The design of the emulsion target follows closely the experience gained with the
CHORUS experiment.
The thickness of each target stack is 60 mm, equal to the amount of material
between each tracking section in CHORUS.
Each of the modules contains four emulsion target blocks, mounted in a two by
two matrix.
In a plane perpendicular to the beam these blocks form a square with
sides of 72 cm and match the size of the silicon tracker to which they can be
fixed mechanically with high precision.
This ensures excellent alignment of the two systems.
The blocks are built from two stacks of target emulsion plates, which have the
same dimension as the CHORUS emulsion ( 
).
The dimensions match the existing emulsion pouring and development facilities
and the automatic microscope stages.
The emulsion stacks are made of 72 target emulsion plates
(Fig. 5).
Each target plate is composed of a thin backing layer, of about 100 
thickness
with 350 emulsion layers on both sides.
Ionizing particles induce a latent image of their track in the emulsion material.
After development, this provides 300 black grains per mm of track trajectory,
which contain position information with a 0.3 precision.
In routine scanning the position accuracy to locate the grains is 1 .
Special emulsion plates are added to the downstream end of the target stacks, to
provide higher angular resolution for the connection of the tracks found
in the electronic detectors with the target stack.
These special plates consist of two 100 emulsion layers separated by an
800 thick backing layer.
Scanning systems can measure the angle of tracks from the grains found on both sides of
the backing layer with a precision of about 1 mrad.
Figure:
Schematic drawing of a target emulsion stack and silicon trackers.
The entire emulsion target, together with the associated electronic trackers, is placed in
a ``cool box'' with controlled temperature ( 
5 
C) and humidity.
This is to prevent
the loss of the latent image in the emulsion (fading) during the running time of the
experiment.
Possible background sources are well under control. The effect of cosmic-ray tracks is strongly reduced by keeping the undeveloped emulsion plates vertical. The background induced by muons associated with the neutrino beam and with the SPS X7 test beam are kept at an acceptable level. In particular, those muons are used for emulsion alignment purposes. The possible background produced by showering tracks in neutrino events is negligible in CHORUS and it is expected to remain at a very low level in this experiment (less than one shower track per microscope view).
Events are located in the emulsion target by a scan-back method
[37, 38, 39].
The electronic tracking detectors provide a precise prediction of the
exit point and direction of all suitable tracks of the event.
A track is suitable for scan-back when its momentum is above 1 GeV/c and its
angle is smaller than 400 mrad with respect to the beam axis.
These tracks are then searched for in the special plates.
The position resolution is determined by the precision of the silicon microstrip
detectors and by the alignment accuracy.
We estimate that the alignment limits the matching precision to 20
(at one standard deviation).
The angular prediction accuracy can be matched to the 1 mrad resolution of the special
plate and is dominated by multiple scattering in the silicon detectors.
A schematic drawing of the track location method in the special emulsion plates is shown in
Fig. 6.
The precision of the prediction is not much altered by the presence of a magnetic
field, for particle momenta above 1 GeV/c as long as the distance between the
tracking detector and the emulsion plate does not exceed 20 mm.
When compared to the CHORUS situation, the volume of phase-space defined by the
position and angular accuracy of the predicted track is two orders of magnitude smaller,
allowing for a higher beam intensity and longer exposure time before background
of randomly matching tracks becomes a problem.
Over 90% of the events have suitable tracks for location.
These are followed in the upstream
plates, plate by plate, until the interaction vertex is found.
High momentum tracks are used to perform correction of local track distortions
caused by the emulsion handling.
This procedure is not affected by the magnetic field since it is applied to
track segments of 
length.
As already discussed, we aim at an improvement of the scanning time with respect to CHORUS, due to a more accurate prediction of the track coordinates in the emulsions. A reduction of about a factor 100 in the special plate surface to be scanned is achievable because of the precise predictions of the silicon detectors.
Figure:
Event location in the CHORUS experiment.
We mentioned that systems have been developed to fully automatize
the emulsion scanning procedure.
At present, two types of automatic microscopes exist, which differ in implementation features
[38][39].
The systems function as follows.
An emulsion plate is mounted on a movable stage, with mechanical precision of
0.5 .
The position is controlled by a PC through a motor controller.
The plate is moved to the relevant position with respect to the microscope objective.
The focal plane of this objective can image small slices of the emulsion layer, such that
for a single track an independent measurement of its position can be obtained
each 6 .
By a vertical movement of the objective, multiple slices of the emulsion layer can be
brought into focus.
The optical image is digitized with the help of a CCD camera and the digital image
is made available to a processor.
The combination of many layers of images, obtained by focusing different depths into
the emulsion layer is input to a track finding algorithm.
On special plates, about 16 independent layers can be measured on each side of the
backing.
Tracks are found by requiring consistency on both sides of the backing.
The target stacks are then used to scan-back tracks predicted with the special plates.
The interaction vertex can be located with this method.
Figure:
Schematic drawing of an automatic microscope stage.
The scanning capability now achieved by CHORUS is about
10 
events / (month 
microscope), including the operational efficiency.
Most of the scanning
time goes into the analysis of the special plates downstream of the target emulsion.
Using the information from these plates, the tracks in the target can be
located with a micron accuracy. The complete emulsion event
reconstruction is performed only for those events with a decay topology
(tau or charm decay).
The speed with which the automatic system can process one event depends ultimately on the
speed with which pixels can be extracted from a CCD and the accuracy of the prediction.
With reasonable extrapolation of recent advances in this technology, an estimate of
the scanning time per followed track is one second per emulsion plate traversed.
We assume that 20 automatic microscopes will be available and
will be routinely operated with the tools and the experience now being gathered.
With the improvements quoted above,
a scanning capability of about 3 10 
events/year is feasible.
As already pointed out, the full measurement of the plates in a target module will
be done only for candidate events.
The scanning of 2 10 muonless and
6 10 muonic events appears feasible in about two years after
the end of the data taking.
In that sample, the fraction of events to be completely reconstructed
can be estimated, from charm production, to be about 5%.
A simplified model of an automatic scanning system is shown in
Fig. 7.
Techniques now exist to fully digitize the volume around the interaction vertex, and
to perform pattern recognition on this information.
The emulsion can then be regarded as a succession of two-dimensional
tracker planes, 6 apart with a point precision of about 1 .
Decay kinks with kink-angles larger than 10 mrad can be measured in the emulsion target.
This technique also allows to measure the direction of the decay parent,
which is extemely valuable for the study of the decay kinematics.
All other charged tracks emerging from the vertex are also visible, and
their track parameters can be measured (Fig. 8).
Corrections for distortions in the emulsion can be performed using through-going
tracks near the vertex (see Fig. 8).
Figure:
Reconstructed event measured in the CHORUS emulsion.
Shown are the grains, with reconstructed tracks,
in the emulsion layers on both sides of the plastic backing.
To indicate the scale, the sensitive volumes (emulsion layers) have a thickness
of 350 each, and the plastic backing 100 .
A background track not related to the event is also displayed.