Amanda II Project, Web Site of the Barwick Group at UCI
Home Amanda Collaboration Overview of Research Scientific Results Papers & Reports Presentations Public Information ACID TWR Software Library Data Archive SimulationTools
AMANDA-II Science Results
AMANDA has been operational since January 1997.  We summarize a few important published results (along with a few preliminary and status results) from the 10 string and AMANDA-II detectors for the science topics listed in the menu:
AM-II schematic

Science Menu

Note: by clicking on the figures with boarders, you can get better images of the graphs.


We will discuss a variety of results, a few of which are still preliminary (and should be treated as a work in progress until the results are published). In general, AMANDA differentiates neutrino flavors by the topology of the light distribution in the event, as the next two figures illustrate.


Schematic of the method to detect muon-neutrinos that is employed by AMANDA. Schematic of the method to detect electromagnetic or hadronic cascades initiated by electron or tau neutrinos.
In general, the method depicted on the left is the one that provides the best sensitivity (because the muon can travel much further than Cherenkov light in the ice) and angular direction (about 2 degrees in AMANDA-II). The detection of cascades are interesting for several reasons. The energy resolution for these event is much better than for muon events (~20% compared to a factor of 3-4). Cascades are initiated by all neutrino species, including neutral current interactions. Finally, at energies above a relatively modest threshold, neutrinos can be detected at any angles. This feature is especially useful at energies in excess of 100 TeV, where the earth begins to attenuatete neutrinos that travel along a signifcant portion of the diameter.


Optical Properties of Antarctic Ice

The optical properties of in situ ice beneath the south pole are measured by a combination of in situ N2 lasers, DC lamps, and YAG laser pulses from the surface.  The properties vary with depth due to climatological variation such as ice ages.  We have not measured the scattering length and absorption length over all relevant optical wavelengths, so they are inferred by a model developed by P. Buford Price (UC-Berkeley) and his collaborators. The two properties that most strongly affect the reconstruction capabilities of AMANDA-II are absorption and scattering.

The right figure shows the absorption length as a function of depth. The bulk of the scientifically-useful optical sensors in AMANDA are embedded between 1500 and 1900 m beneath the surface.

The absorption strongly depends on wavelength. Notice that the absorption also depends on depth at wavelengths where the absorption coefficient is relatively small. For short wavelengths, the absorption coefficient is small and dust contributes significantly, which is responsible for the depth dependence. At 532nm, the absorption coefficient is large, and the value is largely determined by intrinsic properties of ice (i.e, the roll of dust is less obvious).


The left figure shows the average scattering coefficent (1/scattering_length) as a function of depth. Note that the effective scattering length, L_eff, is (approximately) the average length to isotropize the direction of all but 1/e of the photons.  This important parameter for diffusion calculations is related to the geometric scattering length by L_eff=L_geo/(1-<cos(angle)>). The solid curve shows the coefficient of the scattering length for 400 nm light. Other colors behave differently due to the slight dependence of the scattering length on wavelength.

At depths below 1400m, dust is responsible for light scattering in ice. The rapid rise in scattering at shallow depths (relative to the surface) is due to onset of air bubbles trapped in the ice. The dashed blue line shows the intrinsic scattering from dust in the region dominated by air bubbles.

To summarize, the maximum absorption length is slightly more than 100m at AMANDA-II depths, but the scattering length is only 20m for wavelengths that correspond to the longest absorption lengths.


Atmospheric Neutrinos in AMANDA-II

Atmospheric neutrinos measured by AMANDA-B10

The atmospheric neutrino analysis concentrates on signal purity. The distribution show on the right is from our 2000 data sample. It is consistent with expectations from atmospheric neutrinos because 1) the observed events are distributed approximately isotropically, 2) the distribution of the number of OMs participating in the event (which is correlated with energy) is consistent with a soft spectra, 3) the shape of the zenith angle distribution of events is consistent with expectation (see figure for Diffuse Flux), 4) the absolute number of events is within 30% of expectation, consistent with systematic uncertainty of the predictions, and 5) upon visual inspection, the events topology is consistent with upgoing neutrino events. Recently, we have developed an accurate energy estimator, allowing us to plot the differential flux of the sum of atmospheric muon neutrinos and anti-neutrinos.


The figure below shows the differential flux of atmospheric neutrinos. The enormous effective volume allows AMANDA-II to measure the flux to much higher energies than any previous detector. The solid black lines show the theoretical expectation for both horizontal (upper) and vertical (lower) orientations.

The zenith angle distribution is compared with MC predictions in the figure below.  Horizontal events have sin(declination)=0, and upward vertical events have sin(declination)= +1. AMANDA-II has vastly improved sensitivity near the horizon compared to AMANDA-B10 due to the fact that it is much wider. Note that AMANDA-II has very little sensitivity to neutrino oscillations due to its relatively large energy threshold, although we have included such effects in the predicted flux.

Point Source Search using AMANDA-II

The search for point sources uising AMANDA-B10

The point source analysis optimizes the selection criteria on hard spectra (differential energy spectra proportionally to E-2), although it has reasonable sensitivity to softer spectra. The critical features of the point source analysis:  demonstrate good angular resolution and absolute pointing and maintain good effective size for as much of the sky as possible.  For the point source analysis, the AMANDA-II detector achieves:

~25,000 m^2 if the muon energy at the detector is greater than 10 TeV.

space angle resolution of 2 deg (median)

Neutrino flux sensitivity of ~0.22 x10^-7 cm^-2 s^-1

Sensitivity achieved by the AMANDA-II point source analysis. The plot shows 90% CL upper limits (averaged over right ascension) for a source with an assumed differential spectrum of E^-2. The neutrino signal is then integrated for energies above 10 GeV to determine the limit on the neutrino flux. The units for the vertical scale are 10-7 cm-2s-1.

Experimental limits on muon fluxes from high energy neutrinos, and projected sensitivity for AMANDA-II assuming data is analyzed from several years of operation.

AMANDA-II sky map:  

The left figure below, first presented at scientific conferences this spring (03), below shows the neutrino sky as seen by AMANDA-II using data from just the first year of operation (Feb -Oct of 2000). The point source analysis was developed by randomizing the true azimuthal (or right ascension, RA) distribution of events to insure that human expectation does not bias the analysis. Nearly all events in the northern sky are compatible with atmospheric neutrinos (plus a small admixture poorly reconstructed atmospheric muons). While the angular distribution of this data reveals NO evidence for extraterrestrial neutrino sources, it provides important constraints on theoretical models.

A total of 1555 events are shown in the figure to the left (in equatorial coordinates).   Compared to the previous sky map from AMANDA-B10, the coverage near the horizon (declination = 0 degrees) is markedly improved. The angular resolution of AMANDA-II is much improved so the angular dimensions of a search bin are likely to be reduced to 6x6 square degrees. The downgoing atmospheric muon background is responsible for the thick band of events below the horizon. Less than 3% of the events above a declination of 5 degrees are due to atmospheric muon contamination.

The angular distribution of events in the preceeding skyplot was examined for statistically significant excesses. None were found, leading to preliminary limits shown in the contour plot and reported in the table below.

The units of the color legend are 10-7 cm-2s-1.The horizontal units are hours of right ascension and vertical units are degrees of declination.

In the above figure, we show the differential flux limit for two different assumptions of the spectral index, alpha. We compare this result to AMANDA-B10 and to the average gamma ray flux from Mrk 501 as observed in 1997, a period of high activity. Also shown is the intrinsic source flux after correction for IR absorption (de Jager and Stecker).

Although Mrk 501 was not very active in 2000, the figure suggests that AMANDA-II has achieved the requisite sensitivity to observe neutrinos during the active period if the neutrino to photon ratio is one.

The following table provides upper limits for a selection of candidate sources, and compares the results to limits previously published for AMANDA-B10. The units for columns 3 and 4 are 10-8 cm^(-2)s^(-1).
For a more complete list of sources and flux limits, click here.


GRBs (97-00)

Gamma Ray Bursts (or GRBs) are the most spectacular and powerful explosions in the sky. Their duration lasts from milliseconds to hundreds of seconds, and they tend to cluster into two distinct populations.  They are also extremely distant, with redshifts exceeding Z=1, although most of what is known about distances is deduced from the longer duration GRBs. We search for correlation between the GRB event time and location provided by gamma ray satellites.  Unfortunately, one of the most powerful detectors (called BATSE on the Compton Gamma Ray Observatory) ceased operation in May of 2000.  To increase our search exposure, we have examined BATSE bursts that were uncovered offline by additional analysis (termed "non-triggered" bursts) and GRBs identified by the Interplanetary Network (IPN) of satellites.


Simulation of GRB explosion
Simulation of GRB explosion which shows strong jet features (dark) and predicts strong relativistic flows.  High energy neutrino production may occur along the barrel of the jets.

The most critical feature of GRB analysis is the very modest background rejection requirements. GRB emit gamma radiation over short duration, typically less than 100s. Satellites provide time and direction information so the analysis needs to reject only non-neutrino events that happen to occur at the right time and reconstruct with the right direction. The upshot: the effective size of AMANDA-II for GRB analysis approaches 50,000 m2 for muons created by neutrinos. Note that the figure does not include the effects of attenuation by the earth.

The table below summarizes the number of GRB sources observed by AMANDA (as of Aug 03), the observed number of signal events (5th column), the expected background (4th column), and the muon flux upper limit (90% CL).

The result is clear, and widely reported, - neither AMANDA-B10 nor AMANDA-II observes correlated emission of high energy neutrinos from any burst sample.

However, the interpretation of this result in the context of models is nontrivial. We are consulting with scientists worldwide to provide input on how we can best report our results.


From the last row in the table above, we derive a preliminary neutrino flux limit of E^2*(dN/dE) < 4x10^-8 GeV/cm^2/s/sr, assuming a broken power law spectral form. More generically, we follow the method outlined by SuperKamiokande to derive an energy dependent Greens Function for the fluence limit, shown next. The turn-up at high energies is due to the attenuation of the neutrino flux by the earth.

Diffuse Flux

The most sensitive mode to search for diffuse flux of HE neutrinos uses the muon signals.

The current (July 2003) limits are shown in the next figure for AMANDA-B10 and Baikal. Note that the AMANDA limits are published.  Systematic studies have not yet been fully completed.  AMANDA-II and IceCube are also shown.  The curve labeled (down) utilizes the EHE analysis to search for neutrino-induced muons in the downgoing direction (from above the local horizon of the detector). The sensitivity of AMANDA of AMANDA-II is sufficient to search for sources below the "evolved" Waxman-Bahcall limit of ~5x10-8 GeV/cm2/s/sr.

Summary of diffuse flux predictions and experimental limits

The main background to the upgoing muon flux is atmospheric neutrinos generated by cosmic ray collisions.  At high energy energies, direct production by charm decay becomes important, although the predicted fluxes have large uncertainty. As the neutrino energies increase beyond 10^7 GeV, most of the signal comes from above or very near the horizon and perhaps prompt MUONs from charm quark decay become a significant background. The importance of the background depends on the magnitude of the flux and detector energy and angular resolution. Fortunately, the sensitivity of AMANDA-II is not impacted by charm contributions for the most of the favored models of charm production in the atmosphere.

The figure above and to the right are taken from our paper published in PRL. Nch refers to the number of modules that participate in a given event, and crudely related to total energy of the particle.



Cascades Detection by AMANDA-II

High energy neutrinos may interact to produce a large cascade of particles. In this case, the production of Cherenkov light remains localized and the photons propagation radially outward (well, almost).  The effective volume of AMANDA is much smaller for cascades than muons and the angular resolution is very poor, but there are several interesting features of cascades that make them useful to study.  First, the energy resolution of the cascade event can be measured with much better precision relative to the muon signature if the vertex of the interaction is contained.  Second, the backgrounds from atmospheric electron neutrinos is much smaller than muon neutrinos at these energies because the decay of atmospheric muons is suppressed by time dilation. Third, cascades are produced by electron neutrinos and tau neutrinos so the ratio of cascades to muon neutrino events provides insight on the properties of neutrino oscillation. Fourth, even downgoing neutrinos can be observed if the energy is larger than ~10 TeV. Thus, the cascade technique is sensitive to neutrinos from any direction.

The search for a diffuse astronomical source was performed using data from 1997 collected by AMANDA-B10 and 2000, which was collected by AMANDA-II.

Cascade events as a function of reconstructed energy. Experimental data is compared to background predictions (Background MC) and signal expected for a diffuse source emitting electron neutrinos with a differential energy spectrum proportional to E^(-2).

Diffuse flux limits at the source on the sum of all neutrino flavors as derived from AMANDA-B10 using muons (lower blue dotted line) and cascades (lower red).  These limits were derived assuming neutrinos oscillate so the flux of all flavors is equal by the time they reach the earth. 

Correcting for neutrino oscillations increases the deduced flux limits for the muon channel. This is the appropriate value to constrain model predictions.


EHE Physics

Like with accelerator physics, most of the interest neutrino astrophysics is at the extreme energy frontier, which we label "extreme high energy" or EHE. Perhaps the most reliable flux predictions (after atmospheric neutrinos) involve the GZK mechanism.  Neutrinos are produced by the inevitable collisions between cosmic rays and the cosmic microwave background.  Although the physics required by GZK mechanism is straightforward, the GZK mechanism is still in doubt because one of its predictions has not yet been confirmed.  The GZK mechanism predicts are rather strong upper limit on the energy of the cosmic ray, but this upper limit has not yet been clearly identified.

Perhaps the most intriguing aspect of neutrino physics at extreme energies is the potential to study fundamental theories of particle physics.  For example, Jonathan Feng (UCI) and his colleagues predict that extremely energetic neutrinos create micro-black holes as the collide with nuclei of atoms in the earth, if certain models of strong gravity are correct.

Experimentally, neutrinos at EHE energies are difficult to detect. As the neutrino energies increase to 1 PeV (1015 eV), the earth becomes opaque except near the horizon.  We have developed a new technique to search for "downgoing" nearly horizontal muons.  They can be distinguished from the blizzard of downgoing muons from cosmic ray collisions because the energies are much higher than muons generated by cosmic ray collisions. Consequently, the topologies of the events (see next figure on the left) are quite different from the typical event. Notice the large number of OMs that observe Cherenkov light. The effective detection area for muons at these energies is very large, typically 0.2 km2 for AMANDA-B10!  In the next few figures, we show preliminary results from our study of AMANDA-B10 data from 1997.

Using data from 1997, the preliminary limit (90% CL) on the differential flux proportional to E^(-2) is:

E2*(dN/dE)=0.93 x 10-7 GeV/cm2/s/sr.


This limit was derived using only 3 months of data from 1997 (the remaining data from that year was not suitable for this technique).  The horizontal extent of the black experimental limit indicates the energy interval that contains ~90% of the detected events if you assume a differential energy spectrum proportional to  E-2.  For E-2, the energy interval is between about 2x1015 eV and 5x1018 eV, but since few EHE models predict E-2, we studied a much larger energy interval.

The effective detection area of AMANDA-II is even larger than B10.  In addition, we upgraded the data acquisition system of AMANDA-II.  The new system records the complete waveform from all (usable) OMs in the array.  This should dramatically improve the dynamic range of the photon measurement and allow far better energy reconstruction for these high energy events. The expected sensitivity of AMANDA-II is shown on the diffuse limit figure.


The EHE analysis relies on training a neural net (NN2) to separate background from signal. As the above figure shows, the data agree with background expectations (using the program CORSIKA to generate atmospheric particles created by cosmic ray collisions. There is no evidence for an astrophysical source of high energy neutrinos.



A class of dark matter candidates are hypothetical particles known Weakly Interacting Massive Particles (WIMPs).   One of the mostly widely studied WIMP is provided by an extension to the standard model of particle physics known as Supersymmetry (or SUSY to its friends).  If supersymmetry ideas are correct, then the lightest stable supersymmetric particle could be the dark matter. AMANDA can search for WIMP dark matter indirectly by searching for high energy neutrino emission from the core of the earth or the sun.  The idea is that dark matter particles would occasionally inelastically collide with atoms and lose enough energy to become gravitationally bound to the sun or earth.  Eventually, the WIMPs spiral down to the core.  As the density of dark matter WIMPs increases in the core, they begin to interact and annihilate with each other.  Their annihilation produces high energy neutrinos (among other particles). 

Indirect detection of Solar WIMPs
This graphic (from Joakim Edsjo) shows the capture of dark matter WIMPs by the sun, but a similar picture applies to capture of the WIMPs by the core of the earth. The annihilations of WIMPs in the core produce muon neutrinos, and they can be observed by AMANDA if they interact near the detector.
Earth WIMP limits by AMANDA-B10
Figure shows the AMANDA-B10 limit (solid black line) deduced by searching for high energy neutrinos from the center of the earth.  Other experimental limits are shown for comparison. For this science, all experimental techniques are limited by irreducible background from atmospheric neutrinos.  The green dots represent SUSY models that are already excluded by direct search techniques. 

We are just beginning to search for WIMPs from the sun. It is relatively difficult for AMANDA because the sun remains near the horizon where most of the residual background events appear to come from.  Therefore, we expect the first substantial result from AMANDA-II due to its superior background rejection near the horizon.

Supernova Monitor

The extremely low ambient photon flux in deep ice provides the opportunity to monitor the galaxy for supernova explosions.  Supernova events are expected to generate neutrinos at low energies (< 20 MeV), nominally too low of an energy to trigger AMANDA electronics.  However, a nearby supernova blast would generate so many neutrinos that enough of them would interact within 10m of each AMANDA OM and produce Cherenkov light.  The extra photons could contribute the average "noise" rate from each OM.  By summing the signals from each OM, a statistically significant signature of a supernova can be obtained.  We have installed special electronics to read out and sum the "noise" rates from each OM.
SNa reach of AMANDA AMANDA-B10 can monitor about 68% of the stars in our galaxy, and AMANDA-II can reach 95% of the stars in our galaxy.

The AMANDA collaboration is working with SNEWS (supernova early warning system) members to provide timing information to help pinpoint the position of the supernova by triangulation. 


By virtue of it large volume, AMANDA can search for relativistic magnetic monopoles with unprecedented sensitivity.  Our technique relies on fact that the equivalent charge of a magnetic monopole is 68.5, and since the amount of Cherenkov light depends on the square of the charge, the light produced by monopoles is prodigious.  By constraining the search to monopoles that pass through the earth, we simplify the analysis.  The downside of this restriction is that the mass of the monopole must be large to possess enough kinetic energy to pass through the earth.

limit on flux of magnetic monopoles


Atmospheric muons constitute the dominant type of events that trigger the AMANDA telescope. Recently, we have used this data to deduce the muon flux as a function of (slant) depth.

The predicted (Monte Carlo) and observed atmospheric muon flux agree to within 30%, which is consistent with the uncertainties in the prediction. To simplify the comparison of spectra shape, the AMANDA-II data in the figure on the left has been normalized to the most vertical point (smallest depth) of the Monte Carlo prediction.

The depth dependence also agrees with the predictions of Bugaev at al. and previous experimental results.


FUTURE of AMANDA: 2nd Generation DAQ

Last season (January, 2003), the AMANDA collaboration initiated a major upgrade to the data acquisition system. The new system records the complete waveform produced by every OM in AMANDA-II. The figure on the left shows the electronics and the figure on the right provides one example of a waveform. We expect that the new capabilities will greatly improve the sensitivity of AMANDA, especially to neutrinos at the highest energies.

Summary of new capabilities and features of the TWR DAQ:

  1. Full waveform records of every event
  2. Deadtimeless operation up to trigger rates of ~175 Hz ( contrast to 15% deadtime at 90 Hz with 1st generation DAQ)
  3. Linear dynamic range to 5000 photoelectrons (nearly a factor of 100 improvement)
  4. Suitable for multi-level triggering, including software triggers
  5. Simplifies joint trigger with initial deployment phase of IceCube strings

The waveform above shows several distinct pulses. To reduce the data traffic, the AMANDA team had to develop specialized compression procedures. The horizontal flat lines indicate regions where the waveform has been cut out, leaving only the waveform fragments of interest.


The figure on the right indicates that the new electronics can measure the number of photons at each OM with much better accuracy than before. The good agreement between the light blue squares and dark blue stars indicates that AMANDA can measure as many as 5000 photoelectrons in each module, nearly a factor of 100 better than achieved with the original DAQ system.




UCI Anteater

Barwick Group
School of Physical Sciences
University of California Irvine