Astronomy&Astrophysicsmanuscriptno.Huby_2017_onsky_qacits_arxiv (cid:13)cESO2017 January24,2017 On-sky performance of the QACITS pointing control technique with the Keck/NIRC2 vortex coronagraph E.Huby1,⋆,M.Bottom2,B.Femenia3,H.Ngo4,D.Mawet2,5,E.Serabyn2,andO.Absil1,⋆⋆ 1Spacesciences,Technologies,andAstrophysicsResearch(STAR)Institute,UniversitédeLiège,19cAlléeduSixAoût,4000Liège, Belgium 2JetPropulsionLaboratory,CaliforniaInstituteofTechnology,4800OakGroveDrive,Pasadena,CA91109,USA 3W.M.KeckObservatory,65-1120MamalahoaHwy.,Kamuela,HI96743,USA 7 4 CaliforniaInstituteofTechnology,DivisionofGeologicalandPlanetarySciences,1200E.CaliforniaBlvd,Pasadena,CA91125, 1 USA 0 5CaliforniaInstituteofTechnology,DivisionofPhysics,MathematicsandAstronomy,1200E.CaliforniaBlvd,Pasadena,CA91125, 2 USA n January24,2017 a J 3 ABSTRACT 2 Context.AvortexcoronagraphisnowavailableforhighcontrastobservationswiththeKeck/NIRC2instrumentatLband.Thevortex ] coronagraphusesavortexphasemaskinafocalplaneandaLyotstopinadownstreampupilplanetoprovidehighcontrastatsmall M angularseparationsfromtheobservedhoststar. Aims.Reaching the optimal performance of the coronagraph requires fine control of the wavefront incident on the phase mask. I In particular, centering errors can lead to significant stellar light leakage that degrades the contrast performance and prevents the . h observationoffaintplanetarycompanionsaroundtheobservedstars.Itisthuscriticaltocorrectforthepossibleslowdriftofthestar p imagefromthephasemaskcenter,generallyduetomechanicalflexuresinducedbytemperatureand/orgravityfieldvariation,orto - misalignmentbetweentheopticsthatrotateinpupiltrackingmode. o Methods.A control loop based on the QACITS algorithm for the vortex coronagraph has been developed and deployed for the r t Keck/NIRC2instrument.Thisalgorithmexecutestheentireobservingsequence,includingthecalibrationsteps,initialcenteringof s thestaronthevortexcenterandstabilisationduringtheacquisitionofscienceframes. a Results.On-skydatashowthattheQACITScontrolloopstabilizesthepositionofthestarimagedownto2.4masrmsatafrequency [ ofabout0.02Hz.However,theaccuracyoftheestimatorisprobablylimitedbyasystematicerrorduetoamisalignmentoftheLyot 1 stopwithrespecttotheentrancepupil,estimatedtobeontheorderof4.5mas.Amethodtoreducetheamplitudeofthisbiasdownto v 1masisproposed. 7 Conclusions.TheQACITScontrolloophasbeensuccessfullyimplementedandprovidesarobustmethodtocenterandstabilizethe 9 starimageonthevortexmask.Inaddition,QACITSensuresarepeatablepointingqualityandsignificantlyimprovestheobserving 3 efficiencycomparedtomanualoperations.ItisnowroutinelyusedforvortexcoronagraphobservationsatKeck/NIRC2,providing 6 contrastandangularresolutioncapabilitiessuitedforexoplanetanddiskimaging. 0 Keywords. Instrumentation:adaptiveoptics–Techniques:highangularresolution–Methods:observational . 1 0 7 1. Introduction ingcommissioning(Absiletal. 2016).In addition,suchanim- 1 ager at L band benefits from advantageousatmospheric condi- :In June 2015, a new coronagraphic mode available with the v tions and generally more favourable planet-to-star contrast ra- iKeck/NIRC2instrument(Serabynetal.2017)cameonline.The tio thanat shorterwavelengths,especially in the case of young XL band imager is now equipped with a vector vortex coro- self-luminous giant exoplanets. As such, the vortex mode at rnagraph based on an annular groove phase mask (AGPM, Keck/NIRC2 provides a promising tool to directly image exo- a Mawetetal. 2005). It consists of a circular subwavelength planets on relatively compact orbits, thus complementing sec- grating etched onto a diamond substrate (VargasCatalánetal. ondgenerationhighcontrastinstrumentsworkinginthenearin- 2016), placed in an intermediate focal plane wheel and work- frared,e.g.SPHERE(Beuzitetal.2008),GPI(Macintoshetal. inginconjunctionwithadownstreamLyotstop.Combinedwith 2014)orSCExAO(Jovanovicetal.2015). AdaptiveOptics(AO) correctionandadvancedpost-processing Withtheimplementationofthisnewmode,thenecessityof techniques (e.g. Soummeretal. 2012; Amara&Quanz 2012; afinepointingcontroliscrucialfortworeasons:operationeffi- GomezGonzalezetal. 2016), this mode allows high-contrast ciencyandcontrastperformance.First,observingwithavortex imagingofplanetarycompanionsandcircumstellardisksaround coronagraphrequiresthestarimagetobeinitiallycenteredonto stars at very small angular separations, with an effective inner workingangle(IWA)of 0′.′12.A5-σsensitivitylimitof10mag thevortexmask,andifdonemanually,thisprocesscanproveto at0′.′5forastarofmagnitudeK=5hasbeendemonstrateddur- bechallengingandtimeconsuming.AsillustratedinHubyetal. (2015,2016)andrecalledinthenextsection,thisisparticularly ⋆ F.R.S.-FNRSPostdoctoralResearcher true in case of centrally obstructed telescopes, where counter- ⋆⋆ F.R.S.-FNRSResearchAssociate intuitivefluxasymmetrycanappearwhenthestarimageisvery Articlenumber,page1of9 A&Aproofs:manuscriptno.Huby_2017_onsky_qacits_arxiv closetothecenterofthemask.Thiseffectcanleadtothemisin- terpretationoftheoffsetdirectionneededtoimprovethecenter- ingofthestarimageandmakethecenteringprocesslongand/or not optimal. Repeatability and rapidity of the alignment were thus major incentives for the implementation of an automated pointingcontrolsystem. Additionally,reachingtheoptimalperformanceofthecoro- nagraphrequiresahighqualitywavefront.Althoughaberrations due to turbulence are largely removed by the highly efficient Keck AO system, routinely delivering Strehl ratios on the or- der of 85-90% at L band, noncommon path wavefront errors inside the instrument are inevitable and the source of substan- Fig.1.PupilmasksusedinthesimulationsoftheKeck/NIRC2instru- ment,asdefinedinTable1:entrancepupilontheleftandLyotstopon tiallossofcontrastperformance.Indeed,reachingasmallIWA theright.Thecircumscribedcircleinwhitedottedlinehasadiameter with a focal-plane phase mask coronagraphcomes at the price of10.93m. of a high sensitivity to low order aberrations. The most detri- mentalofthereare generallytip andtiltaberrationsduetome- Table1.PupilconfigurationoftheKecktelescopeandLyotstopmask chanicalflexuresand/oropticsrotatingforpupiltracking,which usedforthevortexcoronagraphinNIRC2. are commonon large telescopes. The controland correctionof these aberrations are thus crucial and require the implementa- parameter normalizedvalue tionofadditionaldedicatedsensors.Here,wefocusonloworder Entrancepupildiam. 1.00 aberrationsensors,whicharecriticalinparticularforsmallIWA (circumscribedcircle) coronagraphs. These sensors usually work in conjunction with Centralobstructiondiam. 0.24 othertechniquessensitivetohigh-ordernoncommonpathaber- Entrancepupilspiderwidth 0.0023 rations,suchasspecklenulling,electricfieldconjugation,phase Inter-segmentgap 2.7×10−4 diversity,Zernikewavefrontsensing,orinterferometricmethods Lyotstopexternaldiameter 0.80 (see Bottometal. 2017,andreferencestherein).Asa matterof Lyotstopinternaldiameter 0.27 fact,specklenullingwasrecentlyimplementedonKeck/NIRC2 Lyotstopspiderwidth 0.0061 forusewiththevortexcoronagraph(Bottometal.2016,Bottom etal.inprep). Concerning low order aberration control, several solutions instrument. This paper reports on the practical implementation have been developedand implemented in currenthigh contrast andthe on-skyperformanceofthis pointingsensor.In the next instruments.Inordertocatch mostofthe noncommonpath er- section,theprincipleofthistechniqueisrecalledandthediffer- rors, the extra sensor must be placed as close as possible to ent functions performed by the controller implemented for the the coronagraphic mask. For instance, in the case of the Dif- Keck/NIRC2 instrument are described. In Sect. 3, the experi- ferentialTip-Tilt Sensor (DTTS) integratedto the coronograph mentalcalibrationofthetip-tiltestimatorispresented,andpos- of SPHERE (Baudozetal. 2010), a beamsplitter placed right siblesourcesofbiasareinvestigated.InSect.4,theon-skyper- before the focal plane phase mask sends a few percent of the formanceofthecontrolloopisassessedandcomparedwithre- light towards a separate detector in order to monitor the jitter and drift of the star image. Another solution consists in using sultsobtainedwithouttheQACITScontroller.Lastly,inSect.5, thestatusofthealgorithmisdiscussedandfurtherongoingde- thelightthatisrejectedbythecoronagraph,i.e.thatisstopped velopmentsaredescribed. by the occulting focal plane mask (Coronagraphic Low-Order Wave-Front Sensor, CLOWFS, Guyonetal. 2009), like in GPI (Wallaceetal. 2010), or stopped by the diaphragm in the Lyot plane, in case of a phase mask (Lyot-based Low-Order Wave- 2. ImplementationofQACITSatKeck/NIRC2 FrontSensor,LLOWFS,Singhetal.2014).OnlytheLLOWFS system hasbeencharacterizedon-sky,reportingpointingresid- Throughoutthepaper,thetip-tiltamplitudeisgiveneitherinmil- ualsof0.23masor5.8×10−3λ/D(Singhetal.2015).However, liarcseconds(mas)orinunitsoftheangularresolutionelement it has to be noted that all these sensors are inherently notfully defined as λ/DL, with λ the wavelength and DL the equivalent common path and require a specific optical layout, which can diameterof the Lyotstop, which definesthe angularresolution makethemchallengingtointegratetoanexistingsystem. inthefinalimage.Inpractice,thisdiameterisequalto8.7mand Anotherkindofsolutionisbasedonthesoleanalysisofthe correspondsto80%oftheentrancepupildiameter,definedhere coronagraphicimages.Inthis case,the sensoris fullycommon as the diameter of the circumscribed circle, i.e. 10.93m. The path,butthenumberofmodesthatcanbecorrectediscurrently Lyot stop almost matches the inscribed circle in the hexagonal reducedto tip andtilt. Suchsensor hasbeenfirstproposedand entrancepupil,whichis9mindiameter.Thewavelengthistaken tested in laboratory for the four quadrant phase mask (FQPM) asthecentralwavelengthoftheLbandfilter,i.e.λ=3.776µm, coronagraph(Masetal.2012)incaseofanonobstructedpupil, which givesλ/DL=89.3mas. The entrancepupiland Lyotstop leadingtoanaccuracyof6.5×10−2λ/Dmeasuredinlaboratory. shapesusedinthesimulationspresentedinthispaperareshown Theprinciplewasthenextendedtothecaseofthevortexcoron- inFig.1.Thelattercorrespondstotheincirclepupilmaskavail- agraphwithacentrallyobscuredpupil(Hubyetal.2015).Asa ableintheNIRC2instrument.Thepupilfeaturedimensionsare completelynon-invasivemethodthatdoesnotrequireanysetup reportedinTable1. modification, this technique called QACITS (Quadrant Analy- Inthissection,theprincipleoftheQACITSestimatorcorre- sisofCoronagraphicImagesforTip-tiltSensing)thusappeared spondingto the Keck telescope pupilis described, and the dif- as one of the logical and most accessible solutions for imple- ferent operations performed by the controller implemented on mentation on the new vortex coronagraph of the Keck/NIRC2 Keck/NIRC2aredetailed. Articlenumber,page2of9 E.Hubyetal.:On-skyperformanceoftheQACITSpointingcontroltechnique −0.2 −0.1 0.0 0.1 0.2 Fig.2.Imagesofthetwoasymmetriccomponentsappearinginthefinal coronagraphicimageshapeinpresenceofhorizontaltip-tiltaberration. Positivetip-tiltamplitudeinducesashiftofthestarimagetowardsthe right.Left:A cosψ,contributionofthecircularnon-obstructedpupil. circ Right: A cosψ, contribution of the central obstruction. The circles Fig. 3. Differential intensity measured on simulated images with tip- obsc delimiting the inner and outer regions have a radius of 1.6λ/D and tiltappliedalongthexaxis.Thebluesolidlinecorrespondstotheflux L 2.7λ/D . A comparison of the horizontal profiles of these images is integratedintheinnerareaonly,whilethereddashedlinecorresponds L displayedinFig.6ofHubyetal.(2015). tothefluxintegratedintheouterareaoftheimage.Thelightbluedotted and red dash-dotted lines are the linear approximations for the inner andoutermeasurements,respectively,thatarevalidinthesmalltip-tilt 2.1.TheQACITSestimator regime(i.e.<0.15λ/D and<0.5λ/D ,respectively). L L The QACITS estimator aims to measure the pointing errors affecting the beam incident on a coronagraphic phase mask outerregionoftheimage,andtheyhaveoppositesigns.This directly from the analysis of the coronagraphic image shape contributionvarieslinearlywiththetip-tiltamplitude. (Masetal.2012;Hubyetal.2015).Asaresult,itprobesaberra- Tocompletethequantitativecomparison,thedifferentialin- tionsthatare fullycommonpathwith the coronagraphicmask. tensityamplitudeduetothecircularapertureintheinnerregion Giventhefluxlevelremainingafterthecoronagraph,aberrations (Fig.2,left)isabout4timeslowerthanthedifferentialintensity induced downstream the mask have a negligible impact on the amplitudeduetotheobstructioninthesameregion(right). contrast performance. The QACITS estimator is based on the measurementof differentialintensities, resulting from the inte- The analysis of the image in these two separate concentric regionsallows the partial disentanglementof the two contribu- grationandsubtractionofthefluxinthehalvesoftheimage,as tions.Intheinnerarea,thetwotermsarepresentandhaveoppo- inaquadrantpositionsensor,normalizedbythetotalfluxofthe site signs,thuscompensatingeachotherandmakingthediffer- noncoronagraphicpointspreadfunction(PSF). entialintensitymeasurementambiguous,ascanbeseeninFig.3. Hubyetal. (2015) have previously shown that the electric Thisisalsothereasonwhytheobservercanbemisled,sincethe fieldinthecaseofacentrallyobstructedpupilcanbedescribed flux asymmetry in this region goes in the direction opposite to asthesumoftwocontributionsrelatedrespectivelytothecircu- the actual offset of the star, as illustrated in Fig. 4. Moreover, lar unobstructedpupiland the central obstruction(addednega- foratip-tiltamplitudeofabout0.4λ/D thecentralregionlooks tivelyinamplitude),leadingtotwodistinctcontributionsinthe L likeasymmetricdoughnut,whichcanbeadditionallyconfusing flux asymmetry of the final image. These two components are (seeSect.4.2).Intheouterareahowever,thelinearcontribution expressed as combinationsof Bessel functionsmodulated by a duetothecentralobstructionprevails,avoidingambiguityinthe cosψtermwithψ theazimuthalangle,asitwasestablishedby measurement. Eq.9and19inHubyetal.(2015).Inshort,foratip-tiltampli- In practice, the QACITS estimator used at Keck/NIRC2 is tude T applied along the x axis, the final image I can be de- based on the differential intensity measured in the outer area scribed as the sum of a symmetric term and two asymmetric only,approximatedbyalinearmodel,whichisvalidinthesmall terms: tip-tilt regime (i.e. < 0.5λ/D , see Fig. 3). The linear approx- L I ∝ Isym+T3×Acirccosψ+T ×Aobsccosψ, (1) imation of the inner estimator is also monitored, since it can carryusefulinformation(seeSect.3.2.2).Hereafter,theestima- where A and A arecombinationsofBessel functionscor- circ obsc tors based on the inner and outer parts of the image using the respondingtothecontributionsofthecircularpupilandcentral linear approximation will be designated as the inner and outer obstruction,respectively.Theshapeofthetwoasymmetricterms estimators,respectively.Itshouldbenotedthattheshapesofthe aredisplayedinFig.2.Theycanbedecomposedintwoconcen- differentialintensity curvesdepend on the telescope configura- tricregions,definedbythesigninversionoftheiramplitude.The tion.Inparticular,forasmallercentralobstruction(asisthecase boundariesbetweeninnerand outerregionsare locatedatradii for VLT/NaCo or LBT/LMIRCam with 14% and 11% central of1.6λ/D and2.7λ/D .Theweightofeachcomponentinthe L L obstruction,respectively),theslopesofthelinearmodelapprox- final image is a function of the tip-tilt amplitude, making each imationswouldbelower. termcontributedifferentlytotheasymmetryoftheimageinthe tworegions: 2.2.Closed-loopimplementationatKeck/NIRC2 – thedifferentialintensityduetotheidealcircularpupilis40 timesstrongerintheinnerthantheouterregionoftheimage. TheQACITSalgorithmhasbeenimplementedinIDLandtakes In the final image,this contributiongrowswith the cube of care of the initial centering optimization, the pointing correc- thetip-tiltamplitude. tionandthescienceacquisitions,makingobservationswiththe – the amplitude of the differentialintensity due to the central vortexmodeuser-friendlyandhighlytime-efficient.Beforecall- obstruction is about 3 times stronger in the inner than the ing the QACITS sequence, the observer only has to make sure Articlenumber,page3of9 A&Aproofs:manuscriptno.Huby_2017_onsky_qacits_arxiv Fig. 5. Tip-tilt scanning sequences carried out for characterizing the QACITSestimatorbehaviour.Left:Truetip-tiltvaluesasmonitoredby Fig. 4. Top: Simulated images for different tip-tilt amplitudes applied the secondary component of the binary system, for the scanning se- to the right along the horizontal axis, from left to right: 0.1λ/D , L quences along the x(green crosses) and y (blue squares) axes. Right: 0.16λ/D and 0.38λ/D . Each displayed image is normalized by its L L Comparisonof thesimulatedpeaktransmissioncurve(fluxintegrated maximalvalue.Thedottedcirclesshowtheboundariesfortheinnerand onadiskofdiameter 1λ/D )withtheexperimentalresultsmeasured outerareas.Bottom:Color-codedrepresentationofthefluxasymmetry L fromtheon-sky data. TheeffectiveIWAisreached at aseparationof toemphasize thevisual comparison of theasymmetry inthedifferent 1.4λ/D =125mas(computedfromthesimulatedcurve). images.Theamplitudeofthegradientisequaltothemeasureddiffer- L entialintensityandallimagesareshownwiththesamecolorscale. (2016)andinausermanualavailableonline1.Thepixelsizeof that the star is roughly centered onto the vortex mask (within the coronagraphicimages is 10mas/pixel (Serviceetal. 2016), 1λ/DL), a pointing requirement that is routinely achieved by i.e.about9pixelsperresolutionelementλ/DL.SincethePSFis the AO system once a reference position has been saved. The wellsampled,wehavenotinvestigatedindetailtheeffectoflow QACITScontrollerthenoperatesinthreesteps: samplingon the QACITSestimator.Thiswill be studiedin the future,inparticularforapplicationonotherinstruments. – A calibrationstep,whichtakesanunsaturatedimageofthe starfaroffthecenterofthevortexmask(off-axisPSF),and skyimages.Thepositionofthecenterofthevortexmaskis 3. Vortexmodecharacterization identified by fitting the vortex center glow that is visible in sky images (Absiletal. 2016). The actual offset of the star 3.1.On-skycalibration image is then estimated by fitting a Gaussian profile to the off-axis PSF, and this offset is corrected by sending a tele- The calibration of the model for the QACITS estimators has scope offset commandin units of pixels. At the end of this been performed on-sky. For that purpose, a wide binary sys- tem was observed,with the brightest componentof the system step,thestarisroughlycenteredontothevortexphasemask centered on the vortex mask. The monitoring of the compan- (typicallywithinafew0.1λ/D ).Thisstepusuallyrequires L ∼2mintobecompleted. ionpositionprovidesameanstoestimatethetruepositionofthe – Anoptimizationsequence,whichconsistsofafewiterations starimagebehindthemask.OnOctober29th,2015,acquisitions werethustakenonthebinarysystemofHD46780(L =5.5, oftheQACITSloopthatarerunfasterthanthescientificac- primary L =7.2).The separationwas 737.2masat the dateof the quisitions(usingshorterintegrationtime andsmallerframe secondary size to minimize NIRC2 overheads). By default, the align- observation(orbitparametersfrom Heintz 1993). The long pe- riod of 118.9 years ensures that this separation does not vary ment is considered optimized when the measured residual duringthetimeofobservation. tip-tilt has an amplitude smaller than 0.1λ/D . This crite- L rion can be tuned in the QACITS parameters (the observer The relative position of the two components of the binary canrequirethatseveralconsecutiveestimationsfallbelowa werefirstestimated inan unsaturatedimage,byfittinga Gaus- chosenlimit). With thedefaultsettingsofT = 0.2s(inte- sian profile to each PSF. The vector connectingthe position of int grationtime)and N =10(numberofco-addedimages) the two star images is then used as a reference to estimate the coadd foraframewidthof512pixels(minimalvalueforthevortex true position of the primary star image with respect to the po- centertobeincludedinthesub-image),oneiterationis20s sitionofthesecondaryinthecoronagraphicimages.Giventhat long.Thissequencetypicallytakes1-2min. observationsarecarriedoutinpupiltrackingmode,therotation – Thescienceacquisitionsequence,withtheQACITScorrec- ofthepositionvectorisalsotakenintoaccountbycorrectingfor tionappliedaftereachacquisition.Withthetypicalsettings the parallactic angle. The positions that have been probed dur- of T = 0.5s and N = 50 for a full frame width of ingthetip-tiltscanningsequenceareplottedinFig.5,showing int coadd 1024pixels,onescienceacquisitionis46slong(for25sof thattheinitialpositionthatwasassumedtobealignedwiththe actualintegrationtime). vortexcenterwasactuallyoffsetbyabout−0.25λ/DLintipand tilt, dueto animperfectmanualpositioningat thebeginningof The correction algorithm applied during the science se- thesequence.Basedonthesedata,theoff-axispeaktransmission quence consists of a proportional-integralcontroller, with pro- ofthevortexcoronagraphhasbeencomputedbyintegratingthe portionalandintegralgainsGP = 0.3andGI = 0.1,whichhave fluxinadiskofdiameter1λ/DLcenteredontheactualstarim- been tuned experimentally to ensure the stability of the loop. ageposition.AsshowninFig.5,theseexperimentalresultsare The loop is run at a frequency defined by the time needed for in excellent agreement with the simulated curve, leading to an oneacquisition,i.eabout0.02Hz,hencecorrectingfortheslow effectiveIWAof125mas. drift(seeSect.4.1).TheQACITScallingsequenceandparame- tersforKeck/NIRC2aredescribedinmoredetailsinHubyetal. 1 https://www2.keck.hawaii.edu/inst/nirc2/observing Articlenumber,page4of9 E.Hubyetal.:On-skyperformanceoftheQACITSpointingcontroltechnique Fig.6.ExperimentalcharacterizationoftheQACITSmodelfromon-skydata.Leftandrightplotscorrespondtothedifferentialintensitymea- surementsasafunctionoftip-tiltforthexandydirections,respectively.Inbothcases,thedifferentialintensitywascomputedintheinner(blue crosses) and outer (red pluses) areas of the image. The best fit models in the least-squares sense are shown in solid lines (see Eq. 2 for their definition),andtheresidualsareshownintheplotsbelow. Fig. 7. Same as Fig. 6 with data sets simulated for the same values of tip-tilt as the experimental data (including the offset in the orthogonal direction,asplottedinFig.5),insteadofactualmeasurements. Table2.ParametervaluesforthebestfitmodelsasdefinedbyEq.2.Theresultsarereportedfortheon-skydatasets(seeFig.6)aswellasfor thedatasetssimulatedwiththesametip-tiltsampling(seeFig.7).Theidealmodelcorrespondstotheresultofthefitperformedonthesimulated modelshowninFig.3,includingamuchlargernumber ofsimulateddatapoints(notbiasedbythesamplingorbytheoffsetintheorthogonal directionoftheappliedtip/tilt).Valuesbelow10−3areconsideredinsignificantandmarkedas0. Outerarea(linearmodel) Innerarea(cubicmodel) dataset a σ b σ a σ b σ c σ x σ a b a b c 0 x0 on-skyx 0.109 ±0.008 0.004 ±0.002 0.790 ±0.167 -0.014 ±0.057 -0.082 ±0.020 0.003 ±0.031 simulatedx 0.111 ±0.006 0 ±0.002 0.986 ±0.090 -0.055 ±0.034 -0.095 ±0.012 0 ±0.015 on-skyy 0.113 ±0.013 0.010 ±0.003 0.767 ±0.045 0.128 ±0.015 -0.084 ±0.005 -0.039 ±0.007 simulatedy 0.104 ±0.004 0 ±0.001 0.976 ±0.109 0.014 ±0.032 -0.106 ±0.010 -0.002 ±0.016 idealmodel 0.104 ±0.007 0 ±0.002 1.013 ±0.051 0 ±0.019 -0.146 ±0.007 0 ±0.009 Articlenumber,page5of9 A&Aproofs:manuscriptno.Huby_2017_onsky_qacits_arxiv No aberr. Astig. 1 Coma 1 Trefoil 1 Defocus Astig. 2 Coma 2 Trefoil 2 Fig.8.Simulatedimagesobtainedwiththevortexcoronagraphonthe Kecktelescope,inpresenceofloworderaberrations:100nmrmsdefo- cus,70nmrmsfortheothers.Allimagesaredisplayedwiththesame grayscale. Fig.9.Effectofcoma(70nmrms)ontheQACITSmodel.Ontheleft, the image shows the shape obtained when the star image is perfectly centered on the vortex mask, and the image below isthe color-coded Figure 6 shows the differential intensity measured in these representationofthemeasureddifferentialintensityininnerandouter imagesinthe xandydirectionsasafunctionofthetruetipand regionsoftheimagetoemphasizethevisualcomparisonoftheasym- tilt,respectively.Thedatapointscorrespondingtotiportiltam- metryinbothregions(thefluxgradientineachareahasanamplitude plitudelowerthan0.5λ/DLwerefittedbypolynomialfunctions equaltothemeasureddifferentialintensity). expressedas ax+b,fortheouterarea,and aspossible.However,thissituationcorrespondstothebestcen- (2) a(x−x0)3+b(x−x0)2+c(x−x0),fortheinnerarea. teringofthestar imageonthemaskonlyiftheobservedtarget is a point source, thus assuming that there is no bright asym- All best fit parameters computed in the least-squares sense are metricstructurepresentin theveryclosevicinityofthe central reportedinTable2. star(within2.7λ/D or240mas)andthattheopticalsetupisnot L Tovalidatethemodel,datasetsweresimulatedwiththesame affectedbyotheraberrations.Whiletheformersituationconsti- valuesoftipandtilt(includingtheoffsetintherespectiveorthog- tutesanintrinsiclimitationoftheQACITSestimator,biasdueto onal direction as shown in Fig. 5). These simulated data were opticalimperfectionscanbemitigatedtosomeextent,assuming analyzedusingthesamefittingprocedure.Theresultsareshown thattheircauseisunderstood. inFig.7andthebestfitparametervaluesarereportedinTable2. Theidealmodelcomputedfromsimulationswithoutoffset(only puretiportiltwasapplied,aspresentedinFig.3)andwithafiner 3.2.1. Effectofloworderaberrations samplingisalsoanalyzedinthesamewayforcomparison.The Low order aberrations have been investigated as sources of correspondingbestfitparametersarealsoreportedinthebottom asymmetry in the final image. The resulting image shapes are lineofTable2. shownin Fig. 8. While defocusandastigmatism havean effect Asexpected,thedifferentialintensitymeasuredintheouter ontheshapeoftheimage,theydonotaffectthecentralsymme- area of the image is well approximated by a linear function, try.Ontheotherhand,comaandtrefoildo.Atthesameaberra- in particular for small tip-tilt amplitude (for tip-tilt amplitude tion level, the effect of coma is almost one orderof magnitude > 0.5λ/D ,thedatapointsdivergefromthelinearmodel).The L strongerthan trefoil, and forthat reason, its effect hasbeen in- slopesareinagreementwiththevaluespredictedbysimulations, vestigatedinmoredetail. butsmallglobaloffsetsareobserved(nonzerovalueforparam- eterb).Thiseffectisstrongerforthedatacorrespondingtothey The impact of coma on the QACITS model is shown in direction.Possiblecausesforthiseffectwillbediscussedinthe Fig.9:thedifferentialintensitymeasuredintheinnerareaissig- nextsubsection. nificantlyaffected,whiletheouterdifferentialintensityismostly The differential intensity measured in the inner area shows unchanged.Indeed,althoughitisnotvisuallyobvious,theasym- the expectedsign inversionsaround±0.4λ/D . The modelpa- metryinducedbycomaismostlyconcentratedintheinnerarea L rametervaluesfittedontheon-skydataaregloballylowerthan oftheimage,andtheouterareaisbarelyaffected,asshownby the values that were expected from the simulations (from 14% thesimulatedimageinFig.9.Asaresult,thisaberrationcannot to21%forthefirstandthirdorderterms)butareconsistentwith explainthebiasobservedintheon-skydata. the simulations within error bars, except for the second order term in the y direction. Besides, the coefficients for the second 3.2.2. EffectofmisalignmentoftheLyotstop ordertermswereexpectedtobenullaccordingtotheanalytical modelderivedinHubyetal.(2015),asconfirmedbythesimu- IncaseofaLyotstopmisalignmentwithrespecttotheentrance lations(seethebestfitparametervaluefortheidealmodel).As pupil,the coronagraphicimage becomesasymmetric too, as il- discussed in the next subsection, possible explanation for this lustratedinFig.10. Inparticular,thedifferentialintensitymea- behaviourincludesthepresenceofadditionalasymmetriccom- suredintheouterannulusissignificantlyaffected,resultinginan ponentsintheimage. offsetoftheQACITSmodel,asobservedintheon-skydata.In practice,itmeansthattheQACITScontrolloop,whichisbased ontheouterestimatoronly,willconvergetowardsapositionthat 3.2.Sourcesofbias doesnotcorrespondtothebestcenteredposition.Inthecaseof In its current state, the QACITS closed loop control tends to a4%shiftrelativetotheentrancepupildiameter(asillustrated make the outer part of the coronagraphic image as symmetric inFig.10),thisbiashasanamplitudeofabout0.12λ/D . L Articlenumber,page6of9 E.Hubyetal.:On-skyperformanceoftheQACITSpointingcontroltechnique Fig.10.SameasFig.9inthecaseofamisalignmentoftheLyotstop (shift of 4% of the entrance pupil diameter, along the horizontal di- rection).Theimageobtainedwhentheasymmetryintheouterareais compensatedbytip-tiltisshownintheinset. Fig.11.BiasinducedbytheQACITSouterestimator,asafunctionof theLyot stopshift,inunitsof percent of thetelescope entrance pupil diameter (i.e. 1% corresponds to 11cm at the scale of the telescope Thebiasinducedbytheouterestimatordependsontheam- pupil). The red crosses show the bias induced by the outer estimator. plitudeoftheLyotstopshiftwithrespecttotheentrancepupil. The blue triangles are the inner estimator values as measured at the Thisbiasisestimatedasthetip-tiltamplitudeforwhichthedif- biasedposition.Thelightbluesquaresarethesameestimatesscaledby ferential intensity measured in the outer annulus cancels out. anadjustment factorof0.7.Thisfactoriscomputedasthesloperatio Simulation results reported in Fig. 11 show that the bias in- ofthebest linearmodel fittedontheactual biasandinner estimateat creases linearly with the amplitude of the Lyot stop misalign- smallLyotshiftamplitudes(<2%),shownasdashedlines. mentuptoa2%shift. Based on these simulation results, we propose a method to estimate the amplitude of the bias affecting the outer estimator seeing was estimated to be 0′.′5 during the hour preceding the thankstotheinnerestimator,whichislessaffectedbythemis- acquisitionsequenceonJune9th(noseeingdataduringthese- alignment(as highlightedin Fig.10): when the outer estimator quence) and 0′.′7 on average during the sequence on October loop has convergedon the position where the outer differential 24th(CFHTDIMMseeingmeasuredat0.5µm). fluxisminimal,theinnerestimatorappliedonthatbiasedposi- DuringthenightofJune9th,theQACITScontrollerwasnot tion leads to an estimate of the bias over-estimated by a factor yet operational and the star image was initially centered man- ∼ 1.5 (see Fig. 11). Scaled by 70%, this estimate can thus be ually onto the vortex mask and maintained as well as possi- usedasasetpointtocorrectforthebiasoftheouterestimator. ble by manually adjusting its position (every ∼ 10min) based For Lyot stop shifts smaller than 2.5%, the residual bias when on a visual assessment of the coronagraphicimage shape. The applyingthismethodshouldbe lessthan0.01λ/D , i.e.1mas. L data sequencetaken on HR8799is 15-minlong.The results of Theimplementationofthisadditionalstepwillresultinalonger the QACITS estimators applied in post-monitoring are shown timededicatedtotheoptimizationofthecentering.Engineering in Fig. 12 (left). For this particular data set, a clear drift is ob- time will be needed to implement and test this upgrade of the served,atarateof∼2.7masperminute.Theinnerestimatorisin QACITScontroller. completedisagreementwiththeouterestimatorbecausethetip- tilt amplitude rapidly reaches values outside the validity range of the linear approximation(typically> 0.15λ/D i.e. 13mas). 4. Performanceassessment L Thewaythemodulusoftheinnerestimatordecreasesandthen FirstlightofthevortexmodewiththeKeck/NIRC2instrument increasesduringthesequenceisconsistentwiththeexpectedbe- wasachievedinJune2015(Serabynetal.2017).Duringthis3- haviouraroundthechangeofsignofthederivativefunction(see nightrun,apreliminaryversionoftheQACITSloopwasclosed Fig.3), whiletheimplementedestimatorisbasedsolelyonthe on the second night. Another vortex run took place in Octo- linearapproximation. ber 2015, including one engineeringnight for the implementa- Forcomparison,thesameanalysishasbeenperformedona tionofanimprovedversionoftheQACITSautomatedloop(as sequencetakenonthesametargetfourmonthslater,withanop- described in Sect. 2) and for performing the calibration of the erationalQACITScontrolloop.Thesequencespansover90min model(asreportedinSect.3).Inthissection,wepresentacom- intotal,includingagapwithoutdataduetotheinabilityoftrack- parison of the results obtained on data sets taken before (June ing the star very close to zenith. The QACITS estimators dis- 9th,2015)andafter(October2015)thedeploymentoftheopti- playedinFig.12(right)showasignificantimprovementinsta- mizedQACITScontroller,highlightingthe benefitsof the con- bility:thestandarddeviationofallouterestimatesis2.2masand trolloop. 5.1masinthexandydirections,respectively.Thelargerdisper- sionobservedalongtheyaxisindicatestheprobabledirectionof thedriftthathadtobecorrectedbythecontroller.Theinneres- 4.1.Correctionfortheslowdrift timatesaresomewhatoffset,withameanamplitudeof11.7mas. Inthissubsection,wepresentacomparisonoftheQACITSes- Basedonthesimulationresultsassumingamisalignementofthe timatorappliedinpost-monitoringontwodatasetstakenonthe Lyotmaskwithrespecttotheentrancepupil(Fig.11),thisam- same target, HR8799, and under similar observing conditions: plitudecanbeinterpretedasashiftoftheLyotstopofupto2.7% the target was observed close to transit (airmass ∼ 1), and the atthetimeoftheseobservations. Articlenumber,page7of9 A&Aproofs:manuscriptno.Huby_2017_onsky_qacits_arxiv Fig. 12. QACITS position post-monitoring of the image of the tar- Fig.13.On-skyresultsfromtheQACITSpost-monitoringofdatasets get HR8799 onto the vortex mask in June (left) and October 2015 takenduringthenightoffirstlight(left)andonalaterruninOctober (right). The inner and outer estimators are plotted with blue crosses withthecontrolloopclosed(right).Everypointrepresentsthemeanes- and red pluses, respectively. The dashed circles have a radius of timateofasequence(comprisingbetween8and72acquisitionframes, 0.15λ/D =13mas.Thecolorshadeofthepointsbecomesdarkerwith 10 in average), with the error bars showing the standard deviation of L time.Everydatapointcorrespondsto20sand25sofintegrationtime theestimatesduringthesequence.Thedashedcircleshavearadiusof inJuneandOctober,respectively. 0.1λ/DL=9mas. 5. Conclusionandprospects 4.2.Pointingstatistics The QACITS controllerhas been successfully implemented on Thesamepost-monitoringprocedurebasedontheQACITSesti- the newly commissioned vortex mode of the Keck/NIRC2 in- matorshasbeenappliedtoalldatasequencestakenonthenight strument.Thebenefitsofthisautomatedcontrolaremultiple: ofJune9th,2015,andduringthreeconsecutivenightsdedicated to science targets in October 2015. The mean outer and inner – Theobservationsequenceisfullyautomated,includingtak- estimates of every sequence are plotted in Fig. 13. The com- ingcalibrationframes,initialcentering,andstabilizationof parison of the results with and without the QACITS controller thestarduringtheobservation,makingthevortexmodesuf- is quite explicit: on the June night, the dispersion of the mean ficiently user friendly to be offered to the community (in outerestimatorsis16masand32masinthe xandydirections, shared-riskmodesince2016B). respectively,whileitisreducedto2.3masand2.6mas,respec- – Pointingqualityisnotobserver-dependent,andinparticular, tively,duringthethreeobservingnightsinOctober.Theaverage thepitfallinducedbytheapparentsymmetryofthecorona- standarddeviationwithineverysequenceisalsoindicativeofthe graphicshapewhenthestarisoffsetby∼30masisavoided. – Pointingstabilityof2.4masrmsisachievedonaverage,with improvedstability,asitisreducedfrom4.1masand7.9mason thecontrollooprunningatafrequencyofabout0.02Hz,thus June9th(inthexandydirections,respectively),downto2.0mas correctingforlowfrequencydrifts. and2.8masoverthethreenightsinOctober. – Pointingaccuracyof4.5masisachievedonaverage.Thisac- Additionally,forthe data taken withoutthe automatedcen- curacyiscurrentlylimitedbysystematicerrorsinducedbya tering of the star, the tip-tilt amplitude for the outer estimator probable misalignment of the Lyot stop with respect to the reaches 33mas (0.37λ/D ) on average, which roughly corre- L entrancepupil. Still, the finalaverage accuracyprovidesan spondstotheamplitudeforwhichthefluxasymmetryinthein- improvementofafactor7overtheaccuracyachievedmanu- ner disk changessign (see the model curve in Fig. 3). In other ally,andamethodtoreducethisbiasdowntothe1maslevel words, around this particular tip-tilt value, there is almost no isproposed. asymmetry visible in the inner region of the image, which ap- pears as a symmetric brightdoughnut(see Fig. 4). As a result, GiventhesuccessandbenefitsoftheQACITScontroller,ef- theobservercanbeeasilytrickedbythisapparentsymmetry,and fortsare currentlyongoingto developthesame kindofcontrol canconsiderthatthestarimageiscenteredontothevortexmask, looponotheroperationalinfraredvortexcoronagraphs,namely whileitisactuallyoffsetbyabout30masfromthevortexmask on the VLT/NACO, LBT/LMIRCam and VLT/VISIR instru- center. ments. The QACITS algorithm is also under study for imple- Incontrast,ontheOctobernights,themeanouterestimators mentation on the future mid-infrared ELT/METIS instrument areallsignificantlyclosertozero(asexpectedsincethecontrol (Brandletal.2014),whichincludesavortexcoronagraphinits loopwasbasedontheouterestimatoronly),butthemeaninner baselinedesign.Itcanbenotedthattheimplementationatother estimators still show a systematic offset, of amplitude 6.4mas wavelengths is straightforward: since the tip-tilt amplitude is on average. Assuming that this effect is due to a misalignment measuredinunitsofλ/D ,themodelisnotdependentonwave- L oftheLyotmask,andbasedonsimulationresults(Fig.11),this length.Asamatteroffact,theveryfirstlaboratorytestswitha amplitudeindicatesthattheouterestimatorisaffectedbyabias nonobstructedpupilwereperformedinKbandwiththevortex of4.4masonaverage(correspondingtoashiftoftheLyotstop coronagraphon PHARO at Palomar Observatory (Mawetetal. of∼1.1%,whichlieswithinthe specificationsofthealignment 2010), and observations at M band were recently successfully accuracy).It has to be noted, though,that this systematic error carried out with the vortex coronagraph at Keck/NIRC2. Even is relatively constantover the three nights, with a preferreddi- moregenerally,thebasic principleofthe method,initiallypro- rection. This means that both the science and reference targets posed by Masetal. (2012) for the Four Quadrant Phase Mask areaffectedinthesamewayto someextent,andtheimpacton withanonobstructedpupil,maypotentiallybeadaptednotonly differentialimagingtechniquesisthereforelimited(Mawetetal. to the vortex coronagraphbut also to other small IWA corona- 2017;Serabynetal.2017). graphsbased on a focal plane mask (e.g. the Dual-Zone Phase Articlenumber,page8of9 E.Hubyetal.:On-skyperformanceoftheQACITSpointingcontroltechnique Mask(DZPM,Soummeretal.2003)orthePhaseInducedAm- plitude Apodizer (PIAA, Guyon 2003)). The major adjustment concernsthecharacterizationoftheimagebehaviourinpresence of tip-tiltand morespecifically the definitionofthe underlying modelforthemeasureddifferentialflux,whichmaybederived eitheranalyticallyand/orempirically,basedonexperimentalcal- ibration. Lastly, we intend to use the same kind of method combin- ing the inner and outer estimators in the data processing.Post- monitoring of the data using QACITS can indeed provide a means to perform frame selection based on a centering quality criterion,andnotonlyonafluxcriterionsubjecttoseeingcon- ditions.Besides,theestimateofthetruestarimagepositionbe- hind the coronagraphicmask allows a better registrationof the framesandcanthuspotentiallyincreasethesignaltonoiseratio ofplanetspresentintherotatingfield. Acknowledgements. The research leading to these results has received fund- ingfromtheEuropeanResearchCouncilundertheEuropeanUnion’sSeventh FrameworkProgramme(ERCGrantAgreementn.337569)andfromtheFrench CommunityofBelgiumthroughanARCgrantforConcertedResearchAction. References Absil,O.,Mawet,D.,Karlsson,M.,etal.2016,Proc.SPIE,9908,99080Q Amara,A.&Quanz,S.P.2012,MNRAS,427,948 Baudoz,P.,Dorn,R.J.,Lizon,J.-L.,etal.2010,Proc.SPIE,7735,5 Beuzit,J.-L.,Feldt,M.,Dohlen,K.,etal.2008,Proc.SPIE,7014,701418 Bottom,M.,Femenia,B.,Huby,E.,etal.2016,Proc.SPIE,9909,990955 Bottom,M.,Wallace,J.K.,Bartos,R.D.,Shelton,J.C.,&Serabyn,E.2017, MNRAS,464,2937 Brandl,B.R.,Feldt,M.,Glasse,A.,etal.2014,Proc.SPIE,9147,21 GomezGonzalez,C.A.,Absil,O.,Absil,P.-A.,etal.2016,A&A,589,A54 Guyon,O.2003,A&A,404,379 Guyon,O.,Matsuo,T.,&Angel,R.2009,ApJ,693,75 Heintz,W.D.1993,A&AS,98,209 Huby,E.,Absil,O.,Mawet,D.,etal.2016,Proc.SPIE,990920 Huby,E.,Baudoz,P.,Mawet,D.,&Absil,O.2015,A&A,584,A74 Jovanovic,N.,Martinache,F.,Guyon,O.,etal.2015,PASP,127,890 Macintosh,B.,Graham,J.R.,Ingraham,P.,etal.2014,PNAS,111,12661 Mas,M.,Baudoz,P.,Rousset,G.,&Galicher,R.2012,A&A,539,A126 Mawet,D.,Choquet,É.,Absil,O.,etal.2017,AJ,153,44 Mawet,D.,Riaud,P.,Absil,O.,&Surdej,J.2005,ApJ,633,1191 Mawet,D.,Serabyn,E.,Liewer,K.,etal.2010,ApJ,709,53 Serabyn,E.,Huby,E.,Matthews,K.,etal.2017,AJ,153,43 Service,M.,Lu,J.R.,Campbell,R.,etal.2016,PASP,128,095004 Singh,G.,Lozi,J.,Guyon,O.,etal.2015,PASP,127,857 Singh,G.,Martinache,F.,Baudoz,P.,etal.2014,PASP,126,586 Soummer,R.,Dohlen,K.,&Aime,C.2003,A&A,403,369 Soummer,R.,Pueyo,L.,&Larkin,J.2012,ApJ,755,L28 VargasCatalán,E.,Huby,E.,Forsberg,P.,etal.2016,A&A,595,A127 Wallace,J.K.,Burruss,R.S.,Bartos,R.D.,etal.2010,Proc.SPIE,7736,5 Articlenumber,page9of9