- Email: [email protected]
Results from the chorus νμ → ντ oscillation experiment
ELSEVIER
Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373
RESULTS FROM THE CHORUS EXPERIMENT
v t, --+ Vr
PROCEEDINGS SUPPLEMENTS
OSCILLATION
A. Artamonov a for the CHORUS Collaboration alnstitute of Theoretical and Experimental Physics, Moscow, Russia, E-mail: [email protected] The CHORUS experiment accumulated ,,, 6 x 105v~ CC events, which represents ,,~ 3/4 of the statistics expected at the end of data taking. About 10% of the data were analysed and no r candidate was found. This corresponds to a 90% C.L. upper limit for sin~(2Ot,,) of 4.5 x 10 -3 at large Am 2. Extrapolating this pilot analysis to the full data set confirms that a sensitivity of sin2(20~,~) < 2 x 10-4 can be reached.
1. I N T R O D U C T I O N CHORUS [1] is one of two large neutrino osdilation experiments currently taking d a t a at CERN. It is aimed at a search for v# -+ vr oscillation with a 20 times higher sensitivity t h a n the one currently reached by previous experiments. If v~ -~ uT mixing exists at the present limit [2], CHORUS would detect about 60 vr events. The CHORUS experiment is searching for vT appearance in the CERN SPS neutrino beam consisting mainly of muon neutrinos with a negligible contamination of Vr [3]. The concept of the experiment is the direct observation of Vr induced CC events identified by the characteristic decay topology of the r lepton produced in this reaction. In most cases T decays into one charged particle (85%) appearing as a kink on a negative track coming from the vertex. At SPS energies the decay length of the T lepton is ,~ 1 m m and the kink angle is ~ 50 m r a d . Hence, the events we are searching for are confined to a tiny volume of the order of a m m 3. Thus, we need highly precise and fine grained measurements of the vertex topology such as only nuclear emulsion can provide. 2. D E T E C T O R The emulsion stack is the principal part of the detector. It serves two purposes. First, it is a target for v interactions. With a mass of ,,~ 800 kg it contains the largest amount of emulsion ever ex0920-5632/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S0920-5632(98)00066-8
posed at accelerators. Second, it is a very precise 3-dimensional vertex detector, producing -~ 300 grains~ram for a minimum ionising particle; each grain can be localised with a precision of ,~ l # m . In 4 years of exposure, about a million u interactions will be recorded in the emulsion. Even with the considerable progress achieved during the last years in the emulsion analysis technique, it is impossible to scan the entire emulsion target in the search for u interactions in a vast background of cosmic and beam muon tracks. This search must be g ~ d e d by predictions (on an event-by-event basis) of the position and the angle of tracks from u interaction leaving the emulsion. Precise predictions (ax ", 150#m, a8 "~ 2 m r a d ) are provided by scintillating fibre tracker located downstream of the emulsion. The particle tracks reconstructed in the tracker are extrapolated to the so-called interface emulsion layers as shown in Fig. 1. After being found, they are searched within a reduced area at the emulsion target exit and followed upstream, from plate to plate, to their production vertex. All these operations are performed using fully automated microscopes. But even the precise information about the vertex region is not sufficient to suppress backgrounds which can fake the r decay topology: the decays of charm mesons, the so-called "white kinks" (interactions of hadrons without any visible recoil at the vertex), decays of charged pions and kaons. To provide sufficient background suppression (less than one background event in the
A. Artamonoo/Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373
371
particles except of fast muons. Finally, there is a magnetlsed iron toroidal muon spectrometer. A detailed description of the CHORUS detector can be found in [4]. 3. A N A L Y S I S
View of m~roicope
of tau-neutcinol[~r) I~ emulsion .--'" . ~
IntAIrllc~ion ~ . . . _ ~
b
,,
decay '
- _ ~. _ 12.
Figure 1. Emulsion target and scintillating fibre tracker. full data set) the detector has to be able to determine the charge of secondary particles, reconstruct the event kinematics, identify muons pions and measure their momentum. The CHORUS detector (see Fig. 2) was designed to satisfy these requirements. Behind the
I Figure 2. CHORUS detector. veto planes there are emulsion stacks sandwiched with fibre tracker planes. They are followed by an aircore magnet spectrometer measuring the charge and the momentum of particles before they enter a high resolution lead-scintillating fibre calorimeter, measuring the total energy of all
The partial replacement of the emulsion target, at the end of 1994 and of 1995, allowed us to start a systematic data analysis while the experiment was still running. The analysis was focused basically, in a first stage, on the search for the muonic T decay mode. The reconstruction efficiencies for this mode are higher than for hadronic ones; the signature is quite clean: a single p - with a kink; the background is small. In parallel to the search for r candidates, at this stage of analysis a lot of effort was put into tuning and perfecting the event reconstruction algorithms as well as in development of the emulsion scanning technique.
3.1. Emulsion scanning technique Emulsion scanning is the most delicate and time consuming part of the analysis. The total amount of emulsion to be processed is so large (1.7 tons), that a qualitative improvement of the techniques used by preceding experiments was fi~eded. The solution was an automatic scanning procedure pioneered by the Nagoya group [5]. This is a completely new technology making the scanning process ~ 100 times faster than the manual scanning. In 1995, when the analysis was just launched, only a few automatic microscopes were available in Nagoya. During the last two years the Collaboration has progressively enhanced its scanning facilities. Now 10 automatic systems are running and 8 others will be installed later this year. In addition, a new generation of more powerful automatic systems has been made operational in Japan further increasing the scanning speed. These achievements allowed us to change the analysis strategy initially foreseen [1]. The strong kinematical preselections which were proposed to reduce the scanning load by an order of magnitude (of course at the expense of efficiency) are no longer necessary to complete the analysis in about two years.
372
A. Artamonoo /Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373
spatial image of the vertex region can be reconstructed. The example of a typical vertex "video image" is shown in Fig. 3 (bottom). It corresponds to a tiny volume of 0.02 m m 3. Measured "hits" are corrected for distortion, background hits are removed, track segments reconstructed and fitted to straight lines.
7~ 500
3.2. A n a l y s i s s t e p s ... The analysis starts with the event reconstruction in the electronic detectors. In the case of the T --+ #VV decay search, a single negative muon with P~, < 3 0 G e V / c is required to be identified (this is so-called "1#" event type). The tracks reconstructed in the fibre tracker are extrapolated to the emulsion, automatically located there and followed up to the emulsion plate which presumably contains the neutrino interaction vertex. When the vertex is found, a detailed measurement of the tracks in the vertex region looking for a kink is needed. This is done in a semiautomatic mode (the so called "eye scan") and takes a long time: about one hour per event! To minimise eye scanning, we preselect T decay candidates by using indirect criteria based on automatic measurements. Depending on the event topology, either PT organ impact parameter cut is applied. Only about 13% of all found vertices survive these preselections (of course the price to pay is some loss of efficiency for the signal - about 20%). Now, owing to the latest developments of scanning technique, we are able to further reduce this event sample, taking a video image of the vertex region and analysing it automatically. The capability of the analysis algorithm was confirmed by finding several events with a kink identified as decays of charmed mesons. In future this type of event will be very useful to monitor the kink finding efficiency (currently estimated as of about 55%).
===========================================
2OO
fl
'%,,°
,~ ~0
~.0
100
120
'~
Figure 3. Top: photograph of a neutrino interaction in emulsion (magnification 50, the dimension of the view is 120 x 100#m 2, focal depth is a few microns). Bottom: automatically reconstructed neutrino vertex (scales in micrometers).
An example of a neutrino interaction vertex in the emulsion as seen in a microscope is shown in Eig. 3 (top). The neutrino beam is perpendiculax to the emulsion plate and the focal depth is a few microns. The slow tracks, mainly nuclear fragments, fly away almost perpendicularly to the beam and are the only tracks easily recognised in this thin slice. Fast secondaries have small angles with respect to the beam and their tracks appear as a series of black dots near the vertex. Scanning the emulsion layer-by-layer in depth, the dots in successive layers are associated so that the full
3 . 3 . . . . r e s u l t s ... At present about 80000 events with a muon have been analysed, corresponding to ~ 10% of the expected final statistics. After selection and reconstruction, 14897 neutrino vertices were located and no T-decay candidate was found. The estimated background in this sample is less than
A. Artamonov/Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373 0.1 event. This corresponds to a limit for the mixing parameter of sin2(20~r) <_ 5.4 x 10 -3 (90% C.L.). In the meantime we started the analysis of hadronic r-decay channels. Adding a small subsample of the analysed nmonless events (about 4% of the finally expected statistics) the 90% C.L. limit of 4.5 x 10 -3 was obtained. 3.4 . . . . a n d p r o s p e c t s of C H O R U S By the end of this year we expect to have 800000 neutrino CC events recorded in emulsion. When the analysis of the whole emulsion is completed we expect a sensitivity to a mixing of 2 X 10 - 4 at large Am 2. Comparing it with our present result for the muonic decay mode search it is clear that we have to gain a factor of --~ 25 in sensitivity. A factor of 10 comes from the larger statistics, the rest is expected to come from the inclusion of the hadronic decay channel and from the improvements of the efficiencies (event reconstruction, track following, vertex finding in the emulsion and kink finding). The expected final sensitivity and the current results are shown in the exclusion plot (Fig. 4). Today we have reached the limit established by the Fermilab E531 experiment for large Am 2 [2].
CHORUS July
97
June
96
. s ~,
"-:
J
102
% E
<3
i
. 0 -1 10-4
i iiillll
t
:0 -3
i Iiiiiii
i
10 -2 Siq22~
Figure 4. Exclusion plot.
i i Illlll
lO -1
,
,
ill 1
373
4. S U M M A R Y
By the end of this year CHORUS will complete the data taking phase. Now we have acquired 3/4 (~ 6 x 105uuCC) of the expected statistics. The emulsion exposed in 1994-1995 was developed and is being analysed. Owing to the rapid progress in the emulsion processing technique and the increasing number of automatic scanning systems, we expect to complete the data analysis in 2 years without the need of applying the kinematic preselection foreseen in the proposal. So far ~- 10% of 1# events and ~ 4% of muonless events have been analysed and no r candidates were found. This leads to a limit of sin2(2OuT) < 4.5 x 10 -3 (90% C.L.). The experience gained during the analysis confirms that the nominal sensitivity of sin2(2OuT) < 2 × 10 -4 (at large Am 2) is attainable. REFERENCES
1. CHORUS Proposal: N. Armenise et al. (CHORUS Collaboration), CERN-SPSC/9042, December 1990; M. de Jong et al. (CHORUS Collaboration), CERN-PPE/93131, July 1993, CHORUS Collaboration, ~" CERN/SPSLC/94-23, October 1994. 2. N. Ushida et al., Phys. Rev. Lett. 57, 2897 (1986) 3. B. \ ~ n de Vyver, Nucl. Instr. and Meth. A385, 91 (1997). 4. CHORUS Collaboration, CERN-PPE/97-33. ,~larch 1997, Nucl. Instr. and Meth. (in print). 5. S. Aoki et al., Nucl. Instr. and Meth. B51¢ 466 (1990).
Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373
RESULTS FROM THE CHORUS EXPERIMENT
v t, --+ Vr
PROCEEDINGS SUPPLEMENTS
OSCILLATION
A. Artamonov a for the CHORUS Collaboration alnstitute of Theoretical and Experimental Physics, Moscow, Russia, E-mail: [email protected] The CHORUS experiment accumulated ,,, 6 x 105v~ CC events, which represents ,,~ 3/4 of the statistics expected at the end of data taking. About 10% of the data were analysed and no r candidate was found. This corresponds to a 90% C.L. upper limit for sin~(2Ot,,) of 4.5 x 10 -3 at large Am 2. Extrapolating this pilot analysis to the full data set confirms that a sensitivity of sin2(20~,~) < 2 x 10-4 can be reached.
1. I N T R O D U C T I O N CHORUS [1] is one of two large neutrino osdilation experiments currently taking d a t a at CERN. It is aimed at a search for v# -+ vr oscillation with a 20 times higher sensitivity t h a n the one currently reached by previous experiments. If v~ -~ uT mixing exists at the present limit [2], CHORUS would detect about 60 vr events. The CHORUS experiment is searching for vT appearance in the CERN SPS neutrino beam consisting mainly of muon neutrinos with a negligible contamination of Vr [3]. The concept of the experiment is the direct observation of Vr induced CC events identified by the characteristic decay topology of the r lepton produced in this reaction. In most cases T decays into one charged particle (85%) appearing as a kink on a negative track coming from the vertex. At SPS energies the decay length of the T lepton is ,~ 1 m m and the kink angle is ~ 50 m r a d . Hence, the events we are searching for are confined to a tiny volume of the order of a m m 3. Thus, we need highly precise and fine grained measurements of the vertex topology such as only nuclear emulsion can provide. 2. D E T E C T O R The emulsion stack is the principal part of the detector. It serves two purposes. First, it is a target for v interactions. With a mass of ,,~ 800 kg it contains the largest amount of emulsion ever ex0920-5632/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S0920-5632(98)00066-8
posed at accelerators. Second, it is a very precise 3-dimensional vertex detector, producing -~ 300 grains~ram for a minimum ionising particle; each grain can be localised with a precision of ,~ l # m . In 4 years of exposure, about a million u interactions will be recorded in the emulsion. Even with the considerable progress achieved during the last years in the emulsion analysis technique, it is impossible to scan the entire emulsion target in the search for u interactions in a vast background of cosmic and beam muon tracks. This search must be g ~ d e d by predictions (on an event-by-event basis) of the position and the angle of tracks from u interaction leaving the emulsion. Precise predictions (ax ", 150#m, a8 "~ 2 m r a d ) are provided by scintillating fibre tracker located downstream of the emulsion. The particle tracks reconstructed in the tracker are extrapolated to the so-called interface emulsion layers as shown in Fig. 1. After being found, they are searched within a reduced area at the emulsion target exit and followed upstream, from plate to plate, to their production vertex. All these operations are performed using fully automated microscopes. But even the precise information about the vertex region is not sufficient to suppress backgrounds which can fake the r decay topology: the decays of charm mesons, the so-called "white kinks" (interactions of hadrons without any visible recoil at the vertex), decays of charged pions and kaons. To provide sufficient background suppression (less than one background event in the
A. Artamonoo/Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373
371
particles except of fast muons. Finally, there is a magnetlsed iron toroidal muon spectrometer. A detailed description of the CHORUS detector can be found in [4]. 3. A N A L Y S I S
View of m~roicope
of tau-neutcinol[~r) I~ emulsion .--'" . ~
IntAIrllc~ion ~ . . . _ ~
b
,,
decay '
- _ ~. _ 12.
Figure 1. Emulsion target and scintillating fibre tracker. full data set) the detector has to be able to determine the charge of secondary particles, reconstruct the event kinematics, identify muons pions and measure their momentum. The CHORUS detector (see Fig. 2) was designed to satisfy these requirements. Behind the
I Figure 2. CHORUS detector. veto planes there are emulsion stacks sandwiched with fibre tracker planes. They are followed by an aircore magnet spectrometer measuring the charge and the momentum of particles before they enter a high resolution lead-scintillating fibre calorimeter, measuring the total energy of all
The partial replacement of the emulsion target, at the end of 1994 and of 1995, allowed us to start a systematic data analysis while the experiment was still running. The analysis was focused basically, in a first stage, on the search for the muonic T decay mode. The reconstruction efficiencies for this mode are higher than for hadronic ones; the signature is quite clean: a single p - with a kink; the background is small. In parallel to the search for r candidates, at this stage of analysis a lot of effort was put into tuning and perfecting the event reconstruction algorithms as well as in development of the emulsion scanning technique.
3.1. Emulsion scanning technique Emulsion scanning is the most delicate and time consuming part of the analysis. The total amount of emulsion to be processed is so large (1.7 tons), that a qualitative improvement of the techniques used by preceding experiments was fi~eded. The solution was an automatic scanning procedure pioneered by the Nagoya group [5]. This is a completely new technology making the scanning process ~ 100 times faster than the manual scanning. In 1995, when the analysis was just launched, only a few automatic microscopes were available in Nagoya. During the last two years the Collaboration has progressively enhanced its scanning facilities. Now 10 automatic systems are running and 8 others will be installed later this year. In addition, a new generation of more powerful automatic systems has been made operational in Japan further increasing the scanning speed. These achievements allowed us to change the analysis strategy initially foreseen [1]. The strong kinematical preselections which were proposed to reduce the scanning load by an order of magnitude (of course at the expense of efficiency) are no longer necessary to complete the analysis in about two years.
372
A. Artamonoo /Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373
spatial image of the vertex region can be reconstructed. The example of a typical vertex "video image" is shown in Fig. 3 (bottom). It corresponds to a tiny volume of 0.02 m m 3. Measured "hits" are corrected for distortion, background hits are removed, track segments reconstructed and fitted to straight lines.
7~ 500
3.2. A n a l y s i s s t e p s ... The analysis starts with the event reconstruction in the electronic detectors. In the case of the T --+ #VV decay search, a single negative muon with P~, < 3 0 G e V / c is required to be identified (this is so-called "1#" event type). The tracks reconstructed in the fibre tracker are extrapolated to the emulsion, automatically located there and followed up to the emulsion plate which presumably contains the neutrino interaction vertex. When the vertex is found, a detailed measurement of the tracks in the vertex region looking for a kink is needed. This is done in a semiautomatic mode (the so called "eye scan") and takes a long time: about one hour per event! To minimise eye scanning, we preselect T decay candidates by using indirect criteria based on automatic measurements. Depending on the event topology, either PT organ impact parameter cut is applied. Only about 13% of all found vertices survive these preselections (of course the price to pay is some loss of efficiency for the signal - about 20%). Now, owing to the latest developments of scanning technique, we are able to further reduce this event sample, taking a video image of the vertex region and analysing it automatically. The capability of the analysis algorithm was confirmed by finding several events with a kink identified as decays of charmed mesons. In future this type of event will be very useful to monitor the kink finding efficiency (currently estimated as of about 55%).
===========================================
2OO
fl
'%,,°
,~ ~0
~.0
100
120
'~
Figure 3. Top: photograph of a neutrino interaction in emulsion (magnification 50, the dimension of the view is 120 x 100#m 2, focal depth is a few microns). Bottom: automatically reconstructed neutrino vertex (scales in micrometers).
An example of a neutrino interaction vertex in the emulsion as seen in a microscope is shown in Eig. 3 (top). The neutrino beam is perpendiculax to the emulsion plate and the focal depth is a few microns. The slow tracks, mainly nuclear fragments, fly away almost perpendicularly to the beam and are the only tracks easily recognised in this thin slice. Fast secondaries have small angles with respect to the beam and their tracks appear as a series of black dots near the vertex. Scanning the emulsion layer-by-layer in depth, the dots in successive layers are associated so that the full
3 . 3 . . . . r e s u l t s ... At present about 80000 events with a muon have been analysed, corresponding to ~ 10% of the expected final statistics. After selection and reconstruction, 14897 neutrino vertices were located and no T-decay candidate was found. The estimated background in this sample is less than
A. Artamonov/Nuclear Physics B (Proc. Suppl.) 66 (1998) 370-373 0.1 event. This corresponds to a limit for the mixing parameter of sin2(20~r) <_ 5.4 x 10 -3 (90% C.L.). In the meantime we started the analysis of hadronic r-decay channels. Adding a small subsample of the analysed nmonless events (about 4% of the finally expected statistics) the 90% C.L. limit of 4.5 x 10 -3 was obtained. 3.4 . . . . a n d p r o s p e c t s of C H O R U S By the end of this year we expect to have 800000 neutrino CC events recorded in emulsion. When the analysis of the whole emulsion is completed we expect a sensitivity to a mixing of 2 X 10 - 4 at large Am 2. Comparing it with our present result for the muonic decay mode search it is clear that we have to gain a factor of --~ 25 in sensitivity. A factor of 10 comes from the larger statistics, the rest is expected to come from the inclusion of the hadronic decay channel and from the improvements of the efficiencies (event reconstruction, track following, vertex finding in the emulsion and kink finding). The expected final sensitivity and the current results are shown in the exclusion plot (Fig. 4). Today we have reached the limit established by the Fermilab E531 experiment for large Am 2 [2].
CHORUS July
97
June
96
. s ~,
"-:
J
102
% E
<3
i
. 0 -1 10-4
i iiillll
t
:0 -3
i Iiiiiii
i
10 -2 Siq22~
Figure 4. Exclusion plot.
i i Illlll
lO -1
,
,
ill 1
373
4. S U M M A R Y
By the end of this year CHORUS will complete the data taking phase. Now we have acquired 3/4 (~ 6 x 105uuCC) of the expected statistics. The emulsion exposed in 1994-1995 was developed and is being analysed. Owing to the rapid progress in the emulsion processing technique and the increasing number of automatic scanning systems, we expect to complete the data analysis in 2 years without the need of applying the kinematic preselection foreseen in the proposal. So far ~- 10% of 1# events and ~ 4% of muonless events have been analysed and no r candidates were found. This leads to a limit of sin2(2OuT) < 4.5 x 10 -3 (90% C.L.). The experience gained during the analysis confirms that the nominal sensitivity of sin2(2OuT) < 2 × 10 -4 (at large Am 2) is attainable. REFERENCES
1. CHORUS Proposal: N. Armenise et al. (CHORUS Collaboration), CERN-SPSC/9042, December 1990; M. de Jong et al. (CHORUS Collaboration), CERN-PPE/93131, July 1993, CHORUS Collaboration, ~" CERN/SPSLC/94-23, October 1994. 2. N. Ushida et al., Phys. Rev. Lett. 57, 2897 (1986) 3. B. \ ~ n de Vyver, Nucl. Instr. and Meth. A385, 91 (1997). 4. CHORUS Collaboration, CERN-PPE/97-33. ,~larch 1997, Nucl. Instr. and Meth. (in print). 5. S. Aoki et al., Nucl. Instr. and Meth. B51¢ 466 (1990).