical for use in a large self-shielding
apparatus. Here we consider a CLEAN detector in which a stainless steel tank holds 200 tons of liquid neon, half of
which is exposed to a wavelength shifter to convert the ultraviolet light to the visible. Inside the tank and suspended
in the liquid neon are several thousand photomultipliers. A diagram of the proposed CLEAN design is shown in
Figure 3.
CLEAN will also be sensitive to supernova neutrinos, detected through neutrino-nuclear scattering. In this case
the entire active neon mass inside the wavelength shifter can be used, with a possible modest fiducial volume cut to
reduce radioactive backgrounds.
Of prime importance for determining the sensitivity of CLEAN to neutrino-nuclear scattering is the determination
of the light yield of liquid neon for nuclear recoils. Because the density of excitation in scintillator is typically higher
for nuclear recoils than for electron recoils, the chemically excited species are more likely to interact, increasing the
likelihood that energy will be lost through mechanisms that do not produce light. This quality is often expressed as a
“quenching factor”, the ratio of light emitted for a nuclear recoil to the light emitted from an electron recoil, per unit
deposition energy. While the quenching factor for liquid neon has not yet been measured, we expect it to be similar
in magnitude to the quenching factor measured for liquid xenon. Recent (and widely disagreeing) measurements of
the liquid xenon quenching factor [32,33] are 22 % and 43 %. The amount of quenching for liquid neon should be less
than that for liquid xenon, as the density (1.2) of liquid neon is less than that of liquid xenon, while the scintillation
mechanism in liquid neon is qualitatively similar. For the following simulations, we assume that the quenching factor
for liquid neon is 0.25. Clearly the scintillation quenching factor in liquid neon would have to be accurately measured
in order to properly interpret any supernova data.
As for all neutrino detectors, a prime design consideration in CLEAN is the reduction of radioactive backgrounds.
We expect that any radioactive species suspended in the liquid neon will be removed by passing the neon through
charcoal or similar adsorbant; however there will remain a high rate of gamma rays entering the liquid neon after being
emitted by the surrounding photomultipliers, photomultiplier support structure, wavelength shifter, and stainless steel
tank containing the liquid neon.
Figure 4 shows the expected supernova neutrino recoil spectrum for a CLEAN detector with 100 tons of active
liquid neon, assuming 3750 scintillation photons per MeV, 100 % efficiency for the wavelength shifter, photomulti-
plier coverage of 75 %, and a photomultiplier quantum efficiency of 15 %. Also shown is the expected radioactive
background for 10 seconds of observing time, assuming that the combined gamma and x-ray emission is dominated
by the photomultiplier glass and the wavelength shifter substrate. The simulation assumes photomultipliers 20 cm in
diameter, each with mass 650 g, with 30 ng per g of U and Th and 60 µg per g of K in the glass. The wavelength
shifter is assumed to be evaporated on quartz wafers of 1 mm thickness, with a U and Th concentration of 1 ng per
g. From 104 seconds of simulated data, we find a background of 62 ± 8 events in 10 seconds within the energy range
of 0 to 200 keV. Here the uncertainty corresponds to the ± 1-sigma interval for a particular 10-second observation.
Since liquid neon has no long-lived radioactive isotopes that would limit its practical threshold, the CLEAN detector
could conceivably trigger on as few as two photoelectrons. The accidental coincidence rate in CLEAN will be low,
as the photomultiplier dark count rate will be suppressed at liquid-neon temperature (27 K). Thus we expect that
the full 100 tons of liquid neon could be viewed with a threshold of 2 photoelectrons, equivalent to a recoil energy
of about 5 keV. The expected 62 background events is smaller than the expected supernova signal of 330 Ne elastic
events above 5 keV in 100 tons.
The radioactive backgrounds, while small in comparison to the supernova signal, can be lowered further through
position resolution, since most gamma rays will deposit their energy in the outer edges of the liquid neon sphere.

1
We ignore the slightly different angular distribution of this term.

5
Recently, Monte Carlo simulations have shown that the location of ionizing radiation events in CLEAN can be
determined by analyzing the pattern of photomultiplier hits. These simulations are described in detail in an upcoming
publication [34]. Here we show only some results relevant to supernova neutrino detection. We find that a mild fiducial
radius cut, leaving a mass of 70 tons, virtually eliminates gamma-ray background for the purposes of supernova
neutrino detection (only 2.6 ± 1.6 events in the fiducial volume in 10 seconds). These results are shown in Figure
5. In the case of a fiducial volume cut, detecting a few more photoelectrons will improve the convergence of the
position resolution algorithm. Currently we can analyze events that produce as few as 8 photoelectrons (recoil energy
of 21 keV) if position cuts are applied. In general, as the position cut is increased or the number of photoelectrons is
reduced, algorithm convergence should be carefully checked. The expected 2.6 background events are much smaller
than the expected supernova signal of 140 Ne events above 21 keV in 70 tons.
The signal-to-noise ratio reported here, though already quite large, might be significantly improved by the devel-
opment of photomultipliers with lower inherent radioactivity. Such photomultipliers are under investigation by the
CLEAN and XENON collaborations. In addition, plans are being made to measure the quenching factor for nuclear
recoils in liquid neon, as this measurement is important for the determination of the sensitivity of CLEAN to WIMP
particles, as well as for supernova neutrinos.

IV. CONCLUSIONS

Detectors with reduced radioactive backgrounds may be able to study supernovae via neutrino-nucleus elastic
scattering. This could provide important information on the flux and spectrum of νx (νµ and ντ ). Elastic-scattering
detectors could see a few or more events per ton for a supernova at 10 kpc. This is 20 or more times the number of
events per ton of existing water detectors. The CLEAN experiment, based on the detection of scintillation in liquid
Ne, is a prime example of a detector that would be sensitive to the elastic scattering of supernova neutrinos. Its active
mass of 100 tons may yield almost 400 νx events. In addition, many other detectors, including the largest dark-matter
and double-beta decay experiments, may also be sensitive to supernova neutrinos via elastic scattering.
Observation of neutrino-nuclear elastic scattering will complement supernova signals from other detectors. Water
detectors such as SNO or Super-K will detect νx without νx spectral information. This energy information could be
important for neutrino oscillations. KamLAND may be able to measure ν − p elastic scattering. This contains energy
information just like ν − A scattering. The small cross section for ν − p scattering may be compensated by a large
detector mass. This may yield only slightly smaller statistics than CLEAN. We strongly encourage development of
detectors based on both ν − p and ν − A elastic scattering since they may have different backgrounds, thresholds, and
systematic errors. Furthermore, the very large ν − A elastic cross sections may allow even larger statistics in future
detectors.

ACKNOWLEDGMENTS

We thank John Beacom, Steve Elliott, and Harry Miley for useful discussions and acknowledge financial support
from DOE grant DE-FG02-87ER40365 and NSF grant 0226142. We acknowledge the hospitality of the Aspen Center
for Physics, where some of this work was done.

[1] See for example, J. F. Beacom, hep-ph/9909231.

[2] Super-kamiokande Collaboration, astro-ph/0007003.
[3] A. Drukier and L. Stodolsky, Phys. Rev. D30 (1984) 2295.
[4] Blas Cabrera, Lawrence M. Krauss, and Frank Wilczek, Phys. Rev. Lett. 55 (1985) 25.
[5] K. Langanke, P. Vogel and E. Kolbe, Phys. Rev. Lett. 76 (1996) 2629.
[6] C. J. Virtue, astro-ph/0103324.
[7] V. Barger, D. Marfatia and B. P. Wood, hep-ph/0011251.
[8] John F. Beacom, Will M. Farr, Petr Vogel, Phys.Rev. D66 (2002) 033001.
[9] S. R. Elliot, Phys. Rev. C62 (2000) 065802. J. Engel, G. C. McLaughlin, and C. Volpe, hep-ph/0209267.

6
[10] D. B. Cline et al. Phys. Rev. D50 (1994) 720; P. F. Smith, Astropart. Phys. 8 (1997) 27; J. J. Zach et al., Nucl. Instrum.
Meth. A484 (2002) 194.
[11] C. K. Hargrove et al., Astropart. Phys. 5 (1996) 183.
[12] Mathias Th. Keil, Georg G. Raffelt and Hans-Thomas Janka, astro-ph/0208035.
[13] D. N. McKinsey and J. M. Doyle, Journal of Low Temperature Physics 118 (2000) 153.
[14] J. D. Lewin and P. F. Smith, Astroparticle Phys. 6 (1996) 87.
[15] A. Abusaidi et al., Phys. Rev. Lett. 84 (2000) 5699.
[16] L. Baudis et al., Phys. Rev. Lett. 83 (1999) 41; C. E. Aalseth, et al., (IGEX Collaboration), Yad. Phys. 63 (2000) 1299.
[17] L. DeBraeckeleer, talk at Workshop on the Next Generation U.S. Underground Science Facility, WIPP, June 12-14, 2000,
Carlsbad, New Mexico, USA; C. E. Aalseth et al., in Proc. of TAUP’2001, Gran Sasso, Italy, September 2001, ed. A.
Bettini et al. (2002). See also http://majorana.pnl.gov.
[18] H. V. Klapdor-Kleingrothaus, L. Baudis, G. Heusser, B. Majorovits, H. Paes, hep-ph/9910205.
[19] P. Barbeau, J. I. Collar, J. Miyamoto, and I. Shipsey, hep-ex/0212034.
[20] D. Z. Freedman, D. L. Tubbs, and D. N. Schramm, Annu. Rev. Nucl. Sci. 27 (1977) 167.
[21] SLD Collaboration, K. Abe et al., Phys. Rev. Lett. 86 (2001) 1162.
[22] G. A. Lalazissis, J. König and P. Ring, Phys. Rev. C55 (1997) 540.
[23] K. Takahashi, M. Watanabe, K. Sato and T. Totani, Phys. Rev. D64 (2001) 093004; T. Totani, K. Sato, H. E. Dalhed
and J. R. Wilson, Astrophys. J. 496 (1998) 216.
[24] R. E. Lanou, H. J. Maris, and G. M. Seidel, Phys. Rev. Lett. 58 (1987) 2498; S. R. Bandler et al., J. Low Temp. Phys. 93
(1993) 785.
[25] G. Bonvicini, D. Naples and V. Paolone, NIM A 491 (2002) 402; G. Bonvicini and A. Schreiner, NIM A 493 (2002) 90.
[26] A. de Bellefon for the HELLAZ collaboration, Nucl. Phys. B (Proc. Suppl.) 70 (1999) 386.
[27] M. Nakahata, http://www.mpi-hd.mpg.de/nubis/www lownu2002/transparency/nakahata lownu2002.pdf.
[28] E. Aprile et al., astro-ph/0207670; Y. Suzuki, astro-ph/0008296.
[29] David B. Cline, Hanguo Wang, Y. Seo, astro-ph/0108147.
[30] BOREXINO Collaboration, G. Alimonti et al., Astrop. Phys. 16 (2002) 205. Also at hep-ex/0012030.
[31] L. A. Ahrens et al., Phys. Rev. D35 (1987) 785.
[32] F. Arneodo et al., NIM A 449 (2000) 147.
[33] R. Bernabei et al., EPJC, to be published.
[34] K. J. Coakley and D. N. McKinsey, to be published.

Table I: Yield.
Target Y Y> Y> Y> Y> <E>
5 keV 10 keV 25 keV 50 keV (keV)
4
He 0.85 0.82 0.79 0.72 0.62 240
12
C 2.5 2.2 2.0 1.6 1.1 83
20
Ne 4.0 3.3 2.9 2.0 1.2 46
28
Si 5.5 4.2 3.4 2.1 1.1 31
40
Ar 9.4 6.6 5.0 2.5 0.99 21
76
Ge 18.6 9.6 5.8 1.7 0.30 9.5
84
Kr 19.8 9.5 5.5 1.4 0.20 8.4
114
Cd 26.3 9.7 4.6 0.70 0.041 5.7
130
Te 31.8 10.1 4.3 0.47 0.014 4.8
132
Xe 31.1 9.8 4.1 0.43 0.012 4.8
208
Pb 47.5 7.3 1.7 0.022 0.001 2.6

Yield in events per ton for a supernova at 10 kpc assuming different target materials. Also listed is the number of
events above thresholds of 5, 10, 25 or 50 keV. Finally the average recoil energy < E > is given.

7
 0.12 Kr
0.11
0.1
0.09
Yield (per keV ton)

Xe
0.08 Ne
0.07
0.06
0.05
0.04 C Ar
0.03
0.02
0.01 He
0
0 20 40 60 80 100
E (keV)
FIG. 1. Yield versus recoil kinetic energy E. The solid curves are for noble targets of 4 He, 20 Ne, 40 Ar, 84 Kr and 132 Xe as
indicated, the dashed curve is for 12 C. Finally the dotted curve, only shown for 20 Ne, assumes a reduced νx temperature of
Tνx = 5 MeV.

10

1
Yield (per keV ton)

Si
0.1 Ge

0.01 Cd
Pb
Te
0.001
0 10 20 30 40 50
E (keV)
28 208
FIG. 2. Yield, versus recoil kinetic energy E. The solid curves are for targets of Si and Pb, as indicated, the dashed
curve is for 114 Cd, the dotted curve 76 Ge, and the dot-dashed curve is for 130 Te.

8
 Photomultipliers

Liquid neon
12 meters
27 K

77 K

Dewar support
H2O shielding

7 meters

10 meters

FIG. 3. Diagram of CLEAN

6
Background
Yield (Events per keV)

5 T=8 MeV
T=6 MeV
4

0
0 50 100 150 200
E (keV)
FIG. 4. Yield for full 100 ton fiducial mass of CLEAN versus recoil kinetic energy E. The solid curve is the expected
supernova signal assuming a distance of 10 kpc and a νx temperature Tνx = 8 MeV. The dashed curve assumes Tνx = 6 MeV.
Finally, the thick curve is the predicted background from the Monte Carlo simulation assuming an observing time of 10 seconds.

9
 5

4 Background
Yield (Events per KeV) T=6 MeV
T=8 MeV
3

0
0 50 100 150 200
E (keV)
FIG. 5. Yield for 70 ton fiducial mass of CLEAN versus recoil kinetic energy E. The solid curve is the expected supernova
signal assuming a distance of 10 kpc and a νx temperature Tνx = 8 MeV. The dashed curve assumes Tνx = 6 MeV. Finally,
the thick curve is the predicted background from the Monte Carlo simulation assuming an observing time of 10 seconds.

10