Search for eV Sterile Neutrinos – The STEREO

Experiment [TAUP 2017]

Stefan Schoppmann∗
arXiv:1909.01016v1 [hep-ex] 3 Sep 2019

Max-Planck-Institut für Kernphysik, Saupfercheckweg 1, 69117 Heidelberg, Germany

∗
for the Stereo collaboration
E-mail: stefan.schoppmann@mpi-hd.mpg.de

Abstract. In the recent years, major milestones in neutrino physics were accomplished at
nuclear reactors: the smallest neutrino mixing angle θ13 was determined with high precision
and the emitted antineutrino spectrum was measured at unprecedented resolution. However,
two anomalies, the first one related to the absolute flux and the second one to the spectral
shape, have yet to be solved. The flux anomaly is known as the Reactor Antineutrino Anomaly
and could be caused by the existence of a light sterile neutrino eigenstate participating in the
neutrino oscillation phenomenon. Introducing a sterile state implies the presence of a fourth
mass eigenstate, while global fits favour oscillation parameters around sin2 (2θ) = 0.09 and
∆m2 = 1.8eV2 .
The Stereo experiment was built to finally solve this puzzle. It is one of the first running
experiments built to search for eV sterile neutrinos and takes data since end of 2016 at ILL
Grenoble, France. At a short baseline of 10 metres, it measures the antineutrino flux and
spectrum emitted by a compact research reactor. The segmentation of the detector in six target
cells allows for independent measurements of the neutrino spectrum at multiple baselines. An
active-sterile flavour oscillation could be unambiguously detected, as it distorts the spectral
shape of each cell’s measurement differently.
This contribution gives an overview on the Stereo experiment, along with details on the
detector design, detection principle and the current status of data analysis.

1. Introduction
The Stereo experiment is located at the research reactor of the Institut Laue-Langevin (ILL)
in Grenoble, France. It is a reactor antineutrino experiment searching for a light sterile neutrino
eigenstate in the eV mass range. It is thereby searching for a possible explanation of a measured
deficit in reactor neutrino flux observed at baselines under 100 metres which is known as Reactor
Antineutrino Anomaly (RAA) [1]. Currently, global fits of various experiments hint for a sterile
neutrino with oscillation parameters around sin2 (2θ) = 0.09 and ∆m2 = 1.8eV2 as explanation
to the RAA [2]. Stereo probes this parameter space at a baseline of 9 to 11 metres.

2. Detector Setup and Experimental Signature

Neutrinos are generated in a compact reactor core of 40 cm diameter and 80 cm height and
then measured in a liquid scintillator (LS) detector (Fig. 1). The reactor is highly enriched
in 235 U and runs at a thermal power of 58.3 MWth . The detection reaction of electron
antineutrinos inside the liquid scintillator (LS) of the detector is given by an inverse beta-decay
(IBD), which produces a coincidence signal of a positron event followed by a neutron capture
event. 48 Hamamatsu 8 inch PMTs observe light produced when the positron annihilation
and the radiative neutron capture takes place in the 2000 litre target scintillator as well as the
scintillator filled outer-crown volume. The kinetic energy of the positron is directly connected to
the antineutrino energy, thus allowing for a measurement of the neutrino energy spectrum. The
LS in all volumes are admixtures of LAB, PXE and DIN. In addition, the target LS is doped
with a Gd-β-diketonate complex to achieve a high neutron detection efficiency. DIN is added
to enhance the particle identification by pulse-shape discrimination. The spatial resolution of
the detector is augmented by optically separating the target volume along the neutrino path of
flight in six equally sized cells of 36 cm thickness. This allows a relative measurement of the
neutrino spectrum at six baselines. The expected sensitivity regions after one year of reactor-on
data can be found in Fig. 2.
Since the detector is located inside the reactor building, a high flux of neutrons and gammas
exist. Heavy shielding of about 90 tons was installed to suppress those backgrounds. Shielding
materials consist of lead, B4 C, borated polyethylene and soft iron as well as a mu-metal
layer to shield the detector from external magnetic fields. In addition, a 15 m.w.e. top-
shielding protects against atmospheric radiation. Remaining atmospheric radiation, neutrons
and gammas can mimic the neutrino signature in form of accidental or correlated signal
coincidences. Their magnitude can effectively be estimated by reactor-off phases. During
these periods, a measurement of false neutrino candidates originating only from backgrounds is
possible. Moreover, active countermeasures against background are employed such as the use of
a water Cherenkov muon veto above the neutrino detector, a specific set of cuts on the PMT hit
pattern of each event and selections on the topology of the coincidence signal or the pulse-shape
discrimination (PSD) parameter.

3. Calibration and Detector Performance

Calibration of the PMT and electronics response is done using an LED light injection system
in the target and the outer-crown. Moreover, three calibration systems for the deployment of

Figure 1. Schematic drawing of the Stereo detector and the principle of sterile neutrino
search.
 Figure 2. Projected sensitivity
of the Stereo experiment for one
year reactor-on data. The sensi-
tivity study includes the system-
atic uncertainties from the neu-
trino reference spectra, a signal-to-
background ratio of 1.5 and detec-
tor response contributions and un-
certainties coming from the energy
resolution, energy scale uncertainty
and detection efficiency.

radioactive sources exist: an internal tube system is installed in several target cells, an external
system allows to position sources in any height around the detector, but inside the shielding and
a tube below the detector allows calibration from underneath.
The detection efficiency is determined by an AmBe gamma-neutron source. The capture time
of neutrons is found to be stable over time at (16.2 ± 0.2) µs and it is in agreement with the
capture time from neutrino candidate events of (16.5 ± 0.6) µs. From the AmBe calibration
data, it is also possible to infer an excellent energy containment by observing the capture peaks
of neutrons by hydrogen and gadolinium with good resolution. The fraction of neutron captures
by gadolinium is determined as 86% at the centre of the target.
The energy scale calibration and non-linearity estimation is achieved by the utilisation of
gamma sources between 0.5 and 4.4 MeV. For this analysis, cell-to-cell variations in acceptance
and light-crosstalk between neighbouring cells are taken into account. Thus, the measured
charge Qi is assumed as the sum of energy depositions in all cells:

Ejdep · fj · Lj→i .
X
Qi = α i (1)
j=cells

In this equation, Ejdep denotes the energy deposited in cell j, fj denotes the light yield in cell
j, Lj→i the light-crosstalk from cell j to i and αi the light-acceptance of cell i. All appearing
coefficients are obtained in-situ from calibration data as:

Ejdep · Cj · Lj→i
X
Qi = (2)
j=cells

where Cj denotes the measurable combination of the above parameters.

The light yield is determined as 270 PE per MeV and a good homogeneity of the energy
response is found across the fiducial volume using a 54 Mn gamma source. A small 3% difference
is found when comparing the energy response at the border with that of the centre. The findings
are reproduced with simulations at the percent level. Furthermore, the response is found stable
over time and a good resolution is found by investigating captures of spallation neutrons by
hydrogen throughout the entire target volume (Fig. 3).
Figure 3. Position (left) and width (right) of energy depositions from neutron captures by
hydrogen over time.

The neutrino candidate event rates over time can be found in Fig. 4. All rates are corrected
for dead times and their dependence on atmospheric pressure. The change in rate for reactor-on
and reactor-off periods is clearly seen. The black data points are obtained by requiring for the
total energy of the prompt event 1.5 MeV < E < 8 MeV and less than 1.1 MeV deposited in
the outer-crown. The delayed event is required to have an energy between 5 and 10 MeV in
the entire detector and more than 1 MeV has to be detected in the target. Moreover, the delay
between the prompt and delayed events may not exceed 70 s. For the prompt event signal, the
PSD parameter is required to deviate less than 2.5 σ from the central PSD value for that event
energy.
The red data points depict the rates when using an additional set of more stringent cuts.
These are defined by taking into account the event topology of the PMT hit pattern in order to
reduce the amount of background coincidences caused by stopping muons. Moreover, accidental
coincidences are suppressed by a large fraction when requiring a distance between the prompt
and the delayed events of less than 0.4 m. A further reduction of non-IBD events can be achieved
by requiring a highly localised energy deposition. This is done by demanding less than 0.7 MeV
of energy in the cells next to the main cell of the interaction. Applying the entire set of cuts
yields a factor of two improvement on the signal-to-background ratio while introducing a signal
inefficiency in the few percent order.
When comparing reactor-on and reactor-off data, it is found that the shielding is sufficient:
accidental background rates are below the design level and no increase of correlated events by
neutrons from the reactor is found. Furthermore, a neutrino candidate rate of 300 neutrinos per

Figure 4. Neutrino can-

didate rates for different
event selections as ex-
plained in the text.
day is found, well in agreement with expectations.

4. Summary and Outlook

Since its start of data taking in November 2016, the Stereo experiment has collected 70 days
of reactor-on and 25 days of reactor-off data. The ongoing analysis of this data testifies a good
detector performance and is currently further refined, e.g. by cut optimisation and more detailed
cosmic background studies. Following the finalisation of the energy scale, an oscillation analysis
is going to be performed. At the same time, another 150 days of reactor-on data taking is
upcoming until 2018, allowing for improved statistics.

References
[1] Mention G, Fechner M, Lassere T, Mueller T A, Lhuillier D, Cribier M and Letourneau A 2011 Phys. Rev. D
83 073006
[2] Kopp J, Machado P A N, Maltoni M and Schwetz T 2013 JHEP 1305 050