Detection of Atmospheric Muon Neutrinos with the IceCube 9-String Detector

Detection of Atmospheric Muon Neutrinos with the IceCube 9-String Detector


The IceCube neutrino detector is a cubic kilometer TeV to PeV neutrino detector under construction at the geographic South Pole. The dominant population of neutrinos detected in IceCube is due to meson decay in cosmic-ray air showers. These atmospheric neutrinos are relatively well-understood and serve as a calibration and verification tool for the new detector. In 2006, the detector was approximately 10% completed, and we report on data acquired from the detector in this configuration. We observe an atmospheric neutrino signal consistent with expectations, demonstrating that the IceCube detector is capable of identifying neutrino events. In the first 137.4 days of livetime, 234 neutrino candidates were selected with an expectation of events from atmospheric neutrinos.

atmospheric neutrinos, IceCube, AMANDA, neutrino astronomy,
95.55.Vj, 95.85.Ry, 96.50.sf

IceCube Collaboration

I The IceCube Detector

The IceCube neutrino detector is being deployed in the deep ice below the geographic South Pole Ahrens et al. (2004a). When completed, the detector will consist of two components. The InIce detector is a cubic kilometer of instrumented ice between 1.5 and 2.5 kilometers below the surface. A cubic-kilometer has been long noted as the required scale to detect astrophysical neutrino sources above the atmospheric neutrino background (see, e.g. Gaisser et al. (1995)Waxman and Bahcall (1997)Stecker et al. (1991)). The IceTop detector is a square-kilometer air-shower array at the surface. This analysis concerns data from the InIce detector exclusively.

The InIce detector consists of an array of light-sensitive Digital Optical Modules (DOMs) Achterberg et al. (2006), deployed 17 meters apart in strings of 60. Strings are arranged on a hexagonal grid with a spacing of 125 meters. The DOMs house a 10-inch Photomultiplier Tube (PMT) and electronics to acquire, digitize and time stamp pulse waveforms from the PMT. With a waveform fit for fine timing, the timing resolution for individual photon arrivals is expected to be less than 2 nanoseconds Achterberg et al. (2006). In 2006, the DOMs were operated in Local Coincidence (LC) with their neighbors, meaning that a triggered DOM’s waveform was only transmitted to the surface if an adjacent DOM on the string also triggered within ns. The surface data acquisition system forms triggered detector-wide events if 8 or more DOMs are read out in 5 .

The detector is being deployed in stages during austral summers from 2004 to 2011. After the 2005-2006 season, the InIce detector consisted of 9 strings, termed IC-9.

IceCube is optimized for the detection of muon neutrinos in the TeV to PeV energy range. It is sensitive to these muon neutrinos (and muon anti-neutrinos) by detecting Cherenkov light from the secondary muon produced when the neutrino interacts in or near the instrumented volume. Neutrino-induced muons are separated from air-shower-induced muons by looking only for muons moving upward through the detector. Up-going muon events must be the product of a neutrino interaction near the detector, since neutrinos are the only known particles that can traverse the Earth without interacting.

Ii Atmospheric Neutrinos

Neutrinos produced in cosmic-ray air showers at the Earth are known as atmospheric neutrinos and form the chief background to potential astrophysical neutrino observation. The atmospheric neutrino spectrum is relatively well-understood Gaisser and Honda (2002)Barr et al. (2006) and has been measured up to GeV by AMANDA Münich et al. (2005). Atmospheric neutrinos from the decay of charmed mesons can contribute significantly above GeV, depending on the model (see e.g. Costa et al. (2002)Bugaev et al. (1998)Martin et al. (2003)). This prompt component is not well known due to uncertainties in the charmed meson production, but with the present exposure of IC-9, this prompt component is negligible and it is presently neglected.

Iii Results

Data acquired from the IC-9 detector in 2006 between June and November has been searched for up-going neutrino candidates. The search proceeds by a series of cut levels intended to remove down-going events as shown in Table 1. Initially, hit cleaning is applied which removes all DOM hits which fall out of a 4 time window, and all DOM hits without another DOM hit within a radius of 100 meters and within a time of 500 ns. After hit cleaning, we re-trigger, insisting that at least 8 DOM hits survive hit cleaning. Simple first-guess reconstruction algorithms running at the South Pole were used to filter out clearly down-going events. Events with fewer than 11 DOMs hit were also filtered to meet bandwidth constraints from the South Pole. The remaining events were transmitted to the data center in the Northern hemisphere via satellite and constitute the filter level of the analysis. At the data center, we reconstructed the direction of events using a maximum-likelihood technique similar to the AMANDA muon reconstruction Ahrens et al. (2004b). Events which were reconstructed as down-going were discarded. Despite the fact that remaining events appear up-going, the data is still dominated by misreconstructed down-going events. These down-going events are removed by additional quality cuts. Events which pass these quality cuts constitute the neutrino candidate dataset.

Simulated events fall into the three categories shown in Table 1. “Single shower” events arise from single cosmic-ray air showers in the atmosphere above IceCube and result in a single muon or bundle of collinear muons in IC-9. “Double shower” events come from two uncorrelated air showers which happen to occur within the 5 event window. The CORSIKAHeck et al. (2001) air shower simulation program was used for the simulation of down-going single and double air-shower events. Finally, “atmospheric neutrino” events are muon neutrino events from pion and kaon decay in air showers in the northern hemisphere. The atmospheric neutrino model of Barr et al. (2004) and its extension up to TeV energies Barr () as well as the cross-section parametrization of Lai et al. (2000) were used to model the up-going muon rates due to atmospheric neutrinos.

Criterion Satisfied Data Single Shower Double Shower Atmospheric Neutrinos
Trigger Level 124.5 124.5 1.5
Filter Level 6.56 4.96 0.45
Up-going () 0.80 0.49 0.21
Up-going () - -
Up-going ( and ) - -
Table 1: Event Passing Rates (Hz). Shown are the event passing rates through different processing levels for the simulated event categories and for experimental data. The trigger level comprises the events triggering the detector after hit cleaning and re-triggering. The filter level comprises events which passed the online filtering conditions. Rates are also shown for events which reconstruct as up-going with and without the final quality cuts applied (see the text for cut definition). Note that the rates from air-shower events have been multiplied by so that the simulation and data agree at trigger level. This is consistent with an approximately 20% uncertainty in the absolute cosmic-ray flux. For the final sample, statistical errors are given for the data and systematic errors are given for the atmospheric neutrino simulation.

The events which are reconstructed as up-going are completely dominated by down-going muons from single and double-shower cosmic-ray events. Misreconstructed events are typically of low quality as measured by two parameters, the number of direct hits , and the direct length . A direct hit is a photon arrival in a DOM which is detected between -15 and +75 ns of the time expected from the reconstructed muon with no scattering. is the total number of direct hits in an event. The direct length represents the length of the reconstructed muon track along which direct hits are observed. An event with a large number of direct hits and a large direct length is a better quality event because the long lever arm of many unscattered photon arrivals increases confidence in the event reconstruction. The strength of the quality cuts can be represented by a dimensionless number which corresponds to cuts of and meters. In addition to these quality cuts, we impose a cut requiring that events have no more than 46 DOMs hit, which eliminates only about 1% of the final event sample. The purpose of this cut is to leave the high-multiplicity data blinded for an anticipated search for a high-energy diffuse neutrino flux. Figure 1 shows how many events remain as we turn the cut strength up and increase the signal-to-noise ratio. The accurate simulation of mis-reconstructed down-going events requires excellent modeling of both the depth-dependent ice properties and DOM sensitivity. In this initial study we observe a 60%-80% rate discrepancy for misreconstructed events up to a cut level of about or so. Nevertheless, over more than four orders of magnitude, the background simulation tracks the data, the number of wrongly reconstructed tracks is reduced, and for , the data behaves as expected for atmospheric neutrinos. From simulation, we expect neutrinos with energies between about and GeV, peaked at 1000 GeV, to survive the analysis cuts.

Figure 1: Data vs Cut Strength. Shown is the remaining number of events as the cut strength (defined in the text) is varied. Curves are shown for the data and the total simulation prediction. Also shown is the prediction due to atmospheric neutrinos alone. The selection from the text corresponds to a cuts strength of , and is denoted by an arrow. At this point, the data are dominated by atmospheric neutrinos.

In 137.4 days of livetime we expect atmospheric neutrino events to survive at and 234 events are measured. Above a zenith of 120 degrees, where the background contamination is small, we measure 142 events with an expectation of due to atmospheric neutrinos. The principal systematic uncertainty in this atmospheric neutrino expectation is due to the approximately theoretical uncertainty in the atmospheric flux normalization Barr et al. (2006). The other significant systematic error is due to uncertainties introduced by the modeling of light propagation and the detection efficiency of IceCube DOMs. The uncertainty in the atmospheric neutrino rate due to this modeling is estimated at 20% and is obtained in this initial study by examining changes in the neutrino passing rate when varying the cuts to account for the background simulation disagreement in Fig. 1.

Figure 2 shows the measured zenith distribution for the final event sample along with the atmospheric neutrino prediction. The zenith angle distribution agrees well with atmospheric neutrino simulation for vertical events above about 120 degrees. The observed excess is believed to be residual contamination from down-going single and double cosmic-ray muons. This excess disappears if we tighten the cuts beyond , suggesting that the recorded events at the horizon are of typically lower quality than expected from atmospheric neutrino simulation. This reinforces the belief we are seeing residual background at the horizon. Above about , with low statistics, the data at the horizon are consistent with a pure atmospheric neutrino signal. Figure 3 shows the azimuth distribution with the IC-9 geometry in the inset. The azimuth distribution has two strong peaks corresponding to the long horizontal axis of the IC-9 detector. The cut of 250 meters on event length constrains near-horizontal events that can be accepted along the short axis of IC-9 since the string spacing is 125 meters. We expect more uniform azimuthal acceptance in future seasons as the detector grows and becomes more symmetric.

Figure 2: Distribution of the reconstructed zenith angle of the final event sample. A zenith of 90 degrees indicates a horizontal event, and a zenith of 180 degrees is a directly up-going event. The band shown for the atmospheric neutrino simulation includes the systematic errors from the text, and the error bars on the experimental data are statistical. Note that uncertainty due to the atmospheric neutrino flux is an uncertainty in normalization and is nearly independent of zenith angle.
Figure 3: Distribution of the reconstructed azimuth angle the final event sample. The band shown for the atmospheric neutrino simulation includes the systematic errors from the text, and the error bars on the experimental data are statistical. The inset shows the horizontal locations of the strings making up IC-9 relative to the center of the future array. Note that uncertainty due to the atmospheric neutrino flux is an uncertainty in normalization and is nearly independent of zenith angle.

We can characterize the response of the detector to neutrinos with an effective area which is a function of neutrino energy and neutrino zenith angle . The function is defined as the function which satisfies


where is an arbitrary diffuse neutrino flux and is the corresponding rate of events surviving analysis cuts. Figure 4 shows the effective area of IC-9 to neutrinos with the event selection of , both for neutrinos near the horizon and for nearly vertical neutrinos. The effective area to neutrinos is much smaller than the geometrical area of the detector, due to the smallness of the neutrino cross-section. Above GeV, the Earth starts to become opaque to neutrinos, and the highest energy up-going neutrinos can only be detected at the horizon.

Figure 4: The effective area of IC-9 to a diffuse source of muon neutrinos. The event selection from the text was used, except that there was no multiplicity requirement imposed on the simulated events. The effective area to muon neutrinos and muon antineutrinos have been averaged to produce the figure. The effective area is shown as a function of neutrino energy, both for vertical and horizontal neutrino events. Horizontal events have a zenith of 90 degrees and vertical up-going events have a zenith of 180 degrees.

Iv Conclusions

In 2006, IceCube was approximately 10% deployed and acquiring physics-quality data. Atmospheric neutrinos serve as an irreducible background to astrophysical neutrino observation, as a guaranteed source of neutrinos for calibration and verification of the detector, and may be studied as a probe of hadronic interactions at energies inaccessible to terrestrial labs. In the first 137.4 days of livetime we have identified 234 neutrino candidates with the IC-9 detector. For events above 120 degrees, this neutrino sample is consistent with with expectations for a pure atmospheric neutrino sample. Selection of events was done within six months of the beginning of data acquisition, demonstrating the viability of the full data acquisition chain, from PMT waveform capture at the DOM with nanosecond timing, to event selection at the South Pole and transmission of that selected data via satellite to the North. During the 2006-2007 season, 13 more strings were deployed, bringing the total number of strings for the InIce detector to 22. The deployment of IceCube will continue during austral summers until 2010-2011, while the integrated exposure of IceCube will reach a sometime in 2009.

We acknowledge the support from the following agencies: National Science Foundation-Office of Polar Program, National Science Foundation-Physics Division, University of Wisconsin Alumni Research Foundation, Department of Energy, and National Energy Research Scientific Computing Center (supported by the Office of Energy Research of the Department of Energy), the NSF-supported TeraGrid system at the San Diego Supercomputer Center (SDSC), and the National Center for Supercomputing Applications (NCSA); Swedish Research Council, Swedish Polar Research Secretariat, and Knut and Alice Wallenberg Foundation, Sweden; German Ministry for Education and Research, Deutsche Forschungsgemeinschaft (DFG), Germany; Fund for Scientific Research (FNRS-FWO), Flanders Institute to encourage scientific and technological research in industry (IWT), Belgian Federal Office for Scientific, Technical and Cultural affairs (OSTC); the Netherlands Organisation for Scientific Research (NWO); M. Ribordy acknowledges the support of the SNF (Switzerland); A. Kappes and J. D. Zornoza acknowledge support by the EU Marie Curie OIF Program.


  1. preprint: APS/123-QED
  2. thanks: Deceased
  3. thanks: on leave of absence from Universität Erlangen-Nürnberg, Physikalisches Institut, D-91058, Erlangen, Germany
  4. thanks: on leave of absence from Università di Bari, Dipartimento di Fisica, I-70126, Bari, Italy
  5. thanks: affiliated with Dept. of Chemistry and Biomedical Sciences, Kalmar University, S-39182 Kalmar, Sweden
  6. thanks: NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA
  7. thanks: affiliated with IFIC (CSIC-Universitat de València), A. C. 22085, 46071 Valencia, Spain


  1. J. Ahrens et al. (IceCube), Astropart. Phys. 20, 507 (2004a).
  2. T. K. Gaisser, F. Halzen, and T. Stanev, Phys. Rept. 258, 173 (1995).
  3. E. Waxman and J. N. Bahcall, Phys. Rev. Lett. 78, 2292 (1997).
  4. F. W. Stecker, C. Done, M. H. Salamon, and P. Sommers, Phys. Rev. Lett. 66, 2697 (1991).
  5. A. Achterberg et al. (IceCube), Astropart. Phys. 26, 155 (2006).
  6. T. K. Gaisser and M. Honda, Ann. Rev. Nucl. Part. Sci. 52, 153 (2002).
  7. G. D. Barr, S. Robbins, T. K. Gaisser, and T. Stanev, Phys. Rev. D74, 094009 (2006).
  8. K. Münich et al. (IceCube), Proceedings of the 29th International Cosmic Ray Conference (2005).
  9. C. Costa, F. Halzen, and C. Salles, Phys. Rev. D66, 113002 (2002).
  10. E. V. Bugaev et al., Phys. Rev. D58, 054001 (1998).
  11. A. D. Martin, M. G. Ryskin, and A. M. Stasto, Acta Phys. Polon. B34, 3273 (2003).
  12. J. Ahrens et al. (AMANDA), Nucl. Inst. Meth. A524, 169 (2004b).
  13. D. Heck et al., Proceedings of the 27th International Cosmic Ray Conference pp. 233–236 (2001).
  14. G. D. Barr, T. K. Gaisser, P. Lipari, S. Robbins, and T. Stanev, Phys. Rev. D70, 023006 (2004).
  15. G. Barr, private communication.
  16. H. L. Lai et al. (CTEQ), Eur. Phys. J. C12, 375 (2000).
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
The feedback must be of minumum 40 characters
Add comment
Loading ...
This is a comment super asjknd jkasnjk adsnkj
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test description