Surfaceand Coatings Technology,64 (1994) 183—193

Theeffectof thermalannealingon thepropertiesof aluminafilmspreparedby metalorganicchemicalvapourdepositionat atmosphericpressure

V.A.C. Haanappel,H.D. vanCorbach,T. Fransen*andP.J.GellingsUniversityof Twente,Departmentof ChemicalTechnology,P. 0. Box217, 7500AE Enschede(Netherlands)

(ReceivedOctober8, 1993; acceptedin final form November17, 1993)

Abstract

Thin films depositedat330°Cby metal organicchemicalvapourdepositionon stainlesssteel, type AISI 304, were annealedin anitrogenatmospherefor 1,2 and4 hat 600, 700 and800°C.The film properties,including theprotectionof theunderlyingsubstrateagainsthigh temperaturecorrosion,thechemicalcompositionofthefilm andthemicrostructure,wereinvestigated.

Corrosionexperimentsperformedat450°Cin a hydrogensulphide containinggas,showed that the cracksin the aluminafilmsalmostcompletelydisappearedafter a post-depositionheattreatment,probablyasa result of stressrelaxation.The porosity of thealuminafilms wasnotaffectedby this heattreatment.X-ray diffraction measurementsof thesefilms, depositedat 330°C,revealedanamorphousstructure.Owing to the thermal annealingprocess,the amorphousalumina films were convertedto ‘i-alumina, andOH-groupsdisappeared.

1. Introduction aluminium-tri-sec-butoxide(ATSB) is baseduponlow temperatureneeded.MOCVD is a processof pyi

Thedepositionof thin aluminafilms by metal organic lytic decompositionof a metal alkoxide,also calledchemical vapourdeposition(MOCVD) is increasingly “precursor”,to producea non-volatilesolid thatdeposapplied[1 13]. Applicationsof aluminafilms are found on a suitably placed substrate[13]. This method 1~in the field of semiconductordevice applications as the ability to produce films with desirablepropertiintermetal dielectrics for silicon integratedcircuits, as Other advantagesinclude the relatively low cost ofsolarselectivecoatingsandprotectivecoatingsfor solar equipmentand low operatingexpenses[13].cells [2—4]. At present, silica is used extensively in The aim of this study was to investigatethe effectsemiconductordevices [5]. However, silica layershave thermalannealingon somepropertiesof alumina filia high permeabilityto alkali ions, e.g. sodium, which depositedby MOCVD. The aluminafilms weredep(can easily reach the underlying substrateresulting in ited by thethermaldecompositionof ATSB in anitrogdevice instability [6]. In order to overcometheseprob- atmosphereunderatmosphericpressure.Propertiessulems, thesesilica layerscanbe replacedby, for example, as protection against high temperaturecorrosionalumina films which have a very low permeabilityto steels,structureand chemicalcompositionwere invesalkali ions and other impurities [7,8]. Alumina films gated as a function of the annealingperiod and t

can also be usedas protective coatingsin the field of annealingtemperature.high temperaturecorrosion[9]. This study is focusedon the formation andcharacterisationof thin aluminafilms which canprobablyreplacesilica films to improve 2. Experimental detailsthe corrosionprotection.

Severalmethodshavebeenreportedfor the prepara- Alumina films were deposited by means of atmtion of alumina films, such as plasma enhanced spheric pressure MOCVD using ATSB (JanssMOCVD [10], low pressure MOCVD [11], non- Chimica) and dried nitrogengas.The substrateusedreactiver.f. magnetronsputtering[12] andatmospheric the experimentswas a stainlesssteel, type AISI 3(pressure MO CYD (this paper). The choice for the which wascut from an electropolishedmetal sheetaformation of alumina films by MOCVD by meansof ultrasonicallycleanedin hexaneandethanol.____________ A schematicdiagram of the experimentalset-up VQ

*Author to whom correspondenceshouldbe addressed, shownin a previouspaper[14]. TheATSB is introduc

184 V.A.C. Haanappelet al/Effectof thermalannealing on MOCVD alumina films

in the furnaceby passingnitrogengasthroughthe ATSB were measuredby Fourier transforminfrared spectrbubbler with an equilibrium vapourpressureof ATSB scopy (FTIR), using a Nicolet 20 SX FTJR apparanof 0.13 kPa (1.0mm Hg) at 138 ~C.The ATSB flow rate Owing to the non-transparencyof the specimens,t]was adjustedby controlling the flux of nitrogen gas spectrometerwas equippedwith a diffuse reflection cthroughtheATSB bubblerat atmosphericpressure.This [9]. Themicrostructureof the films was investigated]gasmixture was addedto the main nitrogenflow before X-ray diffraction (XRD), usinga PhilipsPW 1710 X-r;entering the reaction chamber.The flow rates were diffractometer with monochromaticCu Kc~radiati(controlled by mass flow controllers (type Brooks anda high resolutiontransmissionelectronmicrosco~5850TR).The depositionreactionwas carriedout in a model JEOL 200 CX.quartz tube with a diameterof 45 mm. The uniformtemperaturezonein the reactorwas 120 mm. The speci-menswereattached,parallelto thegasflow, to a ceramic 3. Resultstubewith a thermocoupleinsidein orderto measurethesubstratetemperatureand control the furnacetemper- The kinetic aspectsof the depositionof the alumiiature(standardconditionswere: substratetemperature, films on AISI 304 by the MOCVD processhavebe330°C;ATSB temperature,138 °C;flow rate,6.5 1 mm 1 discussedextensively in a previous paper [14]. T(STP)). The depositionrateof the films on the metallic experimentalactivation energy for the heterogeneosubstratewasdeterminedby weighingthe samplesbefore reactionwasfound to be 83 ±5 kJ mol ~.

and after the deposition.In the experimentsspecimens The effect of temperatureand exposuretime ofwere usedwith aluminafilms of 0.20±0.05 mg cm2 thermal treatmentprocesson the corrosion attack

After the depositionprocess,the sampleswereexposed shown in Fig. 1. The relative weight gain is the ratto a nitrogen atmospherewith temperaturesranging betweenthe weight gain of a coatedand an uncoatfrom 600 to 800°C,during 1, 2 and4 h periodswithout sample after the corrosion experiments. From theremoving them from the furnace. Following the depos- figures it is clear that the amount of corrosion produition processthe furnace was heated to the desired (relativeweight gain)wasnot extensivelyreducedby t.

annealingtemperaturein 20 mm. After completion of post-depositionannealing.At timesevenan increasethe thermal annealing period, the temperaturewas the relativeweight gainwasmeasured.Scanningelectrdecreasedto below 300°Cbefore the samples were microscopy(SEM) examinationsshowedthat corrosiremovedfrom the furnace, products(Fe-rich sulphides)were formed on the out

Corrosion experimentswere performed in order to surfaceof the aluminafilm.investigate the protective capacity, i.e. the porosity, of In general,the surfaceof the specimensshowedthe films against aggressivegas components,such as extensivereductionof the number of cracks owingoxygen and sulphur, at high temperaturesand were heat treatment. In some casesno cracks were fouricarriedout in a closedsystemfor 24 h. The systemwas Corrosion productswere locally formed on the surfafirst flushedwith argonfor 20 h with a flow rateof 16 1 probablybecauseof the presenceof porosityin the flUh’. The argon was saturatedwith water at 15°C.A For sampleswith andwithout a post-depositionthermmixture of 5% H

2S in H2 with a flow rate of 4 1 h~ treatment,it was observedthat after higher depositiwas then introducedinto the argonflow. After 2 h, the temperaturesfewer corrosion productswere foundfurnacewas heatedto the test temperature(i.e. 450°C) the specimens(Fig. 2).to stabilize the flow. The whole systemwasflushed for Thecompositionandthedepthprofile of the annealanother2 h and then closed. It had previously been alumina films were investigated by Auger electnfound that no relevantdifferencein corrosionrateand spectroscopy(AES) in combinationwith ion sputterircorrosion products was obtained between a closed From the steadystateAugerspectra(see surveyscansystem and a systemwith a continuousflow, if the Fig. 3), it is clear that next to aluminium and oxygenreactivegasconsumptiondid not exceed20%. small amount of carbonwas also detected,even afi

Themorphologyandthe compositionof the corroded annealingfor 4 h at 800°C.After oneor moresputteri]specimenswere investigatedby meansof optical and cycles(one cycle correspondsto 1 mm sputteringatscanningelectronmicroscopes(JEOL-35CF), the latter rate of 44 nm min 1) carbon was no longer fourequippedwith anenergydispersiveX-ray analysis(EDX) Figure 4(a) shows the AES depthprofile of the alumisystem(Kevex Delta, classIII). film (thickness,0.55 ttm) depositedat 330 C (witho

To characterizethe chemical composition and the thermal treatment).The AES depth profiles of the 1depth profile, the alumina films were analysedusing a at 600°C(0.7 l.tm) andthe 4 h at 800°C(0.5 urn) treatPerkin—Elmer PHI 600 scanning Auger multiprobe. specimensare shownin Figs. 4(b)and 4(c) respectiveAlternate sputtering and Auger analysis were used for From these observations it was found thatin-depthanalysis.Thefunctionalgroupswithin the films alumina—alloy interface was rather sharp for

1/AC.Haanappelet al/Effect of thermalannealingon MOCVD alumina films 1

Annealing Temperature Annealing Temperature600 °C 700 °C

100111 1 100111 1

I I I ~10RI~w

Uncoated 0 1 2 4 Uncoated 0 1 2 4Hours Hours

(a) 300°C ~ 330 °C~iI~370°C (b) 300°C ~ 330 °c 370°C

Annealing Temperature800 °Cj 100111

15

Uncoated 0 1 2 4Hours

(c) 300°C 330 °CEIIIIi370°C

Fig. 1. Bar diagramof the relative weight gain (%) of the specimensas a function of the thermal annealingtime and temperature:annealitemperature(a) 600°C;(b) 700°Cand(c) 800°C.

untreatedsamplesor samplestreatedat low annealing of the films annealedat 700°Care shownin Fig. 5. T~temperaturesandshort annealingtime.After longertime specific absorptionbandswere found at 900—950cmperiodsand/orhigher annealingtemperatures,the inter- and 3000—3700cm1, indicativeof the Al—O and 0—facewasno longer very sharp.From Fig. 4(c) it is clear bonds respectively [9,15]. This figure shows thatthat a certainamountof chromiumpenetratedinto the 0—H absorptionbanddecreasesin intensity at a givaluminafilm. The iron and nickel depthprofile did not annealingtemperaturewith increasingannealingtinchange after the thermal treatment, even after 4 h Theabsorptionbandaround900—950cm1,characterat 800°C. tic for the Al—O band,increasesin intensity. No pr

The chemical composition of the alumina films nounced difference was observed regarding tobtainedin the processdescribedwas also analysedby maximum of the Al—0 peak owing to the differemeansof X-ray photoelectronspectroscopy(XPS). thermal treatment procedures.The maximum of t

To identify functionalgroupsin the aluminafilm, such Al—0 peak of an unannealedsample lies arouias OH-groupswhich cannot be detectedby EDX or 910cm’ andshifts to around928 cm1 after annealirAES, FTIR was applied.The FTIR absorbancespectra The microstructureof the alumina films was inves

186 V.1.( . Haanappel I 0/ Lf/~ct of thermal annealingon MOCVD alumina til,n~

- ~5t~$~J ~ ~• ~ ‘ ‘~ . . .

I hi

ci di

IiIr4_L_~

(C)

Fig. 2. SEM image of the surfacemorphology of the ~pLcImensafter 24 h of sulphidation at 4~i((. Deposition temperature,300°C:unannealed);(b) annealedfor 2 h at 600°C.Deposition temperature,330°C:(c) unannealed;(di annealed for 2 h at 700°C.Deposititemperature,370°C:(e) unannealed;(f) annealedfor 2 hat 800°C.

gated by XRD. It was previously found [15] that 20; = 43.7 °, 50.9 ° and 74.7 ~. After the post-depositialumina films depositedat variousdepositiontemper- thermaltreatmenta small peakwasfound at 20; = 44.~aturesandanalysedby XRD show an amorphousstruc- indicativeof the presenceof y-alumina. This small pe;ture. The XRD peaksof the aluminafilms annealedat wasfound for all heat-treatedsamples.700°Cfor 1, 2 and4 h, andof one samplewithout any Additional depositionexperimentswereperformedheat treatment,are shownin Fig. 6. On the unannealed polishedAISI 304 specimens.It wasobservedthat orspecimensonly the characteristicXRD peaks of the at the annealingtemperatureof 700°Cfor 2 and 4underlying substrate, AlSI 304, could be found: and at 800°Cfor 1, 2 and 4 h did the alumina filt

V.A.C.Haanappeleta//Effect of thermalannealing on MOCVD aluminafilms

2000 Maierial: Al2O~/ AISJ-304

Deposition Temperature: 330 °C

1500

1000H— —.500

-1000

-1500

-2000 0

5(10 01)0 1500 2000

Electron Kinetic Energy(cv) PHI

Fig. 3. AES survey scananalysisof thealumina film on AISI 304, depositedat 330°C(unannealed).

crack and delaminate during the cooling period. From owing to a certain porosity of the alumina films and t]

examinationswith SEM it wasfound that delamination existenceof cracksin the films.took place interfacially. Previously, it was found that It was found earlier [15] that the porosity of t]crackinganddelaminationof the aluminafilm occurred alumina film dependssignificantly on the depositiiafter passinga critical thicknessof the film during the temperature.Figures 2(a), (c) and (e) show the amouMOCVD process[16], owing to internal tensilestresses of corrosionproductsthat reducewith increasingdepcat the outer side of the film. The films bendconcavely, ition temperature.Thiscorrespondswell with the resulas shownin Fig.7(a). by Boldyrev et al. [17] who found that an increase

The aluminafilms, which crack anddelaminatefrom the depositiontemperaturepromotesthermal compathe underlying substrateafter the thermal annealing tion resultingin a denserfilm structure.The mobilityprocedure,bend convexly (see Fig. 7(b)) as a result of adsorbedatomsis low—probablyas a result of the lcthe internal compressivestressesconcentratedat the depositiontemperature.This meansthat those atorouterpart of the aluminafilm. Crackingand delamina- are not able to move from their adsorption sitestion of the films after cooling did not occur when energetically more favourable sites. The existencesubstrateswere used from the electropolished metal poresin the aluminafilms might also be explained1sheet,which might be explainedby the higher rough- an incompleteconversionof theATSB molecules,espenessof the underlying substrategiving larger adhesive ally at lower depositiontemperatures,as already djforces. cussedelsewhere[15]. Also the water vapour releas

by the dehydrationof boehmite(AlO(OH)) during t]

depositionmay affect the porosity of the alumina fili4. Discussion From previously discussedresults [9,15] it is assum

that, owing to the presence of OH-groups, tiFrom the weight gain measurementsit is clear that as-depositedalumina film consistspartly of boehmii

the different post-depositionannealingproceduresdid According to Stumpfef al. [18], ‘i-alumina was foutnot result in anysignificant improvementof the protec- as the crystalline phase after thermal annealing.tivenessof the alumina films against high temperature A1O(OH) were incorporated in the film, it woucorrosionof steels.On the outersurfaceof thespecimens, decomposeto alumina andwater at high temperaturecorrosionproductswereformedafter exposurefor 24 hin a hydrogen sulphide containing gas atmosphere, 2AlO(OH) ~ Al,O~(1 — x)H1O+ xH.,O

188 1/AC. Haanappelet a//Effect of thermalannealing on MOCVD aluminafilms

The transition of A1O(OH) to y-alumina was also explainedby the shortperiod(up to 4 h) at temperaturfound by Goton [19], Bugoshet al. [20] andSaalfeld up to 800°Cof the thermalannealingprocessbeing ti

[21]. Goton [19] found dehydroxylationof boehmite slow to compactthe aluminafilm. Higher temperaturinto A1203 (with 0.1 mol H2O per mole of Al203). andlongerperiodsmayhavemoreeffecton theporosiBugoshet a!. [20] andSaalfeld[21] found that heating of the aluminafilms.of A1O(OH) results subsequentlyinto ~‘-, 0- and o~- Summarizing, it is clear that only the depositialumina, temperaturesignificantly influencesthe porosity; this

ln contrast to the depositiontemperature,the post- in contrastto the post-depositionheattreatment.deposition thermal treatment did not influence the However, the crackdensitywas reducedsignificaniporous structure of the alumina film. This might be after the thermaltreatmentprocess.Thesecracksin t

25~M:terla1 M0/AiSl~

Cr

0 100 200 300 400 500 600 700 800

(a) sputterdepth(not) PHI

x10

4

Material: A1

505 / AISI.3042 Deposition Temperature: 330 ‘C

Annealing Temperature: 600 ‘CAnnealing Time: I hrSpuner Rate: 44 nm/hr ~

Fe

0.5/ ~ii~~iij0 100 200 30!) 400 5(1)) 601) 700 800

(b) sputterdepth (nm) P1-Il

1/AC. Haanappe/et al/Effect of thermal annea/ingon MOCVD aluminafl/ms

,d04

3.5 -r~Material: Al

50, / AISI-304Deposition Temperature: 330 ‘C

1) Annealing Temperature: 800 ‘C3 Annealing Time: 4 hr

Sputter Rate: 44 nm/hr

Fe

1.5

Cr -

0.5

- 100 200 300 - - - 400

(c) Sputter Depth (urn) PHI

Fig. 4. AES depthprofile analysisof the alumina film on AISI 304, depositedat 330 °C:(a) unannealed;(b) annealedfor 1 h at 600 °C;annealedfor 4 h at 800 °C.

‘t 00 3~O0 sáoo 2~O0 2óao i~oo iáoo ~oo 0WAVENUM~E19

Fig. 5. FTIR absorbancespectraof alumina films annealedat700°Cfor 0, 1,2 and4 h.

thin aluminafilms are generally relatedto the internal of the alumina films, after having passed a critistress,consistingof thermalandintrinsic(growth)stress, thickness, took place during the deposition and aaccordingto: during the cooling period. Owing to this high intrin:

— ‘2~ stress the alumina films bend concavely upward (t0~tnternaI— 0’lntrinsic + 0’thermal Fig. 7(a)), causedby a high intrinsic tensilestressin t

It wasfound previously [16] that thin aluminafilms outer surface of the film. Similar observationswdepositedby MOCVD sufferfrom high intrinsicstresses. found for vapourdepositedtantalum [22] and chiSEM analysesindicatedthatcrackinganddelamination mium films [23]. Calculationrevealedthat the therma

190 1/AC.Haanappe/et al/Effect of thermalanncaltngon MOCI 0 aluminaf/ms

>,

C

C

C__

a

40 50 60 70 80 90LI)

20

Fig. 6. XRD patternsof alumina films annealedat 700°Cfor (a) 0 h. ( i(b) I h, Ic) 2 h and(d) 4 h.

induced part of the internal stressis almost negligible[16]. ________________________________

A model for the generationof internal stressduring ‘~~‘

the depositionprocessor thermal annealingprocessis ______ ______illustrated in Fig. 8 [13]. If the film is under residual ~r-’~ -~. -~

tensilestress,the film and also the underlying substratewill bend concavely, or, if under residual compressivestress,convexly.This form of the delaminatedfilm afterthe post depositionthermal treatment(bending con-vexly) canbeexplainedby an internalcompressivestresswhich is concentratedat the outer part of the aluminafilm. If only the thermal stress in the alumina film on FIg. 7. Delaminatedaluminafilm (a)duringthedepositionproce’.~IAISI 304 is considered,the stresscan be calculatedby film bendsconcavelyupwardand(b) after thermal annealtnr or -

the simplified equation: at 700°C.

athermal—Efx ATx Ac~ (3)

w~reEfi~e :u ____ __ _______

and Ac~is the difference betweenthermal expansion SUBSTRATE SUBSTRATEJ

coefficients.Since the annealedalumina films on polished AISI

304 bend convexly (after delamination),it is suggested ~ ~that the major part of the stresscan be consideredtobethe thermalstress(residualcompressivestress).Hence,the films without thermal annealingbend concavelyasa result of high residualtensile stress.This meansthatafter long annealingperiods at high temperatures,the ~°°~‘~°~

relaxationof the intrinsic stressis almostcomplete.Forfilms with thicknessesbelow the critical value,the stressis determinedby the sum of the intrinsic and thermal FILM UNDER RESIDUAL FILM UNDER RESIDUALinduced part. During the cooling period the alumina TENSILE STRESS COMPRESSIVE STRESS

films form cracks. After annealing, almost no cracks Fig. 8. A model for thegenerationof internalstress:(a) residual tenswere found. This can be explainedby the fact that the stressin thefilm; (b) residualcompressivestressin thefilm [10].

1/AC. Haanappelet al/Effect of thermalannealingon MOCVD aluminafilms 1

total internal stressof the oxide film is now determined change,no refractive index could be determined.TIonly by the thermalstress. effect of the underlyingdiffusion zoneon the total vali

A model of stressrelaxationas a function of time and of index of refractionwasnot known and could not Itemperaturehasbeendevelopedby Desu[24]. Also the eleminatedin a simple way. Thereforeno values of tIeffect of depositionand annealingtemperatureon the refractiveindex are shown.film stressand the peakpositions of the characteristic The chemical structureof the films was analyzedIbendingandstretchingfrequencies,obtainedfrom FTIR FTIR andXRD. To identify groupssuchas OH in tispectraof the SiO2 films, wasdiscussedby Desu[24,25]. alumina film, which could not be detectedby EDXFor the alumina films it was found that the deposition AES, FTIR was used.The presenceof OH-groupsw;temperatureaffects the peakposition of the Al—O band suggested[15] by the incorporationof AIO(OH)[15]. From the FTIR spectraobtained it was found H2O in the alumina film, which is probablydue tothat the maximumof the characteristicA1—O peak,after incompleteconversionof the pyrolytic decompositionthe different post-deposition thermal treatment pro- the precursor.After 4 h of annealingthe OH-peakhtcedures, shifts towards a constant value around almost disappearedwhich implies that the hydrog928 cm ‘. This probablymeansthat no largedifference impurity is significantly reduced.As a result of the hi1in the internalstressis now present.If the highintrinsic temperaturesof the annealingprocess,reaction (1) wstress is significantly reduced,the thermal stressnow proceedto form y-alumina.The next stepis thedehydrdominatesthe internalstress.Thealuminafilms delami- tion of the formed alumina, as shown in Fig. 9 [26natedfrom the polishedAISI 304 only at 700°Cand which shows various tranformationsequencesfor ti800 °C—thismay be explainedby the higher thermal formationof alumina. It wasalso clear that the charastress,causedby the larger difference betweenanneal teristic Al—O IR absorptionbandchangedin shape.F)andfinal temperature. the unannealed specimens (deposition temperatul

It is known that the alumina films on polishedAISI 330°C),the Al—O peak is smooth and broad, b304 delaminateat 700°C,but still show adhesionat becomessharperwith longerperiodsandhighertempetemperaturesof 600°C.Fromeqn.(3) it canbecalculated aturesof the anneal process.Morssinkhof [9] relatthat the cohesivestrength lies between4.6 x 108 and this effect to a decreasein the amount of OH, whi5.3 x 10~N m

2. also has a peak at lOSOcm’. Kang and Chun [

The chemicalcomposition of the annealedalumina relatedthe effectto the fact that in generalthe JRspectfilms was investigatedby AES andFTIR measurements. of amorphousmaterialsare smoothand broad, wh~The steadystateAuger spectra(surveyscan) revealed thoseof crystallinematerialsare relatively sharp.TIaluminium, oxygen and also small amountsof carbon. correspondswell with the results obtained from tiHoweverafter one or moresputteringcyclescarbonwas XRD measurements.The XRD analysisshows thataftno longer found, which might be explained by the the annealingprocessof the aluminafilms, small peaadsorptionof carbon containingcompounds,such as were observed,indicative of the presenceof y-alumircarbondioxideandhydrocarbons,on the outsideof the XRD analysisdid not revealany characteristicalumiialumina layer. This correspondswell with previous peaksof the unannealedspecimensbecausethe degrresults [15]. From the AES depth profiles of the (low ofcrystallinity wastoo smallto bedetected[15]. Temptemperatureand short period) annealedspecimensit and Reisman [7] performedpost-depositionthermwas clear that the concentrationprofiles of the alumin- treatmentsof 100—180nm thick alumina films on silicium, oxygen and the alloy elements(iron, chromium substratesfor 15 mm at 1000°Cand60 mm at 800 ~:

andnickel) areequal to thoseof the specimenswithout No X-ray diffraction spectrawere found for any phasannealing. At longer annealingperiodsand/or higher in the alumina film, which is probably becausetheannealing temperatures the interface oxide/alloy films were too thin for XRD measurements.Theformbecomesless sharp,especiallythechromiumprofile. The tion of y-aluminaafter thepost-depositionthermaltretchromiumpenetratesmoreinto the alumina films after ment correspondswell with the results by Aboaf [27longer annealingperiodsand higher temperatures.The Morssinkhof[9] andBoldyrev et a!. [17]. For exampnickel and iron profiles seem to be unaffectedby the Aboaf[27] foundthat the heattreatmentof the alumiithermal treatment.Morssinkhof [9] reportedthat this films in argonat 600°Cfor 1 h only resultedin diffubehaviourcanbeexplainedby the structureof y-alumina electrondiffraction patterns.After 1 h at 800°C,theand the expected completemiscibility of Cr

203 and alumina phasewas also detected.After 3 h at 800°Al203 with a substitutionof Al by Cr in the oxide.This the amorphousphasehadcompletelydisappeared.meansthat the interfacestructurechangedsignificantly Summarizing, the thermal annealing procedurafter severepost-depositionheattreatmentsandproba- describedin this paperprobablyallow stressrelaxatibly affects, for example,the cohesivestrengthandrefrac- in the MOCYD aluminafilms. Theporosity of the fllrtive index.Becauseof this interfacial compositionprofile was not affected by this post-depositionthermal tret

192 1/AC. Haanappelet al/Effect of thermalannealing on MOCVD alumina films

100 200 300 400 500 600 700 800 900 1000 1100 1200

I : I

~- Kappa K AlphaGibbsite CHI

~ ‘6 Delta Theta Alpha

Boehmite ______________Bayerite ETA -0 Theta ~Alpha

~ RHO——

Diaspore Alpha

I I I I I I I I I I I400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

K

Fig. 9. Transformationsequencefor theformation of alumina [26].

ment, but the cracks in the film disappearedalmost experiments,J. Boeijsmafor theXRD measurements,completely. As a result of the thermal treatmentafter Eijkel (MESA) for the ellipsometry measurementsatthedepositionprocess,thehydrogenimpurityalso disap- to T. Kachlicki (CMO) for the TEM analysis.all atpearedand the A1O(OH) (boehmite)was almostcom- University of Twente.pletely convertedto dehydrated7-Al203.

5. Conclusions References

The pyrolytic decompositionof ATSB carriedOut at 1 V.F. Korzo, Zh. Priki. Khim., 49 (1) (1975) 74.atmosphericpressure in the low temperaturerange 2 C. DhanavantrtandRN. Karekar, Thin Solid Films, 127 (1985)

290—420°Cin nitrogen, resulted in the depositionof 3 R.A. Jacobson,Phys. Thtn Ftlms, 8 (1975) 51.4 R.S. EhIe, B.J. Baliga and W. Katz. J. Electron. Mater.. 12

thin alumina films, containingporesandcracks. (1983) 587.From the corrosionexperimentsit was observedthat 5 C.J. Kang and J.S. Chun,Thin Solid Fihns, 189 )l990) 173.

the cracksin the aluminafilms completelydisappeared 6 J.F. Verweij, Insulatingfilms on semiconductors(1979), Inst. Ph

during the post-deposition heat treatment. This is Conf. Ser.,50 (1980)62.explainedby a reduction of the internal stress. The 7 D. Temple andA. Retsman,J. Electron.Mater., 19 (9) (1990) 9(

8 R.A. Abbott andTI. Kamins,Solid State Electron., 13 (1970) 5porosity of the alumina films was not affected by this 9 R.W~. Morssinkhof,The depositionof alumina films on steels

heattreatment. MOCVD, Ph.D. Thesis,Universityof Twente,Enschede,1991.

From the FTIR spectra it was shown that the 10 K.P. Pande,V.K.R. Nair and D. Gutierrez. 2. App!. PhI-s.. 54OH-groupsdisappearedafter 4 h of thermal annealing (1983) 5436.

at 600, 700 or 800°C. During the annealing the II J. Sarate,J. Kwon andY. Yodogawa,2. Electrochenr.Soc.. 132

oxide/alloy interfacechangesas a result of diffusion of (1985) 890.12 F. RamosandMT. Vieira, Mater. Manuf. Process..7)2))1992)2chromium into the alumina matrix, as shown by the 13 M. Ohring. The Materials Scienceof Thin F//ms. AcademicPo

AES depth profiles. XRD measurementsshowedthat Inc. Harcourt BraceJovanovich,SanDiego, CA. 1992.

after the thermal annealingprocessesthe alumina films 14 V.A.C. Haanappel,H.D. vanCorbach,T. FransenandP.J.Gellinwereconvertedto y-alumina. Corrosion resistantcoatings producedby metal organtc vap

deposition using aluminium-tri-sec-butoxide,Thin Solid Fillin press.

15 V.A.C. Haanappel.H.D. vanCorbach,T. FransenandP.J.GellinAcknowledgments Properties of alumina films preparedby atmosphertcpressi

metal organicchemicalvapourdeposition,Surf. Coat. 7cc/mo!..

This research was supported by the Innovative (1994)145.ResearchProgram on Technical Ceramics(IOP-TK) 16 V.A.C.Haanappel,H.D.vanCorbach,T. FransenandP.J.Gellin

Cracking and delamination of metal organtc vapour depostiwith financialaid from the DutchMinistry of Economic aluminaandsilica films, Mater. Sc! Eng.,A167 (1993)179.Affairs. Thanksare dueto G.J.M. Weierink for the FTIR 17 V.F. Boldyrev, V.M. Koleshko, B.S. Reznikov. iv. Nekaryuls

measurements,A.H.J. van denBerg (CMO) for theAES L.D. Buiko and K.D. Yashin. Neorg. Mater., /2(1976)2181.

1/AC.Haanappelet al/Effectof thermal annealing on MOCVD aluminafilms P

18 H.C. Stumpf, AS. Russell,J.W. Newsomeand CM. Tucker, md. D.M. BarnettandD.A. Smith (eds.),Mater. Res.Soc. Symp.PrcEng. Chem.,42 (1950) 1398. 130, MaterialsResearchSociety,Pittsburgh,PA, 1989, p. 249.

19 R. Goton, Ph.D. Thesis,No. 146, Universityof Lyon, 1955. 24 SB. Desu,Jpn. J. Appl. Phys.,30 (12B) (1991)2123.20 J. Bugosh, R.L. Brown, JR. McWhorther, G.W. Sears and 25 SB. Desu, in TM. Besmannand B.M. Gallois (eds.),Mater. Ri

R.J.Sippel,md.Eng.Chem.,Chem.Prod. Res.Dev.,1(3)(1962)157. Soc.Symp.Proc. /68, Materials ResearchSociety, Pittsburgh,P.21 H. Saalfeld, Neues.Jahrb. Miner. Abh., 95 (1960)1. 1990, p. 221.22 R.M. Fisher,J.Z. DuanandJ.B. Liu, in M.F. Doerner,W.C. Oliver, 26 K. Wefers andC. Misra,Oxidesandhydroxidesof alumina,Alcs

G.M. Pharr and FR. Brotzen (eds.),Mater. Res.Soc.Symp.Proc. Technical Paper No. 19 (revised), Alcoa ResearchLaboratork188, Materials ResearchSociety, Pittsburgh,PA, 1990, p. 257. EastSt. Louis, IL, 1987.

23 R.M. Fisher,J.Z. Duan andAG. Fox. in J.C. Bravman,W.D. Nix, 27 J.A. Aboaf,2. Electrochem.Soc.,114 (1967)949.