Groundbased Gravitational Wave Detectors Technology and Science Peter
Ground-based Gravitational Wave Detectors: Technology and Science Peter R. Saulson Syracuse University LIGO-G 1000038 -v 3 1
Outline 1. Gravitational waves and ground-based gravitational wave detectors 2. Challenges 3. LIGO, GEO, and Virgo: how it has been done so far 4. Highlights of scientific results 5. Looking ahead to advanced detectors Advanced LIGO and Advanced Virgo LIGO-G 1000038 -v 3 2
What are black holes like? What happens when two collide? LIGO-G 1000038 -v 3 3
Place some freely-falling test masses in a ring LIGO-G 1000038 -v 3 4
More simply … LIGO-G 1000038 -v 3 5
Sensing relative motions of distant free masses Michelson interferometer LIGO-G 1000038 -v 3 6
A transducer from length difference to brightness Wave from x arm. Wave from y arm. Light exiting from beam splitter. As relative arm lengths change, interference causes change in brightness at output. N. B. : This differs from Michelson’s alignment, which gave “fringes” that shift side-to-side. LIGO-G 1000038 -v 3 7
Black hole coalescence gravitational waveform LIGO-G 1000038 -v 3 8
Gravitational waveform lets you read out source dynamics The evolution of the mass distribution can be read out from the gravitational waveform: I is the mass quadrupole moment of the source. Coherent relativistic motion of large masses can be directly observed. So we can really see how black holes behave when they form or when they collide. We can also investigate many other astronomical systems. LIGO-G 1000038 -v 3 9
LIGO and LISA probe different bands of the spectrum Difference of 104 in wavelength: Like difference between X-rays and IR Rotating Neutron Stars LIGO-G 1000038 -v 3 10
Ground-based detection is almost impossible What is required for LIGO to succeed: • interferometry with free masses, • with strain sensitivity of 10 -21 (or better!), • in the presence of much larger noise. How is the “impossible” being made possible? LIGO-G 1000038 -v 3 11
Interferometry with free masses What’s “impossible”: everything! Mirrors need to be very accurately aligned (so that beams overlap and interfere) and held very close to an operating point (so that output is a linear function of input. ) Otherwise, interferometer is dead or swinging through fringes. Michelson bolted everything down. LIGO-G 1000038 -v 3 12
Michelson and Morley’s apparatus, 1887 LIGO-G 1000038 -v 3 13
Interferometry with free masses How it became possible: Servos » let mirrors swing freely on pendulum suspensions (at high frequencies), » while keeping them aimed properly and at the right interferometric operating point (at low frequencies. ) LIGO has dozens of servo loops. LIGO-G 1000038 -v 3 14
Strain sensitivity of 10 -21 Why it is “impossible”: Natural “tick mark” on interferometric ruler is one wavelength. Michelson could read a fringe to l/20, yielding hrms of a few times 10 -9. LIGO-G 1000038 -v 3 15
Strain sensitivity of 10 -21 How it became possible (I): Make L long. LIGO has 4 km long arms. Lopt is even longer, since light is stored for multiple round trips through the arms, which have the form of Fabry-Perot cavities. LIGO-G 1000038 -v 3 16
Strain sensitivity of 10 -21 How it became possible (II): Make DL small. Don’t rely on eye to judge sideways motion of interference fringes. Align beams, so that entire spot varies in brightness together. Then, interference is judged by a power measurement, and can be shot noise limited. LIGO-G 1000038 -v 3 17
Strain sensitivity of 10 -21 How it became possible (IIa): Make DL small. Minimize shot noise by using lots of light. LIGO uses a 10 W Nd: YAG laser. Low-loss optics, dark-fringe operating point, let most light survive passage through the interferometer. Supplement the laser’s power with recycled light, for even lower shot noise. Result: Can see motions of 10 -10 of a fringe. LIGO-G 1000038 -v 3 18
LIGO Optical Configuration Power Recycled Michelson Interferometer with Fabry-Perot Arm Cavities End Test Mass 4 km Fabry-Perot arm cavity Recycling mirror Laser Input Test Mass 225 W 5 W signal LIGO-G 1000038 -v 3 15000 W 50/50 beam splitter 19
Large mechanical noise How large? Seismic: xrms ~ 1 mm. Thermal: » mirror’s CM: ~ 3 x 10 -12 m. » mirror’s surface: ~ 3 x 10 -16 m. LIGO-G 1000038 -v 3 20
Large mechanical noise How measurement will succeed nevertheless: • Seismic noise input can be filtered. • Thermal noise can’t be filtered, but its spectrum can be controlled. • This leaves noise that is strongly frequency-dependent: » seismic noise falls with frequency, especially after passing through isolators, and » thermal noise is peaked at resonances, especially when Q’s are high. • Fringe lock servo holds interferometer at operating point, ensuring the linearity necessary for filtering the remaining noise. LIGO-G 1000038 -v 3 21
Brownian motion In 1827, Robert Brown noticed incessant jiggling of tiny particles suspended in water. LIGO-G 1000038 -v 3 22
What Einstein showed In the first of his “miracle year” papers of 1905, Einstein showed that: » » These experimental patterns are just what you expect from molecules bumping into the particle. The rate of Brownian “drift” lets you calculate the size of individual molecules, and thus determine Avogadro’s number. LIGO-G 1000038 -v 3 23
Atoms are real! • Soon, Avogadro’s number was determined: • This is a graphic demonstration that heat is the microscopic motion of molecules. • This also explains the relationship between friction and heat. Thus, Brownian motion gives us a direct window into the microscopic world of atoms. LIGO-G 1000038 -v 3 24
Fluctuation-Dissipation Theorem (Callen et al. 1951 -2) For a linear system in thermal equilibrium, Y(f) is the admittance v/F, and its real part is a measure of the dissipation in the system. To reduce the noise, we can either » lower the dissipation, or » lower the temperature. LIGO-G 1000038 -v 3 25
Real part of admittance Y determines strength of thermal noise LIGO-G 1000038 -v 3 26
Relation to Equipartition Theorem: each d. o. f. has, on average, an energy of k. BT/ 2. For a simple harmonic oscillator, this means or rms motion of 3*10 -12 meters, for 10 kg mirror in 1 Hz pendulum at room temp. (!) LIGO-G 1000038 -v 3 27
Temporal character LIGO-G 1000038 -v 3 28
Same lesson in frequency domain LIGO-G 1000038 -v 3 29
Q as a Figure of Merit Q: dimensionless measure of the ratio of elastic restoring force to dissipative force. Measured by width of resonance or by ringdown time of vibration. Related to “loss angle” f, phase lag between force and displacement, by f = 1/Q. In round numbers: garden variety Q = 103, good Q = 106, outstanding Q = 108. LIGO-G 1000038 -v 3 30
Interferometer suspensions LIGO-G 1000038 -v 3 31
LIGO I noise spectrum LIGO-G 1000038 -v 3 32
State of the art Q First LIGO suspensions have Q around 106 for both pendulum and internal vibrations. Internal mode Q comes from mirror substrates and coatings. Pendulum mode Q from steel wires (Q~103) plus dissipation dilution effect: most restoring force in pendulum comes from tension in wires, free of internal friction. We now know how to do better, using: • fused silica fibers to hold the mirrors, instead of steel wires, • ultra-low dissipation glass substrates and dielectric coatings. LIGO-G 1000038 -v 3 33
Mirror Suspensions 10 kg Fused Silica, 25 cm diameter and 10 cm thick magnet LIGO-G 1000038 -v 3 34
Seismic Isolation stack of masssprings LIGO-G 1000038 -v 3 35
LIGO Vacuum Equipment LIGO-G 1000038 -v 3 36
LIGO Beam Tube LIGO-G 1000038 -v 3 37
LIGO (here, LIGO Livingston Observatory) A 4 -km Michelson interferometer, with mirrors on pendulum suspensions. Site at Hanford WA has both 4 -km and 2 km. During S 5 run, operated at design sensitivity: hrms = 10 -21. LIGO-G 1000038 -v 3 38
LIGO Hanford Observatory LIGO-G 1000038 -v 3 39
S 5 noise budget, Livingston 4 km interferometer LIGO-G 1000038 -v 3 40
GEO 600 LIGO-G 1000038 -v 3 41
GEO all-silica pendulum LIGO-G 1000038 -v 3 42
EGO • • • LAPP - Annecy INFN - Firenze/Urbino INFN - Frascati IPN - Lyon INFN - Napoli OCA - Nice • • • ESPCI - Paris LAL - Orsay INFN - Perugia INFN - Pisa INFN – Roma NIKHEF – Amsterdam (joining) Inaugurated July 2003 LIGO-G 1000038 -v 3 43
Virgo Super-Attenuator The rma LIGO-G 1000038 -v 3 l no ise 44
Virgo’s sensitivity In recent commisioning (red curve), noise has come very close to design goal. Note especially the outstanding low-frequency performance. LIGO-G 1000038 -v 3 45
The S 5 and VSR 1 science runs After several years of commissioning, LIGO reached design sensitivity, and carried out a long observing run, “S 5”. Previous science runs had durations of only one or two months. Produced a number of upper limit papers. S 5 collected one year of integrated triple coincident data at design sensitivity, from November 2005 through September 2007. Virgo reached good sensitivity and began on overlapping run, “VSR 1”, beginning May 18, 2007. All data since that date have been analyzed and are being reviewed and published together by the LSC and Virgo collaborations. LIGO-G 1000038 -v 3 46
The Crab Pulsar [ Abbott et al. , Ap. JL 683, L 45 ] » » Spinning down rapidly; energy loss ~ 4 × 1031 W A significant fraction of that could be going into GWs Searched for signal in first 9 months of LIGO S 5 data Assuming that GW signal is locked to EM pulses, null search result implies that no more than 4– 6% of the spin-down energy is going into GW emission LIGO-G 1000038 -v 3 Chandra image Insight into the Crab Pulsar 47
Insight into gamma ray bursts • GRB 070201 » Short, hard gamma-ray burst » Consistent with being in M 31 » Leading model for short GRBs: binary merger involving a neutron star • Looked for a GW signal in LIGO [ Abbott et al. , Ap. J 681, 1419 ] » Searched for both inspiral and burst signals » No GW signal found IPN 3 -sigma error region from LIGO-G 1000038 -v 3 Mazets et al. , Ap. J 680, 545 • Conclusion: it was not a binary merger in M 31 possibly an SGR giant flare instead 48
Other observational results • Upper limits on cosmic stochastic background of gravitational waves Now probing below other experimental constraints in LIGO band Ref: Nature 460, 990 (2009) • Upper limits on rates of gravitational wave bursts Rule out all previous “indications” from observations by resonant bars Ref: Phys. Rev. D 80, 102001 (2009) • Upper limits on signals from compact binary coalescences Not yet sensitive to expected rates, but setting up machinery for best target of Advanced LIGO/Virgo searches Ref: Phys. Rev. D 80, 047101 (2009) • And many more … https: //www. lsc-group. phys. uwm. edu/ppcomm/Papers. html LIGO-G 1000038 -v 3 49
Coming Soon: Advanced LIGO and Advanced Virgo Advanced LIGO: • ~10 x lower noise • ~4 x lower frequency • tunable Initial LIGO Through these features: • ~20 x higher laser power • Signal recycling • Active seismic isolation • Fused silica multi-stage suspension • Ultra-low dissipation test mass substrates and coatings LIGO-G 1000038 -v 3 Advanced LIGO 50
Advanced LIGO and Virgo will dramatically increase sensitivity Advanced Virgo: • Virgo already has outstanding (passive) seismic isolation • Advanced Virgo will have new monolithic suspensions for lower thermal noise • Higher laser power and signal recycling are key at high frequencies LIGO-G 1000038 -v 3 51
Reach of advanced interferometers Advanced LIGO and Advanced Virgo should see lots of signals. Most realistic models: • Neutron star binaries » Nexpected ~ 40/yr • Black hole binaries » Nexpected ~ 20/yr • BH/NS binaries LIGO Range » Nexpected ~ 10/yr (and even pessimistic models predict some detections. ) LIGO-G 1000038 -v 3 Image: R. Powell Advanced LIGO Range 52
Status of Advanced LIGO and Advanced Virgo • Advanced LIGO was approved by NSF in 2008. • Fabrication of new components is well under way. • Installation of new hardware will start in late 2010. • Expecting science data in mid-decade. • Advanced Virgo was approved Fall 2009. It should achieve comparable sensitivity to Advanced LIGO, on a comparable time scale. • In the meantime, we have been observing, in science run S 6/VSR 2. LIGO-G 1000038 -v 3 53
- Slides: 53