The effect of diffusion barriers, stress and lateral diffusion on thin-film phase formation

De Waal, Hendrik S.(Hendrik Schalk) (1999-11)

Thesis (Ph.D.) -- University of Stellenbosch, 1999.

Thesis

ENGLISH ABSTRACT: Instead of studying phase formation in conventional thin film binary diffusion couples, various unconventional thin film sample structures were used to investigate the influence of certain parameters on phase formation. Iron-silicide formation on Si(lOO) through metal alloy diffusion barriers, Tisilicide formation on stressed Si(lOO) substrates and phase formation in various lateral diffusion couples was studied. A variety of amorphous alloys were used as diffusion barriers through which Fe reacted with Si(100). It was found that the diffusion barrier did not change the effective concentration at the reaction interface enough to cause the formation of the semiconducting j3FeSiz phase instead of EFeSi as first phase. There is much technical interest in growing the semiconducting j3FeSiz phase epitaxially on Si(100) for use in amongst others infra red detectors and light emiting devices. The use of the diffusion barriers did however control the reaction in such a way as to allow the formation of a uniform film of polycrystalline j3FeSiz at 800°C. In the case where Fe reacted directly with Si(lOO) without a diffusion barrier the Fe film reacted very non-uniformly with the substrate. The uniform polycrystalline thin film of j3FeSiz was then converted to epitaxial FeosSi[CsCl] using a 0.8J/cmz laser pulse from an excimer laser. This was the first time that a continuous film of epitaxial FeosSi[CsCI] has been formed on Si(100). The j3FeSiz films with large grain sizes formed the best quality epitaxial FeosSi[CsCI] after laser annealing. The grain size was dependant on the type of diffusion barrier used. The influence of stress on Ti-silicide phase formation was also studied. TiSiz is commonly used in integrated circuits and with the move to smaller devices and linewidth's stress is playing a larger role. Varying amounts of compressive and tensile stress was induced in Si(100) substrates by depositing different thicknesses of SiOz or Si3N4 on the backside of the Si wafers. These back side films deformed the wafers causing stress throughout the substrates. The stress in the substrates was determined by measuring the radius of curvature of the samples using a laser technique and then relating the radius of curvature to stress using Stoney's equation. It was found that when Ti reacted with a substrate under compressive stress. the Si diffusion process was slowed down so that TiSiz[C49] grew 50% slower than when Ti reacted with substrates under tensile stress. In the samples with the most compressive stress in the substrate the Si diffusion was limited so effectively that the Ti-rich TisSi3 phase formed instead of TiSiz. This is explained using the Effective Heat of Formation (EHF) model. Lateral phase formation must take place in order to form some of the complex device structures present in the modem integrated circuit. In this study the nuclear microprobe, electron microscope and optical microscope was used to study phase formation in Ru-Al, Pt-Al and Cr-Si lateral diffusion couples. A typical lateral diffusion couple consisted of an island of one material deposited onto a thin film of anoth~r. Jt .was found that if the island did not consist of the diffusing species there was no lateral reaction. In the aluminide systems the AI-island geometry always showed lateral reaction. In the Ru-AI system the RU4AI\3 phase was the only phase to grow laterally with reaction limited kinetics while in the Pt-Al system two phases namely PtgAlz1 and PtAlz grew laterally with diffusion limited kinetics. The EHF model is used to explain how diffusion limited growth causes the simultaneous formation of more than one phase. In the Cr-Si system the Si-island geometry grew laterally forming CrSi2. Oxidation made it impossible to do any kinetic measurements. It is shown how lateral diffusion studies can be used to give information on diffusing species and diffusing mechanism. A summary of all the work done on lateral diffusion couples was also compiled in this study.

AFRIKAANSE OPSOMMING: In teenstelling met die ondersoek vankonvensionele dunlagie faseformasie is verskeie onkonvensionele dunlagie strukture gebruik om die invloed van sekere parameters op faseformasie te bestudeer. Ystersilisied formasie op Si(lOO) deur metaal allooi diffusie-versperringslagies, Tisilisiedformasie op gespanne Si(lOO) substrate en faseformasie in verskeie laterale diffusie stelsels was bestudeer. Verskeie amorfe allooie was gebruik as diffusie-versperringslagies waardeur Fe met Si(lOO) gereageer het. Daar is gevind dat die diffusie-versperringslagie nie die faseformasie op so 'n manier kon beinvloed dat (3FeSiz as eerste fase sou vorm in plaas van cFeSi. Daar is baie industriele belangstelling daarin om (3FeSi2 epitaksieel op Si(100) te groei vir gebruikstoepassings soos infra-rooi detektore en lig-stralende toestelle. Die gebruik van 'n diffusie-versperringslagie het weI die reaksie op so 'n manier beheer dat 'n uniforme lagie polikristallyne (3FeSi2 gevorm kon word by 800°C. In die geval waar Fe direk met Si(lOO) gereageer het, het die Fe op 'n nie-uniforme wyse met die silikon substraat gereageer. Die uniforme polikristallyne (3FeSi2 lagie was toe omgesit na epitaksieel FeosSi[CsCI] deur die gebruik van 'n 0.8 J/cm2 puIs van 'n excimer laser. Dit is die eerste keer dat 'n ononderbroke lagie epitaksieel FeosSi[CsCI] gevorm is op Si(lOO). Die (3FeSi2 lagies met groot korrels het die beste gehalte epitaksieel FeosSi[CsCI] gevorm na die laserpuls. Die korrelgrootte was atbanklik van die tipe diffusie-versperringslagie wat gebruik was. Die invloed van spanning op Ti-silisiedformasie was ook bestudeer. TiSi2 is algemeen gebruik in geYntegreerde stroombane en met die neiging na kleiner lynwydtes en toestelgrotes speel spanning 'n al hoe groter rol. Verskillende hoeveelhede druk-en trekspanning is in Si(lOO) substrate geYnduseer deur middel van die deposisie van Si02 ~n Si3N4 op die agteroppervlak van die substrate. Die lagies wat op die agterkant gedeponeer is vervorm die Si sodat 'n spanningsveld ontstaan deur die substraat. Die spanning in die substrate is bepaal deur die meting van die krommingsstraal van die monster m.b.v. 'n laser tegniek. Die spanning kan afgelei word van die krommingstraal deur gebruik te maak van Stoney se vergelyking. As Ti gereageer het met 'n Si substraat wat onder drukspanning was, is die diffusieproses so vertraag dat die TiSi2[C49] 50% stadiger gegroei het as wanneer Ti met Si substrate onder trekspanning gereageer het. In die monsters met die meeste drukspanning in die substraat is die Si diffusie so doeltrefend afgesny dat die Ti-ryke TisSi3 fase gevorm het in plaas van TiSi2• Hierdie word verduidelik aan die hand van die Effective Heat of Formation (EHF) model. Laterale faseformasie moet plaasvind om van die komplekse strukture te vorm wat op moderne geYntigreerde stroombaan bestaan. In hierdie ondersoek word die kemmikrosonde, elektron mikroskoop en optiese mikroskoop gebruik om faseformasie te bestudeer in die Ru-AI, Pt-AI and Cr-Si laterale diffusie sisteme. 'n Tipiese laterale diffusie monster bestaan uit eilande van een material gedeponeer op 'n dunlagie van 'n ander material. Daar is gevind dat indien die eilande nie die diffunderende spesie bevat nie, sal daar geen laterale faseformasie plaas vind nie. In al die aluminium sisteme het die aluminium eilande altyd laterale reaksie getoon. In die_ Ru-AI sisteem is dit gevind dat RU4AI13 die enigste fase is wat lateraal groei en dat die fase met reaksie-beperkte kinetika groei, terwyl in die Pt-AI sisteem twee fases lateraal groei (PtgA121 en PtAlz) met diffusie-beperkte kinetika. Die EHF model word gebruik om te verduidelik hoe diffusie beperkte groei aanleiding gee tot die gelyktydige formasie van meer as een fase. In die Cr-Si sisteem het net die monsters met Si-eilande laterale reaksie getoon. CrSi2 was die enigste laterale fase om te vorm in die sisteem en oksidasie het dit onmoontlik gemaak om enige kinetiese metings te doen. Daar word gewys hoe laterale diffusiestelsels gebruik kan word om inligting te bepaal oor diffunderende spesies en diffusie meganisme. 'n Opsomming van al die werk gedoen op laterale diffusiestelsels is saamgevat in hierdie ondersoek.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/51465
This item appears in the following collections: