Web Newsletter
Second Page Third Page Back to Front Page

LIGO'S First Science Run - A Special Report

The LIGO Hanford Observatory. For the gravity-wave watchers of LIGO, it's all about the Science Runs. All the planning and preparation, all the building and fine-tuning, is geared toward the goal of getting to the Science Runs. They are the thrown-wide nets in which gravity-waves will eventually be ensnared. Here is captured the data which is then meticulously analyzed in the hopes of revealing our quarry at last.

Not the first Science Run, perhaps. Like all new ways of seeing, the first attempts are as important as lessons in training the eye and adjusting to the equipment as in trying to encompass everything right from the start. But every new adventure has a start, and for LIGO it was on August 23 when the observatory fired up its interferometers for a run that lasted well over two weeks. When it drew to a close on September 9, it had firmly succeeded in seizing a data set that is now under study for scientific results.

But this success arrived despite a decidely shaky beginning. Fred Raab, head of the LIGO Hanford Observatory in Washington State, describes how the Science Run, first scheduled for June, got bumped down the calender by a jolt felt round the world.

Accident Delays Run, But Also Gives Opportunity For Improvement

- Contributed by Fred Raab

LIGO's first Science Run (S1) was originally scheduled to begin on June 29, but accidental damage to H2, the 2-km interferometer here at Hanford, occurred just prior to the start leading to a frustrating postponement. The damage was triggered by a magnitude 7.2 earthquake that took place June 28 on the border of China and Russia. When the seismic waves reached Hanford around 15 minutes later, the ground vibration overwhelmed the control system for the interferometer. Such large earthquakes, those greater than seven on the Richter scale, occur about a dozen times a year somewhere in the world. When the first seismic waves from such events arrive at the observatory, the ground shaking becomes strong enough that the control forces needed to hold the suspended mirrors in position are greater than we can apply using our precision voice-coil actuators. Typically the ground shaking will decay to within the range of the voice-coil actuators within 15 minutes to an hour after the first waves arrive, and we can establish control of the interferometer again. But this time the control system became confused, unlike the dozens of previous times we have been hit by similar seismic waves, so that it failed to bring one of the dozens of mirrors here under control. As we diagnosed the problem from the control room and took steps to bring the errant mirror back under control, the reflected laser beam crossed a suspension wire for one of the mirrors and the subsequent heating led to failure of the wire. That mirror, part of a telescope used to expand the laser beam and steer it into the main interferometer, fell onto its earthquake stops. Repairing the damage would eventually require entering the vacuum chambers, removing the mirror and its suspension cage to the optics lab, rebuilding the optics module and vacuum prepping the repaired module, and then reentering the vacuum chambers to replace the optic into its original position and alignment. At the same time, we needed to analyze this newly discovered failure mode of the interferometer and take appropriate actions to ensure a similar accident would not occur again. S1 would have to be postponed for approximately two months to complete repairs.

Making The Best of the Situation

Delaying S1 was a huge disappointment. There was no way to avoid the inconvenience to several scientists already on their way to the observatories. But we could help minimize the impact to the commissioning program by scrambling to reschedule significant commissioning work on the remaining working interferometers. At Hanford, we decided to implement the common mode servo on H1, the 4-km interferometer. The common mode servo is the last and most crucial of the control systems that reduce the laser frequency noise. Without a working common mode servo, the H1 interferometer would not have the scientific reach of H2 or L1, the 4-km interferometer at Livingston, both of which had working common mode servos. We expected that H1 would not be scientifically important for S1, but that the run would test reliability of H1 as we had done for the other interferometers during earlier engineering runs. With a two-month delay of S1, it looked possible that we could get the common mode servo working on H1, allowing it to have a similar scientific reach to the other interferometers.

Well, it all worked out. Repairs were completed to H2. Preliminary modifications to hardware and software were put in place to prevent similar accidents in the future. We were able to get the common mode servo working on H1. Unfortunately we did not have much time to tune the interferometers for operating stability. In fact, delivering working machines for the rescheduled run required an "all-nighter" up to the start of the run. But everything was working when the curtain came up on S1 at the LIGO Hanford Observatory at 8:00am PDT on August 23.

S1 Performance at Hanford

A key measure of performance for the interferometers is the fraction of time during the run that each machine was in "science mode," that is, ready to observe a possible astrophysical event. We call this fraction the "duty cycle" of the machine. Ideally each machine would have a duty cycle of one-hundred percent, so it would always be available to make an uncompromised detection of an astrophysical event. In the real world, however, the duty cycle is limited by events that cause the machine to lose lock, by the time it takes the machine to lock again, by the time needed to tune up alignments or other operating parameters that can drift during the run, and by environmental conditions that make controlling the interferometers unlikely or impossible. We were able to keep H1 in science mode for 58 percent of the S1 run, and H2 was in science mode for 73 percent. These are excellent results considering that we did not get much opportunity to optimize machine parameters for duty cycle prior to starting the run. Combined with the duty cycle achieved at Livingston, we were able to obtain nearly 100 hours of data with all three LIGO interferometers simultaneously in science mode. We also set a new record for the longest sustained lock of an interferometer, holding H1 in lock for more than 21 hours.

Another key measure of performance is the depth of the search, namely the distance at which one could have detected a particular source. This varies by very large factors for different types of sources, often depending on details that we do not yet know. As a benchmark, we typically use the distance to which we could detect the inspiral of two 1.4-solar-mass neutron stars. The exact range for each interferometer has yet to be determined by a full analysis of the S1 data, which is ongoing. But preliminary estimates indicate that such a source would have been detectable throughout our galaxy. We believe these initial LIGO interferometers should be able to search millions of galaxies by the time they are fully commissioned, but this is a good start. We expect that S1 data will provide new direct experimental insights into the strengths of possible gravitational-wave sources.

A Case of the Jitters

The long running times in science mode also allowed us to gain useful insights into the work needed to improve sensitivity or to obtain more robust operation. In the latter category, we expect large earthquakes around the world or smaller earthquakes closer to the observatory to cause the interferometer to fall out of lock. In fact this occurred several times throughout S1. An unexpected phenomenon was losing lock when the well pump for our water system turned on. Now that we have identified this problem, we can study and fix it. An interesting check of our system's sensitivity was the observation of Brownian motion in the wires that support our mirrors. Brownian motion, observed as a random jitter of small objects viewed microscopically, was explained by Einstein as arising from collisions with atoms that were jittering back and forth due to their thermal energy. (This was before Einstein became preoccupied with explaining the origins of gravity.) The wires suspending our mirrors are made of atoms that jitter with thermal energy, so we expect the wire itself to jitter with its thermal energy. As the fine wire jitters, the much more massive mirror recoils by a tiny fraction of the wire’s motion. The plot below shows a spectrum of the wire motion versus frequency.

Spectrum of wire motion versus frequency.

The observed mirror motion measured by the interferometer is shown as red circles and the theoretical estimate of the recoil of the mirror from the jittering wire is shown as the blue curve. The eight peaks correspond to the two free segments of wire on each of four mirrors in the arms of interferometer H2. The total mirror motion (area under each peak) is approximately 1/50th of the diameter of a proton. The relatively good agreement near the peaks of these lines verifies that various aspects of the seismic isolation, suspension and controls for the interferometer work well, and can help focus our commissioning effort on other aspects that could still be limiting our sensitivity.

                                                             Next Page - The View from Livingston...