This website is no longer maintained. Its content may be obsolete. Please visit http://home.cern/ for current CERN information.
(In collaboration with
C. Avila (Universidad de Los Andes),
A. Brandt (FNAL), and H. da Motta (LAFEX).)
Very forward tracking is typically done using
detectors located in Roman pots,
which
are stainless steel containers that allow the detectors
to function close to the beam but outside the machine vacuum. Particles
traverse thin steel windows at the entrance and exit of each pot.
The pots are remotely controlled and
can be moved close to the beam (within a few mm)
during stable beam conditions and retracted otherwise.
Roman pots have been successfully used in many experiments such as
UA4, UA8, CDF, E710, and ZEUS, and have typically been used as a leading
proton tagger and momentum spectrometer. They
make use of machine magnets along with points measured on
the track of the scattered proton to calculate the proton momentum
and angle ( 
).
Roman pots and their associated detectors do not require major research
and development as they use existing technologies. All that is required
is some optimization for the specific beam optics at the different
pot locations. Since Roman pot spectrometers have typically been
concerned with protons carrying a large fraction of the beam momentum,
the associated track multiplicities for these detectors are usually
very small ( 
), care should be taken in the detector design
to account for the higher track multiplicities expected at
certain locations in the FELIX layout.
A conceptual diagram of a Roman pot is shown in Fig. 5.18. The bellows allow for movement of the pot close to the beam. Thin stainless steel windows bracketing the active area of the pot reduce multiple scattering. Pressure compensation is necessary to combat the forces caused by the imbalance in pressure between the vacuum and the inside of the pot which is at atmospheric pressure. The final positions of the pot which should be accurate to about 20 microns.
Figure 5.18: A conceptual diagram of a Roman pot.
The current scheme envisages some
16 Roman pots in each arm, distributed between 20 and 430 m
from the interaction point. They would be located
just after UA1, at D1 
(with 1 in front
+ 1 behind), D1 
(1+1), D1 
(0+1), D1 
(1+1) and D1 
(1+1 followed
by 1 to provide a longer lever arm) situated between 20 and 50 m from the IP.
At D2, there would be (1+2).
The two final pots would be located in the free space near
Q11 at 420 m
and 430 m (see Table 5.1). The role of the pots is
somewhat different in the various locations,
and so there may be differences in construction and detector technology.
Figure 5.19: A schematic view of the "bay" design.
Roman pots have typically been used either with pressure compensation as in Fig. 5.18 (E710-type Roman pot), or with a large brute force motor (CDF approach). An alternative approach, which could be less costly than either of these two methods, involves reducing the pressure imbalance by minimizing the area of the detector and keeping the bulk of the pot within vacuum. This design involves a ``bay'' for the detector when it is in the retracted position, as illustrated in Fig. 5.19. The combination of the smaller bellows and reduced forces should result in a significantly less expensive apparatus. Such a design is currently being studied by the DØ collaboration [10].
Each Roman pot contains a position detector, which is used to determine the (x,y) coordinate of the deflected proton at the pot position. These detectors should be very reliable as they will be located in the tunnel and consequently difficult to access.
The demands on the detectors depend on the location. The Roman pots at Q11 or D2 will see clean events, with no more than a single track from the collision. On the other hand, it is important to be able to have signals from these detectors which can be used in the Level 1 or Level 2 triggers to tag elastic and diffractive protons. Closer to the IP, the multiplicity of tracks of interest will rise, and so will backgrounds from splash from particles interacting in the beampipe or calorimeters.
Roman pots in the D1 region or closer are essentially standard tracking stations. As such, the detectors will be planes of Si pixels, similar to those described in Section 5.2.
We consider the use of scintillating fiber detectors in the D2 and Q11 regions. In addition to the fast response, the sensitive area can be brought closer to the beam than may be the case with Silicon detectors. We will, however, supplement the scintillating fiber detectors with Si strips in order to improve resolution. It may also turn out that economies of scale again dictate that these silicon planes should be pixel detectors.
A particular option is a
scintillating fiber detector with a
multi-anode photo-multiplier tube (MAPMT) as shown in detail in
Fig. 5.20.
This simple detector is comprised of stacked ribbons
of four fibers oriented so that the scattered proton
will pass through all four fibers to maximize the light output. Each detector
will have two views rotated by 
relative to each other, with each
view consisting of two sets of 20 parallel ribbons
. The ribbons in each
set have a one-third ribbon width spacing and the sets are displaced
from each other by two-thirds of the ribbon width, in order to minimize
the resolution. The use of
mm square scintillator fibers would allow a theoretical
resolution of about 80 
m.
Figure 5.20: Details of the position detector and MAPMT described in the
text.
The scintillating fibers are connected to clear fibers that are bundled together in groups of four and connected to one channel of the MAPMT as shown in the bottom of Fig. 5.20. Four fibers/channel will give about 10 photoelectrons and fit comfortably within the pixel size of the MAPMT. The upper right side of Fig. 5.20 shows a front view of the 16 channel HAMAMATSU H6568 MAPMT. This MAPMT presents good gain uniformity among its 16 anodes and has a pitch of 4.5 mm.
The MAPMT's can be read out by a simple trigger board, with one trigger board required for each of the spectrometers.
Table 5.3: Estimated cost per Roman pot.
The Level 1 trigger logic
is formed in gate array chips which
combine the (x,y) coordinates in the pots along with a lookup-table
incorporating the transport matrix equations
to give bins of 
and |t| which can be used in a Level 1 trigger.
The total time required for the decision is only a couple of
hundred nanoseconds
longer than the sum of the proton transit time and the time for the return
of the signal to the interaction region.
A number of variations on the Scintillating fiber design are being considered. One which should be mentioned is the possibility of using capillary bundles filled with liquid scintillator instead of plastic scintillating fibers.
New types of photodetectors (APD, HPD [12], etc.) will also be tested for this application because they may permit a reduction of the readout cost.
A rough estimate of the cost per pot is shown in Table 5.3
(see also Fig. 5.18). The costs of Si pixel detectors has been
reviewed in Subsection 5.2.6.
Si strip detectors are somewhat cheaper,
about 60 CHF/cm .
The cost of the scintillating fiber detector is dominated by the MAPMT which is about $100 per channel. The fiber and scintillator costs are minimal. Other costs including cables, trigger boards and high voltage would bring the estimated cost to about $150 per channel.
