Embed Size (px)
Citation preview
Enhanced Material Removal Rate For III/V Semiconductors
G. Bahar Basim, Sebnem Ozbek, Wazir Akbar
Ozyegin University Department of Mechanical Engineering,
Nisantepe Mevki, No 34-36, Cekmekoy, Istanbul, 34794, TURKEY
CMPUG Meeting, April 13th, 2017
Portland, OR, USA
Introduc)onØ GaN is thenext important semiconductormaterial a4er silicon that canbe
operatedathightemperatures.
Ø It is the key material for the next genera<on of high frequency and highpowertransistors.
Ø Galliumnitride(GaN)basedsemiconductorsareu<lizedin;§ bluelasersdiodes(LDs)
§ light-emiFngdiodes(LEDs)
§ high-power,high-frequencyfieldeffecttransistor(FET)devices
Ø GaNprovidesdirectandwideband-gapenergyandhighelectronmobility1
Ø InordertomakeLEDsfromGaN,itisrequiredtoplanarizethewafersurfaceandhencechemicalmechanicalplanariza<on(CMP)isneeded
Ø GaNisaveryhardfilmandremovalratestendtobeverylowwhenCMPisapplied2
1. Hideo A., Hidetoshi T., Koji K., Haruji K., Kazuhiko S.,and Toshiro D., ChemicalMechanical Polishing of Gallium Nitride with Colloidal Silica, Journal of TheElectrochemicalSociety,158(12)H1206-H1212(2011).
2. Tavernier P. R.,Margalith T., Coldren L.A.,DenBaars S.P., and Clarke D.R., ChemicalMechanical Polishing of GalliumNitride, Electrochemical and Solid-StateLe_ers,5,(8),G61-G64(2002).
Introduc)on
Material Bandgap (eV) Electron Mobility
(cm2/Vs)
Hole Mobility
(cm2/Vs) Critical Field Ec
(V/cm)
Thermal Conductivity σT
(W/m•K)
Coefficient of Thermal
Expansion (ppm/K)
InSb 0.17, D 77,000 850 1,000 18 5.37
InAs 0.354, D 44,000 500 40,000 27 4.52
GaSb 0.726, D 3,000 1,000 50,000 32 7.75
InP 1.344, D 5,400 200 500,000 68 4.6
GaAs 1.424, D 8500 400 400,000 55 5.73
GaN 3.44, D 900 10 3,000,000 110 (200 Film) 5.4-7.2
Ge 0.661, I 3,900 1,900 100,000 58 5.9
Si 1.12, I 1,400 450 300,000 130 2.6
GaP 2.26, I 250 150 1,000,000 110 4.65
SiC (3C, b) 2.36, I 300-900 10-30 1,300,000 700 2.77
SiC (6H, a) 2.86, I 330 - 400 75 2,400,000 700 5.12
SiC (4H, a) 3.25, I 700 3,180,000 700 5.12
C
(diamond) 5.46-5.6, I 2,200 1,800 6,000,000 1,300 0.8
_Proper)esofGaNascomparedtoothersemiconductormaterials_
Introduc)onPROPERTY / MATERIAL
Cubic (Beta) GaN
Hexagonal (Alpha) GaN
Structure Zinc Blende Wurzite Stability Meta-stable Stable
Lattice Parameter(s) at 300K 0.450 nm a0 = 0.3189 nm c0 = 0.5185 nm
Density at 300K 6.10 g.cm-3 6.095 g.cm-3
Nature of Energy Gap Eg Direct Direct
Energy Gap Eg 3.2 eV 3.4eV
Cubic (Beta) GaN Hexagonal (Alpha) GaN
_Crystallographicproper)esofGaN_
BackgroundCommonapproachestoincreasematerialremovalrate;Ø Increaseapplieddownforce1
Ø Pre-polish theGaNsurfacebyusingadiamondbasedabrasiveslurry thatenhancesthemechanicalabrasions[surfacefinishwasveryrough]1-2.
Ø Usecolloidalsilicabasedslurriestoimprovesurfacequalitybyenablingpolishingonthe gallium reach face of GaN (β-Face) yet the nitride reach face (α-Face) showedextremely low polishing rate due to chemical inertness [important evidence thatso0ersilicapar2clewereabletoremovethehardGaNmaterialwithoutinducingsub-surfacedamage]2.
Ø A damage-free surfacewas achievedwith chemicalmechanical polishingwith KOHsolu<on with so4 polishing pad, while it required applica<on of rela<vely highpressures (2–6 kg/cm2) and a long <me of polishing (~40hrs) finally achieving asurfaceroughnessof0.57nm3.
1. Hideo A., Hidetoshi T., Koji K., Haruji K., Kazuhiko S.,and Toshiro D., ChemicalMechanical Polishing of Gallium Nitride with Colloidal Silica, Journal of TheElectrochemicalSociety,158(12)H1206-H1212(2011).
2. Tavernier P. R.,Margalith T., Coldren L.A.,DenBaars S.P., and Clarke D.R., ChemicalMechanical Polishing of GalliumNitride, Electrochemical and Solid-StateLe_ers,5,(8),G61-G64(2002).
3. S.A.Ghosh.StructuralandElectricalCharacteriza<onofGaNThinFilmsonSi(100).AmericanJournalofAnaly2calChemistry,2011,2,984-988.
Ø ColloidalsilicabasedslurrycanbeusedforCMPofGaNwhichgavearemovalrateof17nm/hat0.4kg/cm2(5.7PSI)downforce.
Ø AnatomicallyflatsurfaceofRa=0.1nmlevelfinishwasachieveda4er40hrsofCMP.
LiteratureReview
SurfacemorphologiesofGaNsubstrateduringtheCMPprocess
Ø Ra surface roughnessmeasured in 5 × 5μm2plo_edasafunc<onofCMPprocess<me.
Objec)veØ PromotematerialremovalrateofGaNCMP
Ø Comparesurfacenatureandcorrela<ngCMPresponsesofα-faceandβ-faceGaNcoupons
Ø Evaluatetheimpactof;§ pH§ Slurryflowrate§ Slurrysolidsloading§ DownforceonMRRresponse.
Ø Evaluatetheimpactof;§ Alterna<veprocesschemistries&§ TemperatureonMRRresponse.
Ø Designatoolconfigura<ontoadoptthenewprocesscondi<ons
Ø Implement the findings to alterna<ve hard to polishmaterials andIII/Vsemiconductors.
MaterialsandMethodsMaterials
Ø GaNWafers:
§ 1.0x1.0mm2GaNcoupons(High-QualityFree-StandingGaNsubstrates)
DesktopTegrapol-31PolisheratOzyeginUniversity
Substrate
1.0x1.0mm2GaNcoupons
AbrassivePar)cles SilicaslurryRota)onalSpeed(rpm) 150Pad SUBAIV-IC1000stackedpolishingpad
AppliedPressure(psi)
~9-44
SlurryFlowRate(ml/min) 20-200
Condi)oningCondi)oner
In-situ3M
CMPAnalysis
Tegrapol-31polisherbyusingSubaIV-IC1000stackedpolishingpad
ContactAngleandSurfaceEnergy
Analyses
KSVATTENSIONThetaLiteOp)cContactAngleGoniometer
SurfaceRoughnessAnalyses
Nanomagne)csAFM
MaterialsandMethodsMethods
OutlineComparesurfacenatureandcorrela<ngCMPresponsesofα-faceandβ-faceGaNcoupons
Ø ContactangleØ SurfaceenergyØ WorkofadhesionØ FTIRspectrumØ CMP
WeXability,surfaceenergyandWaMeasurementsonGaN
Ø The we_ability response of the GaN coupons change based on the crystallographicorienta<on. βfacehashighercontactangleindica<nghigherhydrophobicityyettheαfaceishigherinenergyduetocloseratomicpacking.
ContactanglemeasuredwithDIwaterdroplet
α face 97o β face107o
96.72107.15
69.6252.83
15.73 10.28
0
20
40
60
80
100
120
αSide βSide
CA WA SFE
Ø Anabsorp<onpeakat650cm−1duetoGaNbondstretch.
Ø GaNabsorbsinfraredlightatnear1100cm−1withshoulderat1200cm−1duetoSi-Obondstretchingvibra<onandat816and446cm−1duetoringstructure.
SurfaceFTIRanalysesbasedoncrystallographicorienta)on
β -face 650cm-1
α -face1100-1200cm-1
_MaterialRemovalRateEvalua)onasaFunc)onofpH_
CMPEvalua)ononGaN
Ø Materialremovalrateisincreasingasafunc<onofpH.Ø ThemaximumMRRisdeterminedatpH9.
Cubic (Beta) GaN Hexagonal (Alpha) GaN
0
500
1000
1500
2000
2500
3 6 9
MaterialRem
ovalRate[Å/m
in]
SlurrypHValues
αSideMRR[Å/min]
Slurry pH Value α Side MRR [Å/min] β Side MRR [Å/min]
3 62.9 1313.9
6 94.3 1887.6
9 125.8 2269.6
MaterialremovalmechanismonGaNsurfaceBasicchemicalinterac<onmechanismexplainingthereac<onontheα-faceofGaNinthecolloidalsilicaslurry;Alongthe[1120]direc<onforN-polarGaNselec<veetchingtakesplacethrough
(i) forma<onofanitrogenterminatedlayerwithonenega<velychargeddanglingbondoneachnitrogenatom,
(ii) adsorp<onofhydroxideions,(iii) forma<onofoxidesand
(iv) dissolu<onoftheoxides.
* Tavernier P. R., Margalith T., Coldren L.A.,DenBaars S.P., and Clarke D.R.,ChemicalMechanical Polishing ofGalliumNitride, Electrochemical and Solid-StateLe_ers,5,(8),G61-G64(2002).* Sabrina L. Peczonczyk,† JhindanMukherjee,†Azhar I. Carim,†andStephenMaldonado,WetChemical Func<onaliza<onof III−VSemiconductorSurfaces:Alkyla<on of Gallium Arsenide and Gallium Nitride by a Grignard Reac<onSequence,Langmuir,28,4672−4682(2012).
OutlineEvaluateCMPperformanceasafunc<onof;
Ø pHØ SlurrysolidsloadingØ SlurryflowrateØ Downforce
CMPperformanceevalua)onasafunc)onofpH_GalliumNitrideMRRandsurfaceroughnessresponseasafunc)onofpH_
Ø MaterialremovalrateincreaseswithincreasingpHvalueoftheslurryreaching~125Å/min(750nm/hr)atpH9verifyingTevernier’sremovalmechanism.
RMS: 1.11 ± 0.11 RMS: 1.23 ± 0.11 RMS: 1.56 ± 1.04
00.20.40.60.811.21.41.61.8
0
20
40
60
80
100
120
140
pH3 pH6 pH9
RMS[nm]
MaterialRem
ovalRate[Å/m
in]
Ø CMPac<vatesthesurfaceatpH6&pH9whichcanbea_ributedtotheac<vatedOHgroups
Surfacenaturecharacteriza)onasafunc)onofpH
91.7 92.13 104.44
52.82
88.72 89.27
0
50
100
150
pH3 pH6 pH9
ContactA
ngle[θ
]
18.39 17.5 11.46
48.2119.65 20.14
0
50
100
150
pH3 pH6 pH9SurfaceFree
Energy[J/m
2]
71.34 82.962.73
126.9793.72
75.45
0
50
100
150
pH3 pH6 pH9
WorkofAdh
esion
[J/m2]
Pre Post
_CA,surfaceenergyandWaanalysespreandpostCMPapplica<on_
Ø GaNpeaksatnear1100cm−1withshoulderat1200cm−1duetoSi-Obondstretchingvibra<onarereducedpostCMPatpH6&9.
Surfacenaturecharacteriza)onasafunc)onofpH_FTIRspectrapreandpostCMPapplica<on_
pH3
pH6
pH9
Pre
Post
Pre
Post
Pre
Post
CMPEvalua)ononGaN_MaterialRemovalRateEvalua)onasaFunc)onofSlurrySolidsLoading_
(OxideSlurry,pH9,Downforce30N,SlurryFlowRate20ml/min)
Ø Materialremovalrateismaximumat10wt%solidsloading.
126.83
374.93359.16
137.07
34.83
0
50
100
150
200
250
300
350
400
5 10 15 20 25
MRR
[A/m
in]
SolidsLoading[%wt]
CMPEvalua)ononGaN_MaterialRemovalRateEvalua)onasaFunc)onofSlurryFlowRate_
(OxideSlurry,pH9,Downforce30N,SlurrySolidsLoading10%wt)
Ø Materialremovalrateincreaseswithincreasingslurryflowratestabilizinga4er20-25ml/min.
Ø 20ml/minissetastheop<malslurryflowrate.
220.1 220.16
62.89
0
50
100
150
200
250
20 50 100
MRR
[Å/m
in]
FlowRate[ml/min]
_MaterialRemovalRateEvalua)onasaFunc)onofDownforce_(OxideSlurry,pH9,SlurryFlowRate20ml/min,SlurrySolidsLoading10%wt)
CMPEvalua)ononGaN
Ø Materialremovalrateincreaseswithincreasingdownforceun<ltheslurryflowrateisblockedatveryhighpressures.
Ø 14.5psiissetastheop<malpressure.
Holder design for downforcestudies
0 0
62.89
345.96
220.1
0
50
100
150
200
250
300
350
400
4.8 6.2 8.7 14.5 43.5
MaterialRem
ovalRate[Å/m
in]
Pressure[psi]
SummaryØ GaNhasverylimitedmaterialremovalrate(MRR)inCMP.
Ø TypeofsurfacecrystallinestructureaffectstheMRRresponses.
Ø Contactanglemeasurementswereshowntocorrelatetotheremovalrateresponsesasafunc<onofthesurfacecrystallinestructure.
Ø In order to increase theMRRwhileminimizing surface defect forma<on slurry pH,opera<onschemistryandtemperaturewereevaluated.
Ø TheMRRincreasedwithincreasingpH,decreasingslurrytemperatureandusingso4
polishing pads or inducing long pad break-in with harder pads delinea<ng thechemicalnatureoftheGaNpolishingprocess.
Ø MRR values of ~880 Å/min (5280nm/hr) were achieved as compared to 17nm/hrreportedintheliterature.
Ø AnewprocesschemistryadjustmentandtooldesignwasproposedfortheGaNCMP.
