ANTARES Status February 2007

ANTARES Status February 2007 Introduction Since 29 January 2007, the ANTARES detector consists of 5 operating lines. These lines plus the instrumentation line, MILOM, are operating very successfully with a small number of dead OM channels and data is currently being collected and analysed. Data with two lines have been taken since September 2006 and with one line since March 2006. Many performance results are available with the one and two line configurations and a major effort is underway to obtain results with the five line detector. During the last submarine operation much extra information was obtained on the state of the junction box outputs, however it was not possible to test all outputs and choices must be made without a complete knowledge of the status. Production of the remaining components to complete the twelve line detector is well advanced and will be detailed later in the document. Lines 6 and 7 are being assembled and will be ready for deployment in March. The full 12 line are planned to be deployed before the end of 2007 with the last connection planned for January 2008. In September 2006, the ANTARES collaboration grew with the addition of a group from the Univ. Politécnica de Valencia who will work closely with the existing Spanish group of IFIC in Valencia and a group from the Institute of Space Sciences in Bucharest, Romania. Detector in Operation Figure 1 illustrates the location of the detector lines and cables on the ANTARES site. Figure 1. Situation of the ANTARES site on January 29 2007. The installed lines are indicated by red boxes and labels and the future lines in purple. At the bottom the location of the junction box is indicated with the Main Electro-Optical Cable to the shore. The routes of the interlink cables from the junction box to the lines, as measured after deployment, are shown by the multicoloured lines. The junction box was deployed in December 2002, the MILOM instrumentation line was deployed and connected in spring 2005 and lines 1-5 were deployed and connected between March 2006 and January 2007. Line 1 has been in the sea for 11 months, line 2 for 5

months and the remaining lines for 1-2 months. The MILOM has been in the sea for 22 months. Out of the 75 optical modules on each of the 5 detector line there are a few dead channels: Line 1 has four dead channels, Line 2 has four, Line 3 has one, Line 4 has zero and Line 5 has six, making a total of 4%. In detail there are different symptoms for the observed dead channels and in some cases for the recently connected lines they might be recoverable. Only three channels in Lines 1+2 became dead during operation the rest were dead from the first turn-on. All of the 130 electronics containers on lines bases and the storeys function. In line 1 there was a loss of optical transmission after 3 months of operation. This loss was compensated for using an optical amplifier and the situation has remained stable for the last six months. All optical fibres in the lines are monitored on a regular basis and although there are some fluctuations in the measurements there is no evidence for any other serious optical transmission losses in the 5 lines. It should be noted that line 1 has only a provisional correction for the problem of optical fibre movement observed in the line 0 test. Unlike line 2 and the subsequent lines, the EMC cables penetrators do not have the efficient blocking solution and it is likely that some optical fibres move as a function of time. The estimate of line 1 lifetime was 1 year when the decision was taken to deploy it with that technology. The fact that with the amplifier it is still completely operational is clearly very positive, however it could still be that the fibres will move in the future and cause additional optical transmission losses. The MILOM has an original problem with a leak in an optical beacon and has one optical fibre whose transmission losses are corrected with an amplifier. The situation of the MILOM has been stable for the last 6 months. The calibration devices in the detector lines work according to specifications. All optical beacons operate well and are being used to calibrate and understand the detector. In the lines 1 and 2, the acoustic positioning system suffers from some electrical noise pickup. This noise has been understood and removed in the later lines by introducing a decoupling capacitor while for the first two lines all of the problematic hydrophones function but with some reduction in sensitivity due to the degraded signal to noise ratio. On all the five lines the full acoustic positioning system satisfies the specification accuracy. Figure 2. Integrated number of days of effective data taking over 11 months. The number of days takes into account all losses: periods with no operation and all data taking inefficiencies. Figure 2 gives and indication of the data taking efficiency since the connection of Line 1. The plot shows the integrated number of days during which data was taken taking into account all losses. These losses include periods when the detector was off; when no data were taken; losses due to software problems and data taking losses due to high bioluminescence

rates. It can be seen that over the 11 months of operation the data taking efficiency has risen continuously following numerous improvements in the data acquisition system. During December 2006 the efficiency averages 85% and the last month of data taking, when the 5 lines were in operation, the efficiency averages 63%. Junction Box Situation The junction box has 16 outputs of which 12 are required for the detector lines, 1 for the instrumentation line and 1 for the IFREMER interdisciplinary platform, leaving 2 spares. During the early submarine connection operations there were a number of problems in connections of the interlink cables to the junction box. So far all connections of the interlink cables to the connectors on the bases of the lines have had a 100% success rate. To understand the problems with the junction box connections there have been extensive discussions and meetings with the manufacturer of the undersea connectors, Ocean Design Industries and during some of the submarine operation in January an Ocean Design representative was present on the boat. At the present time 6 outputs are connected to lines and are in perfect operation. Previously 2 other connectors have been tested to be good and 4 connectors have never been tested. The other 4 have had problems in some way. One had mechanical damage which is repaired and remains to be retested. Two have been connected without a successful mate and subsequently have an indication of an electrical short circuit which opened the safety circuit breaker on their output lines. One connector was connected without a successful mate but remains electrically OK. After inspections and discussions it is judged that all these three later connectors are bad, possibly damaged by the same cause. It is believed that the reasons for the connector difficulties could be due to inadequate cleaning of accumulated silt from the connector before connection. The most recent connection operation employed an improved cleaning technique and had a 3/3 success rate. At present it is only known that a total of 8 outputs are good and it will not be possible to know anymore until the next submarine operation which is planned for September. Although the final decision has not been taken, the collaboration is developing a method to multiplex two lines to one junction box output. A rough estimate of the extra cost of this multiplexing is ~50Keuro per pair of lines multiplexed and at present the plan is to do this for two pairs of lines. The extra money will come from the common fund budget. The impact of the necessary modifications on the schedule is under study and could cause, depending on delivery dates, a delay of a couple of months in the completion of the detector beyond January 2008. Line Production Figure 3 gives the production status of the components for the detector lines. Lines 6 and 7 are presently being assembled in parallel at two sites. For many components the items are already available for all lines, only in a few cases does delivery have to be carefully coordinated with the electronics module integration and line assembly. For some time the EMC cables have been the critical path for production but since a couple of month the manufacturer has been able to deliver cables at a rate which does not limit the line production. A new instrumentation line will be assembled, deployed and connected in 2007 using parts from the existing instrumentation line (MILOM) which will be recovered from the site sometime in the spring of 2007. The new instrumentation line will contain the existing instruments of the MILOM in some cases with duplications for redundancy. The new line will contain two optical modules and two cameras to observe bioluminescence. A test system for acoustic neutrino detection is under construction. In the full detector two lines will contain hydrophones for this system: the new instrumentation line and one of

the later optical detector lines. The production of the elements of this system is in progress and on time for the line deployments. Figure 3. Status of component production for lines. Lines 1-5 are in operation in the sea. For each of the other detector lines (L6-L12) and for the instrumentation line (IL) the components are indicated in green if all items are delivered, in orange if the delivery is partial, in yellow if ordered and in white if not yet ordered. Current Planning The assembly of line 6 and 7 is well advanced at the present time, with 2/5 sectors already assembled and the remainder expected to be finished in early March. Below is the current schedule for the deployment and connection of these and the remaining lines of the detector. Mar 2007 Lines 6+7 deployment Jul 2007 Lines 8+9 deployment Sep 2007 New instrumentation line deployment Oct 2007 Lines 6, 7, 8, 9, IL connection Oct-Dec 2007 Lines 10, 11, 12 deployment Dec 2007 or Jan 2008 Line 10, 11, 12 connection In the past the ERC has requested that the present planning is tracked compared to earlier plannings and the requested comparison is given in as follows. At the ERC/FRC meeting in July 2006 the main production milestones presented were: 4 or 5 lines operational Jan 2007 12 lines operational Dec 2007 The dates for the same operations presented to the review in February 2005 were 4 lines operational Apr 2006 12 lines operational Mar 2007 The present planning is 5 lines operational Jan 2007: ACHIEVED 12 lines operational Jan 2008

Since the beginning of the deployments of the real detector lines in 2006 the project has been able to keep to the production milestones within one or two months. Data The bioluminescence rate on the ANTARES site is very variable. During the first weeks of the 5 line detector operation the counting rate has been at its minimum. Figure 4 shows a typical singles rate counting sequence for Line 5 and figure 5 shows the counting rate averaged over 1 minute for all the optical modules in all the 5 lines. Figure 4. Online counting rate module for floor 23 of Line 5 on Feb 9. The difference in rates between the three optical modules is due to the as yet missing threshold calibration. Figure 5. Distribution of counting rate on 22 Feb in each of the five lines, giving the counting rate per optical module in khz for each of the 75 OMs in each line. The lines 3-5 are lacking calibration and so have variations in rates due to different thresholds. The holes usually correspond to dead channels but in this data, in line 5, two storeys were absent from the readout and also appear as holes.

A typical down going cosmic ray muon event with the 5 line detector is shown in figure 6. The existing data is dominated by downward going cosmic ray muons and these events frequently contain more than one track. In figure 7 the hits in a high multiplicity event are shown with the interpretation that the event is a bundle of parallel downward going muons. Figure 6. Sample of typical data event of downward going cosmic ray muon seen in all 5 line of detector. The colours indicate the time of arrival of the light. All hits in the event are shown. Figure 7. Example of an event with many parallel downward going muons seen in the five lines. The vertical axis is the vertical position of the hit OM and the x axis the time relative to the first trigger hit. The cartoon in the bottom right indicates hypothesised topology of the event.

Two candidate events for upward going muons originating from neutrino interactions are shown in figure 8. Run 25922, frame 53569, θ=30 Run 25929, frame 61770, θ=36 Figure 8. Two candidate events for upward going neutrinos showing only the hits used in the fit.

APPENDIX Replies to specific points raised at ERC/FRB meeting July 2006 At the last ERC/FRB meeting a series of questions were posed and below we reply to these points. 1) Repair policy for ANTARES detector Here we set out the repair policy which has been planned since the ANTARES Technical Design Report. Now that the project has 5 lines operational in the sea it will be possible to re-evaluate this policy in one year s time when significant statistics are available. Until this re-evaluation can be made the collaboration wishes to stay with the following policy in order to plan space and manpower requirements. The repairs will have a cost as given below, but as for all expenses the repair costs will be presented each year for approval and the numbers below are only intended as an indication. Design philosophy The design of the ANTARES detector lines allows for remote release from the seabed after installation to allow repairs. The lines are segmented into five sectors with independent readout in a way to avoid a single point failure which could cause the loss of a whole line. There are, however, a number of failure modes which could cause the failure of a whole sector. Since the Technical Design Report the repair policy has been to only recover a line if two or more sectors, corresponding to 40% of the line are non-functional. The detector design also allows for recovery and repair of the junction box. Spare Parts In purchasing the electronics components in general provision has been made for 12% spare parts. Some spare components have been consumed due to failures which have occurred in production and it is likely that when production is complete typically between 5 and 10% spares will exist for the electronics. Often more components have been purchased when it has been known that the components would become obsolete and unavailable in the future. The quantity of spares needed to repair a line depends on the failure mode and here some examples are given. An upper extreme example are water leaks which would destroy all the electronics in a sector and so require 3% of total electronics as spares to repair a line with two flooded sectors. The lower extreme example is broken optical fibres which would require 0% spare electronics modules. A possibly more typical example could be electronics elements in the MLCM which might require just one replacement per sector repaired. Without experience it is impossible to have any accurate estimates of the spares needed over a long period. The spares which will be available when the 12 line are constructed could be enough to repair 2-4 lines. Time scale for repairs The estimate for the minimal time to repair two lines in a one year period is four months. The repair would be planned taking into account the future availability of the submarine ROV with the two lines being recovered in the same period four month in advance of a possible connection. The line integration facility at CPPM must remain operational as must one electronics integration site in Italy. With these known facilities and the same trained team of technicians available the repair would take 6 weeks per line. Taking into account the time to prepare the line for redeployment, the total time out of the sea could be the four months. Further ~1 month would be required for the recalibration of a repaired line.

Cost of repairs The cost of all repairs are dominated by a submarine operation which costs ~150Keuro. The total cost of repairing one line in a year would be ~300Keuro and for two line in a year ~350Keuro. A junction box repair would cost ~500Keuro. These costs should be referenced to the full detctor cost of 20Meuro with the consideration that a certain recovery of detector efficiency can be justified as the same fraction of the full detector cost Conclusion on repair policy The collaboration at present rests with the intention to repair a line if 40% of channels are un-operable even though with experience this may change. For this it is necessary to maintain the facilities and manpower for one electronics integration site in Italy and one line integration site in France. Clearly repairs in the early years of detector operation give more significant gains than at the end of the detector lifetime. It should be remembered that Line 1 was deployed with the prediction that it s lifetime would only be ~1 year and so even though it is in fact still completely functional this made not remain the case. No repairs will be made before the completion of the detector in early 2008. It is proposed to maintain the above repair policy for the years 2008 and 2009 and then re-evaluate the situation. 2)Effect of bioluminescence on physics sensitivity The ANTARES detector hardware and software were designed with a lower rate of bioluminescence than has been observed during some periods of the first year of detector operation. With the design software and strategy losses occur in both the data acquisition and the event reconstruction A first estimate of the effect of bioluminescence on the physics sensitivity due to the losses in the present reconstruction software has been made. For this, a full simulation of the detector response to neutrinos has been made assuming different singles rates and burst fractions. The burst fraction is defined as the dead-time caused by a bioluminescence burst. The result - expressed as the effective volume is shown in Figure. Figure A1 Effective volume as a function of the neutrino energy for different assumed singles rates and burst fractions. From left to right: 60 khz, 120 khz and 240 khz singles rates. Colour coding: black 0%; red 20%; and purple 40% burst fraction. As can be seen, the effective volume decreases with increasing singles rates and burst fractions. For a neutrino energy of 10 TeV, the effective volume decreases by a factor of about 3 (4) at 240 khz singles rate and a burst fraction of 20 (40)%, with respect to 60 khz singles rate and no bioluminescence bursts. For a neutrino energy of 100 TeV, the reduction of the effective volume is about 30 (50)%. Above 5 PeV, the reduction is limited to about 5 (10)%. The actual singles rates and burst fractions have been measured in situ. For this analysis, 1 year of data with the "Mini Instrumentation Line with Optical Modules" and recent data of the first complete detector line have been used. The cumulative probability distributions of the singles rates and burst fractions are shown in Figure A for three different periods in 2006.

Figure A2 Cumulative probability distribution of the singles rates (left) and burst fractions (right). Colour coding: purple March-May 2006; blue June-August 2006; and red September-November 2006. The measurements shown in figure A2 were made with the first complete detector line. It should be noted that there is a strong correlation between the singles rate and the burst fraction. The effect of bioluminescence on the overall efficiency has been evaluated using a parameterisation of the effective volume as a function of the singles rate and the quoted measurements. Averaged over 1 year, the effect can be summarised as a reduction of the overall efficiency by less than 50%, assuming a neutrino flux Φ E 2. In order to cope with the bioluminescence, the strategy for the data acquisition (DAQ) has been to maximise the data transmission bandwidth and to improve the signal/noise ratio and the speed of the software trigger (see below). The bandwidth for each detector storey has been improved from about 10 Mb/s to about 50 Mb/s using the zero-copy feature of the VxWorks operating system. The current bandwidth corresponds to more than 300 khz per photo-multiplier tube. The present software trigger can operate up to 300 khz without loss of efficiency. Possible strategies to take data beyond the present limit of 300 khz include increasing the discriminator thresholds, lowering the high voltage of the photo-multiplier tubes and applying a local coincidence logic. The various data taking strategies will inevitably yield different reconstruction efficiencies. The optimal data taking and reconstruction strategy during periods with singles rates in excess of 300 khz has not been worked out yet. Effort are in progress to modify software at all levels to optimise the overall detector efficiency. The previously mentioned efficiency calculations do not take into account the established improvements in the DAQ system and the possible improvements in the reconstruction algorithms. Already for the DAQ significant improvement have been made as already shown in figure 2. Over the whole of 2006, the average data taking efficiency was 30%, corresponding to about 120 days of observation time but by the end of the year the improvements in the DAQ system resulted in a data taking efficiency of about 85%. A similar effort on the reconstruction software could yield significant gains. 3) Physics perspectives for point source detection The physics perspective of the Antares neutrino telescope has been evaluated and compared to existing and future limits. The results are summarised in figure A3. As can be seen from this figure, the ANTARES neutrino telescope will be able to probe the Southern Hemisphere for sources of high energy extra-terrestrial neutrinos with an expected sensitivity that is 10 times better than that of the Macro (and Super-K) experiment.

Figure A3 Flux limits (90% C.L.) as a function of the declination. The Antares sensitivity is shown for 1 year of data taking for two different methods: binned (thin green line) and unbinned (thick green line). As can be seen in Figure A4, the Galactic Centre and large parts of the Galactic plane are visible to ANTARES 70% of the time. This region is a major source of high energy gamma radiation (0.1-10 TeV) as demonstrated by the H.E.S.S. observations. Neglecting possible attenuation of high energy photons, calculations of neutrino fluxes based on these measurements show that the detection of these sources is probably beyond the sensitivity of the ANTARES detector, though dedicated analyses leave room for improvements in the effective area or optimisation of the signal to background ratio. On the other hand, since strong gamma-ray absorption or cascading in some of the discovered sources are not excluded or even are likely to occur, the Galactic plane represents a highly interesting, exclusive target for observations with the ANTARES detector. Figure A4 Sky map of TeV -ray sources observed by the H.E.S.S. experiment and their visibility to the Antares neutrino telescope (grey coding: white 0%; light grey 25-75%; dark grey 75-100%) and the Amanda/IceCube neutrino telescope (solid line, labeled South Pole Visible ). Finally, one shouldn't discard the possibility of the existence of high energy neutrino sources that have eluded observations with other messengers up to now. A likely location of such sources would be the Galactic plane.

4) Computing farm The main purpose of the data-acquisition (DAQ) system is to convert the analogue signals of the photo-multiplier tubes (PMT) to digital data; to transport the data to shore; to filter the data; and to format the data for the physics analyses. The Antares DAQ system is based on the All-data-to-shore concept. In this, all PMT signals are digitised off shore and all data are sent to shore where they are processed in real time by a farm of commodity PCs ( the computing farm ). The main challenge is the filtering of the rare physics signals from the background of 40 K decays and bioluminescence. It has been shown that one can readout the complete Antares detector with this system up to a singles rate of 300 khz. The trigger rate can be limited to about 10 Hz by requiring a minimum of 5 time-position correlated L1- hits (L1-hit is a local coincidence or a large pulse). The purity (fraction of events that are due to a genuine muon) is larger than 50%. A position accuracy of 10 cm and a timing accuracy of 1 ns are required to achieve the projected angular resolution of 0.2 degrees. The positions of the PMTs are determined from data taken with compasses, tilt-meters and acoustic transponders. The taking of these data does not cause any dead-time and does not require additional computing power. The timing calibration is done using LED (or laser) beacons, so-called internal LEDs and 40 K coincidences. The taking of these data causes at present about 6 hours of dead-time per week. A session was dedicated to online computing at the September 2006 collaboration meeting. During this session, significant improvements in the signal/noise ratio of the software trigger have been reported (about 10-100) as well as improvements in terms of speed (a factor of about 100 at 300 khz singles rate). These improvements are mainly due to a combination of direction sensitive pattern recognition and the "clique" algorithm. It has also been shown that with the Antares DAQ system one makes optimal use of the multiple-core processor technology because the design of the data filter is parallelised completely. A designated Gamma-Ray Burst (GRB) trigger that listens to the Swift (and HETE and Integral) satellite is operational. With this trigger, all raw data are stored on disk during a GRB event. These data are then processed remotely. The GRB trigger and the processing of the raw data do not require additional computing power in the shore station. Since the start of operation, about 90% of the satellite triggers have been recorded this way (missing only those that occurred during periods of deployment operations etc.). Software triggers that can track a point source including the Sun and the Galactic Centre are possible. It has been shown that with these triggers, the sensitivity to a point source can be increased significantly. An increase of the bandwidth from the shore station to the global network from 10 Mb/s to 100 Mb/s is anticipated for the completion of the detector. In summary: The standard muon trigger requires 32 dual-cpu, dual-core PCs for singles rates up to 300 khz; 16 of such PCS are operational and the remaining 16 PCs have been purchased already; In total 8 Ethernet Switch modules will be installed of which 4 are already operational; this infrastructure enables to connect up to 130 PCs; Each dedicated trigger (e.g. Monopole, point-source) requires about 25 additional dual- CPU, dual-core PCs at the nominal singles rate; the total price tag is about 65 k per dedicated trigger; The maximum number of dedicated triggers that is presently foreseen amounts to 4, corresponding to a total additional investment of about 350 k, including the cost for the infrastructure.