FRC Status Report of the CAST Experiment The CAST Collaboration

CAST FRC-D 2004-9
FRC Status Report of the CAST Experiment
The CAST Collaboration
Athens-CERN-Chicago-Darmstadt-Frankfurt-FreiburgMoscow (INR)-Munich (MPE-MPI)-Patra-Pisa-SaclaySouth Carolina-Thessaloniki-Vancouver-Zagreb-Zaragoza
September 2004
1. Cryogenics and magnet operation in 2004
Cryogenics
Following on from the successfully commissioning of the CAST cryogenic
system in 2001-2 and the installation of the new 1.8K pumping group during
the 2002-3 shut down, the cryogenics system continues to operate well
within our normal tolerances of better than 99.5% reliable production. The
initial ramping of the magnet at the start of the 2004 data taking campaign
was only interrupted by a water supply fault and one or two training ramps
required to regain full field strength. The thermal stability of the system
remains good and the automated control system has been fine tuned to allow
fully automated recovery from any quenches, which may occur due to utility
"outages". Thus far in 2004 we have experienced approximately 11
interruptions to cryogenic production; 4 directly due to interruptions to the
cooling water supply to the magnet supply converter. These were eventually
traced to a faulty relay within the cooling water supply system; 3 additional
quenches with no definitive explanation may also be attributable to the same
cause due to the relays initial deterioration; 1 due to the activation of a
circuit breaker, again in the general water supply system; 1 due to the failure
of the electricity supply in the building, thought to have been caused by
workmen installing new cabling for the LHC; The final 2 interruptions are
directly related to the cryogenic production system; The first caused by the
failure of a bearing on the main compressors oil pump. This required a
system stop of 5 hours in order to replace the pump with a spare from stock.
The second interruption was due to the loss of signal from the compressors
oil level gauge, which was found to be due to vibration causing a break in
the signal line.
The only ongoing concern relating to the Cryogenic / Magnet system is the
recurrent leak of helium into the isolation vacuum, believed to be coming
from a super fluid helium leak from the cryogenic pipe work within the
-6
-4
MFB / Magnet. The leak rate has been monitored at 10 to 10 for several
weeks and on no occasion has the leak exceeded the pumping capacity of the
turbo pump connected to the isolation vacuum. Further investigation of this
leak will be conducted during the Phase I/II shutdown.
The ECR cryogenics group is currently helping with the ongoing planning
for CAST’s phase II operation, contributing to the progress of the
experiment.
2. TPC
The Time Projection Chamber (TPC) of the CAST experiment has been
operational during most of the 2003 data taking phase of the experiment. At
the end of 2003, the TPC was dismounted in order to proceed with the
winter shutdown. Concerning the TPC, the works during that period
regarded the building and installation of a differential pumping system and
the installation of the shielding. Since May 2004 the TPC is back on the
magnet and it is taking data in its full performance, with substantially
improved conditions compared with last year.
2003 data taking
At the end of the 2003 CAST data taking phase, the TPC detector had
gathered 151 hours of data in “axion-sensitive” conditions (tracking the Sun
with the magnet energized) and 172 days of background data.
Approximately half of these tracking data corresponds to the data set 6,
taken in especially favorable conditions during the period June-August, and
which has been used to derive the 2003 result on the coupling constant gaγγ
presented below.
The main problem that the data analysis had to face in 2003 was the fact that
the background data depended substantially on the magnet position. This
fact prevented a straightforward subtraction of the total background
spectrum from the tracking spectrum, needed to extract a limit on the axionphoton coupling. To cope with this, a “weighting method” was designed to
obtain an effective background from the background data taken only in the
magnet positions where tracking is performed, appropriately weighted with
the relative tracking exposure. This effective background could be safely
compared with the tracking spectrum, but with the payload of a reduction of
the background statistics, since not all the available background material is
finally used (Fig. 1). The method proved successful, and the subtracted
spectrum shown in Figure 2 was obtained.
From that spectrum the best fit value was obtained (gaγγ)4 =(-1.0±3.0)×10-40
GeV-4 (lower solid line in Fig. 2). Also an upper limit of gaγγ=1.51×10-10
GeV-1 is given with 95% CL (corresponding expected signal represented by
the upper solid line in Fig. 2). The limit has been obtained by renormalizing
the Bayesian probability in the physically allowed region. This limit has
been calculated for different values of the axion mass so the complete
contour line in the axion parameter space has been calculated. It is shown in
Figure 3.
Shutdown period: improvements for 2004 operation
The TPC work during the shutdown has been focused in two major items
that will contribute to a better data quality during the ongoing 2004 data
taking phase of CAST. The first is a differential pumping system for the
TPC X-ray windows and the second one is the detector passive and active
shielding.
Differential pumping for the TPC
The leak through the windows of the gaseous detectors towards the magnet
is a limiting factor for the proper operation of CAST. In fact, during 2003,
three weeks of shutdown for the TPC happened due to slight deterioration of
the windows that produced unacceptable leaks towards the magnet. The aim
of the differential pumping system, following the successful experience of
the micromegas detector, is to reduce the effective leak towards the magnet
due, for instance, to the formation of pinholes in the metallic layer of the
detector windows or just diffusion through them. This is accomplished by
the continuous pumping of an intermediate volume, separated from the
magnet also by a thin window. This intermediate volume is under a
relatively poor vacuum (~10-5 mbar in the case of the TPC) but the leak rate
through the second window is very small due to the small pressure
difference.
Such a system was designed, built and installed during the last shutdown
period. After a thorough test the system proved to work satisfactorily. With
the TPC back in place, the effective leak towards the magnet is 1.5×10-7
mbar·l/sec of argon, a factor 240 better than in the 2003 operation. This
situation allow us to have the necessary margin to tolerate small
deteriorations of the windows and therefore have a continuous and
homogeneous operation during the whole 2004 without the need of
intervention on the detector.
TPC Shielding
A passive shielding for the TPC had been designed and built, but due to
problems of weight constraints of the mechanical CAST structure it had not
been installed during 2003 but for testing purposes. Those problems have
been solved in the last shutdown (see below), and the shielding has been
finally installed for the 2004 data taking phase. It consists of an innermost
copper cage of 5 mm of thickness, surrounded by a lead layer of 2.5 cm, 1
mm of cadmium and finally a wall of 20 cm of polyethylene. The full set is
closed by a plastic bag, which is inside slightly pressurized by injection of
clean N2. This reduces the radon concentration from the air in the space
close to the detector, further reducing the detector background.
The 2004 data is giving a very satisfactory result of the effect of the
shielding (Fig 4.). Regarding the average background level, it is compatible
with the expectations drawn from 2003 tests. The TPC background between
1 and 10 keV is ~ 3.85×10-5 counts/KeV/sec/cm2, a factor of 4.4 below the
level reached by the TPC without any shielding. Also it represents a factor
of ~2.4 of improvement with respect 2003 operation (the TPC was then
shielded by only a copper box and N2 flux).
But more importantly, the shielding was supposed to reduce the worrisome
dependence of the background on the magnet position, and the result
concerning this issue shows no dependence so far. This allows us to use the
full background data in the analysis, reducing the statistical error bars and
therefore increasing the sensitivity.
In addition to the passive shielding, a muon veto has been installed on top of
the shielding. Though muons interacting directly in the TPC are very clearly
identified and rejected, there may be secondary particles (photons, neutrons,
…) produced by muons traversing the shielding that may mimic X-ray
interactions in the TPC. The background reduction capability of the veto is
~4% taking into account the prompt coincidences.
2004 data tacking
The TPC is installed on the magnet and taking data routinely with its full
performance since the 25th of May. The interventions described above have
produced an important improvement in the sensitivity, basically due to a
more homogeneous data taking all along the 2004 phase, and the effect of
the shielding both concerning overall background level and reduced
sensitivity on the magnet position.
At the moment of writing this text, 116 hours of tracking data (~ double of
the 2003 period) and 71 days of background have been gathered. The 2004
background and tracking spectra are show in Figure 5. Also for the data of
this year we show the very preliminary subtracted spectrum in Figure 6. The
lower line represents the best fit value: (gaγγ)4 =(1.4±1.1)×10-40 GeV-4 and the
upper one corresponds to the 95% CL upper limit: gaγγ=1.28×10-10 GeV-1 .
Figure 1 Energy spectra with TPC. Data corresponding to sun tracking (red)
and background (black) obtained during part of the 2003 operation period.
Figure 2. Subtracted spectrum (sun tracking minus background). The lower
line is the best fit axion signal and the upper one is the corresponding
expected axion signal at 95% CL.
6
Figure 3. Exclusion plot for the axion-photon coupling constant vs. axion
rest mass. The CAST TPC value obtained during the 2003 operation period
is compared with previous limits, including also future prospects of CAST.
Figure 4. Energy spectra corresponding to different shielding conditions.
7
Figure 5. 2004 TPC data. Sun tracking and background energy spectra.
Figure 6. Subtracted spectrum. The lower line is the best fit axion signal
and the upper one is the corresponding expected axion signal at 95% CL.
8
3. The X-ray Telescope and the pn-CCD Detector
Modification of the tracking system of CAST
The tracking system of CAST allows following the sun with the magnet for
2x1.5 h during sunset and sunrise every day. Thereby the pointing accuracy
of the magnet relative to the center of the sun has to better than ±1 arcmin at
any time. To measure the absolute pointing direction angular encoders are
installed close to the tracking motors, but far away from the vertical and
horizontal rotation axis of the magnet. Therefore, changes of the magnet
structure can not be measured during tracking. For this reason, we improved
the system by installing two additional angular encoders directly on the
vertical and horizontal rotation axis of the magnet. These new devices allow
us to measure the non-linearity in the tracking system, which can be taken
into account, and partially are corrected by the tracking software. To do so, a
correction matrix had to be derived from a systematic survey at different
magnet positions (see below).
Influence of Pointing on the Effective Area
100.0
Vignetting for the theoretical axion distribution
90.0
Vignetting for a point source
80.0
70.0
Loss [%]
60.0
50.0
40.0
30.0
20.0
10.0
0.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
10.0
Off-axis angle [arc minutes]
Figure 7. The relative efficiency loss of the X-ray telescope, for different
off-axis angles, is shown for a realistic extended axion source and for a point
source.
9
Alignment of the X-ray telescope and pn-CCD detector
The most critical parameter with respect the efficiency of the X-ray
telescope/pn-CCD system is the alignment of the optical axis of the
telescope system relative to the optical axis of the magnet bore. To reduce
the uncertainty of the effective area below 5 %, the optical axis of the whole
telescope/pn-CCD system has to be parallel to the optical axis of the magnet
with an accuracy better than 1.4 arcmin (= 0.023o see Figure 7). We used a
laser beam parallel to the magnet bore to align the X-ray telescope relative to
the magnet (Figure 8). With the laser, the telescope system was aligned
relative to the magnet bore and subsequently the pn-CCD detector relative to
the X-ray telescope. After and during the alignment procedure the positions
of all components were verified by the aid of the surveyors using reference
points on the magnet, telescope, and pn-CCD detector housing to cross
check the results derived from the laser measurements and to get a reference
data set. In addition,
Figure 8. Experimental setup for the alignment of the X-ray telescope and
pn-CCD relative to the optical axis of the CAST magnet.
10
Figure 9. Spot of the parallel laser beam observed with the pn-CCD
detector. The position and size of the spot correspond to the expected axion
image of the sun.
this setup was and will be used in future to determine the expected position
of the image of the sun on the pn-CCD chip with an accuracy close to the
spatial resolution of the CCD ( pixel size 150 μm = 20 arcsec). Figure 9
shows the resulting laser spot centered on the CCD chip after the alignment.
As a verification of the final alignment we measured the off axis behavior of
the focal spot for different off-axis angles in vertical and horizontal
direction. The results are shown in Figure 11 and demonstrate an almost
11
perfect linear dependency between the off-axis angle of the laser and the
position of the laser spot on the pn-CCD chip.
Stability of the alignment at different magnet positions
A miniature X-ray generator is mounted on the TPC side to monitor the
telescope performance. This source is aligned relative to the optical axis of
the magnet and can be moved in horizontal direction with a positioning
accuracy of ±1µm. It provides a near-parallel X-ray beam through the
magnet bore which illuminates the X-ray telescope aperture. Figure 10
shows the X-ray image and the energy spectrum of the calibration beam
observed with the pn-CCD. The spot size is larger than the expected size in
the focal plane of the telescope, since the X-ray beam is slightly divergent.
The X-ray beam allows to monitor the temporal stability of the focus and to
study a possible dependence on the orientation of the CAST magnet during
datataking.
12
Figure 10. X-ray image and energy spectrum of a near parallel X-ray beam
through the CAST magnet. The beam is produced by a ~100 MBq
pyroelectric source emitting mostly in the Cu-K fluorescent line.
13
Figure 11. Off-axis position of the laser spot as observed with the pn-CCD
detector depending on the off-axis angle of the parallel laser beam.
2
Effective area [cm ]
To study the influence of the magnet motion on the alignment of the optical
axis of the telescope relative to the magnet axis, we made systematic
measurements with this X-ray source at different magnet positions. The
results of these measurements show, that the spot position on the CCD chip
is stable with an accuracy better than 30 arcsec. This test will be repeated at
the end of the 2004 data taking phase to verify the stability of the alignment.
Effective area of the telescope
The effective area of the X-ray telescope is a function of the off-axis angle,
micro roughness of the mirror surfaces and the photon energy. To
characterize the effective area depending on energy we made calibration
measurements at the PANTER test facility in Munich for a telescope
aperture of 48 mm (see status report of 2003). In order to transfer these
results to an aperture of 42 mm, which is the diameter of the CAST magnet
bore, we made ray-tracing simulations for both, a 48 mm and 42 mm
diameter of the bore. These simulations include the mirror system with the
mirror support structure and the magnet tube geometry assuming a perfectly
straight beam pipe. As a result we get the energy dependant on-axis effective
area and the vignetting function depending on energy and off-axis angle.
The energy dependant absolute effective area of the CAST X-ray telescope
is shown in Figure 12. The simulated off-axis behavior is shown in
14
Figure 13 for off-axis angles in radial and tangential direction. The
asymmetric experimental setup with one of six sectors being illuminated
causes a difference between the tangential and the radial off-axis behavior.
On-Axis Effective Areas at PANTER and at CAST
12.0
Calculated effective area in PANTER (48 mm diameter aperture)
Measured Effective area in PANTER (48 mm diameter aperture)
10.0
Expected effective area at CAST (42 mm diameter aperture)
8.0
6.0
4.0
2.0
0.0
0.5
1.5
2.5
3.5
4.5
5.5
6.5
7.5
8.5
9.5
Energy [keV]
Figure 12. The effective area of the X-ray telescope depending on the
photon energy.
15
Influence of Pointing on the Effective Area
100
Tube
90
Loss [%]
Telescope (tangential)
80
Telescope (radial)
70
Tube + Telescope (radial)
60
Tube + Telescope
(tangential)
50
40
30
20
10
0
-10
-10
-8
-6
-4
-2
0
2
4
6
8
10
Off-axis angle [arc minutes]
Figure 13. The effective area of the X-ray telescope depending on the
radial and tangential off-axis angle. The blue and red lines indicate the total
influence of the telescope and tube geometry on the effective area of the
telescope.
Improvements to the detector during the winter shutdown 2003/2004
We made intensive tests to reduce the abnormal electronic noise of the pnCCD detector. Unfortunately, up to now the source for the electronic noise
or a correlation to electronic equipment or the other detectors in use in the
CAST experimental area could not be found, nor could the noise level be
reproduced in our laboratory in Munich. Due to this problem the energy
range of the pn-CCD is restricted to energies above 0.75 keV compared to
250 eV when the detector is being operated in Munich. The pn-CCD
detector was calibrated in the range of 0.5-9 keV using an X-ray tube. The
linear gain of the detector is shown in Figure 14, with an energy resolution
of 160 eV at 6 keV.
16
Figure 14. Energy calibration of the CCD detector.
The detector shielding was enhanced by adding a 17 to 25 mm thick layer of
low-activity ancient Pb free of 210Pb on the outside of the Cu shielding used
during the data taking period in 2003. The additional shielding has reduced
the background level by a factor of 1.5 to a mean flux of (7.5 ± 0.2) x 10-5
counts/cm2/keV/sec in the energy range of 1-7 keV (Figure 15), which is
consistent with background measurements in Munich.
The 55Fe calibration source installed on a manipulator close to the pn-CCD
detector was modified to reduce the count rate of the source by adding an Al
filter. Thus, the emergent Al fluorescent emission line can also be used for
calibration.
2003 data taking period
The X-ray telescope of CAST has acquired in total 121 h of data under axion
sensitive conditions and 1233 h of background data. The background data
was observed under different magnet conditions (magnetic filed on/off,
different pointing positions) and allows a systematic study of the
background. Contrary to the background observed with TPC the background
of the CCD only weakly depends on the magnet orientation. Systematic
17
Figure 15. Background spectrum measured with the pn-CCD detector.
The additional Pb shield added in 2004 has reduced the background level by
a factor of 1.5 to a mean differential flux of (7.5 ± 0.2)x10-5
counts/cm2/sec/keV in the energy range of 1-7 keV.
Inlay: Calibration spectrum of an 55Fe source.
studies of background variations and their influence on the upper limit of gaγγ
are currently in progress by the aid of Monte Carlo simulations. During the
data taking phase in 2003 a continuous monitoring of the pointing stability
of the X-ray telescope was not possible, since an appropriate tool for this
purpose was missing in the experimental setup of CAST. Though, using the
results of the alignment measurements, we are able to restrict the region on
the CCD chip where we expect the potential axion signal to an area of 54.2
mm2. Using data from this area we derived a preliminary upper limit of
gaγγ(95%)<1.2x10-10GeV-1 including the detector and the telescope
efficiency. For the 2004 data we will be able to use the full potential of the
X-ray telescope by reducing the signal region further to the size of the area
of the spot of the sun (A=6.4 mm2).
18
4. The MICROMEGAS detector
Starting in May 2003 and until the shutdown in November 2003, the
Micromegas detector had been taking data, showing a remarkable stability in
its performance. An upper limit was derived for the coupling constant of the
axion to two photons. The detector was changed for the 2004 data taking
resulting in an increased performance and a better and more automated
system.
Data taking period of 2003
During the 2003 data period the Micromegas detector showed the stability
shown in Figure 16. The implementation (August 2003) of a high sampling
VME Digitizing Board (MATACQ Board) in order to read and record the
time structure of the pulse coming from the mesh, paid off giving a
significant advance in the analysis.
In total, 810 h of background data and 77 h of tracking data were
accumulated. The achieved background rejection was of the order of 7×10-5
counts keV-1 s-1 cm-2. Figure 17 shows the subtracted spectrum, tracking
minus background, representing approximately a third of the complete data
set, giving a limit on the axion-to-photon coupling of gaγγ=1.69×10-10 GeV-1
at 95% CL, following the same procedure as described for the TPC. The
limit obtained for the axion-photon coupling given by the complete set of
data is gaγγ=1.46×10-10 GeV-1 at 95% CL
Improvements for 2004 during the shutdown period and present status
For the 2004 data taking an improved detector was installed. The energy
resolution of the Micromegas detector is taken from the VME Digitizing
Board and is about 20 % (FWHM). The energy response from the strips,
which is used for background rejection, has been improved with respect to
2003 as can be seen in Figure 18. This figure shows the cluster energy
spectrum for the Fe55 source for 2003 and 2004. Thanks to the improved
resolution on the strips, the level of background is ameliorated by a factor of
~2.5 with respect to results obtained from 2003 data as shown in Figure 19.
A faster VME Digitizing Board has improved the data transfer and an
automatic source controller for calibrations, along with more information on
display, allow for an easier and more direct control and monitor of the DAQ
system. In total, 1700 h of background and 123 h of tracking data have been
accumulated up to the beginning of September 2004.
19
mV/keV
Pulse_height Calibration
54
52
50
48
46
44
42
40
38
36
34
0
2
4
6
8
10
12
14
16
18
Days after 20/10/2003
Figure 16. Stability of the detector : gain (mV / keV) as a function of time.
Subtracted43_3
-1
counts s cm-2 keV
-1
x10
Subtracted43_3
Entries
-13093
Mean
4.497
RMS
4.315
-4
0.4
0.2
0
-0.2
-0.4
-0.6
0
2
4
6
8
10
12
14
[keV]
Figure 17. Subtracted spectra for a third of the Micromegas 2003 data. The
black line represents the best fit to the data and the red line the gaγγ value
excluded at 95 % C.L
20
counts
2000
counts
2250
2003
1750
2000
2004
1750
1500
1500
1250
1250
1000
1000
750
750
500
500
250
0
250
0
1
2
3
4
5
6
7
8
9
10
0
0
1
2
3
4
5
6
7
8
9
10
cluster energy (keV)
cluster energy (keV)
Figure 18. Cluster energy spectrum using a Fe55 source for the 2003 and for
the 2004 data set with Micromegas detector.
sec -1cm-2keV -1
Compared background spectrum
×10-3
0.3
0.25
0.2
0.15
0.1
0.05
0
0
1
2
3
4
5
6
7
8
9
10
11
12
keV
Figure 19. Background spectra for the 2003 data set (in empty histogram)
and for the 2004 data set (in full histogram) .
21
5. CAST High-Energy Calorimeter
A high-energy calorimeter dedicated to parasitic searches for
solar/cosmological axion-like particles has been installed during May-June
2004 behind the previously operating Micromegas X-ray detector (which is
transparent to photon energies above a few tens of keV). Since a new boson
(such as the axion) may couple to nucleons, it can substitute for a photon in
nuclear and plasma processes. In particular, a pseudoscalar axion will be
predominantly coupled to magnetic nuclear transitions. Therefore, a realistic
high-energy boson search should continue to rely on the Sun, with its wealth
of nuclear reactions and high luminosity, as a source of such particles. Weak
experimental limits already exist from the observed flux of solar gammarays below 5.5 MeV, to which axion decay (a→γγ) following (p+d→3He+a)
might contribute. Other unexplored interesting channels exist and a generic
search should not be limited to pseudoscalar particles (i.e., M1 transitions
like the one mentioned). It should also be kept in mind that astrophysical
constraints still allow axions to partake of a non-negligible fraction (few %)
of the total solar luminosity and that this type of higher-energy search has
not been performed with a helioscope before. In particular, the absence of a
511 keV excess signal (from e++e-→γ+a) in these additional CAST detectors
at times of solar alignment may impose tighter bounds than similar searches
for anomalous production of single photons in accelerator experiments. The
addition of the calorimeter thus extends the sensitivity of the CAST
experiment from a few tens of keV up to ~200 MeV.
The calorimeter construction and design relies on low backgroundradioactivity techniques applied to the inorganic scintillating crystal
calorimetry. The 0.6 kg (Ø45 x 50 mm) CdWO4 crystal was chosen after
careful screening of many types of crystals (BGO, BaF2, etc.) for its
extremely low intrinsic radiocontamination, stopping power for high energy
photons (>20% full peak efficiency below 10 MeV) and excellent pulse
shape discrimination (PSD). In addition, low background techniques such as
ancient lead shielding, radon displacement, PMT with low intrinsic 40K
content, active muon veto, and high radiopurity of all materials are
incorporated into the detector design. Due to space constraints on the
crowded CAST magnet platform, strict limitations were placed on the
shielding and data acquisition systems. With this goal in mind, a powerful
and effective PSD algorithm was developed that allows for efficient offline
rejection of alpha particle induced events, neutron recoils, and spurious PMT
22
and electronic noise events, thus reducing the absolute need for more
extensive shielding.
The first solar tracking run upon the 2004 restart took place on 10 June and
the calorimeter has been fully operational since. With 69 days of successful
solar tracking out of the 73 possible the detector operation has proved
extremely efficient, totaling ~103 hours of tracking data and more than 1000
hours of background data at the time of this writing. A preliminary analysis
of a few weeks of data is shown below, with the energy spectrum
accumulated during tracking (9.28 hrs) compared with the time normalized
background spectrum (130.5 hrs total). It should be noted that corrections to
local background conditions in different magnet positions is not included
yet. The achieved counting rate following very weak software cuts is
already ~1.65 Hz over the energy range 200 keV – 150 MeV, which is
excellent for a crystal of this mass, operated at sea-level and without heavy
shielding.
Figure 20. Comparison of energy spectra acquired during solar tracking
(9.28 hours) and background measurements (130.5 hours normalized to
tracking time) with very moderate software cuts applied to both data sets
(rejecting only spurious noise events, electronic pulser events and some
alpha particle-induced events). Counting rate over full energy spectrum
(200 keV to 150 MeV) after cuts ~1.65 Hz. Correction to local background
conditions in different magnet positions is not included yet.
23
6. Grid measurements
The CAST magnet is pointed to the sun with the support of an azimutal
mounting. The pointing accuracy for the tracking system is specified to be
below ± 1 arcmin. Caused by the constructive constraints (cryogenic
system) it was not possible to move the centre of gravity of the magnet to the
crossing point of the 2 rotation axis. Therefore the mounting has resulted in
some imperfection of positioning the magnet to the sun due to an
overestimation in the mechanical mounting. These imperfections of the
pointing system can be overcome by using an adapted pointing software.
The basic elements for the software is the so called grid measurement. The
basic idea of the correction is, that the adjustment of the magnet for several
pointings is correlated to an astronomical coordinate system. This matrix of
correction has been determined in 2002. We also have verified by
observation of the sun with a telescope in the visible, that the system works
very well.
During the last observation period we have found some malfunctions of the
moving system, and we had to modify some components; during the last
shut-down phase we have increased the weight of the movable part by
~10%. Therefore, we had again to verify the used matrix of correction.
Before we started our observations we again performed a grid measurement.
This time we have found some imperfection: a slip of the motor encoder and
some shift and drift in the horizontal rotation. In the meantime we have
solved these malfunctions of the moving system. Based on the results of a
little grid measurement performed in June 2004, we can use the matrix of
correction from 2002 and the available software. The deviations of
adjustment for the grid measurement 2002 and the little grid measurement
2004 is < 0.01° and this is within our error box.
Since we have not solved fundamentally the mechanical problems in the
tracking system, we have to improve the concept for the 2nd phase of CAST.
To do this task we have to study the dynamical behaviour of the moving
system and structure of the magnet.
7. Second phase of CAST
Introduction
For axions, the ratio between mass and photon-coupling is given, up to
model-dependent numerical factors, by the corresponding pion properties.
This defines the "axion line" in the parameter plane of mass and photon
coupling strength. The gas-filling phase of CAST will allow us to extend our
24
sensitivity up to masses of about 1 eV and thus to "cross the axion line".
This will be the first laboratory experiment to reach a sensitivity where the
existence of "invisible axions" is tested, albeit only in a narrow range of
masses around 1 eV. All previous experiments, including the CAST vacuum
phase, were only able to search for generic axion-like particles with masses
much smaller than would be expected if they are indeed responsible for
solving the strong CP problem.
Experimental aspects
CAST plans to operate with low-pressure helium gas in the magnet cold
bore from 2005 onwards. This requires the installation inside the cryostat of
thin windows to contain the helium in the isothermal region of the cold
bores, so that good vacuum in the beam pipe extensions leading to the
detectors can be preserved and a low heat leak along maintained. The cold
windows need to be helium gas-tight, capable of operating at 1.8K and able
to withstand rapid temperature and pressure fluctuations in the event of a
quench. Finally, the windows must transmit efficiently low energy X-rays
above ~ 1keV.
The density of the cold bore gas filling will be varied in order to tune the
magnet to different axion rest masses. In order to operate safely at 1.8˚K
(avoiding condensation), He-4 can be used up to a maximum pressure of ~ 6
mbar and He-3 up to ~ 60 mbar. Initially, low pressure He-4 will be used
together with a relatively simple gas system where the risks due to the
effects of quenches are small. With the experience gained and after in situ
tests - the final gas system can be completely specified and built - ready for
the more challenging higher-pressure He-3 running starting about a year
later.
Schedule
In an extended winter shutdown, the preparations and modifications will be
made for the start of the 2nd phase. He-4 running is planned in the second
half of 2005 and will then consist of about 150 sun-tracking runs divided
into 30 discrete pressure settings between 0 and 6 mbar (extending the axion
mass coverage up to ~0.35 eV/c2). The subsequent He-3 running will consist
of about 300 sun-tracking runs divided into discrete pressure settings
between 6 to 60 mbar (extending the axion mass coverage up to ~ 0.8 eV/c2).
25
8. Modifications to CAST for the second phase
Mechanical intervention on the magnet movement system
In the extended shutdown leading to the second phase, modifications to the
mechanical lifting system should be made following the recommendations of
the study made by TS-MME group.
Installation of cold windows
The cold windows are a crucial element in the success of the 2nd phase. No
such windows exist in commerce and a collaborative R&D project together
with METOREX (FI) has been launched to develop such windows. A small
8mm diameter prototype has been successfully tested at 1.8˚K at CERN
(Cryolab Note 03-04). The final window will have a diameter of ~ 50 mm
(cold bore diameter 43mm), METOREX is now fabricating two 50 mm
diameter prototype windows, which will be tested by early autumn 2004.
Helium gas system
The gas system must be able to maintain a precise and constant pressure
during data taking and also be able to respond rapidly in the event of a
quench in order to protect the integrity of the cold windows. One key
parameter involved in the system design is the transient temperature
response of the gas filling of the cold bore in the event of a quench. This
crucial parameter has never been measured in an LHC magnet. A test will be
made to fill the cold bore with low pressure He-4, provoke a quench and
monitor the pressure rise in the gas. The passive and active safety systems
for He-3 running will be designed to cope with the measured response time
and the pressure limitations of the cold windows. The He-3 system must
necessarily be a high-purity and completely leak-tight recirculation system
in view of the high cost of He-3.
The vacuum lines between cold windows and detectors can no longer be
assured by the cryo-pumping of the cold bore after the installation of the
cold windows. Active pumping of the external pipe work will be necessary.
The vacuum pumping system must be interlocked and integrated with the
gas system and be interlocked to a He-3 leak detector in case of leaks
developing on cold windows. A selected doctoral student supervised by
experts from AT-ACR, AT-ECR and AT-VAC will make the final design.
26
Plots without absorption
4.2 keV
All energies
Figure 21. These curves show the steps needed to change the buffer gas
pressure inside the magnetic pipes for the 2nd phase of CAST.
27
APPENDIX
28
Cryolab Note 03-04
16/02/2004
Tests d’étanchéité
Sur PG-W X-Ray Windows
S.Prunet, T.Niinikoski, J.M.Rieubland, A.Vacca, L.LeMao
Demande CAST :
Du 20 Septembre 2003, par Martyn Davenport
Le but du test est de vérifier l’étanchéité à l’He de deux fenêtres identiques, la
fenêtre doit répondre aux exigences suivantes :
-
Le diamètre de la fenêtre est celui du LHC Dipôle cold bore ( 40 mm).
La fenêtre doit être transparente aux rayons X supérieurs à 1 KeV.
La fenêtre doit résister à une pression d’ 3He inférieure ou égale à 300 mbar
dans un sens, à une température de 1.8 K.
Cette fenêtre est fixée sur bride CF63
Pour les tests réalisés au Cryolab :
-
Les tests de taux de fuite sont réalisés avec de l’ 4He et non de l’ 3He.
Pour les premiers tests, le diamètre de la fenêtre est de 8 mm.
Cette fenêtre fabriquée par Metorex est constituée d’un film de Polyimide de 300
nm d’épaisseur sur lequel est déposé une grille de Polyimide par photolithographie, un
dépôt de nitrure d’aluminium de 30 nm est ensuite réalisé de chaque coté. Pour empêcher
le passage de lumière un dépôt de 40 nm d’aluminium est par la suite appliqué d’un coté.
Pour terminer, une grille de renfort en tungstène est plaquée d’un coté. Pour le test cette
fenêtre est collée sur une bride CF 63.
Nous disposons de deux fenêtres identiques de 50 mm2 (Fenêtre A et B) de 8 mm
de diamètre.
29
Ces fenêtres supportent un delta P maxi de 1100 mbar uniquement dans un sens.
On nomme chaque coté de la fenêtre HP (haute pression) et BP (basse pression). La
variation maximum de pression supportée est de 100 mbar/s, pour prévenir les effets
dynamiques et de frottement.
Fenêtre A montée sur une bride CF 63
30
1. Présentation du cryostat.
1.1 Schéma de principe du test.
Pompage du bain 4He
V2
Canne de transfert
Réservoir d’ 4He
50 l
P=300 mbar
V1
Ecrans thermiques
Détecteur
de fuites
Niveau Hélium
BP
X-Ray Window
HP
Pt 100
C100
Figure 1
Fenêtre montée dans le cryostat, en phase de refroidissement.
Grâce à un système de presse étoupe, la position de la fenêtre est réglable. En début
de refroidissement, la fenêtre est en position haute (fig 2 a) puis progressivement, la
fenêtre est immergée dans l’ 4He liquide et elle est placée en position basse (fig 1).
La mesure du taux de fuite est réalisée avec le détecteur de fuite et l’alimentation
en 4He se fait par le réservoir de 50 L.
31
fenêtre
b)
a)
Figure 2
Déplacement en hauteur de la fenêtre pour permettre un refroidissement
progressif
1.2 . Principe du test.
Pour l’ensemble des tests, le coté BP est pompé à l’aide du détecteur et le coté HP
est pressurisé à 300 mbar avec de l’ 4He. Cette pression d’ 4He correspond aux
spécifications de CAST.
Le taux de fuite d’ 4He, la pression et la température des brides sont enregistrés
simultanément pendant les cycles en température et en pression.
Les deux fenêtres sont testées à 300 K, 4.2 K et 1.8 K avec le delta P de 300
mbar. Un cyclage en pression est aussi réalisé à 4.2 K.
La fenêtre est refroidie dans un premier temps avec les vapeurs froides d’hélium
et se trouve en position haute (Voir fig.1 a).
Puis progressivement lorsque la fenêtre est 20 K, l’ensemble brides/fenêtre est
descendu en position basse (Voir fig. 2 b).
Le transfert d’hélium se poursuit jusqu'à immersion totale du système. L’ensemble se
trouve ainsi à 4.2 K.
32
Pour les mesures à 1.8 K, un pompage du bain d’hélium est réalisé pour atteindre une
pression de 20 mbar.
1.3 Photos du test
Emplacement
de la fenêtre
a) Cryostat
V2
V1
b) Partie supérieure du cryostat
Figure 3
33
Emplacement de
la fenêtre
Partie inférieure du cryostat
Figure 4
34
2
Test de la fenêtre A.
2.1 Mesures a chaud.
La fenêtre A est testée avec la BP reliée au détecteur donc sous vide et la HP pressurisée
à 300 mbar. Ceci à une température de 300 K.
Le signal détecteur est enregistré sur 16 heures entre le 20/01/04 et le 21/01/04.
Signal détecteur
E-10 mbar.l/s
Test fenêtre A
A chaud
Température fenêtre
en K
12
294
10
293
8
292
6
291
4
290
2
289
0
16:19:12 17:45:36 19:12:00 20:38:24 22:04:48 23:31:12
0:57:36
2:24:00
3:50:24
5:16:48
6:43:12
Détecteur
T en K
288
8:09:36
Temps
Figure 5
Enregistrement du signal du détecteur de fuite,
La fenêtre est sous 300 mbar de delta P à température ambiante.
Le signal détecteur sur une moyenne de 12 heures est de 3.46 E-10 mbar.l/s à une
température de 291.8 K.
35
2.2 Mesures à 1.88 K
Après la mesure à chaud le cryostat à été rempli d’hélium, la fenêtre a été ainsi
refroidie à 4.2 K. Et sans attendre, le pompage du bain a été réalisé pour atteindre la
température de 1.88 K.
Test fenêtre A
Refroidissement de 292 K à 1.88 K
Signal détecteur
E-9 mbar.l/s
Température fenêtre
en K
12.00
300
10.00
250
8.00
200
6.00
150
4.00
100
2.00
50
0.00
14:09:36
Détecteur
T en K
0
14:31:12
14:52:48
15:14:24
15:36:00
15:57:36
16:19:12
16:40:48
Temps
Figure 6
Enregistrement du signal détecteur et refroidissement progressif à 4,2 K puis à
1,88 K par pompage sur le bain.
A cette température (1.88 K), le bain du cryostat est à 22 mbar, l’hélium du coté HP
se condense et la pression passe de 300 à 22 mbar.
36
Test fenêtre A
A 1.88 K
Signal détecteur
E-9 mbar.l/s
Température fenêtre
en K
2.00
4.5
1.90
4
1.80
3.5
1.70
3
1.60
2.5
1.50
2
1.40
Détecteur
T en K
1.5
1.30
1
1.20
0.5
1.10
1.00
15:50:24
16:04:48
16:19:12
16:33:36
16:48:00
17:02:24
17:16:48
17:31:12
17:45:36
0
18:00:00
Temps
Figure 7
Enregistrement du signal détecteur pendant le refroidissement à 1. 88 K.
Le signal détecteur sur une moyenne de 2 heures est de 1.36 E-9 mbar.l/s à une
température passant successivement de 4.2 K à 1.88 K puis 4.2 K.
Le refroidissement à 1.88 K ne provoque aucune fuite.
Remarque :
Pendant le montage du système dans le cryostat, le détecteur a été remis à l’air, ce qui
lui a fait perdre une décade en bruit de fond. Il est passé de 4 E-10 à 2 E-9 mbar.l/s.
37
2.3 Cyclage en pression.
Pour tester la résistance de la fenêtre, le coté BP est sous vide et branché sur le détecteur.
Le coté HP est pressurisé avec une pression de 300 mbar puis pompé à quelques
mbars à la vitesse de 1 mbar/s. Ce cycle doit être répété 10 fois.
L’ensemble du système est plongé dans l’hélium à une température de 4.2 K.
En fermant très lentement la vanne V2, le signal du détecteur est passé
brusquement de 2 E-9 mbar.l/s à 4 E-5 mbar.l/s en quelques minutes.Une faible montée
en pression du coté HP est observée simultanément.
Test de cyclage en pression
Fenêtre A
Signal détecteur
en mbar.l/s
T fenêtre
4.50E-05
30
4.00E-05
25
3.50E-05
3.00E-05
20
2.50E-05
2.00E-05
Détecteur
T en K
15
1.50E-05
10
1.00E-05
5.00E-06
5
0.00E+00
13:55:12
14:02:24
14:09:36
14:16:48
14:24:00
14:31:12
14:38:24
0
-5.00E-06
Temps
Figure 8
Observation de la fuite pendant le cyclage en pression.
La fenêtre A est détériorée et n’est donc plus étanche. Cette monté en pression est
probablement due à un effet Taconis (oscillation thermomécanique) qui a perforée la
feuille de polyimide.
Pour résoudre ce problème, cinq chicanes ont été mises en place (montées avec un
jeu de 0.2 mm dans le tube et distantes de 100 mm), dans le tube coté HP, pour empêcher
l’oscillation thermomécanique.
L’ensemble du système est réchauffé et la fenêtre A est démontée. L’observation
visuelle de la fenêtre ne montre pas une détérioration la surface de la fenêtre.
38
3
Test de la fenêtre B.
a.
Mesures a chaud.
La fenêtre B est testée avec la BP reliée au détecteur donc sous vide et la HP pressurisée
à 300 mbar.
Le signal détecteur est enregistré sur 14 heures entre le 26/01/04 et le 27/01/04.
Test fenêtre B
à chaud
Signal détecteur
E-9 mbar.l/s
10
Température fenêtre
en K
294
9
293
8
7
292
6
5
291
Détecteur
T fenetre
4
290
3
2
289
1
0
16:48:00
19:12:00
21:36:00
0:00:00
2:24:00
4:48:00
288
7:12:00
Temps
Figure 9
Enregistrement du signal du détecteur de fuite,
La fenêtre est sous 300 mbar de delta P à température ambiante.
Le signal détecteur sur une moyenne de 12 heures est de 2.33 E-9 mbar.l/s à une
température de 292K.
39
b. Cyclage en pression
Comme pour la fenêtre A, pour tester la résistance de la fenêtre, le coté BP est sous
vide et branché sur le détecteur.
L’ensemble est refroidit progressivement de 292 K à 4.2 K. Contrairement au test de
le fenêtre A, les cycles en pression sont réalisés avant le refroidissement à 1.75
K
Test fenêtre B
Signal détecteur
E-9 mbar.l/s
Température fenêtre
en K
12
300
10
250
8
200
6
150
Détecteur
T en K
4
100
2
50
0
10:33:36
11:02:24
11:31:12
12:00:00
12:28:48
12:57:36
13:26:24
13:55:12
0
14:24:00
Temps
Figure 10
Enregistrement du signal détecteur pendant le refroidissement progressif de la
fenêtre à 4.2 K.
Le coté HP est pressurisé avec une pression de 300 mbar puis pompé à quelques
mbars à la vitesse de 1 mbar/s. Ce cycle doit être répété 10 fois.
L’ensemble du système est plongé dans l’hélium à une température de 4.2 K.
Le test à été réalisé en deux fois, une série de 8 cycles le 27/01/04 et une série de 2
cycles le 28/01/04.
40
Série de 8 cycles :
Test de cyclage en pression
Fenêtre B
Signal détecteur
E-9 mbar.l/s
Pression HP
en mbar
6
350
300
5
250
4
200
Signal Détecteur
3
Pression HP en mbar
150
2
100
1
50
0
13:12:00
0
13:40:48
14:09:36
14:38:24
15:07:12
15:36:00
16:04:48
16:33:36
Temps
Figure 11
Enregistrement du signal détecteur pendant les 8 cycles de pression.
Série de 2 cycles :
Test de cyclage en pression
Fenêtre B
Signal détecteur
E-9 mbar.l/s
Pression HP
en mbar
350
6
300
5
250
4
200
Détecteur
3
P HP en mbar
150
2
100
1
0
9:36:00
50
9:43:12
9:50:24
9:57:36
10:04:48
10:12:00
10:19:12
10:26:24
10:33:36
10:40:48
0
10:48:00
Temps
Figure 12
Enregistrement du signal détecteur pendant les 2 cycles de pression.
Pendant l’ensemble des 10 cycles, le signal détecteur varie légèrement mais ne
dépasse jamais 3 E-9 mbar.l/s.
Le cyclage en pression n’a pas provoqué de fuite.
41
3.3 Mesure à 1.75 K
Le pompage du bain a été réalisé pour atteindre la température de 1.75 K.
A cette température, le bain du cryostat est à 15 mbar, l’hélium du coté HP se
condense et la pression passe de 300 à 15 mbar.
Test fenêtre B
A 1.75 K
Signal détecteur
E-9 mbar.l/s
Température fenêtre
en K
10
4.5
9
4
8
3.5
7
3
6
2.5
Détecteur
5
T en K
2
4
1.5
3
1
2
0.5
1
0
10:19:12
0
10:48:00
11:16:48
11:45:36
12:14:24
12:43:12
13:12:00
13:40:48
14:09:36
Temps
Figure 13
Refroidissement à 1.75 K après les cycles en pression.
Pendant le refroidissement, le signal détecteur passe de 2.7 E-9 mbar.l/s à 1 E-9
mbar.l/s.
La baisse en température ne provoque pas de fuite de la fenêtre B.
42
4
Résumé des mesures
T = 300 K
T = 4.2 K
T = 1.8 K
10 Cycles à
4.2 K
Fenêtre A
3.46 E -10 mbar.l/s *
1.36 E-9 mbar.l/s
1.36 E-9 mbar.l/s
non testé
Fenêtre B
2.33 E -9 mbar.l/s *
3 E-9 mbar.l/s
2.7E-9 à 1 E-9
mbar.l/s
3 E-9 mbar.l/s
* Variation du bruit de fond du détecteur.
43
Conclusion
Les deux fenêtres A et B étaient étanches à chaud avant les tests à froid.
La fenêtre A n’a pas supporté les oscillations en pression à 4.2 K, probablement du à
l’effet Taconis. La modification du tube avec les chicanes a permis de ne pas reproduire
ce phénomène pendant le test de la fenêtre B.
Les 10 cycles en pression sur la fenêtre B n’ont pas provoqué de fuite. Les tests sur
les deux fenêtres à 1.8 K n’ont pas provoqué de fuite.
Sur l’ensemble des mesures le signal du détecteur est toujours resté très proche de la
valeur de son bruit de fond :
- Les variations de signal sont probablement induites par le dégazage des
surfaces intérieures des tubes de pompage et des vannes métalliques, soumis à
la variation de température.
- Il n’y a pas de diffusion à travers la surface de polyimide, prouvée par
l’absence de corrélation claire entre le taux de fuite et la température (figure 6
et 10).
- L’absence de fuite par « trous » se manifeste aussi par ce manque de
corrélation (figure 6 et 10) et par l’absence de fuite en superfluide (figure 7 et
13).
- Pour terminer, le manque de corrélation entre le taux de fuite et la pression
(figure 11 et 12) prouve que la fenêtre B est parfaitement étanche.
44