OG 10.3.4


D. Berley1,a, M. Cavalli-Sforza2,b, M.L. Chen1, D. Coyne2, S. Delay3, D. Dorfan2, R.W. Ellsworth4, C. Espinosa5, G. Gisler5, J.A. Goodman1, T.J. Haines5, C.M. Hoffman5, S. Hugenberger3, L. Kelley2, I. Leonor3, J. Macri6, M. McConnell6, R.S. Miller5, A. Mincer7, M. Murray5, P. Nemethy7, J.M. Ryan6, M. Schneider2, B. Shen8, A. Shoup3, C. Sinnis5, A.J. Smith8, G. Sullivan1, T.N. Thompson5, T. Tumer9, K. Wang8, M.O. Wascko8, S. Westerhoff2, D.A. Williams2, T. Yang2, and G.B. Yodh3

1Department of Physics, The University of Maryland, College Park, MD 20742-4111, USA

2Santa Cruz Institute for Particle Physics, University of California, Santa Cruz, CA 95064, USA

3Department of Physics, University of California, Irvine, CA 92717, USA

4Department of Physics and Astronomy, George Mason University, Fairfax, VA 22030, USA

5Physics Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA

6Department of Physics, University of New Hampshire, Durham, NH 03824, USA

7Department of Physics, New York University, 4 Washington Place, New York, NY 10003, USA

8Department of Physics, University of California, Riverside, CA 92521, USA

9Institute of Geophysics and Planetary Physics, University of California, Riverside, CA 92521, USA

aPermanent address: National Science Foundation, Arlington, VA 22230, USA

bPermanent address: Institut de Física díAltes Energies, Universitat Autònoma de Barcelona, E-08193 Bellaterra, Spain


The Milagrito detector is being operated as the second stage of the Milagro detector. Milagrito is a large(~1500m2) single layer water-erenkov detector designed to study extensive air showers above a few hundred GeV. The detector with 228 PMTs has been operational since early 1997 and is taking data at a rate of several hundred events per second. The primary physics goals of Milagrito are to measure the high energy spectra of gamma-ray bursts, survey the northern sky and to use the shadow of the sun to study solar magnetic fields. The high sensitivity of Milagrito will permit the detection of lower energy air showers than with any previous ground-based array. In addition to its low energy threshold, the detector will have excellent angular resolution because it measures the arrival time of a larger fraction of incident particles than conventional detectors. This talk will present the preliminary results from Milagrito, including results on detector operation and physics studies.

Introduction to Milagro

Milagro will be a large water-erenkov detector designed to study steady and transient sources in the energy range from several hundred GeV to 100 TeV. It was conceived out of a desire to have a detector with a low threshold similar to an air-erenkov telescope, but with the ability to operate continuously and look at a large fraction of the sky simultaneously. In order to achieve this sensitivity, a detector must detect a larger fraction of the shower than conventional air shower arrays, which typically cover only a small fraction (< 1%) of the ground with detectors and are only sensitive to the electrons and positrons in the shower. At the end of the shower development curve, showers are rich in photons, which are four to five times more plentiful than electrons. Some sensitivity to the photon component of the shower can be gained by covering scintillators with a thin layer of lead to convert photons, but still most photons are not detected. Moving a traditional air shower array to high altitude will also help somewhat. The Tibet (Amenomori, 1995) array at 4300m has a threshold of several TeV, still an order of magnitude above that for air-erenkov telescopes.

Milagro uses erenkov light in water to detect both electrons, positrons and photons. It gains almost two orders of magnitude in fractional coverage compared to traditional air shower detectors by being sensitive to particles over its entire area. It gains further by having two meters of water (with a radiation length of 37cm) above a layer of PMT's, so that incoming photons are converted to electrons which then emit erenkov light.

Simulations show that Milagro should be sensitive to gamma-rays above several hundred GeV with an effective area that is nearly 100 m2 and that this area rises to the geometric area of 5000 m2 by about 1 TeV. This gives us the opportunity to study transient sources.

The core of the Milagro detector will consist of 790 photomultiplier tubes deployed in a 5,000 m2 covered pond at 2650m above sea level. Photomultiplier tubes will be deployed in three layers in purified water, under a light-tight cover. The first layer of 450 tubes that view the top ~2 m of water will be used to measure the time of arrival of an air-shower wave front. A second layer of 170 upward-facing tubes, called the hadron layer, will be located at a depth of 6.5m. These tubes will be used to make a calorimetric measurement of the energy deposited by the shower in the pond and study the hadronic and muonic content of the shower. A third layer of 170 tubes, optically isolated from the other layers at 7m below the water surface, views the 1500m2 area of the bottom of the pond. This layer will be sensitive to muons, allowing a study of the muon content of cosmic ray showers and the rejection of cosmic-ray background.

Fig. 1. The Milagrito configuration


Milagro is being built in three phases. During the first phase, which we have completed, we cleaned and prepared the pond, installed a new cover and liner, had utilities installed at the site and built the underlying support structure for the detector. We then instrumented this bottom layer with 28 PMT's, partially filled the pond and took data in this configuration during the spring of 1996. We call this phase "Milagrisimo". (See paper OG 4.3.10 in this conference for results from Milagrisimo). In the summer of 1996, we removed those tubes, finished the support structure and installed 228 PMT's on the bottom of the pond as shown in Figure 1. The pond is now partially filled with water replicating the top layer of Milagro and is being operated as an air shower detector; this mini-Milagro is being called "Milagrito". This configuration has more than 1/2 the effective area of the entire Milagro detector and a similar threshold and angular resolution, but does not have muon or hadron detection capability.

Milagrito Electronics

The PMT signals must have both their relative arrival time and their pulse height measured to obtain the best possible angular resolution. The dynamic range of the pulse height measurement must be about 104 and must be accurate at low pulse heights. We perform both operations using a custom built front-end board that operates in a "time-over-threshold'' mode. The width of the output pulse is equal to the time that the PMT signal spends over some predefined threshold. Since the PMTs are prone to pre-pulsing at high light levels we use two discriminators per PMT channel. The first discriminator fires when the pulse height exceeds ¼ of a photo-electron (PE) and the second when the pulse height exceeds 7 PE. The 2 signals are multiplexed onto a single line.

The digital signals from the front-end boards are routed to a LeCroy 1877 multi-hit TDC module. This module records the time and polarity of each edge. The TDC modules are readout by a Fermilab Smart Crate Controller (FSCC) and the data is shipped to a VSB dma controller and then into two dual-ported memory boards. The memory boards are readout via a Silicon Graphics Challenge computer over the VME bus. At an event rate of 1 kHz the dead-time in the entire system is ~1%.

Fig. 2. Milagrito trigger rate versus the number of PMTs in coincidenceFig. 1. Milagrito trigger rate versus the number of PMTs in coincidence

Data Taking with Milagrito

Milagrito data taking began in February of 1997 with about 1 meter of water above the PMTs. Once we have taken several months of data in this configuration the water level will be increased. This will allow us to study the detector characteristics with different overburdens of water. The water which was used to fill the pond was drawn from our onsite well. This water was passed through a series of filters before entering the pond and is recirculated nearly continuously through the filter system. The attenuation length of this water has improved with time and we will study the response of the system as the water continues to clean up. The trigger for Milagrito is a simple coincidence of N PMTs within a 300ns window. Figure 2 shows this rate as a function of the number of PMTs required. It is clear from the plot that there is a break in the slope of this graph near 50 PMTs. A possible explanation for this break could be triggering from large angle muons which produce significant light in the pond. Even with trigger settings as high as 100 PMTs there is evidence of large angle (>85o ) muons triggering. Currently we are operating at a trigger setting of 100 PMTs. This produces a trigger rate of ~300 Hz. This rate produces ~0.4Mb/s of raw data which is then processed in the online computer. Since we are still developing our reconstruction algorithms we write both compressed raw data and processed data to tape. We write approximately 20Gb/day of data on to DLT tape.

Fig. 3. Event display for Milagrito event. Vertical lines are arrival time multiplied by the speed of light.Fig. 2. Event display for Milagrito event. Vertical lines are arrival time multiplied by the speed of light.

The system operates continuously with a high duty factor. Since the system is unattended at a somewhat remote site, we have a separate online monitoring system which watches the performance of the system and monitors environmental conditions. Should either the online computer or our environmental monitoring system detect a problem, they send a message which pages the shift person in Los Alamos who can attend to the system.


We have built and installed a laser calibration system which consists of a nitrogen dye laser which is attenuated by a computer controlled filter wheel. This light is then put through a fiber optic switch which can send light to any one of the ten fibers leading into the pond. In the pond, diffusion balls which spread the light are suspended from floats. They are positioned so that each PMT can see more than one light source. The laser emits a short pulse whose amplitude is monitored by a photodiode which is also used to trigger the electronics. This system can provide both timing and charge calibrations for both the PMTs and the electronics.

Data Analysis

Milagrito is a new type of detector which necessitates new methods of data analysis. We are currently studying the response of this detector and are developing algorithms for shower reconstruction. It is also necessary for us to understand the response of the instrument. Specifically, we must understand both the shower curvature and instrumental time slewing in detail before we can reach our desired angular resolution. Figure 3 shows an event display for a typical air shower. The vertical lines are the arrival time (multiplied by c). The shower plane can clearly be seen. We have started making preliminary distributions of some parameters online. In Figure 4 we show a preliminary zenith angular distribution taken for a single run. The data can be fit out to about 60o with a cosnq where n is seen to be 4.6. This is considerably flatter than the distribution for a conventional scintillator array such as Cygnus where n~8.5 (Biller, 1993). Note that some of the large angle events in the spectrum may not be air showers, but may be muons or misreconstruction.

Future Directions

Fig. 4. Preliminary zenith angular distribution for detected Milagrito events. The fitted line is Cosine to the power 4.6

We are developing event reconstruction algorithms. We plan to take data continuously for the remainder of 1997 and the first part of 1998. In summer 1998, we will remove the Milagrito tubes and install the full three layer Milagro. Still there are many unresolved questions which will be studied with Milagrito. We need to design and deploy an air shower array around the pond to help find shower cores. This is needed to find shower cores and to help in energy determination. This past winter we tested a large (3m diameter) prefabricated plastic water tank instrumented with a single 8" PMT. Initial results from this study were very encouraging and may point the way for them to be used as the counters in this array. Another area which is under study is the exact design of the muon layer. We will also use the results of our Milagrito studies to determine the optimum depth of the Milagro EAS layer.


The Milagrito array is currently taking data at 300 events per second. The experiment will be used to study detector performance and produce physics results. Preliminary results from this array will be presented at the conference.


Amenomori, M. et al., Proc. XXIV ICRC, Rome, vol. 2, 346 (1995)

Biller, S.D. et. al., Nuclear Instruments and Methods, A328, (1993) p. 507-577