Table Of Content

aLIGO Input Optics The Advanced LIGO Input Optics Chris L. Mueller,1,a) Muzammil A. Arain,1,b) Giacomo Ciani,1 Ryan. T. DeRosa,2 Anamaria Effler,2 David Feldbaum,1 Valery V. Frolov,3 Paul Fulda,1 Joseph Gleason,1,c) Matthew Heintze,1 Eleanor J. King,4 Keiko Kokeyama,2 William Z. Korth,5 Rodica M. Martin,1,d) Adam Mullavey,3 Jan Poeld,6 Volker Quetschke,7 David H. Reitze,1,e) David B. Tanner,1 Luke F. Williams,1 and Guido Mueller1 1)University of Florida, Gainesville, FL 32611, USA 2)Louisiana State University, Baton Rough, LA 70803, USA 3)LIGO Livingston Observatory, Livingston, LA 70754, USA 4)University of Adelaide, Adelaide, SA 5005, Australia 5)LIGO, California Institute of Technology, Pasadena, CA 91125, USA 6)Max-Planck-Institut für Gravitationsphysik, 30167 Hannover, Germany 6 7)University of Texas at Brownsville, Brownsville, TX 78520, USA 1 0 (Dated: January 22, 2016) 2 The Advanced LIGO gravitational wave detectors are nearing their design sensitivity and should n begintakingmeaningfulastrophysicaldatainthefallof2015. Theseresonantopticalinterferometers a will have unprecedented sensitivity to the strains caused by passing gravitational waves. The input J optics play a significant part in allowing these devices to reach such sensitivities. 0 Residing between the pre-stabilized laser and the main interferometer, the input optics is tasked 2 with preparing the laser beam for interferometry at the sub-attometer level while operating at continuous wave input power levels ranging from 100 mW to 150 W. These extreme operating ] t conditions required every major component to be custom designed. These designs draw heavily e d on the experience and understanding gained during the operation of Initial LIGO and Enhanced - LIGO. In this article we report on how the components of the input optics were designed to meet s theirstringentrequirementsandpresentmeasurementsshowinghowwelltheyhaveliveduptotheir n i design. . s c si I. INTRODUCTION servatories. This paper focuses on the input optics y of aLIGO. h Aworldwideefforttodirectlydetectgravitational The main task of the input optics (IO) subsys- p radiation in the 10Hz to a few kHz frequency range temistotakethelaserbeamfromthepre-stabilized [ with large scale laser interferometers has been un- laser system1 (PSL) and prepare and inject it into 1 derway for the past two decades. In the United the main interferometer (IFO). The PSL consists v States the Laser Interferometer Gravitational-Wave of a master laser, an amplifier stage, and a 200W 2 Observatories (LIGO) in Livingston, LA (LLO) and slave laser which is injection-locked to the ampli- 4 in Hanford, WA, (LHO) have been operating since fied master laser. The 200W output beam is fil- 4 5 the early 2000’s. Initial and enhanced LIGO pro- tered by a short optical ring cavity, the pre-mode 0 duced several significant upper limits, but did not cleaner, before it is turned over to the IO (see Fig- . havethesensitivitytomakethefirstdirectdetection ure1). ThePSLpre-stabilizesthelaserfrequencyto 1 ofgravitationalwaves. Duringthistimeofoperation a fixed spacer reference cavity using a tunable side- 0 6 a significant amount of effort was invested by the band locking technique. The PSL also provides in- 1 LIGOScientificCollaborationtoresearchanddesign terfacestofurtherstabilizeitsfrequencyandpower. v: Advanced LIGO (aLIGO), the first major upgrade The IFO is a dual-recycled, cavity-enhanced i of initial LIGO. In 2011 the initial LIGO detectors Michelson interferometer2 as sketched in Figure 2. X were decommissioned and installation of these up- The field enters the 55 m folded power recycling r grades started. The installation was completed in cavity (PRC) through the power recycling mirror a 2014 and the commissioning phase has begun for (PRM). Two additional mirrors (PR2, PR3) within many of the upgraded subsystems at the LIGO ob- the PRC form a fairly fast telescope to increase the beam size from 2 mm to 50 mm (Gaussian beam ∼ ∼ radius) before the large beam is split at the beam splitter(BS)andinjectedintothetwo4kmarmcav- a)Electronicmail: [email protected]fl.edu ities formed by the input and end test masses. The b)PresentAddress: KLA-Tencor,Milpitas,CA95035,USA reflected fields recombine at the BS and send most c)PresentAddress: GetCurrentAddress1 d)PresentAddress: GetCurrentAddress2 ofthelightbacktothePRMwhereitconstructively e)Present Address: LIGO Laboratory, California Institute of interferes with the injected field3. This leads to a Technology,Pasadena,CA91125,USA power enhancement inside the power recycling cav- aLIGO Input Optics 2 ETM ML Slave Laser 4km ~16m from IO PRM PR2 ITM 4km ITM ETM AOM EOM FI PR3 ~25m BS ~ SR2 IO: Input Optics VCO PRM: Power Recycling Mirror ~25m ~16m PMC BS: Beam Splitter Ref. ITM: Input Test Mass ETM: End Test Mass SR3 SRM SRM: Signal Recycling Mirror Ampl. to output optics to Input Optics Figure 2. Sketch of the main interferometer which con- ML: Master Laser FI: Faraday Isolator sists of two 4 km arm cavities, the beam splitter, and EOM: Electro−optic modulator Ref.: Reference Cavity thefolded55mlongpowerandsignalrecyclingcavities. AOM: Acousto−optic modulator PMC: Pre−mode cleaner Theinputopticsislocatedbetweenthissystemandthe VCO: Voltage Controlled Osc. Ampl.: Amplifier PSL shown in Figure 1. Figure1. Sketchofthepre-stabilizedlaser(PSL)system. Red: Main beam, Green: Pick-off beam. The figure II. OVERVIEW OF THE INPUT OPTICS shows the low power master laser, the phase-correcting EOM,theamplifierstage,ahighpowerFaradayisolator, and the high power slave laser. The pre-mode cleaner Figure 3 shows a sketch of the first part of the suppresseshigherorderspatialmodesofthelaserbeam. input optics. This part is co-located with the PSL TheVCOdrivestheAOMwhichshiftsthefrequencyof on the same optical table inside the laser enclosure, the pick-off beam to stabilize the sideband and with it outside of the vacuum system. It prepares the laser the frequency of the laser to the reference cavity. beamfortheinjectionintothevacuumsystem. The beam from the PSL is first routed through a half waveplateandapolarizingbeamsplitter. Thesetwo elements form a manual power control stage which is used mainly during alignment processes on the optical table. The following mirror transmits 2.5% of the light. This light is used by the arm length ity and provides additional spatial, frequency, and stabilization (ALS) system during lock acquisition amplitude filtering of the laser beam. The second outputoftheBSsendslightintothe55mlongfolded of the main interferometer5. signalrecyclingcavity4 (SRC)whichalsoconsistsof Most of the light is sent through an electro-optic abeamreducingtelescope(SR2,SR3)andthesemi- modulator(EOM)whichmodulatesthephaseofthe transparent signal recycling mirror (SRM). laser field with three different modulation frequencies. Two of these frequencies are used by the interferometer sensing and control system (ISC) to sense Thispaperisorganizedasfollows: SectionIIgives most of the longitudinal and alignment degrees of an overview of the IO; its functions, components, freedom of the mirrors inside the IFO and to stabi- and the general layout. Section III discusses the re- lizethelaserfrequencyandthealignmentofthelaser quirements for the IO. Section IV presents the core beamintotheinterferometer. Thethirdfrequencyis of this paper; it will describe individual IO compo- used to control the input mode cleaner (IMC). The nents,theirperformanceinpre-installationtestsand two lenses L1 and L2 mode match the beam to the the detailed layout of the IO. Section V discusses in-vacuum input mode cleaner. The next steering the expected and measured in-vacuum performance mirror directs the beam through another half wave as known by the time of writing. The final inte- plateinsideamotorizedrotationstageinfrontoftwo grated testing of the IO subsystem at design sensi- thinfilmpolarizers. Thissecondpowercontrolstage tivity requires the main interferometer to be nearly is used during operations to adjust the power to the fully commissioned to act as a reference for many of requested level. The periscope raises the height of the required measurements; this will be discussed in thebeamandsteersitintothevacuumsystem. The Section VI. top mirror is mounted on a piezo actuated mirror aLIGO Input Optics 3 togALS thebeamwhichisreflectedfromtheIFO.Thelatter EOM L1 L2 two beams are routed to IOT2R7, an optical table PBS onthe rightsideofHAM2, while thefirst beam and HWP RWP the field which is reflected from MC1 are routed to IOT2LontheleftsideofHAM2. Thepositionofthe DCPD fromgPSL RFPD forward going beam through IM4 is monitored with TFP Diag. an in-vacuum quadrant photo detectorwhilea large BD fraction of this beam is also sent to an in-vacuum photo detector array which is used to monitor and DCPD togVacuum Periscope stabilize the laser power before it is injected into the IFO. Most of the IFO reflected field goes back ALS:gArmglengthgstabilizationgsystem L1vgL2:gLenses to the FI where it is separated from the incoming BD:gBeamgdump RFPD:gFastgphoto,detector beam. This field is routed into HAM1 where it is DCPD:gPhoto,detector RWP:gRotatingghalf,wavegplate Diag:gtogDiagnostics TFP:gThingfilmgpolarizer detected to generate length and alignment sensing EOM:gElectro,opticgmodulator :gWatergcooledgbeamgdump signals. HWP:gHalf,wavegplate In HAM3, a small fraction of the intra-mode cleaner field transmits through MC2 onto a quad- Figure 3. The IO on the PSL modulates the phase of rantphotodetectortomonitorthebeampositionon the laser beam with the EOM, mode matches the light MC2. The forward and backwards traveling waves into the input mode cleaner (located inside the vacuum inside the PRC partly transmit through PR2 and system), and controls the power injected into vacuum areroutedintoHAM1andtoanopticalbreadboard system. inside HAM3, respectively. These beams are used by ISC for sensing and control of the interferometer and for diagnostic purposes. The breadboard holds mount to fine tune the alignment of the beam into two lenses which image the beam with orthogonal the vacuum chamber. Gouy phases onto two quadrant photo detectors to Between the lenses is a wedge to pick off a small monitorbeampositionandpointinginsidethepower fraction of the laser beam for diagnostic purposes. recycling cavity. IOT2R and IOT2L host photo de- A fast photo detector monitors the residual ampli- tectorsanddigitalcamerastomonitorthepowerand tude modulation at the phase modulation frequen- beam sizes in each of the picked-off beams. IOT2L cies while a second photodetector monitors the DC alsohoststhephotodetectorswhichareusedbythe power. A fraction of the main beam also transmits interferometer sensing and control system to gen- through the bottom periscope mirror and is used to erate length and alignment sensing signals for the monitor the power going into the vacuum system as input mode cleaner. well as the size, shape, and quality of the beam. While the figure shows all key components in the Following the periscope, the main beam is sent correct sequence, we intentionally left out the de- through a metal tube which includes a mechani- tailed beam routing, the baffles used to protect all cal shutter and through HAM16 into HAM2; all in- componentsfromthelaserbeamincaseofmisalign- vacuum IO components are mounted on seismically ments, and the beam dumps to capture all ghost isolated optical tables inside HAM2 and HAM3. As beams. shown in Figure 4, the beam passes over the FI to A complete document tree which contains all de- a second periscope which lowers the beam to the in- sign and as-built layouts as well as drawings of all vacuum beam height. The next element in the IO is componentsisavailablewithintheLIGODocument the suspended IMC, a 33 m long triangular cavity. Control Center8 (DCC) under document number The two flat input and output mirrors, named MC1 E12010139. and MC3 respectively, are located in HAM2 while the third curved mirror, MC2, is located in HAM3. FollowingMC3aretwosuspendedmirrors, IM1and IM2, whichsteerthebeamthroughtheFI.IM3and III. INPUT OPTICS REQUIREMENTS IM4 are used to steer the beam into the PRC. IM2 andIM3arecurvedtomodematchtheoutputmode TheaLIGOinterferometercanbeoperatedindif- of the IMC to the mode of the main interferometer. ferent modes to optimize the sensitivity for differ- Two of the steering mirrors, IM1 and IM4, trans- ent sources10. These modes are characterized by mit a small fraction of the light creating three dif- the input power and the microscopic position and ferent auxiliary beams which are used to monitor reflectivity of the signal recycling mirror. The re- the power and spatial mode of the IMC transmit- quirements for the aLIGO input optics are specified ted beam, of the beam going into the IFO, and of to simultaneously meet the requirements for all an- aLIGO Input Optics 4 HAM 2 HAM 3 to IOT2L IM1 QPD MC3 MC2 IM3 QPD to HAM 1 ISC−Sled MC1 FI to HAM 1 from PSL PRM IM4 QPD PR3 PR2 IM2 to BS MC: Mode Cleaner mirror (suspended) IM: Input Optics Mirror (suspended) QPD: Quadrant Photo Detector PSL−PD FI: Faraday Isolator to IOT2R BS: Beam Splitter (Core Optics) Figure 4. A sketch of the in-vacuum components and beam directions within the input optics in HAM 2 and HAM 3. Theredbeamistheforwardgoingmainbeamwhilethegreenbeamsareauxiliarybeams. Themainitemsinthe in-vacuum input optics are the input mode cleaner (IMC) which is formed by the three mirrors MC1, MC2, MC3; theFaradayisolator(FI);andthefoursuspendedsteeringmirrorsIM1-4ofwhichIM2and3matchthespatialmode oftheIMCintothemaininterferometer. TherecyclingcavitymirrorsPRM,PR2,andPR3arenotpartoftheinput optics. The ISC sled in HAM3 belongs to the interferometer sensing and control subsystem and provides alignment signals for the recycling cavity. ticipated science modes and address all degrees of inallopticalcomponentsbutalsolimitstheallowed freedom of the laser field. Requirements in aLIGO thermal lensing in the EOM, the FI, and the power are defined for three distinct frequency ranges: DC, control stages; the reflective optics and fused silica the control band up to 10 Hz, and the signal or de- lenses are much less susceptible to thermal lensing. tection band from 10 Hz to a few kHz. The require- Efficient power coupling is also dependent on good ments in the detection band are defined in terms of mode matching between the recycling cavities and linear spectral densities and include a safety factor the arm cavities in the main interferometer. of ten for all technical noise sources. To first or- Power Control: The injected power into the in- der, a perfectly symmetric Michelson interferometer terferometer has to be adjustable from the control is insensitive to all input noise sources which is an roomfromminimumtofullpowerfordiagnosticand often overlooked reason for its use in the first place. operationalpurposes,toacquirelockofthemainin- However, all degrees of freedom of the injected laser terferometer, and to operate between different sci- field couple via some asymmetry to the output sig- ence modes. The rate of power change (dP/dt) has nal. Thisdrivestherequirementsinthecontrolband to be sufficiently small to limit the radiation pres- which are usually defined as RMS values. The more sure kick inside the IMC and the main interferome- critical requirements for the IO are: ter to a level that can be handled by the length and Power: Thehighpowersciencemodesrequireto alignment control system. It has to be sufficiently inject125Wofmodematchedlightintotheinterfer- fast to not limit the time to transition to full power ometerwithlessthananadditional5%inhigheror- after lock acquisition; i.e. it should be possible to dermodes. ThePSLhastodeliver165Woflightin change from minimum to maximum power within a anappropriateTEM mode. Consequently,thenet few seconds. 00 efficiencyofTEM opticalpowertransmissionfrom Notethatminimumpowerherecannotmeanzero 00 thePSLoutputtothemaininterferometerhastobe power because of the limited extinction ratio of po- above 75%. This sets limits on accumulated losses larizers. Going to zero power requires actuation of aLIGO Input Optics 5 the aforementioned mechanical shutter which can ties are not a good reference and are made to follow only be accessed manually between the laser enclo- thefrequencyreferenceinsideofthePSL.Theinput sure and HAM1. The emergency shutter is part of mode cleaner acts as a frequency reference during the PSL laser system and cuts the laser power at lock acquisition and as an intermediate frequency the source. Furthermore, the power control system reference during science mode. It is integrated into within the IO is not used for actively stabilizing the thecomplexandnestedlaserfrequencystabilization laserpowerwithinthecontrolorthedetectionband. system. Basedontheexpectedcommonmodeservo Power fluctuations: Fluctuations in the laser gain the frequency noise requirements for the IMC power can couple through many different channels are set to: to the error signal. The noise scales with the asym- metries in the interferometer. Two different mecha- mHz nismsareexpectedtodominatethesusceptibilityof δν(f =10Hz)<50√ Hz the interferometer to power fluctuations. The opti- mHz cal power inside the arm cavities will push the test δν(f ≥100Hz)<1√ . masses outwards. Any change in power will cause Hz fluctuations in that pressure which can lead to dis- These requirementscan be expressedequivalently placement noise at low frequencies. The suscepti- as length fluctuations of the IMC: bility to radiation pressure noise scales with differ- m ences in the power build up inside the arm cavi- δ(cid:96)(f =10Hz)<3·10−15√ ties and it is assumed that these differences are be- Hz m low 1%. At high frequencies, the signal itself lim- δ(cid:96)(f ≥100Hz)<6·10−17√ . its the allowed power fluctuations. In lock the two Hz arm cavities are detuned by a few pm which causes RF modulation frequencies: The main laser field some light to leak out to the dark port11. Gravita- consists of a carrier and multiple pairs of sidebands. tional waves will modulate these offsets causing the The carrier has to be resonant in the arm cavi- light power to fluctuate. Obviously, power fluctua- ties and the power recycling cavity; the resonance tions in the laser itself, although highly filtered by condition in the signal recycling cavity depends on the interferometer, will cause similar fluctuations. the tuning and specific science mode. One pair of The relative intensity noise in the detection band √ sidebands must be resonant in the power recycling has to be below 2×10−9/ Hz at 10 Hz increasing √ cavity while the second pair must resonate in both with f to 2 × 10−8/ Hz at 1 kHz and remaining the power and signal recycling cavity. The modu- flat after this12. Furthermore, the expected seismi- lation signals of f = 9.1MHz and f = 45.5MHz cally excited motion of the test masses limits the 1 2 are provided by the interferometer sensing and con- allowed radiation pressure noise in the control band √ trol system. A third modulation frequency of f = to 10−2/ Hz below 0.2 Hz. Above 0.2 Hz, the re- 24.1MHz is required to sense and control the in3put quirementsfollowtwopowerlaws;initiallyf−7 then mode cleaner. The last pair of sidebands should be f−3, before connecting with the detection band re- rejected by the input mode cleaner so as not to in- quirement at 10 Hz. terfere with the sensing and control system of the The IO does not provide any active element to main interferometer. change or stabilize the laser power within the con- RF modulation depth: The required modulation trol or the detection band. The PSL uses a first depths depend on the final length and alignment loop which stabilizes the laser power measured√with sensing and control scheme. This scheme is likely a photo-detector on the PSL table to 2×10−8/ Hz to evolve over the commissioning time but the cur- between 20 and 100 Hz and meeting the aforemen- rent assumption is that a modulation index of 0.4 tioned requirements above 100 Hz. The PSL also for a 24dBm signal driving the EOM is more than measures the power of the injected field with the sufficient. Note that this only applies to the two photo detector array shown (PSL-PD) in Figure 4. modulation frequencies which are used for sensing TheIOhastosupplythisauxiliarybeamandmain- and control of the main interferometer; the modula- tain a sufficiently high correlation with the injected tionindexforthethirdfrequencyneedonlybelarge beam and minimize the chances of additional power enough to control the IMC. fluctuations within any of these two beams. Theclassicphasemodulation/demodulationsens- Frequency fluctuations: In the detection band, ing scheme for a single optical cavity measures how the laser frequency will ultimately be stabilized to muchthecavityconvertsphasemodulationintoam- the common mode of the two arm cavities which plitude modulation when near resonance. Unfortu- are the most stable references available in this fre- nately all phase modulators also modulate the am- quency range. At lower frequencies the arm cavi- plitudeofthelaserfield. Thisamplitudemodulation aLIGO Input Optics 6 cansaturatetheRFamplifiersandmixersinthede- Additional requirements: It is well known that tectionchainandgenerateoffsetsintheerrorsignals parasiticinterferometersandscatteredlighttogether whichhavetobecompensated. aLIGOrequiresthat with mechanically excited surfaces can add fre- the amplitude modulation index is less than 10−4 of quency and amplitude noise to a laser beam. The the phase modulation index.13 IOadoptedapolicytolimittheaddednoiseto10% oftheallowednoise;notethattheallowedfrequency RF modulation noise: Changes in the amplitude andamplitudenoisepriortotheinputmodecleaner and phase of the RF modulation signals can change is significantly higher than after the mode cleaner. the dark port signal by changing the power build- This drives requirements on the residual motion of up of the carrier in the arm cavities or through the optical components, the surface quality of all cross coupling in the length and alignment sens- optical components and their coatings, and on the ing and control schemes. These effects were ana- placement and efficiency of the optical baffles. The lyzed by the ISC group10. The analysis uses speci- requirement to align the IO drives requirements on fications from a commercial crystal oscillator man- actuationrangesforallopticsand,lastbutnotleast, ufacturer as the expected oscillator phase and am- theIOhastomeetthestringentcleanlinessandvac- plitude noise. These specifications for phase noise √ uumrequirementsofaLIGO.Theserequirementsare are 10−5rad/ Hz at 10 Hz falling with 1/f3/2 to √ discussed throughout the paper when relevant. 3×10−7rad/ Hz at 100 Hz and then a little faster √ The official requirements were originally captured than1/f to2×10−8rad/ HzatakHzabovewhich in LIGO document T02002016 and have been de- the requirement stays constant. The specifications √ rived, updated and clarified in several other docu- for amplitude noise are 10−7/ Hz at 10 Hz falling ments which are all available under E1201013.9 with 1/f between 10 and 100 Hz and then with √ 1/ f until 1 kHz above which it stays constant √ at 3 × 10−9/ Hz. These specifications have been IV. INPUT OPTICS COMPONENTS AND FINAL adoptedasrequirementsalthoughtheanalysisshows LAYOUT that they could be relaxed at higher frequencies. Beam jitter: Changes in the location and direc- This section will first discuss the individual com- tion of the injected beam can be described as scat- ponents and their measured performance. This will tering light from the TEM into a TEM mode. be followed by a description of the optical layout 00 10 This light scatters back into the TEM mode in- whichincludesadiscussionofbeamparametersand 00 sideamisalignedinterferometerandcreatesnoisein mode matching between the various areas. the gravitational wave signal.14 This is an example where noise in the detection band, here beam jitter, couples to noise in the control band, here tilt of the A. Electro-optic modulators inputtestmasses. Itisexpectedthatthetestmasses will all be aligned to better than 2 nrad RMS with The EOM must use a material capable of with- respecttothenominalopticalaxisoftheinterferom- standing CW optical powers of up to 200 W and eter. Under this assumption, the relative amplitude √ intensities up to 25kW/cm2. At these power levels oftheinjected10-modehastostaybelow10−6/ Hz theinducedthermallensing,stressinduceddepolar- at 10 Hz falling with 1/f2 until 100 Hz above which √ ization, and damage threshold of the electro-optic the requirement stays constant at 10−8/ Hz. material must be taken into consideration. Rubid- Optical isolation: The FI isolates the IMC from iumtitanylphosphate(RTP)waschosenmanyyears back reflected light from the main interferometer. ago over other electro-optic materials, such as ru- The requirements for the isolation ratio are based bidium titanyl arsenate (RbTiOAsO4 or RTA) and on experience gained during the initial years of op- lithium niobate (LiNb03), as the most promising erating LIGO and also VIRGO. Virgo operated for modulator material after a literature survey, dis- a long time without a FI between the mode cleaner cussions with various vendors, and corroborating and the main interferometer and encountered prob- lab experiments.17,18 RTP has a very high damage lems due to the uncontrolled length between the threshold, low optical absorption, and a fairly high IMC and IFO15 (a parasitic interferometer). Ini- electro-opticalcoefficient. EnhancedLIGO(eLIGO) tial and enhanced LIGO never encountered any ma- allowed for testing of the material and design over a jorproblemswithinsufficientopticalisolationinthe one-year period at 30 W input power.19 FI. The requirements of 30dB for the optical isola- The aLIGO EOM uses a patented design20 which tion in the FI were set based on the experience in is very similar to the one used in eLIGO; both con- initialLIGO,takingintoaccountthehigherinjected sist of a 4 × 4 × 40mm long wedged RTP crystal power. (see Figure 5). The 2.85◦ wedges prohibit parasitic aLIGO Input Optics 7 EOM Resonances Lower Sideband IMC Sideband Upper Sideband LLO 0.4 LHO 0.35 Modulation Index0000..12..2355 0.1 0.05 0 8 9 10 22 23 24 25 44 45 46 47 Frequency (MHz) Frequency (MHz) Frequency (MHz) Figure 6. The measured modulation indices for both Figure 5. Images of the EOM. The housing uses two theLivingstonandHanfordEOMwitha24dBmdrive. modules;thecrystalandtheelectrodesareplacedinthe Thedataisshowntogetherwithabestfittotheexpected lowermodulewhiletheuppermodulehousesthecoilsfor circuit response. the three resonant circuits. The left picture shows the inside of the lower module: the aLIGO EOM consists of a wedged RTP crystal with three pairs of electrodes. The two 15 mm electrodes on the outside are used for The residual amplitude modulation (RFAM) pro- themainmodulationfrequenciesf =9.1MHzandf = duced by the EOM was also characterized. The 1 2 45.5MHz. The 7 mm electrodes in the middle are used AM/PM ratio for each of the three sidebands was for f3 = 24.1MHz. The crystal and the electrodes are measured to be 1.0·10−4, 1.2·10−5, and 4.1·10−5 clamped between two macor pieces. The right picture forthe 9.1MHz, 24.0 MHzand45.5MHzsidebands shows thefinalmodulator(both modules)on afive axis respectively. All three measurements come out to alignment stage. be at or below the requirement of 10−4 derived by Kokeyama et. al.13 The temperature dependence of the RFAM generation was briefly investigated and interferometers from building up inside the crystal was found to be able to push the AM/PM ratio at and allow for separation of the two polarizations of 9.1 MHz up as high as 3·10−4. This may need to the injected laser field with an extinction ratio of be addressed with a temperature stabilization servo betterthan105. Thisseparationavoidspolarization inthefutureifRFAMisfoundtobeanissueduring rotation which could otherwise convert phase mod- detector commissioning, but the design of the modulation to amplitude modulation. The AR coated ulator was left unchanged until such an issue arises. surfaces have a reflectivity of less than 0.1%. For Detailed design drawings, assembly instructions, aLIGO we use two 15 mm long pairs of electrodes andtestreportsareavailableunderLIGOdocument for the two main modulation frequencies and one number T1300084.21 7 mm long pair for the auxiliary frequency used to controltheIMC.Eachelectrodepairformsacapac- itor which is part of a resonant circuit in the form of a π network where the additional inductor and B. Faraday isolator capacitor are used to simultaneously match the res- onancefrequencyandcreatetherequired50Ωinput TheFIisamuchmorecomplicatedopticaldevice impedance. comparedtotheEOM.Itismoresusceptibletother- Afterinstallationandalignmentatbothsites,ini- mal lensing and its location after the mode cleaner tial tests confirmed that the RTP crystals do not amplifiestherequirementtomaintainagoodspatial produce a significant thermal lens. An optical spec- mode. TheFIhastohandlebetween20to130Wof trum analyzer was used to measure the modulation laser power without significantly altering the beam indexasafunctionofmodulationfrequencyforeach profile or polarization of the beam. Like the EOM, ofthethreeresonantcircuits. Theresultsareshown the aLIGO FI is also very similar to the FI used in Figure 6. The modulation indices for f and f in eLIGO19. Both were designed to minimize and 1 2 meet the requirements at both sites while the f mitigatethermallensingandthermalstressinduced 3 modulation index is still a little low, especially at depolarizationbycompensatingtheseeffectsinsub- LLO. However, early commissioning experience in- sequent crystals22,23. dicatesthatthemodulationindicesaresufficientfor The aLIGO FI design consists of a Faraday ro- theaLIGOlengthandalignmentsensingschemeand tator, a pair of calcite-wedge polarizers, an element itwasdecidedtousetheEOMasisfornowandpo- with a negative dn/dT for thermal-lens compensa- tentiallyimprovetheresonantcircuitslaterifneces- tion, and a picomotor-controlled half-wave plate for sary. restoring the optical isolation in-situ. In addition, a aLIGO Input Optics 8 G G DKDP crystal were procured from the manufactur- TG QR TG erswiththeirstandardpolishingsandcoatings. The C W P H W P H DKDP C W P twocalcitepolarizerseachhaveathicknessof∼5mm andarewedgedat8.5◦ toallowtheorthogonallypo- larized beams to separate sufficiently. The calcite H wedges have an extinction ratio of at least 105 and more than 99% optical efficiency. -22.5° -22.5° +67.5° -22.5° = 0° The magnetic field is created by a stack of seven +22.5° -22.5° -67.5° -22.5° = -90° magnetized Fe-Nb magnetic disks25 each having a bore of 24 mm and a thickness of 19.7 mm. This CWP2 stack produces a maximum axial field of 1.16 T Faraday Rotator (LLO) and 1.55 T (LHO) near its center which falls off towards the end. The difference in the magnetic CWP1 DKDP field is caused by the selection of the magnetic ma- λ/2 plate terials and the thermal treatment of the individual magnets.26TheTGGcrystalsandquartzrotatorare installed about 3 cm apart from each other before being fine tuned to produce 22.5◦ of rotation by ad- justing their depth in the magnet. The entire FI is mounted on a 648 mm x 178 mm breadboard for convenienttransferintotheHAMchamberafterout- of-vacuum optimization. After undergoing a thorough cleaning procedure, theFIwasassembledandalignedwiththemainPSL beam in the laser enclosure. The optical table in Figure7. AdvancedLIGOFaradayisolator(fromtopto the enclosure is made from stainless steel while the bottom): optical layout, design, final product. opticaltableinHAM2ismadefromaluminum. The differences in magnetic susceptibility are significant enoughtorequiretheFItoberaisedwithan 11cm ∼ heatsinkisconnectedtotheholdersofthemagneto- thick granite block visible in the bottom picture in optical crystals to drain excess heat into the FI Figure 7. The bottom periscope mirror in Figure breadboard. The Faraday rotator is based on an 3 was removed and the beam was sent via several arrangement developed by Khazanov et. al.24, that mirrorsthroughtheFI.Thissettingensuredthatthe uses a pair of ∼1 cm long Terbium Gallium Garnet beam parameters, beam size and divergence angle, (TGG) crystals as magneto-optical elements, each are very similar to the ones expected in-vacuum. nominally producing a 22.5◦ rotation of the elec- The thermal lensing of the FI was determined tric field when placed in a magnetic field of about from beam-scan measurements of a sample of the 1T. They are separated by a 1 cm long piece of ∼ beam after it was transmitted through the isolator quartzthatrotatesthepolarizationfieldreciprocally forincidentpowersashighas120WatLLOand140 by67.5◦±0.6◦ . Thisarrangement(shownschemat- W at LHO. At both sites, the diagnostic beam was ically at the top of Figure 7) allows thermally in- focusedwithalensof1mfocallengthandthebeam duced birefringence produced in the first magneto- profilewasrecordedwithCCDorrotatingslitbeam opticalelementtobemostlycompensatedinthesec- scansasafunctionofpowerfordifferentDKDPcrys- ondone. TheHWPisazero-orderepoxy-freequartz tals. The thermal lens at the location of the FI was half-wave plate. It is set to rotate the polarization then computed using an ABCD matrix algorithm. byanadditional22.5◦ tohave0◦ netrotationinthe Figure 8 shows the thermal lensing measurements forward going and 90◦ in the backward going direc- for the TGG crystals and different DKDP crystals tion. at LHO and LLO. The length of the DKDP crys- All crystals were selected to minimize absorp- tal was chosen to compensate the a priori unknown tion, thermal beam distortion and surface rough- thermal lensing in the TGG crystals. Based on ex- ness. Those made of harder and non-hygroscopic periencefrominitialandenhancedLIGO,theexpec- materials; the half-wave plate, quartz rotator, and tationwasthatDKDPcrystalsbetween3.5mmand TGGcrystals,areallsuper-polished(surfacerough- 5.5mm would beneeded to compensatethe thermal ness below 0.5 nm) and received a custom low loss lensing in the TGG crystals. However, the absorp- IBS AR coating with a rest reflectivity of less than tioninthenewlypurchasedTGGcrystalswaslower 300 ppm. The softer calcite polarizers and the than expected and even our shortest crystals over- aLIGO Input Optics 9 Figure 8. Left graph: Thermal lens for various DKDP crystals measured in-air as a function of laser power. Right graph: Isolationratiomeasuredin-airatbothsites. ThepoweristheinjectedpowerwhilethepowerinsidetheFIis twice as high. Therefore 70 W incident power corresponds to ∼125 W injected power during science mode when the near impedance matched interferometer reflects less than 10% of the light. compensated. While the low absorption in TGG is and tooptimize the isolationratio in-situ forthe in- obviously good, it required to shorten the originally jectedpower. Thisin-situoptimizationisimportant ordered DKDP. We choose 3.5 mm for both isola- becauseitisplannedtooperateaLIGOwith 20W ∼ tors instead of the more optimum 3 mm because of injected power during the first science runs. of concerns that a thinner DKDP crystal might Historically, the FI has always been one of the fracture inside the vacuum chamber under thermal main sources of optical losses. The aLIGO FI con- stress. Both isolators meet the thermal lensing re- sistsofsevenopticalelementswithatotaloffourteen quirements for aLIGO. optical surfaces each contributing to the losses. The The isolation ratio was also measured as a func- TGGcrystalsandtheDKDPcrystalsarealsoknown tion of input power. To do so the transmitted beam for absorbing a non-negligible fraction of the light, was reflected back under a negligibly small angle to hence the thermal lensing (which also reduces the allow to separate the return beam from the incom- powerinthefundamentalmode). Thenextculpritis ing beam. The powers in the beam going into and thepolarizationrotationbetweenthetwopolarizers. through the FI, and in the return beams in both Ideally,theFIwouldhave0◦ intheforwardand90◦ polarizations were measured to determine and opti- rotationinthebackwarddirection. Thisisonlypos- mize the optical efficiency and the isolation ratio as sible when the TGG crystals provide exactly 45◦in a function of laser power. The results for both FI both directions and the quartz rotator and the half- are shown in Figure 8. For these measurements, the wave plate combined give exactly ∓45◦ in the for- powerinsidetheFIistwicetheincidentpower. 70W wardandbackwarddirection,respectively. Onlythe incidentpoweror 140WinsidetheFIismorethan isolation ratio or the optical efficiency can be opti- ∼ themaximumpower weexpect duringsciencemode mizedin-situbyrotatingthehalf-waveplatetocom- when 125 W are submitted to the near impedance pensate the aforementioned changes in the Verdet matchedmaininterferometerandlessthan10Ware constant when moving from low to high power and reflected. These results show that the in-air tested fromairtovacuum. Themeasuredopticalefficiency FI meet the aLIGO requirements. of the FI was 96.7% (±0.4%) for up to 70 W input power at LLO and 97.7% (±0.4%) for up to 140 W Oneproblemwiththisapproachofoptimizingthe at LHO. isolation ratio before installation into the vacuum system is that the temperature dependence of the Prior to installation of the LLO FI, the half-wave Verdet constant will cause the rotation of the two plate was temporarily adjusted to maximize the op- TGG crystals to shift after installation. This effect tical efficiency rather than the isolation ratio. By also causes a power dependent shift in the rotation measuring the residual power dumped in transmis- angle. Themotorizedrotationstageallowstoadjust sion of the FI an upper limit was placed on the ho- the half-wave plate to compensate for these changes mogeneity of the polarization rotation at 36 ppm. aLIGO Input Optics 10 Measuring the isolation ratio in this configuration gets progressively stronger at higher stages with the also allows for a measurement of the missing rota- middle and upper stages having, respectively, an tion in the Faraday rotator which came out to be actuation authority at DC that is 20 times and ∼ ∼1.6◦ at LLO. ∼1500 times that of the mirror. Staging the actuation strength in this way prevents the applied force from spoiling the seismic isolation provided by the suspension. C. Input Mode Cleaner Length sensing of the IMC is accomplished with the PDH technique by adding a 24.0 MHz sideband The input mode cleaner (IMC) is a resonant tri- to the beam via the EOM and sensing the ampli- angular cavity consisting of the three mirrors MC1, tude modulation induced when the cavity is off res- MC2, and MC3 which form an isosceles triangle as onance. This signal provides an accurate compari- shown in Figure 4. The purpose of the IMC within son between the round trip length of the IMC and the input optics is multifaceted. It suppresses spa- the frequency of the laser which is used to quiet tial non-uniformities of the input laser beam while the laser frequency above 15 Hz and to quiet the transmitting the diffraction limited Gaussian mode. ∼ cavity length below 15 Hz. Controlling the cavity It passively suppresses frequency and pointing noise ∼ length employs hierarchical control in which control and serves as a reference for additional active sup- at lower frequencies is offloaded to the higher stages pression. In addition, the IMC filters the polariza- of the suspension. The mirror stage is offloaded to tion of the input beam before being sent to the FI. the upper stages at frequencies below 7 Hz, and The input and output coupler (MC1 and MC3 re- ∼ the middle stage is offloaded to the top at frequen- spectively) of the IMC are nominally flat with a cies below 100 mHz. transmissivity of 6000 ppm while the apex mirror ∼ Angularsensingoftherelativealignmentbetween (MC2) has a nominal radius of curvature of 27.27 theinputbeamandtheIMCisachievedwithdiffer- m and a transmissivity of 5 ppm. MC1 and MC3 ential wavefront sensing,28,29 a variant of the PDH areseparatedby46.5cmwhilethedistancebetween technique. This technique provides independent er- MC2andMC1/MC3is16.24m. ThisgivestheIMC ror signals for all four relative degrees of freedom a free spectral range of 9.099 MHz, a finesse of 515, and is used to force the cavity to follow the input and a cavity pole of 8.72 kHz. beam with a bandwidth of 500 mHz. In addition, The reflected beam from the IMC is brought out- ∼ the quadrant detector behind MC2 is used to servo of-vacuum to the IOT2L table where it is detected two degrees of freedom of the input beam with a with a narrowband photodetector for length sens- bandwidth of 10 mHz. ing via the Pound-Drever-Hall27 (PDH) technique. ∼ Some of the reflected beam is picked off and sent to two wavefront sensors28,29, separated by 90◦ of Gouy phase, for angular sensing of the IMC. The D. Auxiliary mirror suspensions light leaking through MC2 is sent to an in-vacuum quadrant detector for additional angular informa- The HAM Auxiliary Suspensions (HAUXes), de- tion. In addition, a sample of the transmitted light picted in Figure 9, are single pendulum suspensions of the IMC is brought out-of-vacuum to the IOT2L with the addition of blade springs for vertical isola- table for diagnostics. tion. Themainstructure, madeofaluminum, fitsin The three mirrors of the IMC are made of fused an envelope of 127x217x441 mm (DxWxH), weighs silicaandhaveamassof2.9kg. Theyhangfromthe approximately 6 kg and consists of a base, two side aLIGOsmalltriplesuspensions30 whichprovideiso- walls,twohorizontalbarssupportingfourA-OSEMs lation from seismic noise proportional to f−6 above (a particular variation of the sensors/actuators de- the three resonant frequencies near 1 Hz for all described in the previous section), a stiffening slab grees of freedom except vertical and roll. The ver- connecting the two walls and a top part support- tical and roll degrees of freedom are isolated with ingthebladesprings. Thestructureisdesignedand blade springs which provide isolation proportional testedtohavetheloweststructuralresonanceabove tof−4 abovethetwobladespringresonancesnear1 150 Hz, so as not to interfere with the delicate con- Hz. Eachstageofthesuspensions,includingthemir- trol loops of the LIGO seismic isolation platform on ror, have small permanent magnets attached which which it is installed. can be actuated upon with electromagnets attached Two 250 mm long, 150 µm diameter steel mu- tothesuspensionframe,knownasOSEMs.31,32 The sic wires run from the tips of two 77 mm long, OSEMs also incorporate shadow sensors which use 500 µm thick tapered maraging steel blades down the magnets as flags to sense the important degrees to a lightweight circular aluminum holder contain- of freedom of each stage. The actuation strength ing the optic. The resonant frequencies of the op-