reflow是什么意思low在线翻译读音例句

Reducedpropagationlossinsiliconstrip

andslotwaveguidescoatedbyatomic

layerdeposition

rela,,ti,2A.Saynatjoki,en,1

,ld,,en1,3

1DepartmentofMicroandNanosciences,AaltoUniversitySchoolofElectricalEngineering,

POBox13500,FI-00076Aalto,Finland

2InstituteofPhotonicsandQuantumElectronicsandInstituteofMicrostructureTechnology,

KarlsruheInstituteofTechnology(KIT),76131Karlsruhe,Germany

3DepartmentofPhysicsandMathematics,UniversityofEasternFinland,POBox111,

FI-80101Joensuu,Finland

*rela@aalto.fi

Abstract:Whensiliconstripandslotwaveguidesarecoatedwitha50nm

amorphoustitaniumdioxide(TiO

2

)film,measuredlossesatawavelength

of1.55mcanbeaslowas(21)dB/cmand(72)dB/cm,respectively.

Weuseatomiclayerdeposition(ALD),estimatetheeffectofALDgrowth

onthesurfaceroughness,anddiscusstheeffectonthescatteringlosses.

BecausethegapbetweentherailsofaslotwaveguidenarrowsbytheTiO

2

deposition,

usefulforpreciseadjustmentiftheslotistobefi,anonlinear

organicmaterialorwithasensitizerforsensorsapplications.

2011OpticalSocietyofAmerica

OCIScodes:(130.3130)Integratedopticsmaterials;(220.4241)Nanostructurefabrication;

(230.7370)Waveguides;(240.5770)Roughness;(310.2785)Guidedwaveapplications.

Referencesandlinks

a,,s,,“Guidingandconfininglightinvoidnanostructure,”Opt.

Lett.29,1209–1211(2004).

,a,cci,,“Experimentaldemonstrationofguidingandconfining

lightinnanometer-sizelow-refractive-indexmaterial,”.29,1626–1628(2004).

s,on,B.Sanchez,,rom,o,l,“Slot-waveguide

biochemicalsensor,”.32,3080–3082(2007).

man,“Label-freebiosensingwithaslot-waveguide-basedringresonatorinsilicononinsulator,”IEEE

Photon.J.1,197–204(2009).

rela,A.Saynatjoki,ainen,en,“Featuresizereductionofsiliconslotwaveguides

bypartialfillingusingatomiclayerdeposition,”.48,080502(2009).

rela,moinen,en,joki,en,rg,nen,-

nen,“Atomiclayerdepositedtitaniumdioxideanditsapplicationinresonantwaveguidegrating,”.49,

4321–4325(2010).

ld,,J.-,,,o,a,ich,,

,“Siliconorganichybridtechnology-aplatformforpracticalnonlinearoptics,”97,

1304–1316(200关于描写长江的诗句 9).

,u,tis,,ts,,son,o,obu,

ich,,ld,“All-opticalhigh-speedsignalprocessingwithsilicon-organichybrid

slotwaveguides,”ics3,216(2009).

n,,,,el,ld,,“Radiationmodesand

roughnesslossinhighindex-contrastwaveguides,”mElectron.12,1306–1321(2006).

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11529

,“Atheoreticalanalysisofscatteringlossfromplanaropticalwaveguides,”Opt.

QuantumElectron.26,977–986(1994).

,,H.-,l,,ing,“Effectofsizeandroughnesson

lighttransmissioninaSi/SiO

2

waveguide:Experimentsandmodel,”.77,1617(2000).

,,ing,,a,“Fabricationofultralow-lossSi/SiO

2

waveguides

byroughnessreduction,”.26,1888–1890(2001).

in,r,ing,“Siliconwaveguidesidewallsmoothingbywetchemicaloxidation,”

aveTechnol.23,2455–2461(2005).

,age,te,agne,,n,,,“Correla-

tionofscatteringloss,sidewallroughnessandwaveguidewidthinsilicon-on-insulator(SOI)ridgewaveguides,”

aveTechnol.27,3999–4008(2009).

t,,,,“Propagationlossinsingle-modeultrasmallsquaresilicon-on-

insulatoropticalwaveguides,”aveTechnol.24,891–896(2006).

wen,ttens,,,man,out,,“Focused-ion-

beamfabricationofslotsinsiliconwaveguidesandringresonators,”.20,2004–2006

(2008).

t,,,,,“Sizeinfluenceonthepropagationlossinducedby

sidewallroughnessinultrasmallSOIwaveguides,”.16,1661–1663(2004).

bsiliconphotonicsshuttleservice,/.

VEbyPhotonDesign,/.

,,n,ld,,“Nonlinearsilicon-on-insulatorwaveguidesfor

all-opticalsignalprocessing,”s15,5976–5990(2007).

uction

Siprocesses

areavailableforfabricatingelectronicsusingthesiliconplatform,anditishighlypromising

toexploitthesametechnologyalsoforphotonicsatwavelengthsnear1.55ider

twobasicwaveguidetypes,ehighindex-contrast

structures,wherelightisconfinedeitherinthehigh-indexmaterialofthestrip,orinthelow-

indexmaterialthatfillsthegapofslotwaveguides[1,2].Ourstripandslotwaveguideswere

fabricatedonsilicon-on-insulator(SOI)wafers,wheretheindexcontrastbetweensilicon(Si)

witharefractiveindexofn

Si=3.48andsilicondioxide(SiO2

)withn

SiO

2

=1.46amountsto

n≈quiresverytighttolerancesforthefabrication,andputsstringentconditions

th193nmdeep-UVlithographythese

twaveguides,awell-definedslotwidthofabout100nm

uirementsforanaccuratesmoothslotareespeciallyhigh

forsensorapplications,wheretheslotsareusuallyfilledwithairorawaterysolution[3,4].To

solvetheseproblems,wepreviouslysuggestedusingconformalatomiclayerdeposition(ALD)

forcoatingslotwaveguideswithamorphoustitaniumdioxide(TiO

2

)[5].Wealsoinvestigated

thewaveguidingpropertiesofamorphousALD-TiO

2

slabwaveguides,andmeasuredarefrac-

tiveindexofn

TiO

2

=2.27at1.55m[6].

Inthispaper,wedemonstratetowhichextentthelossesofstripandslotwaveguidescan

bereducedwhencoveringthemwithathinTiO

2

layer,andinparticularweshowthefirst

experimental

brieflydiscusstheinfluenceofsurfaceroughness,andthebeneficialeffectanALD-grownlayer

hasonthesurfaceprofihowALDcoatingwithTiO

2

makesotherwisenon-guiding

slotwaveguidesworkproperly,andhowthemodepropertieschangeasafunctionofcoating

ucedlossandthepossibilitytocontroltheslotwidthareofspecialadvantage

fornonlinearsilicon-organichybrid(SOH)slotwaveguides[7,8].

eroughnessandloss

Waveguidelossesaremainlyduetosidewallsurfaceroughnessandtoleakageintothesub-

strate,iftheopticalfieldisnotsuffition,surfacestatesattheuncovered

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11530

evethatabsorbingsurfacestates

arenotprominentinourcase,becausethenthethicknessoftheTiO

2

covershouldnotinfluence

r,inSection5weshowexperimentallythatalargercoverthicknessactually

reducestheloss,andthiscanbeexplainedbytheresultingreductionofthesurfaceroughness.

Therefore,weconcentrateinthefollowingontheinfluencesofwaveguideroughnessandof

leakageintothesubstrate.

Surfaceroughnessleadstoscatteringlosses,andadetailedtheoreticaldescriptionbacked

upwithexperimentshasbeenpublishedforthecaseofhighindex-contrastthree-dimensional

stripwaveguidesbyPoultonetal.[9].Intuitivelyunderstandableguidelinesweregivenfor

fabricatinglow-losswaveguides.

Forthesimplercaseoftwo-dimensionalslabwaveguides,acoupled-modeformalismwas

developed[10].Whilethistheoryisonlyapplicableforstripshavingahighaspectratioand

thereforecloselyresemblingasymmetricslab,itisstillinstructivetoseethetendencyfromthe

closed-formrelationforthelosscoefficient

(unitcm

−1

)[10],

=2(d)

n2

1

−n2

2

2

k3

0

4n

1

0

R(−n

2

k

0

cos)d.(1)

Here,(d)isthenormalizedmodalfield(

+∞

−∞

2(y)dy=1)atthecore-claddingboundary

y=dcladdingrefractivein-

dicesaredenotedasn

1

andn

2

,respectively,k

0=/cisthefree-spacewavenumberforan

angularfrequencyandthevacuumspeedoflightc,theso-calledspatialpowerspectrum

R()istheFouriertransformofthesurfaceroughnessautocorrelationfunctionR(u),andis

rmulahasbeenusedinseveralpublicationsassuming

anexponentialautocorrelationfunctionandtherebyexplainingmeasuredlossesinSiwave-

guides[11–14].Ithasalsobeendemonstratedthatsiliconoxidationcanbeusedtoreducethe

surfaceroughnessandlossesofstripwaveguides[12,13],althoughapartofthereductionprob-

ablycr,

oxi丰乐亭记阅读题答案 dationisnotasuitablemethodforsmoothingsiliconslotwaveguidesasthealreadytoowide

slotswouldgetevenwider.

Siliconnanophotonicwaveguidesaretypicallyfabricatedonsilicon-on-insulator(SOI)

waferswithaburiedoxide(BOX)thicknessof2

tripsaredimensionedsuchthat

eventhefundamentalmodeisonlyweaklyguided,itsmodalfieldtunnelsthroughtheBOX

ultinglosswascalculatedforthecasesofsquareSiwave-

guides[15]andslotwaveguides[16].

Ontheotherhand,scatteringlossesforsquareSistripwaveguidesarefoundtohaveamaxi-

mum,whentheeffectiveindexn

eff=/k0

ofthefundamentalmodeiscloseton

eff=1.7[17].

Forn

eff<1.7,thewaveguidefieldsarelessconfinedtothestrip,whichmeansthattheelec-

tricfieldatthecore-claddingboundary((d)inEq.(1))andthereforethelosscoefficient

becomesmaller.

layerdepositionofTiO

2

forreducingsurfaceroughness

Atomiclayerdeposition(ALD)ates

areplacedinadepositionchamberwheretemperature,pressureandotherparametersareap-

propr(ormore)

chemicalsarethensuppliedsequentiallytotheinertgasflowingthroughthedepositioncham-

bertoformasinglemonolayerofmaterialwhenreactingwitheachotheronthetargetsurface.

ber=0,1,2,...of

depositioncyclesdeterminesthefiositionofTiO

2

,weusedanALD

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11531

tialsurfaceroughnessis

exponentiallydistributedhavingacorrelationlengthofL

c=15nmandaneffective(RMS)

roughnessof=ghtprofilesh(x)andtheRMSroughnessesareshown

forvariousthicknesses(0,10,...,50)

othingeffectisstrongestfor

thehighestspatialfrequenciesvisibleintheinitialheightprofile,whilethesmoothingafter

applyingthefirst10nmALDlayerislesseffective.

processwithtitaniumtetrachloride(TiCl

4)ledexplanationis

foundin[6].

Theconformalchangeofaninitialsurfaceheightgeometryh(x,0)duetothecyclicgrowth

ofALDmonolayersisdescribedasafunctionofpositionxandcyclenumber.Becauseof

thelargenumberofALDmonolayersperfinallayer,thecyclenumbercanberegardedasa

continuous“time”e

∂h(x,)

∂

=v

1+

∂h(x,)

∂x

2

.(2)

ThequantityvistheALDgrowthrate,andvistheresultingthicknessofthefi

visualizingtheALDgrowthofmaterialslikeamorphousTiO

2

orAl

2

O

3

,atypicalgrowthrate

ofv=0.1nm/surfacetobe顺其自然的经典句子 covered,weassumedanexponentially

distributedinitialheightprofileh(x,0)withanautocorrelationlengthL

c=15nmandaroot-

mean-square(RMS)roughnessof=edEq.(2)withafinitedifferencescheme.

Theheightprofilesh(x)andtheRMSroughnessesareshowninFig.1forvariousthick-

ditional10nmlayermakesthesurface

othingeffectisstrongestforthehighestspatialfrequenciesoftheinitial

heightprofile,whilethesmoothingforlowerspatialfrequenciesislesseffective.

ements

Thelossreductionwasdemonstratedexperimentallywithvarioussiliconstripandslotwave-

estigatedwaveguideswerefabricatedonSOIwafersusingtheePIXfabshuttle

service[18].

TheSOIwafershadadevicelayerthicknessof220nmanda2g

couplerswereusedtocouplelightinandoutofthestripwaveguides,whichwereeitherdi-

rectlymeasuredorcoupledwithappropriateadiabatic,virtuallylosslesstransitionstoslot

sesfromcouplingexternalfibersthroughgratingcouplerstothesilicon

stripwaveguideswereestimatedbycomparingtransmissionthroughreferencewaveguides

(∼450nmwidestripwaveguides)tothedirectfiber-to-fisesfromgrat-

ingcouplersofchipsfromthesamewaveguidefabricationrun(butwithoutALDcoating)were

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11532

geofaslotwaveguidecross-sectionthatisALD-coatedwitha30nmthick

layerofTiO

2

.

detses

fromgratingcouplerswereestimatedtobethesamefortheTiO

2

coatedchips(25.5dB),

andtheactuallossesfromwaveguideswereestimatedbydeductingthereferencedifference

andthegratingcouplerlossesfromthemeasuredtransmittedpoweranddividingtheresultby

twaveguides,alsothelossfromshortstripwaveguidesections

gthsofstripwaveguideswere5.5mmandthelengthsofslotwaveguides

were4mm(+1.9mmstripwaveguidefeedingsections).Foralllossmeasurements,weexcited

thequasi-TEfundamentalmodeatawavelengthof1.55m.

Fortheinvestigatedstripandslotwaveguidesweestimatedtheactualgeometryfromscan-

ningelectronmicroscope(SEM)ipwaveguidewidthsvariedintherange

ualrailandslotwidthsmeasured

fromtheSEMimageare214nmand191nm,alltheinvestigatedwave-

ualdimensionaldataforthe

slotwaveguideswereinterpolatedwithastraight-linefitusingthevaluesforrailwidths,slot

widths,andcoatingthicknesses,

errorin

thefittingcomparedtothedimensionsfromtheSEMimageswas5nm,whichisalsoclose

imumdeviation

fromthedimensionsisestimatedtobe9nm.

Asameasureofthefieldconfinement,wecalculatetheeffectiveindicesn

eff

usingthefilm-

modematching(FMM)method[19]forallinvestigatedwaveguidesandlayerthicknesses,

usingtheactualwaveguidegeometry(w

r

andw

s

inTable1)andrefractiveindicesn

Si=3.48,

n

SiO

2

=1.46,andnTiO

2

=eff

isasimpleandusefulindicatorforthefieldconfine-

ment,theeffectivecross-sectionareaA

eff

[20,Eq.(1)]wouldbebettersuitedtoestimatethe

degreeofinteractionwiththird-ordernonlinearmaterialsinwaveguideandcladding.

mentalresults

Atawavelengthof1.55m,wemeasuredthetotallossresultingfromleakageintothesilicon

substrateandfromroughness-inducedscatteringforanumberofdifferentlydimensionedstrip

(Fig.3)andslotwaveguides(Fig.4).Figure3ashowsthestripwaveguidelossesasafunctionof

initialSistripwidth,ossreductionduetotheALD

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11533

edRailWidthsw

rd

,DesignedSlotWidthsw

sd

,EstimatedActualRail

Widthsw

r

,EstimatedActualSlotWidthsw

s

,andtheEffectiveIndicesCalculatedforthe

estimatedActualDimensionswithThreeDifferentALD-TiO

2

Thicknessest

Now

rd

[nm]w

sd

[nm]w

r

[nm]w

s

[nm]n

eff

n

eff

n

eff

(t=20nm)(t=30nm)(t=50nm)

122.4851.5361.736

222.4801.5201.705

322.4551.4951.657

422.4541.4781.621

522.4531.4671.592

623.4831.5671.771

723.4731.5491.739

823.4611.5201.690

923.4561.4981.653

181.4551.4821.622

1124.5081.6001.805

1224.4951.5811.774

1324.4751.5481.725

1424.4641.5231.686

1524.4581.5041.654

1625.5381.6361.840

1725.5231.6151.810

1825.4981.5801.760

1925.4801.5521.721

201.4681.5301.689

2126.5721.6731.875

2226.5551.6511.845

2326.5271.6151.796

2426.5041.5851.757

2526.4871.5601.724

covercanlargelybeduetotheincreaseofwaveguidedimensionsforaparticularinitialstrip

width,itisbettertoplotthelossasafunctionofeffectiveindexvaluesn

eff

calculatedforeach

waveguidestructure,asisdoneinFig.3bforstripwaveguides,andalsoforslotwaveguides

sdecays

quicklywithincreasingn

eff

,startingattheBOXrefractiveindex,belowwhichnolightisguided

creasingn

eff

themodeexperiencesabetterguiding,andleakagelossreduces.

However,becausethefield(d)atthecore-claddingboundaryincreases,roughness-induced

scatteringlossesgainweight,sothatashallowminimumappearsatn

eff≈f≈1.8,a

connectedtothemaximumroughness

lossreportedin[9,Fig.5,Eq.(40),(41)],buthereitisalsoinfl

increasingn

eff

,i.e.,foranincreasedpropagationconstantchosenbyabroadersiliconstrip

(assumingthestripheightremainsconstantbecauseoftechnologicalconstraints),theroughness

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11534

edpropagationlossesasafunc-

tionof(a)initialstripwidth,and(b)quasi-TEmodaleffectiveindexn

eff

=/k0

.Parame-

ter:TiO

2

lossesbecomelessandlessimportant[9].Themaximumn

eff

issetbytherequirementthatthe

waveguideshouldremainsinglemoded.

Forthestripwaveguides,Fig.3(b)showsthatthelargestreductionoflossesfrom11dB/cm

ofanuncoatedoneto5.7dB/cmofawaveguidecoveredwith50nmofTiO

2

isreachedwhen

n

eff≈ailsofobservedlossreductionfeaturescannotbeeasilyexplainedfrom

basicmodefieldsimulations,anddescribingthemmayrequirerigorousscatteringmodelswith

llestlossof2dB/cmwasfound

forthebroadeststrip(450nm,n

eff=2.45).Inthiscase,a50nmthickTiO2

coverdidnot

mateanuncertaintyof1dB/cmforthemeasurements,

basedonthevariationoftransmissionvaluescausedinalargepartbytheuncertaintiesinthe

measurementsetup.

Forslotwaveguides,themeasuredlossesasafunctionofn

eff

a50nmthickTiO

2

cover,minimumlossesof7dB/cm,withanestimateduncertaintyof

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11535

veguideswithvariousdimensions,whichcanbefoundinTable1withthe

edslotwaveguidepropagationlossesasafunction

ofquasi-TEmodaleffectiveindexn

eff

=/k0

fora)allmeasuredslotwaveguides,b)only

ters:Waveguidenumbers(onlya)),

TiO

2

50nm.

2dB/cm(alsolargelyduetotheuncertaintyofthemeasurementsetup),werereachedfor

n

eff=1.69(wr=207nm,ws=230nm).

Toillust器的拼音 ratethemeasurementaccuracy,w上官仪的诗 ecomparedfiveslotwaveguides(waveguides

5,10,15,20,and25)withsimilardimensionsasgivenaboveandfoundthelossvalues

(71,74,79,65,67)dB/cmwithoutanycover(onaverage(71+8

−6

)dB/cm).Witha50nmthick

TiO

2

cover,wemeasuredlossesof(10,9,9,8,7)dB/cm(onaverage(9

+1

−2

)dB/cm).Foranother

setoffiveslotwaveguideswithsimilardimensions(waveguides2,7,12,17,and22)wemeas-

uredthelossvalues(81,78,87,83,91)dB/cmwithoutcover(onaverage(847)dB/cm).With

a50nmthickTiO

2

coverwefound(12,12,13,15,15)dB/cm(onaverage(13.51.5)dB/cm).

ForbothstripandslotwaveguidestheTiO

2

layerthicknessof50nmleadstolowerlosses

thanthethicknessesof30nmand20nm,especiallynearthelossminimumatn

eff≈

comparinggeometricallydifferentslotwaveguideshavingthesameeffectiveindex,onehasto

eofthedatapointswiththesameeffective

indexinFig.4,thelowerlosscanbeexplainedwiththelowermodalfieldstrength(d)at

theair-TiO

2

interface,orwiththemodegeometryleadingtoalowersubstrateleakageloss.

Thethicker50nmcoatingincreasestheeffectiveindextoarangewherethesubstrateleakage

isnotsignificantanymore,andthusreducesthelossdeviationbetweenthedifferentinitial

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11536

edpropagationlossofslotwaveguideswithn

eff

>1.55(noleakageloss)as

ckercoatinggivesastronger

lossreduction,ters:Slotwaveguidenumbers(see

Table1),TiO

2

50nm.

geometries.

Tobetterexaminethedependencyoflossesontheslotwidth,weplotthelossofslotwave-

guideswitheffectiveindicesn

eff>1.55asafunctionoftheairslotwidthremainingafter

coating(Fig.5).Forthechosenn

eff

range,substrateleakageisnotsignificantanymore,and

moreeffectivesurfacesmoothening,athicker

coatingwithaheightof50nmleadstolowerlosses,especiallyfornarrowairslots.

sions

LossesinsiliconstripandslotwaveguidescanbereducedwithaTiO

2

covergrownbyatomic

layerdeposition(ALD).Forwaveguideswithloweffectiverefractiveindexn

eff

,i.e.,witha

relativelysmallfieldconfinement,themeasuredlossesstemmostlyfromleakagethroughthe

2tterfieldconfinement(largern

eff

),scatte-

dthata50nmthickTiO

2

coverreducesthelossofa230nmwidestripwaveguidefrom>50dB/cmto5.7dB/cm,orfrom

11dB/cmto5.7dB/cmwhencomparedtoa320nmwideuncoatedwaveguidewithsimilarn

eff

of∼450nmwidestripwaveguidesthelossisassmallas(21)dB/sof

slotwaveguidesreducesfrom71dB/cmto(72)dB/cm.

Asafurtheradvantage,thewaveguidingpropertiesforstripwaveguidesandtheslotwidth

twaveguidesthisisespecially

convenient,ifasensorsensitizeroranonlinearmaterialisusedtofillapreciselycontrolledslot.

Acknowledgments

WeacknowledgesupportbytheAcademyofFinland(grant128826and134087forA.S.)and

bytheFinnishFundingAgencyforTechnologyandInnovation(TEKES).ledges

supportfromGraduateSchoolofModernOpticsandPhotonics,EmilAaltonenFoundation,

andfromWalterAhlstrrkwasfurthersupportedbytheGermanRe-

searchFoundation(DFG)throughDFGCenterforFunctionalNanostructures(CFN),Karlsruhe

SchoolofOptics&Photonics(KSOP),andthroughtheInitiativeofExcellence,allatKarlsruhe

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11537

InstituteofTechnology(KIT),bytheEuropeanprojectsEURO-FOS(grant224402)andSOFI

(grant248609),andbytheGermanMinistryofEducationandResearchintheframeworkof

thejointprojectMISTRAL(BMBFgrant01BL0804).Wearegratefulf醉花间冯延巳翻译赏析 ortechnologicalsup-

portbytheEuropeanNetworkofExcellenceePIXnet(nowePIXfab,siliconphotonicsplatform

ofIMECandCEA-LETI).

#144638-$15.00USDReceived22Mar2011;revised14May2011;accepted25May2011;published31May2011

(C)2011OSA6June2011/Vol.19,No.12/OPTICSEXPRESS11538