The Virgo detector: status and first experimental

resultsNicolas Arnaud NIKHEF June 20th, 2003

Outline

• The quest for gravitational waves (GW): a long history

• Detection principle Interferometric detectors

• Description of the Virgo interferometer Optical scheme Main features of the instrument Foreseen sensitivity

• Experimental control of the Central Interferometer (CITF)

CITF description and CITF commissioning goals Experimental results (spring 2001 summer 2002)

• Virgo versus the other GW interferometric detectors The LIGO interferometers (USA) + TAMA (Japan)

• Main GW sources and filtering techniques

«J'ai été d'abord conduit à supposer que la propagation de la gravitation n'est pas instantanée mais se fait à la vitesse

de la lumière (…) Quand nous parlerons donc de la position oude la vitesse du corps attirant, il s'agira de cette position ou de cette vitesse à l'instant où l'onde gravifique est partie de

ce corps (…)» [Italics of the author]• 50’s-60’s: back in the footlights GW theoretical framework developped (Pirani & Isaacson)

Do gravitational waves exist?

• GW existence predicted by Einstein in 1918

• A difficult first appearance Validity of the General Relativity linearization ?!

«GW travel at the speed of mind » Sir A.S. Eddington

• First «imagined» by Poincaré in 1905

• The breakthrough: the binary pulsar PSR 1913+16 (1974) Indirect evidence that GW exist Hulse & Taylor (Nobel 1993) [& Damour]

20 years of measurementYes they do!

GW main characteristics

• Perturbations of the Minkowski metric• Quadrupolar emission

• Extremely weak!!! Luminosity G/c5 10-53 W-1

Ex: Jupiter radiates 5.3 kW as GW during its orbital motion over 1010 years: EGW = 2 1021 J Ekinetic 2 1035 J

• A good source of GW must be: asymetric compact (R ~ RSchwartzchild = 2GM/c2) relativistic

No Hertz experiment possible!Astrophysical sources required

GW detectable effect

GW effect : differential modification of lengths

L L + L

LL h )(2)( tt

The detector sensitivity volume should ultimately extendbeyond the Virgo cluster (~ 20 Mpc 65106 light years)

h: dimensionless amplitudeh 1 / distance

Two main categories of detectors: • resonant bars• giant interferometers, Earth-based or space-based Virgo LISA

A very large GW frequency domain

• Extremely Low Frequencies 10-18 10-15 Hz

• Very Low Frequencies 10-9 10-7 Hz

• Low Frequencies 10-4 10-1 Hz

• High Frequencies 1 104 Hz

LISA

Earth-based detectorsResonant bars or IFOs

CMB polarization

Pulsar timing

Frequency Range GW ‘Probe’

• First GW detectors: Joe Weber’s pioneering work – see Phys. Rev. 117 360 (1960)• Resonator: supraconducting coupled with cylindrical bar a transducer

• Network of bars working for years with high duty cycles• Narrow-band sensitivities limited by noises difficult to beat

Resonant bars

GWdepositenergyinside

the bar

Vibrationsmodulate

DC voltage

x mM x Mm

Interferometric detection

Incident GW Optical pathmodification

Variation of thepower Pdet at theIFO output port

Sensitivity : BS on incident Power length Arm

1 h sens

SuspendedMichelson

Interferometer

Mirrors used astest masses

The Virgo optical scheme

To increase the arm length : 1 m 3 km To add Fabry-Perot cavities (Finesse = 50 Gain = 30) To add a recycling mirror (P = 1 kW on the Beam Splitter)

Sensitivity :Sensitivity : hsens ~ Hz /DetectionPhotodiode

Laser

Gain :Gain : 3000 30 50 ~ 106

10-173 10-2110-2310-22

White

fringe

Laser power: Pin = 20 WSensitivity

in P / 1 h

Dual role:

• Passive seismic isolation

• Mirror active control only 0.4 N needed for a 1 cm motion

The Virgo SuperAttenuator

Length ~ 7 m; Mass ~ 1 tonStructure in inverted pendulum

Seismic Attenuation:~ 1014 at 10 Hz

fres ~ 30 mHz

lg m

k 2π1 f res -

INFNPisa

Virgo foreseen sensitivity

Minimum ~ 3 10-23 between ~ 500 Hz et 1 kHz Hz /

«Seismic Wall»

Thermalnoise

Tail of the0.6 Hz marionetta/mirror resonance

Shot noise

Thermalnoisemirrors

Violinmodes

Full Virgoconfiguration

The Virgo detector

Half-Arm Buildings

1.5 km 1.5 km

North ArmWest Arm

3 km 3 km

Mode-Cleaner144 m

Central BuildingControlBuilding

Virgo in numbers

• Arm length: 3 km 6800 m3 in ultra-high vacuum (10-10 mbar)

• Very high quality mirrors: Diffusion < 5 ppm, absorption < 1 ppm

Reflectivity > 99.995% Radius of curvature 3450 m (4.5 m sagitta)

• Laser power: 20 W

• Seismic noise attenuation: > 1014 above 10 Hz

• Foreseen sensitivity range: 4 Hz 10 kHz Best sensitivity ~ 3 10-23 / Hz around 1 kHz

• Control accuracy Length: down to 10-12 m Angular: from 10-6 to 10-9 radians

Fabry-Perotend mirrors

Status of Virgo• Spring 2001-Summer 2002:Successful commissioning of the central interferometer (CITF)

CITF: Virgo without the 3-km Fabry-Perot armsBut :

Same suspensions Same control chain

Ideal benchmark for the complete Virgo interferometer

• From autumn 2002: upgrade to Virgo • March 2003: first beam in the 3-km arm

• The Full Virgo commissioning will start after summer

• First Physical Data: 2004 or a bit later…

Virgo central interferometer (CITF)

• CITF commissioning = 1rst step of Virgo commissioning• Recycled and suspended Michelson Interferometer• Uses the technology developped for the Virgo control system• CITF commissioning goals:

check the different component performances validate control algorithms test data management (acquisition, storage…)

Armlengths~ 6 m

The CITF is not sensitive enough: no hope to collect data with GW signal!!!

«North»Mirror

«West» Mirror

RecyclingMirror

CITF and working point

Best sensitivity :

• Michelson on dark fringe control arm asymmetry: l2-l1

• Recycling cavity resonant (maximize the stored power) control IFO mean length: l0 + (l1+l2)/2

Very narrow Working Point

In addition: residual low frequency motion of mirrors (0.6 Hz)

CITF active controls needed (local and global)

Goal : Longitudinal control«Locking »

Resonant cavitiesl ~ 10-10 – 10-12 m

Angular control«Alignment »

Aligned mirrors ~10-9 – 10-7 rad

The steps of the Virgo control

• Decreasing the residual motion separately for each mirror Local controls + First alignment of mirrors

• Lock acquisition of the cavities

• Check working point control stability

• Switch on the angular control Automatic Alignment

Switching fromlocal controls

toglobal controls

Control aim: to go from an initial situation withrandom mirror motions to the Virgo working point

Cavity Control

LM1 (r1, t1) M2 (r2, t2)

Characteristic quantity: the finesse F r r - 1r r 2 1

2 1 F

• Linear around resonance

• Linear region width 1 / F

• Slope increasing with F

A finesse of 400 (aligned CITF)is high for a suspended cavity

Pound-Drever error signal

The higher F, the moredifficult the cavity control

FabryPerotcavity

GlobalControl

First control of the Michelson

Fringe interval~ 0.5 m

Error signal

Interferometerpower output

Fringe Counting

Time (s)

Time (s)

Time (s)

AC Power

DC Power

Dark fringe

June 13th 2001

• Pmax ~ 5.8 W Gain ~ 70 (Plaser ~ 80 mW)

• Dark fringe less «dark» unperfect contrast

• Large fluctuations of the stored power:

low feedback gain misalignments

December 16th 2001

IFO outputpower

StoredPower

West correction

Recyclingcorrection

A complex problem:• Two lengths to be controlled instead of one coupled error signals• Narrow resonance of the recycling cavity (high finesse)• Limited force available to act on mirrors• Error signal ~ to the electronic noise outside resonance[weak laser power + Recycling mirror reflectivity = 98.5%]

Main issues: • To select the right resonance [trigger on the stored power]• Simultaneous acquisition of the 2 cavity controls• Fast damping of the 0.6 Hz pendulum resonance excited each time the locking attempt fails

First control of the recycled CITF

CITF main steps

• 5 Engineering Runs• 3 days duration (24h/24h)• ~ 1 TB data collected / Engineering Run ~ 5 MBytes/s ~ 160 TB/an

• The 2 first in Michelson configuration (9/01 and 12/01)• The 3 others Recycled configuration (4/02, 5/02 and 7/02)

Channel type

«Physics»

Control

Monitoring

Data fraction

2 % 61 % 37 %

Engineering Run

ER0 ER1 ER2 ER3 ER4

Duty Cycle 98%

85%

98%

96%

77%All sources of control losses

understood Improvements in progress

CITF sensitivity improvements

ERBest

Sensitivity m/Hz

E08 10-12

(@ 500 Hz)

E15 10-12

(@ 500 Hz)

E210-14

(@ 1 kHz)

E35 10-15

(@ 1 kHz)

E410-16

(@ 1 kHz)

Factor 103

improvement@ 10 Hz Factor 105

improvement@ 1 kHz

June 2001 July 2002

Room formany more

Improvements

Virgo foreseensensitivity

From the CITF to the full Virgo• CITF commissioning completed

• Large improvements in sensitivity in only one year

Gain in ‘experimental experience’ many upgrades for Virgo

CITF Virgo will provide ‘free’ sensitivity improvements:• Arm length: 6 m 3 km gain of a factor 500 in h• Fabry-Perot cavities: factor 30 in addition• Reduction of laser frequency noise

In reality, such gains are unfortunately not automatic:• Some noises do not depend on the laser optical path• Noise hunting is a very long work

Virgo scheme more complicated (4 lengths instead of 2) Control acquisition procedures from CITF (under study)

Virgo can benefit from the other detector experiences

Virgo versus other interferometers

LIGO

TAMA

June-August 2002

Virgo CITF

July 2002

• All sensitivities in m/Hz Comparable plots!

• Improvements still needed!

• Record sensitivity: Tama 10-18 m/Hz @ 1 kHz

• @ 10 Hz, the CITF has the best sensitivity: 10-13 m/Hz

10 Hz 10 kHz

5 kHz1 Hz

10 kHz

October-November 2002

10-20

10-20

10-12

10-7

1 Hz10-20

10-7

One word about LISA

• Earth-based detectors limited by seismic noise below few Hz

• Strong sources certainly exist in the mHz range

• Constellation of 3 satellites• 3 semi-independent IFOs• Optimal combinations to maximize SNR or study noise• Search periodical sources• Expected lifetime: 5 years •Approved by NASA/ESA• To be launched in 2011

Seismic wall

Preparing the GW Data Analysis

• Activity parallel to the experimental work on detectors 1 international conference / year (GWDAW)

• Large number of potential GW sources: compact binary coalescences (PSR 1913+16) black holes supernovae pulsars stochastic backgrounds …

• The corresponding signals have very different features various data analysis techniques

Coincidence detections

Why ?

• Some detectors will be working in the future

LIGO : 4 km

VIRGO : 3 km

GEO : 600 m

TAMA : 300 m

ACIGA : 500 m

• Coincidence = only way to separate a GW (‘global’ in the network) from transient noises in IFOs• Coincidences may allow to locate the source position in sky• Coïncidences with other emissions: ,

now ACIGA

Interferometer angular response

Reduction of a factor ~ 2 in average of the amplitude

• 2 maxima GW perpendicular to detector plane

• 4 minima blind detector! e.g. when the GW comes along the arm bissector

Right ascension

Decl

inati

on

Example of the Virgo-LIGO network

• Spatial responses in a given direction

• Similarities between the maps of the two LIGO interferometers

• Complementarity Virgo / LIGO

Good coverage of the whole sky

Double or triple coincidences unlikely

Summary

• Many interferometers are currently under developpement Worldwide network in the future

All instruments work already although they did not prove yet there can fulfill their requirements

Control of complex optical schemes with suspended mirrors

All sensitivities need to be significally improved to reach the amplitude of GW theoretical predictions

• Many different GW sources various data analysis methods in preparation

• In the two last years, the Virgo experiment became real The different parts of the experiment work well together Successful commissioning of the CITF 2003: CITF Full Virgo First ‘physically interesting’ data expected for 2004 !?!?!

GW: a never ending story

The future of gravitational astronomy looks bright. 1972

That the quest ultimately will succeed seems almost assured.The only question is when, and with how much further effort. 1983

[I]nterferometers should detect the first waves in 2001 or several years thereafter (…) 1995

Kip S. Thorne

Km-scale laser interferometers are now coming on-line, and it seems very likely that they will detect mergers of compactbinaries within the next 7 years, and possibly much sooner.

2002

References about Virgo and GW• Virgo web site: www.virgo.infn.it

• Virgo-LAL web site (burst sources): www.lal.in2p3.fr/recherche/virgo

• Source review: C. Cutler - K.S. Thorne, gr-qc/0204090

• Some other GW experiment websites:

LIGO: www.ligo.caltech.edu GEO: www.geo600.uni-hannover.de TAMA: www.tamago.mtk.nao.ac.jp/tama.html IGEC (bar network): igec.lnl.infn.it LISA: sci.esa.int/home/lisa

• Moriond 2003: moriond.in2p3.fr/J03 «Gravitational Waves and Experimental Gravity» Recent status of all detectors: bars, IFOs and LISA

Detector noise characterization

Gaussian noise characterization: Power Spectrum Density (PSD)

• If the noise is dimensionless, the PSD unit is Hz-1

• RMS in the bandwidth [f1;f2]:

• Amplitude Spectrum Density (unit )

)(A FT 2 (f)S n n FT: Fourier Transform

one-sided PSD (only positive frequencies)

Tdt )n(t n(t)

lim )(A T/2

T/2

T n

with Autocorrelation function

ff df (f)S f ;f RMS 2

1n 2 1

(f)S (f)s~ nn Hz1/

DetectorSensitivity:

Frequency (Hz)

Sn

orsn~

Log-logscalesgraph

Compact binary coalescences

Example: PSR 1913+16 Coalescence expected in a few hundred million years Virgo will (?!?) be sensitive to the last minutes…

Waveform analytically estimated by developments in v/c Wiener filtering used for data analysis Optimal but computationally expensive

Chirp signal: amplitude and frequencyincrease with time untilthe final coalescence

The signal knowledge endsbefore the coalescencewhen approximations usedfor the computation areno more valid. large theoretical work to go beyond this limit!

Impulsive sources (‘bursts’)

Examples:• Merging phase of binaries• Supernovae• Black hole ringdowns

GW main characteristics:• Poorly predicted waveforms model dependent• Short duration (~ ms)• Weak amplitudes

Need to develop filters : robust (efficient for a large class of signals) sub-optimal (/ Wiener filtering) online (first level of event selection)

Zwerger/ Müller

examples ofsimulatedsupernovaGW signals

Pulsars

• GW signal: permanent, sinusoidal, possibly 2 harmonics

• Weak amplitude detection limited to the galaxy

• Matched filtering-like algorithms using FFT periodograms

• Idea: follow the pulsar freq. on large timescales (~ months) compensation of frequency shifts: Doppler effect due to Earth motion, spindown…

• Very large computing power needed (~ 1012 Tflops or more) Hierarchical methods are being developped 1 TFlop Need to define the better strategy:

search only in the Galactic plane, area rich of pulsars uniform search in the sky not to miss close sources focus on known pulsars

• Permanent signal coincident search in a single detector: compare candidates selected in 2 different time periods

• Described by an energy density per unit logarithmic frequency normalized to the critical density of the universe:

• Two main origins: Cosmological Emission just after the Big Bang: ~10-44 s, T~1019 GeV Detection informations on the early universe Astrophysical Incoherent superposition of GW of a given type emitted by sources too weak to be detected separately.

• Detection requires correlations between 2 detectors• After 1 year integration: h0

2 stoch 10-7 (1rst generation) 10-11 (2nd generation)• Theoretical predictions: ~ 10-13 10-6

• Current best limit: stoch 60 @ 907 Hz [Explorer/Nautilus]

Stochastic backgrounds

S stoch stoch ρG

(f)fπ (f) Ωc2

3

GH

c 83 2

0with