source: Backup NB/Talks/MEMPHYSetal/MEMPHYS EOI/CAMPAGNE_MEMPHYS-EOI/undlab_detector.tex@ 596

\section{Underground laboratory and detector} 

\subsection{Results of a feasibility study in the 
central region of the Fr\'ejus tunnels}
\label{sec:undlab}
The site located in the Fr\'ejus mountain in the Alps, 
which is crossed by a road-tunnel connecting France (Modane) 
to Italy (Bardonecchia), has a number of interesting characteristics 
making it a very good candidate for the installation of a megaton-scale 
detector in Europe, aimed both at non-accelerator and accelerator 
based physics.
Its great depth (4800 mwe, see figure~\ref{muonflux}), 
the good quality of the rock, 
the fact that it offers horizontal access, its distance from CERN (130 km), 
the opportunity of the excavation of a second (``safety'') tunnel, 
the very easy access by train (TGV), by car (highways) 
and by plane (Geneva, Torino and Lyon airports),
the strong support from the local authorities
represent the most important of these characteristics.

\begin{figure}[htb]
\centerline{\epsfig{figure=./figures/muonflux_red.eps,width=8cm}}
\caption{\it Muon flux as a function of overburden. 
The Frejus site is indicated by "LSM".}
\label{muonflux}
\end{figure}

On the basis of these arguments, the DSM (CEA) and IN2P3 (CNRS) institutions 
decided to perform a feasibility study of a Large Underground Laboratory 
in the central region of the Fr\'ejus tunnel, near the already existing, 
but much smaller, LSM Laboratory. This preliminary study
has been performed by the SETEC (French) and STONE (Italian) companies 
and is now completed. These companies already made the study and managed 
the realisation of the Fr\'ejus road tunnel and of the LSM 
(Laboratoire Souterain de Modane) Laboratory.
A large number of precise and systematic measurements of the rock 
characteristics, performed at that time, have been used to make a 
pre-selection of the most favourable regions along the road tunnel 
and to constrain the simulations of the present pre-study for the 
Large Laboratory. 

\begin{figure}[htb]
\centerline{\epsfig{figure=./figures/tunnel.eps,width=8cm,}}
\caption{\it Possible layout of the Fr\'ejus underground laboratory.}
\label{fig:tunnel}
\end{figure}

Three regions have been pre-selected : 
the central region and two other regions at about 3 km from 
each entrance of the tunnel. 
Two different shapes have been considered for the cavities to be excavated: 
the ``tunnel shape'' and the cylindrical ``shaft shape''. The main purpose 
was to determine the maximum possible size for each of them, 
the most sensitive dimension being the width (the so-called ``span'') 
of the cavities.        

The very interesting results of this preliminary study 
can be summarized as follows : 
\begin{enumerate}
\item the best site (rock quality) is found in the middle of the mountain, 
at a depth of 4800 mwe;
\item of the two considered shapes : ``tunnel'' and ``shaft'', 
the ``shaft shape'' is strongly preferred;
\item cylindrical shafts are feasible up to a diameter $\Phi$ = 65 m  
and a full height  h = 80 m ($\sim$ 250000 m$^3$);
\item with ``egg shape'' or ``intermediate shape between 
cylinder and egg shapes'' the volume of the shafts could be still increased
(see Fig.~\ref{fig:egg});
\item the estimated cost is $\sim$ 80 M Euro per shaft.
\end{enumerate}

\begin{figure}[htb]
\centerline{\begin{tabular}{cc}
\epsfig{figure=./figures/egg.eps,width=8.5cm} &
\epsfig{figure=./figures/eggsim.eps,width=5.5cm} 
\end{tabular}}
\caption{\it An example of ``egg shape'' simulation, 
constrained by the rock parameter measurements made during the road tunnel 
and the present laboratory excavation.
The main feasibility criterium is that the significantly perturbated 
region around the cavity should not exceed a thickness of about 10 m.}
\label{fig:egg}
\end{figure}

Fig.~\ref{fig:tunnel} shows a possible configuration for this
large Laboratory, where up to five shafts, of about 250000 m$^3$ each, 
can be located between the road tunnel and the railway tunnel, 
in the central region of the Fr\'ejus mountain. 

Two possible scenarios for Water \v{C}erenkov detectors are, for instance: 
\begin{itemize}
\item 3 shafts of 250000 m$^3$ each, with a fiducial mass of 440 kton 
(``UNO-like'' scenario).
\item 4 shafts of 250000 m$^3$ each, with a fiducial mass of 580 kton.
\end{itemize}
In both scenarios one additional shaft could be excavated
for a Liquid Argon and/or a liquid scintillator detector of about 
100 kton total mass.

The next step will be a Design Study for this Large Laboratory, 
performed in close connection with the Design Study of the detectors 
and considering the excavation of 3 to 5 ``shafts'' of about 250 000 m$^3$ 
each, the associated equipments and the mechanics of the detector modules. 

\subsection{Detector: general considerations}

The 20 year long successful operation of the 
Super-Kamiokande detector has clearly
demonstrated the capabilities and limitations of large water \v{C}erenkov
detectors\,:
\begin{itemize}
\item This technique is by far the cheapest 
and the most stable to instrument a very large detector
mass, as price is dominated by the photodetectors and their associated
electronics (this price growing like the outer surface of the detector),
while the active mass, made of water, is essentially free except for the 
purification system
\item These detectors are mainly limited in size by the finite attenuation
length of \v{C}erenkov light, found to be 80 meters at $\lambda = 400$~nm
in Super-Kamiokande, and by the pressure of water on the photomultipliers at
the bottom of the tank, which gives a practical limit of 80~m in height.
At large depths, the maximal size of underground cavities actually limits
relevant dimensions to about 70~m.
\item The detection principle consists in measuring \v{C}erenkov rings produced
by charged particles going faster than light in water. This has several
consequences\,:
\begin{enumerate}
\item neutral particles and charged particles
below \v{C}erenkov threshold are undetectable, so that some energy may
be missing
\item complicated topologies are difficult to handle, and in practice only
events with less than 3 to 5 rings are efficiently reconstructed
\item ring topology, based on their degree of fuzziness, allows to separate
between electromagnetic (e, $\gamma$) rings and ($\mu$, $\pi$) rings
\item the threshold in particle energy depends mainly on photocathode
coverage and also on water purity (due to radioactive backgrounds, such as
radon). Super-Kamiokande has achieved an energy threshold of 5 MeV with
40\% cathode coverage
\item due to points 1 and 2, water \v{C}erenkov detectors are not suited to
measure high energy neutrino interactions, as more rings and more undetectable
particles are produced. A further limitation comes from the confusion between
single electron or gamma rings and high energy $\pi^0$'s giving 2 overlapping
rings. In practice water \v{C}erenkov's stay excellent neutrino detectors
for energies below 1 (may be 2) GeV, when interactions are mostly quasi-elastic 
and the 2 rings from $\pi^0$ well separated.
\end{enumerate}
\end{itemize}
  
\subsection{Detector design}

Three detector designs are being carried out worldwide,
namely Hyper-Kamiokande \cite{uno} in Japan,
UNO \cite{hyperk} in the USA  and the present project MEMPHYS in Europe.
All of them are rather mild extrapolations of Super-Kamiokande, and rely on the 
expertise acquired after 20 years of operation of this detector.
Their main characteristics are summarized in table~\ref{WC:tab-1}.
\begin{sidewaystable}
\centering
%
\begin{tabular}{rccc}
\hline\noalign{\smallskip}
 Parameters                  &        \textbf{UNO} (USA)            &   
\textbf{HyperK} (Japan)          &      \textbf{MEMPHYS} (Europe)\\
\noalign{\smallskip}\hline\noalign{\smallskip}
\multicolumn{4}{l}{\textbf{Underground laboratory}}  \\ 
       location   &   Henderson / Homestake      &   Tochibora    & Fr\'ejus \\
                depth (m.e.w.)  &    4500/4800                 &    
1500                  &        4800  \\
Long Base Line (km)   & $1480\div2760$ / $1280\div2530$ & 290    &   130 \\
                       & FermiLab$\div$BNL       & JAERI         &   CERN \\
\noalign{\smallskip}\hline\noalign{\smallskip}
\multicolumn{4}{l}{\textbf{Detector dimensions}}          \\
type              & 3 cubic compartments   & 2 twin tunnels  & $3\div5$
shafts\\
                  &                        & 5 compartments  &           \\
dimensions            & $3\times (60\times60\times60)\mathrm{m}^3$ 
                                                                        &
$2\times 5 \times (\phi=43\mathrm{m} \times L=50\mathrm{m})$       &
$(3\div5)\times(\phi=65\mathrm{m} \times H=65\mathrm{m}) $ \\   
fiducial mass (kt)& 440                          &       550                  
& $440\div730$\\
\noalign{\smallskip}\hline\noalign{\smallskip}
\multicolumn{4}{l}{\textbf{Photodetectors}}          \\
           type   & 20" PMT              & 13" H(A)PD           & 12" PMT
\\
    number (internal detector)     & 57,000      & 20,000 per compartment
                                                & 81,000 per shaft \\
 surface coverage & 40\% (1/3) \& 10\% (2/3) & 40\%  & 30\%\
                                            
\\
\noalign{\smallskip}\hline
\end{tabular}
%
\caption{\label{WC:tab-1}
\it Some basic parameters of the three Water \v{C}erenkov
detector baseline designs}
\end{sidewaystable}

These 3 projects aim at a fiducial mass around half a megaton, taking into
account the necessity to have a veto volume on the edge of the detector, 1 to 2
meters thick, plus a minimal distance of about 2 meters between photodetectors
and interaction vertices, leaving some space for ring development.
The main differences between the 3 projects lie in the geometry 
of the cavities (tunnel shape for Hyper-Kamiokande, shafts for MEMPHYS,
intermediate with 3 cubic modules for UNO), and the photocathode coverage,
similar to Super-Kamiokande for Hyper-Kamiokande 
and MEMPHYS, while UNO keeps this
coverage on only 1 cubic detector, while the 2 others have only 10\% coverage
for cost reasons. Another important parameter is the rock overburden, similar
for UNO and MEMPHYS (4800~mwe), but smaller for Hyper-Kamiokande (1500~mwe),
which might be a limiting factor for low energy physics, due to spallation
products and fast neutrons produced by cosmic muons, more
abundant by 2 orders of magnitude (see figure~\ref{muonflux}).

The basic unit for MEMPHYS consists of a cylindrical detector module 65~meters
in diameter and 65~meters high, which can be housed
in a cylindrical cavity with 70~meter diameter and 80 meter height, as proven
by the prestudy. This corresponds to a water mass of 215 kilotons, that is
only 4 times the Super-Kamiokande detector. 
Conservatively substracting 2~m for the
outer veto plus 2~m for the fiducial volume, this leaves us with a fiducial
mass of 146 kilotons per module. The baseline design uses 3 modules, giving
a total fiducial mass of 440 kilotons, like UNO, corresponding to factor
20 increase over Super-Kamiokande (4 modules would give 580 
kiloton fiducial mass). The modular aspect is actually mandatory for
maintenance reasons, so that at least 2 of the 3 modules would be active
at any time, giving 100\% duty cycle for supernova explosions. Furthermore,
it would offer the possibility to add Gadolinium in one of the modules, which
has been advocated to improve diffuse supernova neutrino detection.
We estimate an overall construction time of less than 10 years, and of course
the first module could start physics during the completion of the two other
modules.

\subsection{Photodetection}

The baseline photodetector choice is photomultipliers (PMT) as they have
successfuly equipped the previous generation of large water \v{C}erenkov
detectors and many other types of presently running detectors in HEP. The PMT
density should be chosen to allow excellent sensitivity to a broad range
of nucleon decays and neutrino physics while keeping the instrumentation costs
under control.

Our goal for MEMPHYS is to reach in the whole detector
the same energy threshold as Super-Kamiokande,
that is 5 MeV, important for solar neutrino studies, for the proton decay into
$K^+ \nu$ using the 6~MeV tag from $^{15}$N desexcitation, and also very useful
for SN explosions, since the measurement of the $\nu_{\mu}$ and $\nu_{\tau}$
fluxes could be achieved using the neutral current excitation of Oxygen.

Our first approach was to consider 20" Hamamatsu tubes as used by
Super-Kamiokande, but the cost for 40\% coverage becomes prohibitive, as these
tubes are manually blown by specially trained people, which makes them very
expensive. Following a suggestion presented at the NNN05 conference by Photonis
company, we have considered the possibility of using instead 12" PMT's, which
can be automatically manufactured and have better characteristics
compared to 20" tubes\,: quantum efficiency (24\% vs 20\%), collection
efficiency (70\% vs 60\%), risetime (5~ns vs 10~ns), jitter (2.4~ns vs 5.5~ns).
Based on these numbers, 30\% coverage with 12" PMT's would give the same number
of photoelectrons per MeV as a 40\% coverage with 20" tubes. Taking into 
account the ratio of photocathodes (615~cm$^2$ vs 1660~cm$^2$), this implies
that going from 20" tubes to twice as many 12" tubes will give the same
detected light, with a bonus on time resolution and on pixel locations. If the dark
current of better photocathodes does not increase dramatically the trigger rate, 
we can expect MEMPHYS performances be at least as good as Super-Kamiokande.
A GEANT4 based Monte Carlo is under development to quantify the effective
gain. Pricewise, each 20" PMT costing 2500 Euros 
is replaced by 2 12" PMT's costing 800 Euros each. 
The only caveat is to make sure that the savings on PMT's are not
cancelled by the doubling of electronic channels. An R\&D on electronics
integration is presently underway (see Sec. \ref{sec:photo}).

\subsection{Photomultiplier tests}

A joint R\&D program between Photonis company and French laboratories has been
launched to test the quality of the 12" PMTs in the foreseen conditions of 
deep water depth, and to make a realistic
market model for the production of about 250,000 PMTs that would be necessary
to get the 30\% geometrical coverage. 

In parallel, studies on new photo-sensors have been launched. 
The aim is to reduce cost, while improving production rate and performance, as
it is essential
to achieve the long term stability and reliability which is proven for PMTs. 
Hybrid photosensors (HPD) could be a solution\,:
the principle has been proven by ICRR and Hamamatsu with a 5" HPD prototype. 
Successful results from tests of an 13"
prototype operated with 12 kV are now available, 
showing a $3 \cdot 10^4$ gain, good single photon sensitivity, 
0.8 ns time resolution and a satisfactory gain and 
timing uniformity over the photo-cathode area.
The development of HPD has also been initiated in Europe, 
in collaboration with Photonis.