HAWC Technical Design
The HAWC design builds upon our experience with the Milagro detector. Milagro is the first large, uniformly instrumented, air shower array using water Cherenkov technology. The Milagro pond is instrumented with 2 layers of PMTs, a shallow layer for triggering and shower angle reconstruction and a deep calorimetric layer used for hadron rejection. Surrounding the central pond is an array of plastic, outrigger tanks (1 m deep by 3 m diameter) used for core position and shower angle reconstruction. In contrast, the HAWC design utilizes a single deep layer of PMTs with wider separation than used in Milagro. This configuration gives HAWC a much larger active area than Milagro for the same photo-cathode area.
HAWC will re-use the 900 8" Hamamatsu PMTs from Milagro, and deploy each PMT in a 4.6 m deep by 5.0 m diameter commercial plastic water tank. The tanks will be deployed in a dense pattern that provides more than 75% coverage of the 150m x 150m instrumented area. Each tank will contain an 8" baffled upward-facing PMT anchored to the bottom. Figure 17 shows the proposed deployment pattern. Figure 18 shows a single tank cross-section diagram as visualized in GEANT4.
The interior of the tanks is black plastic to minimize late light from reflections. In contrast, the Milagro outriggers are lined with white Tyvek to maximize light collection. We have found that while lining the interior of the tanks increases the photon count, it compromises the time-over-threshold method for pulse amplitude measurement because additional delayed hits lengthen the PMT pulses. Also, the timing of the shower front is not as accurately measured because the pulse amplitude is used to correct the timing for slewing effects.
We previously considered a HAWC design more similar to Milagro with a lined reservoir and curtains separating the PMTs. The reservoir could be made light tight by using a floating cover or enclosing it with a building. With LANL funds, we commissioned an engineering study that examined the complete costs and risks of each option. The complete report can be found on the HAWC proposal web site http://umdgrb.edu/hawc/Proposal. The conclusion was that a floating cover is more robust and less expensive than a building by about 10%, but the installation and repair of PMTs would be difficult. A building was the most expensive and riskiest option, requiring a completely light tight environment despite daily thermal changes and corrosion resistant materials to prevent contaminating the water. The option of using 900 individual tanks was considered as an addendum to the engineering study when we learned of the additional costs and risks associated with the pond. The engineers found no difficulties with the array of tanks and calculated the costs associated with flattening the required area, providing a drainage system, and installing the tanks. These costs, including the tank costs, are less than that of a covered pond. Additionally, the tank based design was found (via Monte Carlo simulations) to achieve approximately the same sensitivity as a function of energy as the pond. Furthermore, the tank design has the following advantages:
1) Science. Deployment and operation can begin immediately after the site is prepared. HAWC will achieve ~4 times greater sensitivity than Milagro with the deployment of only 1/3 of the array (2nd year).
2) Water. The single-pond design would have us fill the pond after construction is complete. With the tank design, we can add water incrementally, as we add tanks. This flexibility allows us to explore more economical ways to provide water to the site.
3) Risk. The solution is scalable, so cost risks can be absorbed by adjusting the number of tanks.
4) Flexibility. The tanks allow us to reconfigure the detector to attack different scientific goals. A larger array with a less dense core would increase the area at the highest energies, or more than 1 PMT could be placed in the tanks in a central area to increase the sensitivity at low energies.
Tank Design
The cylindrical tanks are constructed from opaque polypropylene, similar to Milagro’s outrigger tanks, and are used for commercial and residential water storage. The tank walls are ~1” thick. Each tank weighs ~2000 pounds and can be deployed easily with a forklift. The tanks are 4.6 m high and 5 m in diameter. The PMTs with the encapsulated base are 34 cm high, so the total amount of water covering each PMT is ~4 m or ~10.5 radiation lengths. When full, the tanks will each contain ~84,000 liters of purified water.
A test tank has been installed at the Milagro detector as seen in Figure 19. The test tank has a diameter of 3.6 m and is slightly shorter than the tanks specified in the design, but serves as a suitable prototype. The single photo-electron rate of the PMT is ~10 kHz, consistent with predictions from Monte Carlo simulations. The single muons that self trigger the PMT in the tank have an average of ~30 photoelectrons, also consistent with Monte Carlo simulations.
We do not anticipate problems with the HAWC tanks freezing. The top few inches of the Milagro outriggers do freeze each winter. The ice changes the reflective properties of the surface, which causes more internal reflection and increased single photoelectron rates, but has a negligible effect on the detection or reconstruction of extensive air showers. Furthermore, the Milagro site is much colder than the HAWC site. We will monitor the test tank in the next year to understand how these large tanks behave in cold weather.
Water Delivery System
The project will require ~80 million liters of fresh filtered water. We have budgeted for the water to be supplied from a well located 12km from the Sierra Negra site. A series of pump stations will bring the water up the mountain. We will also pursue the possibility of obtaining the water from a well adjacent to the HAWC site. An aquifer that is fed from glacial runoff has been identified in a geo-electric study performed in 2006 . We plan to dig a test well to verify the availability of water. Should the test well prove productive, the cost of providing the water needed for the project could be substantially reduced. We will also investigate the collection of runoff from the ~100 cm/yr of annual rainfall at the site and in nearby drainages.
The water will be filtered and staged in undeployed tanks and then transferred to empty tanks after they have been positioned and instrumented. The Milagro detector currently uses smaller tanks that are 1 m high and 3 m diameter that are also made of polypropylene as outrigger detectors. The water in the outriggers was filtered on the initial fill, but not subsequently. The cold and dark environment of the outriggers has resulted in no significant degradation of the water clarity in the outriggers over 3 years of operation. However, the larger tanks require longer attenuation lengths and we have budgeted for the HAWC tanks to be filtered. The water clarity in the prototype tank at the Milagro site will be monitored using a UV LED to understand the degradation of the water with time.
Elements of the Milagro water purification system will be used for HAWC with the same filtration setup and a capacity of 760 liter per minute. A carbon filter precedes a series of progressively smaller filtration stages from 10 microns to 1 micron to .3 microns. The water is sterilized using a UV light source. This system will be used to fill the tanks initially as well as allow us to filter 60 tanks at a time with 12 liter per minute for 5 days before switching to the next tank array (total of 15). In addition, the filtration system provides cooling for the electronics. A 40 liter per minute pump will be used by two heat exchangers with a combined cooling capacity of 140kW.
Deployment and Cabling
Prior to deployment, the entire 150m x 150m instrumented area will be graded and prepared, removing large rocks and any other obstructions. An over-excavation of 0.5 meters is planned to allow burying cables, and a storm sewer drainage system. The tanks will be placed in rows with a 1 m gap between every other row. Beneath the 1 m gap will be the cables and water filtration pipes.The counting house will be located at the center of the array to minimize the cable lengths and reduce signal dispersion. The cables will all be the same length of ~200 meters which is also the same length as those in the Milagro pond. The cables will be buried in PVC cable trays to prevent daily heating and cooling of the cables to reduce the effect on the signal timing. As in Milagro, the HV and signal are carried by a single cable that will run from each PMT to a service box in the ground next to the tank. From there on it will run under ground to the lightning protection service box next to the counting house. The HV / signal passes from here through an individual grounded spark gap array. These lightning protection boxes have been successfully used for the Milagro outrigger array over many years and have prevented observed lightning strikes from damaging the front end electronics.
Instrumentation
HAWC will reuse the 900 8” Hamamatsu R5912 photomultiplier tubes, the bases and encapsulations from Milagro. A single RG-59 cable provides high voltage to each PMT and carries the high frequency signal back to the front-end electronics. In HAWC, the cable will be permanently attached to the PMT housing as is now done in Milagro, thus avoiding the problems that Milagro initially encountered with underwater connectors.
The Milagro front-end electronics will be reused with minor modifications. These modules isolate and process the high frequency signals from the PMTs. The pulses are shaped and analog edges are generated at two discriminator levels, ~1/4 PE and ~5 PEs. These analog “edges” are subsequently digitized with multi-hit TDCs. The Time-Over-Threshold (TOT) method is used to measure both pulse arrival time and amplitude with a single multi-hit TDC channel. Additionally, the front-end boards provide summed trigger signals and direct access to the analog pulses for debugging and calibration. The high voltage for the entire experiment will be provided by a single supply with multiple channels. HAWC will trigger when ~30 or more PMTs are hit (~1/4 PE threshold) within 50ns. Simulations show that this gives a rate of ~8kHz. In Milagro the trigger window is 190ns, which we have found to be longer than optimal, so we will shorten the window for HAWC. The trigger pulses are generated in the front-end boards, which output an analog sum of the signals from 16 channels. In Milagro, the signals from each board are then combined using NIM electronics and this sum is discriminated to form a trigger.
New trigger electronics will be built for HAWC that feed the trigger pulses from the front end boards into a custom VME trigger module that uses a programmed FPGA to provide the trigger logic. The new trigger will allow us to develop more complex triggering algorithms, to push the trigger threshold lower, and possibly perform gamma/hadron separation at the trigger level.
For HAWC we plan to use the new CAEN model V1190A VME TDCs. These modules can handle the increased HAWC data rate as well as simultaneous digitization and read-out, thus eliminating the principal source of dead time that Milagro experienced with Fastbus TDCs. A number of these units have been purchased and will be evaluated in Milagro. A time stamp for each event is provided by a GPS clock that is latched by the trigger. The VME TDCs, GPS clock and trigger module will be housed in a single VME crate and read out into a single computer.
As in Milagro, we will have a separate monitoring system to measure various parameters, such as voltage, weather, water depths and temperatures, and rates of individual PMTs. These parameters will be read out every second and stored in a database. We will upgrade the Milagro system by using scalers in a VME based system, rather than CAMAC.
The laser calibration system for HAWC is designed to provide the relative timing and pulse-height calibrations for the PMTs in the HAWC detector. The absolute energy scale will be determined using single muons passing through the detector which produce ~30 photoelectrons. The laser system is patterned after the calibration system in Milagro, but each tank requires a separate optical fiber. The proposed light source design builds on the present Milagro calibration system including the JDS Uniphase “PowerChip NanoLaser” operating at 532nm. The ~1ns light pulses pass through a variable neutral density filter to allow control of the light intensity over 4 orders of magnitude. The laser beam will be directed through a series of optical fan-outs to illuminate one half of the PMTs at any given time. This system will run continuously at a low rate and will be controlled remotely.
Online Computing and Operation
The HAWC hardware will be read out using a single data acquisition computer. Software has been developed for Milagro to reconstruct the data in near real time by distributing it to an array (~4 computers) of client nodes. The stream of reconstructed events is then made available to online analysis programs that search the sky for transients, which can be detected within 5-10s of their occurrence. These clients also monitor the quality of the data. The design for the Milagro online software was utilized by the IceCube experiment for their online data processing and filtering system. The HAWC online systems will be based on the Milagro design, as extended and improved by IceCube.
Data will be buffered locally on disk arrays. After initial filtering to remove obvious background events, the raw and reconstructed data will be transmitted via the Internet to two archive sites, one in the United States and one in Mexico. This operational model demands a high bandwidth network connection to the site of the experiment. Although we will archive the data at two sites, all collaborators will have complete access to both sites and will work together on a single data product.
HAWC will utilize the robust and flexible C++ data analysis framework developed by IceCube for both online and offline data analysis. This common framework facilitates the rapid deployment of new reconstruction and filtering techniques for online processing at the site. Reuse of the IceCube software systems will leverage the considerable software investment made in the course of IceCube construction, permitting the deployment of a robust, well-tested system with considerably less effort and cost than would be required to develop and test a new system.
Computers and disk arrays will be tested for operational reliability at the ~4000m elevation. (Most commercially available computers are certified to operate up to ~3000m.) At higher elevations, CPU cooling becomes an issue. This problem can be mitigated by providing either cooler air or more air flow. Hard disks can fail at high elevations because the read head uses the pressure of the atmosphere to float over the platter. Sealed pressurized drives are readily available for high altitude operation and will be used if necessary.
HAWC is designed to be completely remote controllable. This is critical for reliable 24/7 operation of the detector. This operational model requires reliable power and a high bandwidth internet connection. All scientific collaborators are expected to take “remote shifts”, monitoring and maintaining the experiment. Shift-takers are required to review the operation of the experiment and the data quality daily and respond immediately to electronic pages (text messages) indicating problems. Repairs that require travel to the site will be performed by on-site technicians who maintain the experiment. This operational model is used by Milagro where ~95% uptime has been achieved.
The HAWC Site
The HAWC site is inside the Parque Nacional Pico de Orizaba, a Mexican national park comprising Citlaltepetl or Pico de Orizaba, the highest peak in Mexico at 5610m, and Sierra Negra, a 4600m volcano 7km SW Citlaltepetl. The Large Millimeter Telescope (LMT) is located on top of Sierra Negra, and HAWC will be located on a 200m x 450m plateau near the saddle between the two peaks. The exact geographical coordinates of the site are latitude 18º59’41”N, longitude 97º18’28”, altitude 4100 meters above sea level. The latitude of the Sierra Negra site provides an excellent visibility of celestial objects: HAWC will see 15% more of the celestial sphere within a 45º field of view compared to Milagro. When considering a cone of 45º the survey solid angle reaches 8.4sr, or 2/3 of the entire sky. The Crab culminates at 3º from the zenith and will be visible for slightly more than 6 hours each day; Cygnus reaches 20º zenith angle and the Galactic plane will be covered such as to include over a dozen of the VHE sources observed by HESS. Even the Galactic Center, at 48º from zenith, will be observable at the highest energies. The coverage of HAWC will have a 90% overlap with that of IceCube. The longitude of the site is also favorable, as its visibility has good overlap with observatories in the US, Mexico and Chile, which can be promptly alerted of any interesting activity and can aim to simultaneous observations of objects in the field of view of HAWC.
Climate
High sites with tolerable conditions are scarce. The site is located close enough to the equator to have weather conditions as benign as could be wished for its altitude. Weather conditions have been monitored for over six years at the summit of Sierra Negra, 500 meters above the HAWC site with a horizontal distance of 1km. The median temperature (adding the 6.5º thermal gradient to the 4600m measurements) becomes 4.3ºC for the site, with sub-zero temperatures only 5% of the time (specifically 10% of the time during winter). Water freezing inside the detector will not be an issue. Wind velocities are generally mild, with a median of 4 m/s for the recorded data. Occurrence of wind above 10 m/s is rare. Still, the recent passage of a hurricane Dean some 100 km North of the site provided winds up to 150 km/h, the largest measured in the 6 years of meteorological monitoring. The Large Millimeter Telescope has been designed to withstand winds up to 250 km/h, equivalent to a hurricane of category 5, which is not considered possible for a site over 4km high and 100 km inland.
Site accessibility
The coordinates of the HAWC observatory are defined tentatively, with over 200m of freedom in approximately the EW direction, as the northern base of the Sierra Negra can fit a square of (200m)2 in more than one location. An even a larger rectangle of 450m ´ 200m can be placed between two 20m topographic contour levels.
The HAWC site is located just over 2 hours from Puebla, a city of 2 million inhabitants with a relatively small international airport currently in expansion. Puebla itself is 2 hours by road from either Mexico City (to its East) or the HAWC site (to its West). Most of the distance between Puebla and the LMT HAWC site is through the Puebla-Veracruz motorway, with the last 40 minutes on minor roads. Veracruz is a major international port within 2.5 hours drive to the site.
The LMT project required an access road wide enough to allow transportation of items up to 6m wide. Electricity and internet have been installed up to the top of the mountain. The road and infrastructure would need to be extended 1 km to reach the HAWC site over mostly flat terrain.
The Large Millimeter Telescope
The Large Millimeter Telescope (LMT) is the largest scientific project ever undertaken in Mexico(120M$) and was constructed by a joint US/Mexico Collaboration. LMT is a single dish 50 meter telescope for millimeter-wave astronomy located at 4600 meter and due to operate in the frequency range of 80 to 350 GHz. Sierra Negra was the highest of close to twenty candidate sites monitored for water vapor content in the atmosphere and was selected as the LMT site in February 1997. Construction of the telescope began in 2000, with the antenna inaugurated by President Fox in November 2006. The surface of the telescope is presently being completed, set and tested while the LMT scientific instruments prove their performance in other telescopes.
The construction of the LMT required the development of the site infrastructure. The electric grid was extended 13 km to reach the LMT site to supply up to 1 MegaWatt of power during operation, with a potential peak supply of 5 MW. The road was constructed mostly during 1998-1999 to be able to allow pieces up to 6m wide and has been continuously improved. A fiber optic Internet line running parallel to the electric power has been set and is currently kept to just 2 Mb/s, but will be expanded once LMT enters the operation phase. The LMT installations are already able to lodge scientists in oxygen enriched areas. However, staying at the site is not encouraged and a base camp will be set at a lower altitude.
Sierra Negra Consortium
The construction of the LMT and the development of the Sierra Negra site brought the opportunity for other instruments to benefit from the high altitude site. Nine such facilities are in different stages:
- the Telescopio de Neutrones Solares in a solar neutron telescope installed by the Instituto de Geofísica of UNAM and in operation since 2005. It detected a major solar event in September 2006.
- RT5 is a 5m radio telescope in construction, due to perform daily monitoring of the Sun at 43 GHz during daytime and astronomical observations during nighttime. It can also function as test-bed for LMT instrumentation. RT5 is a joint project of INAOE and the Institutos de Astronomía and Geofísica of UNAM.
- two Cerenkov telescopes formerly part of the HEGRA array will be installed at the top of Sierra Negra, at about 1 km horizontal distance from HAWC. These will monitor blazars during the GLAST era and can also complement HAWC observations.
- the Instituto de Física of UNAM is to build an antineutron detector.
- the University of Puebla (BUAP) has set an array of cosmic ray detectors on the top of the mountain. These are small water tanks with individual PMTs. BUAP is also setting an array of larger tanks in the slope of Citlaltepetl to be complemented with a fluorescence detector at Sierra Negra.
- non astrophysical facilities include a seismological station from BUAP already operational, a greenhouse gas monitor from the Climate Institute and a geo-reference point from INEGI, the latter two in planning stage.
Together with HAWC and LMT, these facilities are members of the Sierra Negra Consortium (CSN, for its abbreviation in Spanish), a non-profit organization charged with organizing the joint operation of the site. The CSN will act as a provider of common services like site access, electricity, Internet, communications (with special consideration to preventing RFI to the LMT), water supply, security (there is a gate house with a 24 hour a day guard), etc. and the respective maintenance; in exchange the consortium members will cover their share of the operations cost, according to the location and characteristics of each experiment.
Site permission
The HAWC site is inside a National Park, with the land formerly owned by the Mexican Federal government. Permission for using the site can be granted by the Secretaría del Medio Ambiente y Recursos Naturales (SEMARNAT), the federal body in charge of the environment. An environmental impact declaration for the installation of HAWC in the Parque Nacional Pico de Orizaba was submitted to SEMARNAT on June 2007 and a conditioned permission was granted in September 2007. The permit allows a construction phase of three years and up to ten years operation, allowing the installation of the experiment, its peripheral infrastructure and water acquisition systems. A separate permit is required for the construction of the 1km access road and power line, which we plan to apply for by November. The conditions imposed to the project through the SEMARNAT permit are directed to minimize the environmental impact and to compensate it through a reforestation program related to the HAWC project. The permit is online at the HAWC proposal web site.
Water availability
This proposal includes funding to pump water for HAWC from a nearby valley. However, given the location and the reasonably high precipitation in the region (100 cm/year with a marked seasonal dependence), we will likely be able to acquire the water locally, through a deep well nearer to the HAWC site or a water capture system. A 3D topographic model of the region was constructed to model water flows in the vicinity of the site and was complemented with geo-electrical studies to determine the most suitable location for an extraction well. The demands on the capacity of the well are reasonable: considering the volume of water required for the 900 tanks the demand is only 2.4 liter/second for acquiring the water within one year using this method alone. Extending the time for water acquisitions relieves the demands on the well with the same proportion.
We already have defined the position for a test well and plan to perform its drilling in the next month. Depending on the water extractable, the test well can be expanded to a full extraction well. This will require widening its diameter, installing a high power pump and about 2km of pipes to transport the liquid to the HAWC observatory.
A second water acquisition option is a capture system placed below a convergence point, a natural nozzle, of water running down Citlaltepetl Sierra Negra, which was identified some 7 km WNW from the the HAWC site. This capture system will be a large concrete parallelepiped, which would be particularly efficient during the rain season (May-October), while the well would provide water in a more continuous