Figure:
Schematic drawing of the 12-detector STAR silicon ladders.
The main purpose of the silicon tracker is to provide a precise interface
between the emulsion target and the other electronic trackers. Therefore, it
must provide a space segment, which gives efficient pattern recognition in the
electronic tracker providing a high resolution constraint.
It also allows good momentum resolution and gives a precise prediction of the track
impact point in the emulsion.
In fact, the ultimate precision of silicon microstrip detectors is about 20 microns,
limited by multiple scattering
in the silicon itself and in their support and by aligment uncertainties.
This allows the track prediction in the emulsion to be within one microscope view
(about 
) at the 5 
level. In comparison with CHORUS, this represents
a reduction of about a factor 100. This is important not only to achieve a much
higher scanning speed, but also to reduce backgrounds associated
with randomly matched tracks.
In order to provide a segment, two X-Y points are needed. The silicon tracker consists of two planes (each plane is defined as providing an X-Y measurement). The first plane is located as close a possible to the emulsion (10 mm) while the second plane is at about 50 mm from the emulsion to provide a sufficiently long lever arm.
Each plane consists of two layers of single-sided silicon microstrip detectors. The first layer has strips pointing in the direction perpendicular to the magnetic bending plane, while the second layer strips are perpendicular to the first one. The two layers are mounted very close to each other, in order to minimize error in the second layer due to multiple scattering in the first. From this point of view, the use of double sided detectors or back-to-back glued detectors would seem preferable. However, there are several reasons to use independent layers of single sided detectors: the construction of the detector modules (``ladders'') is much simpler, the detectors and modules are more robust and the signal to noise ratio of the two layers is the same. In the case of double sided detectors the extra capacitance of the n-side results in a degraded performance.
Each layer consists of 48 half modules. A half module is
made up of 12 individual detectors, 
in surface, 
thick,
glued together to a common carbon fiber back-bone and wire-bonded. One end of the
module is glued to a hybrid board containing the read-out electronics. Two
half modules are glued (at the end opposite to the electronics) to form a full
"ladder" 1.44 m long.
The read-out electronics is at both ends of the ladder outside the fiducial volume.
Twenty-four such ladders, slightly overlapping with each other form a layer
of active dimensions of 
.
The second layer is fixed to the
same support frame as the first one, and has the same structure, except that
the ladders point in a direction perpendicular to the ladders of the first
plane. A plane is made up of 48 ladders, and therefore the silicon tracker of
each one of the 6 independent modules in the detector consists of 96 ladders.
The total number of ladders is 576.
The design and operation of the silicon tracker relies largely on the
experience acquired with the NOMAD-STAR project [40].
In fact, very long (72 cm) ladders
with good signal to noise ratio have already been succesfully and reliably built.
This detector
will be installed in March 1997 right upstream of the NOMAD first drift
chamber. It consists of 4 layers of passive material, boron carbide ( 
,
, 
) each followed by a layer of single-sided
silicon microstrip detectors, plus an additional layer at the end.
The 5 layers of silicon detectors (for a total active surface
of about 1.14 
) allow the measurement of the track coordinates in the
projection parallel to the magnetic field.
Each layer consists of 10 overlapping ladders made of 12
independent detectors (Fig. 9).
A total of 50 long ladders have been built. The
design of the half modules proposed here is very close to the STAR ladders.
The construction of the STAR detector has already provided valuable experience, as
well as infrastructure. A fully equipped silicon laboratory at CERN (shared with
the OPAL and CMS silicon groups) has been used to build the STAR detectors in a very short period of
only 6 months. The performance of the ladders is excellent. The signal to noise
ratio of the ladders is 15-16, and a spatial resolution better than 5 
has been measured.
The number of defective channels is less than 1 %. The
ladders are mechanically robust, and the read-out electronics (the VA1 chip, a
commercial descendent of the VIKING has been used [40]) performs very well. During
production, two ladders a day were built. Detector testing, mounting
of the ladders and debugging was done in parallel. The STAR detectors
manufactured by Hamamatsu are AC coupled, FOXFET biased [52],
single sided wafers of 
transverse dimension and 
thickness, with a read-out pitch of 
and a floating strip at a pitch of
. The detectors were selected applying the following criteria:
good quality (low leakage current and small number of defects), simplicity
(single sided) and availability (the masks and the production
chain were available from Hamamatsu, thus eliminating development costs and guaranteeing
prompt delivery). The detectors were purchased at a very competitive
price, 15 
.
Our baseline design for this detector simply doubles the dimensions of the detectors and the
pitch keeping the basic features of the STAR ladders.
A resolution better than 
is achieved by using a
pitch of 100 with a floating strip at 50 .
Following the STAR experience, the construction of two half-modules a day (one
ladder) is possible. We need 576 days for the full detector, the production of which can be
spread over a period from two to three years. The time scale can be
improved if several groups participate in the construction of the silicon system.
We are confident that the silicon tracker needed for this experiment can be built in a reasonable time and at a moderate cost, in spite of its large dimensions. Considerable experience has been gathered with the construction of the STAR detector and will further improve with its commissioning and operation during the 1997 run.