Fabrication of millimeter-long structures in sapphire using femtosecond infrared laser pulses and selective etching

(1)

ContentslistsavailableatScienceDirect

Optics

and

Lasers

in

Engineering

journalhomepage:www.elsevier.com/locate/optlaseng

Fabrication

of

millimeter-long

structures

in

sapphire

using

femtosecond

infrared

laser

pulses

and

selective

etching

L. Capuano

a,∗

_,

_R.M.

_Tiggelaar

b

_,

_J.W.

_Berenschot

c

_,

_J.G.E.

_Gardeniers

c

_,

_G.R.B.E.

_Römer

a

a Chair of Laser Processing, Department of Mechanics of Solids, Surfaces & Systems (MS3), Faculty of Engineering Technology, University of Twente, P.O. Box 217, 7500 AE, Enschede, The Netherlands

b MESA + NanoLab cleanroom, MESA + Institute, University of Twente, P.O. Box 217, 7500 AE, Enschede, The Netherlands

c Mesoscale Chemical Systems, MESA + Institute, Faculty of Science and Technology, University of Twente, P.O. Box 217, 7500 AE, The Netherlands

a

b

s

t

r

a

c

t

Thispaperanalyzeslaserandetchingparameterstofabricateopenandcontinuousmicrochannelsandstacksofsuchmicrochannelsinthebulkofcrystallinesapphire (𝛼-Al2O3).Thestructuresareproducedusingatwo-stepmethodconsistingoflaserirradiationandselectiveetching.Infraredfemtosecondlaserpulsesarefocused inthebulktolocallyrenderthecrystallinematerialintoamorphous.Theamorphousmaterialis,then,selectivelyetchedinhydrofluoricacid.Amorphoussapphire showsahighetchingselectivityincomparisontoitscrystallinestate,whichmakesthismaterialveryattractiveforausewiththistechnique.However,someofits propertiesmaketheprocessingchallenging,especiallyduringthelaser-inducedamorphizationphase.Thispaperstudiestheeffectoflaserparametersbya step-by-stepapproachtofabricatelongstructures(longestdimensionsuptomillimeters)ofdifferentshapesinsidethebulkofsapphire.Theminimumcross-sectional dimensionsoftheresultingstructures(microchannels)varyfromfewhundredsofnanometersforthesmallestchannelstotensofmicrometersforthelargeststacks ofmicrochannels.Theeffectofthevariationofrepetitionrate,pulseenergyandchannel-to-channeldistanceonthemicrochannelsandstacksofmicrochannelsis studied.SEMmicrographsofpolishedcross-sectionsareusedforperformingaquantitativeandqualitativeanalysisofthemorphologyofthestructuresafterlaser irradiationand,subsequently,afterselectivewetchemicaletching.

1. Introduction

Crystalline sapphire (𝛼-Al2O3) is nowadays used as construction

componentorbase materialin many sectorsof scienceand technol-ogy.Thehardnessofsapphire(9ontheMohsscale[1])andits trans-parencyinthevisiblespectrum(from450nmto2000nm[2]),together withotherphysicalandchemicalpropertiesmakethematerialsuitable inmanyapplicationsintheﬁeldsofsemiconductors(particularlywith higheﬃciencyGalliumNitrideLEDs[3–10]),andinphotonicsin gen-eral[11–14].

Processingofsapphirehasbeendemonstratedusingdiﬀerent meth-ods:directlaserwriting[15,16],mechanicalsawing[17],dry(plasma etching[18–20])andwetetching[21–23].Inthismanuscriptwestudy atwo-stepmethodconsistingoflaserirradiationofcrystallinesapphire withconsequentmodiﬁcationoftheexposedmaterialintoamorphized materialandsuccessiveselectiveremovalofthelatterbywetetching.

Duetothetransparencyofsapphire,thelaserbeamcanbefocused insidethebulk.Iffemtosecondorpicosecondlaserpulsesareusedwith intensitiesintheorderof1013_–1014_W/cm2_[24–26]_absorbed_laser

en-ergy[27]leadstotheamorphizationofthecrystallinesapphire. Amor-phoussapphireisselectivelyetchablebyhydroﬂuoricacid(HF)ata105

fasterratethancrystalline[24–26,28–31].Ifthematerialisexposedto several,overlappinglaserpulses,itispossibletocreateregionsand vol-umesofamorphizedmaterial.

∗_{Corresponding}_author.

E-mailaddress:l.capuano@utwente.nl(L.Capuano).

Fig.1(a)showsacross-sectionalmodelofasinglemicrochannelin thebulkofsapphire,whichformsthetargetedbasicshapeforthisstudy. Suchstructurescanbeexploitedfor,e.g.microﬂuidicdevices– inthe form ofmillimeterlonghollowmicrochannels.Morecomplexshapes canbecreatedbysuperpositionofmicrochannels.

However,whenlaserpulsesaregeometricallyoverlappingwiththe aimtoformlargeamorphizedstructures– suchasone-dimensional mod-ifiedlines(madebyamorphousmaterial, beforeetching)– aseriesof phenomenamayaffecttheformationandmorphologyoftheamorphous sapphire.Thelatter,inturn,affectsthesolubilityoftheformedmaterial. Infact, publicationson etchedchannels in sapphire,often report cross-sectionsin which,afterthewetchemical etching,theobtained structureisnotcompletelyhollow[14,24,32,33].Thatis,inthese struc-tureshollow/openregions,whereamorphizedsapphireisdissolved,can befound,aswellascrystalline/unetchedregions.Theselatterregions canbecharacterizedbyseriesofparallelnanochannels,seeFig.1(b),or byadiscontinuedandirregularstructuresFig.1(c).In2008,Juodkazis etal.[32]studiedthisphenomenonandshowedthatoverlappingsingle pulsemodificationscausesrecrystallizationoftheamorphizedmaterial, which makesitnon-etchable byanacidlike HF.Ontheotherhand, Gottmannetal.[24]showedthechangeinmorphologyofcross-sections ofchannelsobtainedwiththismethod,andfoundthattheresultsoflaser irradiationandwetetchingdependsmainlyonlaserparametersand fo-cusingconditions.Morespecifically,theauthorsshowedthat,withina

https://doi.org/10.1016/j.optlaseng.2020.106114

Received21January2020;Receivedinrevisedform22March2020;Accepted29March2020 Availableonline30April2020

(2)

rangeofparameters(mainlynumericalapertureandenergyperpulse) thecross-sectionofchannelseithershowaseriesofamorphizedparallel nanochannels(Fig.1(b)),orhollowchannels(Fig.1(a)).

BothJuodkazisetal.[32]andGottmannetal.[24]studiedtheeffect ofonlyalimitednumberofprocessingparameters.Despitetheseresults, untilnow,awell-ordered studyregardingthemainfactorsplayinga roleonthefinalmorphologyoftheirradiatedlinesandotherstructures (formedbyoverlappinglines)inthebulkofsapphireislackinginthe literature.Awiderinvestigationis needed,infact,tounderstandthe problemandhaveageneralviewofwhichspecificconditionsdetermine thefinalshapeandappearanceandhowtotunethesettingstoobtain exactanddistincttypesofchannels.

2. Experimentalset-upandanalysistools

Fig.2showsaschematicoftheset-upusedforthelaserirradiation experimentspresentedinthiswork.

AKMLabsY-Fifemtosecondlasersourcewasused, whichemitsa linearlypolarizedlaserbeamatacentralwavelengthof𝜆 =1030nm. Thepulsedurationofthissourceis230fs,measuredwithan autocorre-lator(APEBerlinPulseCheck,Germany).Heattransmissionphenomena insidethelatticeofsapphireoccuronatimescaleintheorderoftens ofpicoseconds[34].Hence,theselectedultrashortpulseduration lim-itstheheattransmissioninsidethesampleduringthelaserpulse.The spatialdistributionofthelaserbeamisnearlyGaussian(M2_<_1.2).

Sincethelasersourcedoesnotoffertheoptionofchangingthe repe-titionratetolessthan1MHz,anelectro-opticmodulator(EOM,Model 360–80byConoptics,USA)ismountedafterthelasersourceforpulse picking.Abeamattenuator(UltrafastVersion,Altechna,Lithuania)is usedtosetthepulseenergyof thelaserpulses.Amicroscope objec-tive(11,101,666,LeicaMicrosystems,Germany,NA=0.7)isusedto focusthebeamtoaspotofabout0.9μmdiameter(calculated).The microscopeobjectiveismountedonalinearstage(ATS100,Aerotech USA),whichmovesthelensinthez-direction,allowingthefocalspot ofthelaserbeamtobepositionedrelativetothesurfaceofthesample. Thesampleisfixedwithavacuumchuckonxy-stages(ALS130-150, AerotechUSA).Themicroscopeobjectiveisalsousedasmagnifying el-ementforimaging;thatis,thelightreflectedbythesamplepassesagain throughtheobjectiveandatubelenstoarrivefinallyonaCMOScamera (DCC1545,Thorlabs,USA).Thisalignmentfacilitatesthepositioningof thefocalspotinthebulk(z-direction)ofthesampleaswellasinthe xy-plane.Italsoallowstoobservetheprogressandqualityofthe irra-diationduringmachining.

Afterirradiation,thesamplesareinspectedusingaKeyenceVHX 5000(Japan)microscopeforopticalandpolarizedmicroscopy,aswell as using a Scanning Electron Microscope (SEM, JEOL JSM 7200F, Japan).AfterHF-etching,samplesareanalyzedatthesamepositions asafterirradiation.

3. Materialsandmethods 3.1. Materials

Circularsapphirewafers(2inchesindiameter)withathicknessof 430μmandcrystalorientation(0001),purchasedfromCrystechGmbH (Germany),wereused.Thecircularsampleswerecutintorectangularly shaped(ofvarioussizes)stripsforeasierhandling.

Thewetchemicaletchant,hydroﬂuoricacidaqueoussolution50% (BASF,Germany)wasusedatroomtemperature.

3.2. Methods

Toinvestigateonedirectionalstructures,thesampleismovedinone direction(xoryinFig.3),whileexposingthesampletosingle femtosec-ondlaser.Thisresultsinamodiﬁedline/track,seeFig.3(a).

Thelaserbeamisfocusedinsidethebulkofsapphireatabout50μm belowthetopsurfaceofaspecimen.Thevelocityvofthestageiskept constantat1mm/s.Byvaryingthepulserepetitionrateofthelaser source,thegeometricaloverlapbetweenthelaserpulsescanbevaried. Inordertoproducestacksoflines,asisshowninFig.3(b),the ap-proachforsinglelinesisappliedrepeatedly.Thatis,stacksoflinesare producedbyoverlappingsinglemodiﬁedparalleltracks.The geomet-ricaloverlapofadjacenttracksislimitedbytheminimalincremental stepofthexystage,whichequalsabout50nm.Sincethesmallest cross-sectionaldimensionisabout600nm,itispossibletovarythelateral ge-ometricaloverlapbetweenadjacentlinesfrom90%downtoseparated lines(nooverlap).

Thegeometricaloverlapisdeﬁnedasthepercentageoftheoverlap ofdiametersofadjacentpulsesandiscalculatedas:

OL = ( 1− 𝑣 𝑓⋅ 𝐷 ) × 100%

WherevisthescanningspeedoftheXYstage,fthepulserepetitionrate ofthelasersourceandDisthe(calculated)diameterofthefocalspot.

Alltheline-basedstructuresareirradiatedclosetoanedgeofthe specimenforeasyinspection.Table1providesanoverviewoftherange oftheotherlaserparameters.

(3)

Fig.2. Schematicofthesetupusedduringthelaser irra-diationphaseoftheexperiments.

Fig.3.(a)Singlelinesareproducedbyexposingthe sampletolaserpulses,whiletranslatingthesampleat aconstantvelocityofv=1mm/s.(b)Astackof micro-linesiscreatedbylaterallyoverlappingsinglelines.

Table1

Setofexperimentalparameters.

Laser Polarization // Parallel to the direction of irradiation ⊥ Perpendicular to the direction of irradiation

Pulse repetition rate 0.001 MHz, 0.010 MHz, 0.050 MHz, 0.100 MHz, 0.200 MHz, 0.500 MHz, 1 MHz, 5 MHz, 10 MHz, 15 MHz

Laser pulse energy 94.5 nJ, 234 nJ, 457 nJ for repetition rates: 0.001 MHz ≤ f ≤ 1 MHz 18.9 nJ, 46.8 nJ, 91.4 nJ for repetition rates: 5 MHz ≤ f ≤ 15 MHz No. of stacked lines 2, 4, 8, 16, 32, 64, 128, 256

Cross-sectionsofthestructureswereobtainedbygrindingthe sam-plealongthedirectionshowninFig.4,andremovingenoughsapphire toexposetheamorphousmaterial.Subsequently,foreaseofinspection ofthecross-sections,thesamplewaspolishedtoopticalquality(average roughnessRa<5nm).Grindingandpolishingwerecarriedoutusinga

Tegramin(Struers,Netherlands)polishingapparatususingsilicon

car-bidepapersforthegrindinganddiamondpastesforthepolishing,with progressiveﬁnersteps.

Afterpolishing,thecross-sectionofeachlineisinspectedbySEM and analyzed. Afterwards, the irradiated samples are immersed in 50%HFforabout2htodissolvetheamorphousmaterial.After rins-ingin demineralizedwateranddrying,SEMinspection ofthe

(4)

cross-Fig.4.Opticalmicroscopyimage(topview)ofchannelsandstacksofchannels obtainedafterirradiationandetchinginHFatroomtemperatureforabout2h. Toexposethe(amorphized)channelsandstudythecross-sections,samplesare grindedandpolished.

sectionsofformedmicrochannels(andstacks),atthesamelocations,is performed.

Thelinesandstacksoflinesareirradiateduponvaryingthe polar-izationoftheincominglaserbeam,therepetitionrateofthelaserwith whichthepulsesaredelivered,theenergyperpulseandthenumberof overlappinglinesincaseofstacks(seeTable1)

4. Resultsanddiscussion

Fig.5showsthegeneralfeaturesofatypicalcross-sectionbefore(i.e. line)andafter(i.e.channel)thewetetchingstep.

Ascanbeobservedfromthisfigure,theshapeofthecross-section ofthemodifiedlineafterlaserirradiationgenerallyfollowstheoriginal shapeofthelaserfocalspot.InFig.5(a)amorphizedsapphireappears darkerintheSEMmicrographsthancrystallinesapphire.Aswas men-tionedinSection1,theamorphousregionisnothomogeneousinmany cases.Thatis,thelaser-affectedvolumeshowsbothamorphousregions

Fig.6.SEMmicrographofacross-sectionofasinglemodiﬁedlineobtained usingapulseenergyEp=91.4nJandapulserepetitionrateoff=15MHz.The irradiationwasperformedalongthedirectionperpendiculartothepolarization ofthelaserbeam.Atahighenergyperpulseandhighrepetitionratesavoidcan appearnearthetopofthelaser-aﬀectedvolume.Thelaserradiatedfromtopto bottomofthepicture.Thevelocityofthestagewasv=1mm/s,thegeometrical pulsetopulseoverlap(calculated)OL=99.999%.

andcrystallineregions.Forthisreason,thecross-sectionsofchannels obtainedaftertheetchinginFig.5(b)fallinthemodelinFig.1(c).

The“microexplosions” causedbytheabsorptionofthelaserenergy [35]inaveryshorttimearecausingcrackingofthecrystallinesapphire. Aswillbeshowninthenextparagraphs,mostofthemodiﬁedlinesare surroundedbycrackswhichmaychangeinlengthfromfewhundreds ofnanometersuptomillimeters,dependingmainlyonthepulseenergy andtherepetitionrate.Thesecracksarenotmodiﬁedinsizeand mor-phologybytheetchingprocess.

Moreover,athighrepetitionrates(typically15MHz)andhighpulse energies(typically91.4nJ),thepresenceofsmallvoidsmaybeobserved inthecross-sectionsofthelines(i.e.priortoetching)aswell,seeFig.6.

Fig.5.SEMmicrographsofatypicalcross-sectionof asinglemodiﬁedline(a)afterlaserirradiationata laserpulseenergyof457nJandarepetitionrateof 0.010MHzand(b)after2hetchinginhydroﬂuoric acidatroom temperature.The irradiationwas per-formedalong thedirection paralleltothe polariza-tionofthelaserbeam.Thelaserradiatesfromtopto bottomofthepicture.Thevelocityofthestagewas

v=1mm/s,thegeometricalpulsetopulseoverlap (calculated)OL=90.000%.

(5)

Fig.7.SEMmicrographs(toprow)andzoomedSEMmicrographs(bottomrow)ofcross-sectionsofmodiﬁedlines(non-etched)afterirradiationobtainedusing: (a)f=0.1MHzandEp=94.5nJ,(b)1MHzandEp=94.5nJ,(c)5MHzandEp=91.4nJ,(d)10MHzandEp=91.4nJ.Theamorphizedparallelnanochannels increaseinsizeandnumberwithincreasingpulserepetitionrate.Inallthepicturesthesampleisirradiatedalongthedirectionperpendiculartothepolarizationof thelaser.Thelaserradiatedfromtoptobottomofthepicture.Thevelocityofthestagewasv=1mm/s,thegeometricalpulsetopulseoverlap(calculated)varied fromOL=99.000%forf=0.100MHztoOL=99.990%for10MHz.

Asmentioned,thelinearvelocityoftheXYstagesmovingthe sam-plewas 1mm/s, whereas thepulse repetitionratewas varied from 0.001MHzto15MHz,correspondingrespectivelytogeometricalpulse topulseoverlapsrangingfrom0%(completelyseparatedsingle modiﬁ-cations)to99.999%.

4.1. Structuresparallelorperpendiculartothepolarizationofthelaser light

Thecross-sectionalshapeoftheirradiatedandetchedstructuresis aﬀectedbytheanglebetweenthe(linear)polarizationdirectionofthe laserradiation[24,28,36–38]andthedirectionofirradiation(direction alongwhichthestagemoves).

Hnatovskyetal.[38]demonstratedthatmodifyingthebulkoffused silicausinglinearlypolarizedlaserpulsesresultsinperiodicstructures orientedinadirectionperpendiculartotheirradiationdirection.

Tayloretal.[39]explainedtheformation,duringprocessing,of pe-riodicstructuresperpendiculartothepolarizationofthelaserradiation, usingthetransientnanoplasmonicsmodel.Thismodeltheorizesthe for-mationofionizationhotspotsduringtheirradiationbyultrashortlaser pulses,whicheventuallyinducesplasma.Thesehotspotsmayleadtoa preferentiallocalionizationofthematerial.Inparticular,ﬁeld enhance-mentontheboundaryofthegeneratedionizedspotsfacilitatesthe gen-erationofplasmainthedirectionperpendiculartothepolarizationof thelaserlight.

Tostudytheeﬀectoftheorientationofthepolarizationonthe irra-diatedlinesandetchedchannels,structureswereproducedbothparallel andperpendiculartothepolarizationofthelaserradiation.

Fig. 8.SEM micrographof across-sectionof a singleirradiatedline (non-etched)obtainedusingEp=91.4nJandf=15MHzalongthedirection perpen-diculartothepolarizationofthelaserbeambeforeetching.Thelaserradiates fromtoptobottomofthepicture.Thevelocityofthestagewasv=1mm/s,the geometricalpulsetopulseoverlap(calculated)OL=99.999%.

Ifthelaserpolarizationisperpendiculartothescan-directionofthe stage, thepresenceof parallelnanochannels isobserved in thefocal region(Fig.1(b)),whicharenotobservediftheirradiationdirectionis

(6)

Fig.9. SEMmicrographsofcross-sectionsafterirradiationoflinesandafteretchingofchannelsobtainedatdiﬀerentrepetitionrates:(a)f=0.001MHz,(b) 0.010MHz,(c)0.050MHz,(d)0.100MHz,(e)0.200MHz,(f)0.500MHz,(g)1MHz,(h)5MHz,(i)10MHz,(l)15MHz.Thelaserradiatedthesamplesfromtop tobottomofthepictures.Thevelocityofthestagewasv=1mm/s,thegeometricalpulsetopulseoverlap(calculated)variedfromcompletelydetachedpulsesat

(7)

Fig.10. SEMmicrographsofcross-sectionsafterirradiation oflinesandafteretchingofchannelsobtainedusingapulse repetitionrateoff=1MHzatdiﬀerentenergiesperpulse: (a)Ep=457nJ,(b)234nJ,(c)94,5nJ.Thelaserradiated fromtoptobottomofthepicture.Thevelocityofthestagewas

v=1mm/s,thegeometricalpulsetopulseoverlap(calculated) OL=99.900%.

paralleltothepolarizationdirection.Theseregularamorphizedparallel nanolinesareobservedforpulserepetitionratesover0.1MHzandare morepronouncedwhenarepetitionrateofatleast1MHzisapplied, seeFig.7.

Upon increasingtherepetition rate,anincrease in both horizon-tal andvertical dimension of the cross sections and in the number ofnanolinesperareaisobserved.However,atapulserepetitionrate of 15 MHz,a circularshape is observed frequently,see Fig.8. This

spherical shape impedes the propagationof the focused laser beam deeperintothesample.Thisphenomenonwillbeexplainedindetailin Section4.2.

Objectiveofthispaperistoformhollow/openmicrochannelsand stacksofhollow/openmicrochannels.Hence,theoriginandgrowthof parallelnanolines/nanochannelsareoutofthescopeofthisstudy.For thisreason,inthenextsections,onlyresultsobtainedwiththe irradia-tiondirectionparalleltothepolarizationofthelaserarediscussed.

(8)

Fig.11. Polarizedopticalmicroscopyimage(topview)ofchannels(dark)andstacksofchannels(dark),surroundedbystressﬁelds(colored).Thechannelswere obtainedafterirradiationandetchinginHFatroomtemperatureforabout2h.Thenumberofstackedchannelsrangefrom1(extremeleft)to128(extremeright). Thechannelsweremadeusingapulserepetitionrateof0.200MHzandapulseenergyof94,5nJ.Thevelocityofthestagewasv=1mm/s,thegeometricalpulseto pulseoverlap(calculated)OL=99.500%.(Forinterpretationofthereferencestocolorinthisﬁgurelegend,thereaderisreferredtothewebversionofthisarticle.)

Fig.12. SEMmicrographsofcross-sectionsafter irra-diationofa stackof 256lines(a)andthestructure leftafteretching(b)obtainedusingarepetitionrate of0.100MHzat94,5nJ.Acracktowardsthecenter ofthestackoflinesishavinga“shieldingeﬀect” on thematerialbelowresultinginlocallynoformation ofamorphoussapphire.(c)isanopticalmicroscopy pictureintopviewofthesameetchedchannel.The laserradiatedfromtoptobottomofthepictures(a) and(b).Foreachline/channelthevelocityofthestage wasv=1mm/s,thegeometricalpulsetopulseoverlap (calculated)OL=99.000%.

4.2. Eﬀectofpulserepetitionrateonlinesandstacksoflines

Fig.9showstheeﬀectsofpulserepetitionrateonthecross-sections ofsinglelines(onlyirradiation)andchannels(afterHF-etchingoflines). Sincethevelocityof thestagesis constantat1mm/s,the geometri-calpitchbetweenthelaserpulseschangesfromsingleisolatedpulses

(atf=0.001MHz)toageometricaloverlapof99.999%betweenlaser pulses(atf=15MHz).

Atf=0.001MHz,thelaserpulsesarenotgeometricallyoverlapping. ThisisalsoreﬂectedinFig.9(a),whereaclearcross-sectionoftheline inthefocalregioncannotbeidentiﬁed.At0.010MHzthegeometrical overlapbetweenthelaserpulsesisover80%(overlapcalculatedwhen

(9)

Fig.13. SEMmicrographofacross-sectionbeforeetchingofastackof2 irradi-atedlines(non-etched)obtainedusingapulseenergyof94,5nJandarepetition rateof0.200MHz.Themismatchinrefractiveindexesbetweencrystallineand amorphoussapphireiscausingaseparationbetweenline1(theﬁrsttobe ir-radiated)andline2.Thelaserradiatesfromtoptobottomofthepicture.The velocityofthestagewasv=1mm/s,thegeometricalpulsetopulseoverlap (calculated)OL=99.500%.

anenergyperpulseof94,5nJwasusedwithaneﬀectiveminimal cross-sectionaldimensionof0.6μmbeforeandafteretching).However,when arepetitionratebelowf=0.100MHzisused(Fig.9(b)and(c)),the obtainedirradiatedlinesshowirregularcross-sections-i.e.amorphous sapphirealternateswithcrystallinetoformanirregularpatternsuchas theoneseeninFig.1(c).

Thisphenomenonmaybe explainedby recrystallizationof previ-ouslyamorphizedmaterial.Thatis,evenaftertheﬁrstlaserpulse,each subsequent,overlappinglaserpulseisirradiatingavolumeofboth crys-tallineandamorphizedmaterial.AswasreportedbyJuodkazisand Mi-sawa[32],thismayrecrystallizepartsoftheamorphizedmaterial, leav-ingitnon-solublebytheetchant.

Anotherhypothesisisthattheoverlapbetweenpulsesisnot suffi-cienttocausehollowchannelswithaconstantwidth,becausethe mate-rialamorphizedbypreviouslaserpulsesisopticallydistortingthe prop-agationandprofileofthelaserbeam.Thiseffectwillbediscussedmore indetailinSection4.4.

Forpulserepetitionratesrangingfromf=0.100MHzto1MHzthe cross-sectionsinFig.9(d)–(g)showopenchannelswithaconstant cross-sectionalongtheirlength.Thesearethetargetedhollowmicrochannels showninFig.1(a).Intheseﬁgures,thecross-sectionalwidthincreases from0.6μmatf=0.100MHzto1μmat1MHz.

Fig.9(h),(i),(l)showsthat,forpulserepetitionratesof5MHzand higher,thecross-sectionsoflines/channelsrevealadisruptedand frag-mentedmorphology.Attheserepetitionrateslargecrackshaveastrong effect,detrimentalonthepropagationandfluenceprofileoftheincident laserbeam,and,inturn,ontheamorphizationofthematerial.

Moreover,atf=15MHz,thelaser-affectedregionsshowa(nearly) circularshapenear thetopof thelaseraffectedvolumeandsmaller featuresbelowit,seeFig.9(l),aswellasFig.8.Thisphenomenoncan possiblybeassociatedwiththefindingsof Gamalyetal.[40].These authorsreportedtheformationofplasmaattheapexofthefocalspot duringtheionizationofthematerialbothathighrepetitionrateandat highpulseenergies.Thisplasmais,mostlikely,preventingexposureof materialbelowittothelaserradiation.

4.3. Eﬀectofenergyperpulse

Forpulserepetitionratesup to1MHz,thesamplewasirradiated usingthreepulseenergies:94.5nJ,234nJ,457nJ.Belowthesmallest pulseenergy,thematerialisnotaﬀectedbythelaser(atthementioned repetitionrates),whereasatthehighestpulseenergyitisnotpossible toirradiatesapphirewithouttheformationofcrackssolargethatcause thebreakingofthespecimen.Forpulserepetitionratesoff=5MHz, 10MHzand15MHzlowerpulseenergieswereappliedbecauseofthe powerlimitationsofthelasersource.Atthesepulserepetitionrates ex-perimentswereperformedatEp=18.9nJ,46.8nJand91.4nJ.

Fig.10shows theeffect ofthepulseenergyon thecross-sections ofirradiatedlinesandetchedchannels.Theshapeoftheamorphized regiondeviatesfromtheshapeofthefocalspotandthecross-sections showmultiplefoci,withincreasingpulseenergy.Thelattercanbe as-sociatedwithKerr-inducedself-focusing.Self-focusinginducedbythe electro-opticKerreffectisachangeoftherefractiveindexcausedby anappliedstrongelectricalfield,inthiscasethelaserradiation.This changeintherefractiveindexcausesthefocalspottoelongate.Ifthe focusingduetotheKerreffectiscounterbalancedbyadefocusingeffect duetothepresenceofplasmainthefocalspot[40],whichlowersthe refractiveindex,thismayresultinspatialfocusing/defocusing (multi-foci)alongthepropagationaxisofthelaserbeam.Thepowerthreshold PcritabovewhichKerreffectistriggeredcanbeexpressedas[35]: 𝑃𝑐𝑟𝑖𝑡=

𝜆2 0

2𝜋𝑛0𝑛2

where𝜆0denotesthelaserwavelength,n0thelinearrefractiveindex

(n0=1.755at1030nm[41])ofthematerialandn2thenonlinear

re-fractiveindex(n₂=3⋅10−16 _cm2_/W_[42]_)._Hence,_P

crit ≈3.02⋅106W,

whichhasthesameorderofmagnitudeoftheappliedpeakpowersin thiswork,especiallyatEp=234nJand457nJ.Thisconﬁrmsthe

possi-bleoccurrenceofself-focusing/multifociattheselevelsofpulseenergy. 4.4. Overlappingofthemodiﬁedlines

Foreachrepetitionrateandpulseenergy,singlelineswere over-lappedlaterallyforproducingstacks.Thenumberofstackedlinesper structureproducedis2n_,_with_n₌₁_to₈_(see_Table₁_)._However,_at_high

pulseenergies,itisoftennotpossibletogoabove8–16linesbecauseof severecrackingthathampersproperirradiation.

Apitchof50nm(correspondingtoanoverlapof90%)asthe lat-eralshiftbetweentheadjacentlinesisused.Atlowpulseenergies,itis possibletocreatestacksofupto256lines,althoughotherfactorsare inﬂuencingtheformationoffullyempty(withnoresidualcrystalline materialleft)stacksofmicrochannels.

Accumulatedstressinsidetransparentmaterialsisknowntocause birefringence[43,44].Stressinducedbylaserprocessingcanfacilitate theetchingofthematerialitself[45].Inthiscase,though,thedifference inrefractiveindicesbetweenunprocessedcrystalline,stress-affectedand amorphousmaterialwillmostlikelycauseadistortionoftheintensity profileoftheincidentlaserbeam.Fig.11showstheetchedchannels surroundedbystressinthecrystallinesapphire(andconsequent bire-fringence)madevisible bypolarizedlight microscopy.Althoughthis imagedoesnotprovidequantitativeresults,itcanbeconcludedthat thesurroundingstressincreasessignificantlywithincreasingnumberof channels.

Cracksadditionallyinterferewiththeformationofamorphous ma-terial.Crackedmaterial,infact,iscomposedofnormalcrystalline sap-phire,voidsandstressedcrystallinesapphire.Italsohas,therefore,a mixofrefractiveindices,whichareaﬀectingthefocusingofthelaser beam.Thisis,inmostcases,havingashieldingeﬀectonthecrystalline material,whichisdirectlybelowthecrack,thuspreventingits amor-phization.

Fig.12 showsanexample ofthis:acrack occurringatabout the centerofthestacksofmicrolineswas“shielding” thematerial

(10)

under-Fig.14. SEMmicrographsofcross-sectionsafterirradiationofstacksoflinesandthestructuresleftafteretchingobtainedatarepetitionrateof0.500MHzanda pulseenergyof94,5nJ.Thenumberofsinglestructuresoverlappedisinorder:(a)1,(b)2,(c)4,(d)8,(e)16,(f)32,(g)64,(h)128.Thelaserradiatedfromtop tobottomofthepictures.Thevelocityofthestagewasv=1mm/s,thegeometricalpulsetopulseoverlap(calculated)OL=99.800%.

neath:thelinesirradiatedafter/nexttothelineduringwhichthecrack wasgeneratedexhibitasmallercross-sectionwhich,inturn,causesan “interruption” inthestructureitself.

Fig.12(c)shows anopticalmicroscopy pictureofthesame chan-nel(topview,afteretching)showingthatthecrack,whichpropagated longitudinally,maintainedtheshieldingeﬀectforalmosttheentire re-mainderoftheirradiatedstack.

Giventhestep resolutionof 50 nmof thesetupandbecausethe minimalcross-sectionaldimensionofsinglelinescomposingthestack is in theorder of few hundred nanometers,itshould be possible to obtainstacksofmicrochannelswithasingleanduninterrupted cross-section.However,amorphizedsapphireofapreviouslyirradiatedline hasalowerrefractiveindexcomparedtothecrystallinesapphire[46]. Thisindexdifferencehasadeflectingeffectontheincomingbeamthat irradiatesthefollowingline.Theresultisthatsubsequentlinesareoften

(11)

irradiatedataslightlytiltedangle,whichyieldsaseparationbetween adjacentlines(Fig.13).

Finally,inFig.14anoverviewispresentedofaseriesofstacksof microchannelsproducedwitharepetitionrateof0.500MHzandapulse energyEp=94.5nJ,startingfromasinglechannel(Fig.14(a))upto 256overlappingchannels(Fig.14(h)).Theshieldingeﬀectcausedby thecracksisprominentinFig.14(g)and(h),whiletheseparationofthe channelscausedbymismatchoftherefractiveindicesisvisiblestarting from(d)andinvaryingdegreesforeveryothercase.Overall,theresults forexperiments withmorethan64 linesshow thepresenceoflarge cracks.

5. Conclusions

Astudyhasbeenperformedonthefabricationofmicrochannelsand stacksofmicrochannelsinsidethebulkofsapphire.Adetailedstep-by stepapproachregardingtheformationofsuchchannelswithan anal-ysisofthemainfactorsplayingaroleontheﬁnalmorphologywas,to thebestofourknowledge,missing.Thestudyincludedtheeﬀectsof polarizationof thelight,repetitionrate,pulseenergyandnumberof stackedlinesonthemorphologyandappearanceofthestructuresafter irradiationandwetchemicaletching.Themainresultsfoundarethe following:

• Theﬁrstpartoftheinvestigationregardedthedirectionofthe irra-diation:parallelorperpendiculartothepolarizationofthelight.It wasfoundthat,ifthesampleisirradiatedalongthedirection per-pendiculartothepolarizationofthelight,theirradiatedlinesdonot showasmoothsingleamorphizedcross-section,butratheraseriesof verticalamorphizedparallelnanolinespropagatingalongthewhole lengthofthechannel.Suchstructureswerenotinlinewiththe ob-jectiveofthestudy;therefore,nextexperimentswererestrictedto irradiationparalleltothelightpolarization.

• Withthisarrangement,thenextphasecomprisedstudyingtheeffect ofrepetitionrateontheobtainedstructures.Itwasfoundthatthe idealwindowoflaserirradiationisbetween0.100MHzand1MHz. Belowthisrangethestructuresdonotshowasingleandconstant cross-section,butfragmented.Uponusingarepetitionratehigher thanthisrange,themodifiedlinesalsoshowirregularcross-sections. Infact,theyaredisruptedandoftencontainacircularshapeontopof thefocalregionwhichisshieldingthelower/deeperlocatedmaterial andpreventingittobemodified.

• Theeﬀectofpulseenergywasinvestigated,anditwasfoundthat, ifapulseenergyofmorethan234nJforpulserepetitionratesof (f=0.001to1MHz)isused,thefocusissplitinmultiplefoci.The splittingismostprobablycausedbytheKerr-eﬀectand,asa con-sequence,thecross-sectionsofthechannelsafteretchingarenotas targeted.

• Finally,stacksof microlineswerestudiedbyvaryingthenumber ofsinglelinescomposingthem.Ingeneral,thepresenceofcracks preventtheformationofhollowstacksofmicrochannelsandlimits itssizeto64laterallyoverlappingchannels.Moreoverapossible non-suﬃcientoverlapping(technicallylimitedbythesetup)often causesseparationofadjacentchannels.

This work demonstrates the possibility of controlling the cross-sectionalshapeof channelsobtainedin sapphireusinga double-step processingtechniquebasedonfemtosecondpulsedlaserirradiationand selectiveetchinginhydroﬂuoricacid.Itisbelievedthatstructureswith ahollow,continuousandconstant(alongthelengthofthestructure) cross-sectioncanbeusedformicroﬂuidicapplications.

DeclarationofCompetingInterest None.

CRediTauthorshipcontributionstatement

L. Capuano:Conceptualization,Methodology,Validation, Investi-gation,Datacuration,Formalanalysis,Writing-originaldraft,Writing -review& editing,Visualization.R.M.Tiggelaar:Conceptualization, Methodology, Validation, Resources,Writing -review & editing, Su-pervision.J.W.Berenschot:Conceptualization,Methodology, Valida-tion,Resources,Writing-review&editing,Supervision.J.G.E. Garde-niers:Conceptualization,Methodology,Validation,Resources,Writing -review& editing,Supervision.G.R.B.E. Römer:Conceptualization, Methodology,Validation,Resources,Writing-review&editing, Super-vision,Projectadministration,Fundingacquisition.

Acknowledgements

Theprojectleadingtothispublicationhasreceivedfundingfrom theEuropeanUnion’sHorizon2020researchandinnovationprogram undertheMarieSkłodowska-CuriegrantagreementNo.675063. References

[1] Dobrovinskaya ER, Lytvynov LA, Pishchik V. Application of Sapphire. In: Sapphire. Boston, MAUS: Springer; 2009. p. 95. doi: 10.1007/978-0-387-85695-7_1 . [2] Kyocera Single crystal sapphire datasheet - Kyocera. Single Cryst Sapphire - Kyocera

2011. doi: 10.1063/1.3050841 .

[3] Lei T, Ludwig KF, Moustakas TD. Heteroepitaxy, polymorphism, and fault- ing in GaN thin ﬁlms on silicon and sapphire substrates. J Appl Phys 1993. doi: 10.1063/1.354414 .

[4] Kelly D, Brindle C, Kemerling C, Stuber M. The state-of-the-art of silicon-on-sapphire CMOS RF switches. Tech Dig - IEEE Compd Semicond Integr Circuit Symp CSIC 2005. doi: 10.1109/CSICS.2005.1531812 .

[5] Yoshida S, Misawa S, Gonda S. Improvements on the electrical and luminescent properties of reactive molecular beam epitaxially grown GaN ﬁlms by using AlN- coated sapphire substrates. Appl Phys Lett 1983;42:427–9. doi: 10.1063/1.93952 . [6] Wong WS, Sands T, Cheung NW. Damage-free separation of GaN thin ﬁlms from

sapphire substrates. Appl Phys Lett 1998. doi: 10.1063/1.120816 .

[7] Craven MD, Lim SH, Wu F, Speck JS, Denbaars SP. Structural characterization of nonpolar (112 ̄0) a-plane GaN thin ﬁlms grown on (11 ̄02) r-plane sapphire. Appl Phys Lett 2002. doi: 10.1063/1.1493220 .

[8] Wu ZH, Fischer AM, Ponce FA, Bastek B, Christen J, Wernicke T, et al. Struc- tural and optical properties of nonpolar GaN thin ﬁlms. Appl Phys Lett 2008. doi: 10.1063/1.2918834 .

[9] Halstuch A, Westreich O, Sicron N, Ishaaya AA. Femtosecond laser inscription of Bragg gratings on a thin GaN ﬁlm grown on a sapphire substrate. Opt Lasers Eng 2018. doi: 10.1016/j.optlaseng.2018.05.006 .

[10] Reklaitis I, Grinys T, Toma š iunas R, Puod ž iunas T, Ma ž ule L, Sirutkaitis V, et al. A new geometrical approach for rapid led processing by using femtosecond laser. Opt Lasers Eng 2015. doi: 10.1016/j.optlaseng.2015.05.002 .

[11] Singh N, Hudson DD, Eggleton BJ. Silicon-on-sapphire pillar waveguides for Mid-IR supercontinuum generation. Opt Express 2015. doi: 10.1364/oe.23.017345 . [12] Wang J, Zhou Y, Watkinson M, Gautrot J, Krause S. High-sensitivity light-

addressable potentiometric sensors using silicon on sapphire functionalized with self-assembled organic monolayers. Sensors Actuators, B Chem 2015. doi: 10.1016/j.snb.2014.11.071 .

[13] Lioubtchenko D, Dudorov S, Mallat J, Tuovinen J, Räisänen AV. Low-loss sapphire waveguides for 75-110 GHz frequency range. IEEE Microw Wirel Componen Lett 2001. doi: 10.1109/7260.928929 .

[14] Kaiser M, Kumkar M, Leute R, Schmauch J, Priester R, Kleiner J, et al. Selective etching of ultrafast laser modiﬁed sapphire. In: Ra čiukaitis G, Makimura T, Molpeceres C, editors. Laser appl. microelectron. optoelectron. manuf. xxiv. SPIE; 2019. p. 14. doi: 10.1117/12.2509415 .

[15] Kudrius T, Š lekys G, Juodkazis S. Surface-texturing of sapphire by femtosecond laser pulses for photonic applications. J Phys D Appl Phys 2010. doi: 10.1088/0022-3727/43/14/145501 .

[16] Shamir A, Ishaaya AA. Large volume ablation of Sapphire with ultra-short laser pulses. Appl Surf Sci 2013;270:763–6. doi: 10.1016/j.apsusc.2013.01.153 . [17] Matsumaru K, Takata A, Ishizaki K. Advanced thin dicing blade for sapphire sub-

strate. Sci Technol Adv Mater 2005;6:120–2. doi: 10.1016/j.stam.2004.11.002 . [18] Park S-H, Jeon H, Sung Y-J, Yeom G-Y. Refractive sapphire microlenses fab-

ricated by chlorine-based inductively coupled plasma etching. Appl Opt 2001. doi: 10.1364/ao.40.003698 .

[19] Zhou S, Liu S. Study on sapphire removal for thin-ﬁlm LEDs fabrication using CMP and dry etching. Appl Surf Sci 2009. doi: 10.1016/j.apsusc.2009.07.063 . [20] Dutta A, Kinsey N, Saha S, Guler U, Shalaev VM, Boltasseva A. Plasmonic inter-

connects using zirconium nitride. In: Conf. lasers electro-optics. CLEO; 2016. 2016, 2016 .

[21] Wuu DS, Wang WK, Wen KS, Huang SC, Lin SH, Horng RH, et al. Fabrication of pyra- midal patterned sapphire substrates for high-eﬃciency InGaN-based light emitting diodes. J Electrochem Soc 2006;153:G765–70. doi: 10.1149/1.2209587 .

(12)

[27] Capuano L, de Zeeuw, Römer GRBE D. Towards a numerical model of picosecond laser-material interaction in bulk sapphire. J Laser Micro Nanoeng 2018:13. doi: 10.2961/jlmn.2018.03.0005 .

[28] Hörstmann-Jungemann M, Gottmann J, Keggenhoﬀ M. 3D-microstructuring of sapphire using fs-laser irradiation and selective etching. J Laser Micro Nanoeng 2010;5:145–9. doi: 10.2961/jlmn.2010.02.0009 .

[29] Gamaly EG, Juodkazis S, Nishimura K, Misawa H, Luther Davies B. Laser-matter interaction in the bulk of a transparent solid: conﬁned microexplosion and void formation. Phys Rev B 2006;73:214101. doi: 10.1103/PhysRevB.73.214101 . [30] Wortmann D, Gottmann J, Brandt N, Horn-Solle H. Micro- and nanostructures inside

sapphire by fs-laser irradiation and selective etching. In: Conf. quantum electron. laser sci. conf. lasers electro-optics. CLEO/QELS; 2008. 2008 .

[31] Moser R, Ojha N, Kunzer M, Schwarz UT. Sub-surface channels in sapphire made by ultraviolet picosecond laser irradiation and selective etching. Opt Express 2011;19:24738. doi: 10.1364/OE.19.024738 .

[32] Juodkazis S, Misawa H. Laser processing of sapphire by strongly focused femtosecond pulses. Appl Phys A Mater Sci Process 2008. doi: 10.1007/s00339-008-4763-0 . [33] Matsuo S, Shichijo Y, Tomita T, Hashimoto S. Laser fabrication of ship- in-a-bottle microstructures in sapphire. J Laser Micro Nanoeng 2007. doi: 10.2961/jlmn.2007.02.0001 .

fabrication in fused silica. Opt Lett 2005. doi: 10.1364/ol.30.001867 .

[39] Taylor R, Hnatovsky C, Simova E. Applications of femtosecond laser induced self- organized planar nanocracks inside fused silica glass. Laser Photonics Rev 2008. doi: 10.1002/lpor.200710031 .

[40] Gamaly EG, Rapp L, Roppo V, Juodkazis S, Rode AV. Generation of high energy density by fs-laser-induced conﬁned microexplosion. New J Phys 2013:15. doi: 10.1088/1367-2630/15/2/025018 .

[41] Malitson IH. Refraction and dispersion of synthetic sapphire. J Opt Soc Am 1962;52:1377. doi: 10.1364/JOSA.52.001377 .

[42] Major A, Yoshino F, Nikolakakos I, Aitchison JS, Smith PWE. Disper- sion of the nonlinear refractive index in sapphire. Opt Lett 2004;29:602. doi: 10.1364/OL.29.000602 .

[43] Fan H, Ryu M, Honda R, Morikawa J, Li ZZ, Wang L, et al. Laser-Inscribed stress- induced birefringence of sapphire. Nanomaterials 2019. doi: 10.3390/nano9101414 . [44] Fernandes LA, Grenier JR, Herman PR, Aitchison JS, Marques PVS. Stress induced birefringence tuning in femtosecond laser fabricated waveguides in fused silica. Opt Express 2012. doi: 10.1364/oe.20.024103 .

[45] Agarwal A, Tomozawa M. Correlation of silica glass properties with the infrared spectra. J Non Cryst Solids 1997. doi: 10.1016/S0022-3093(96)00542-X . [46] Boidin R, Halenkovi č T, Nazabal V, Bene š L, N ěmec P. Pulsed laser deposited alumina