Silicide formation through diffusion barriers

• Main Author: Balde, Maryna
• Other Authors: Pretorius, R.
• Format: Others
• Language: en_ZA
• Published: Stellenbosch : University of Stellenbosch 2011
• Subjects: Diffusion; Silicides; Theses > Physics; Dissertations > Physics
• Online Access: http://hdl.handle.net/10019.1/17321

Dissertation (PhD)--University of Stellenbosch, 2006. === ENGLISH ABSTRACT: The formation of Ni-, Co- and Fe-silicides through different diffusion barrier interlayers was investigated. The diffusion barrier layers examined were Ta, Ti and Cr. In some cases the thickness of the barrier layer and the influence of a capping layer was also investigated. The thin-film structures were prepared on single crystal Si-substrates by Electron Beam Vacuum Deposition. The samples were vacuum annealed for times ranging from 10 to 60 min at temperatures ranging from 340 - 800°C and sample characterization was carried out by conventional RBS, dynamic RBS, channeling RBS and X-ray diffraction (XRD). The use of a thin (20Å) Ta diffusion barrier in the Ni-Si system allowed no reaction even after annealing for 10 min at 400°C, but RBS measurements showed that after annealing for 15 min at 400°C uniform NiSi formed suddenly as first phase. XRD as well as dynamic RBS measurements confirmed this abrupt formation of NiSi instead of the normal first phase Ni2Si. According to the Effective Heat of Formation (EHF) model this shows that the diffusion barrier reduces the effective concentration of the Ni atoms to a value where the effective heat of formation of NiSi is more negative than that of Ni2Si and first phase formation of NiSi is thus thermodynamically favoured. The thickness uniformity of the first phase NiSi that formed through the thin Ta barrier improved at higher annealing temperatures. A thicker (100Å) Ta barrier also retarded the Ni diffusion and first phase, non-uniform NiSi only started to form at 500°C. The uniformity of this NiSi also improved with increased temperature but the use of the 20Å Ta barrier produced more uniform first phase NiSi in the 400 to 700°C temperature range. The use of a thin (30-50Å) Cr barrier also allowed the formation of mainly NiSi at 400°C, although XRD spectra indicated the presence of some Ni2Si. The uniformity of NiSi improved at higher temperature anneals. Similar results were obtained from samples with a thicker (100Å) Cr barrier layer at lower temperatures, i.e. the formation of NiSi as first phase at 400°C, but the first phase NiSi that formed at 500 to 700°C was non-uniform. In the case of Ti-barriers, the thicker (100Å) Ti barrier seems less effective than the thinner Ti barriers in delivering uniform first phase NiSi in the 500 to 700°C temperature range. The use of a thin (30-50Å) Ti barrier produced a mixture of Ni2Si and NiSi as first reaction at 400°C, but a 10 min anneal at 500°C formed uniform NiSi as confirmed by RBS and XRD measurements. The uniformity of the NiSi improved with an increase in annealing temperature up to 700°C. In the case of the thicker Ti interlayer no reaction occured at 400°C and non-uniform first phase NiSi formed at 500°C. All three thin barriers formed NiSi2 at temperatures of 750°C and above, but the thin Ti barrier formed the most uniform di-silicide. The NiSi2 that formed at 800°C through all three of the thicker barriers was non-uniform. The use of a thin (10-30Å) Ta diffusion barrier prevented Co-silicide formation up to 560°C. The effective Co concentration at the growth interface is lowered, thus skipping the usual first phase formation of Co2Si at 450°C. At 560°C a mixture of CoSi and CoSi2 formed, as was confirmed by XRD. The CoSi2 that formed at 640°C (a higher formation temperature than without barrier) was of quite uniform thickness, but XRD measurements indicated that some CoSi was present as well. The use of thicker (100Å) Ta barrier layers retarded the diffusion of Co atoms for temperatures of up to 600°C. Annealing at 700°C formed CoSi2 and some CoSi and at 800°C non-uniform CoSi2 formed. The addition of a Ta capping layer (of different thicknesses) in conjunction with a 30Å Ta diffusion barrier layer did not significantly improve Co-silicide formation. The use of thin (10-30Å) Ti barrier layers resulted in the skipping of the Co2Si precursor phase and the formation of quite uniform first phase CoSi at 520°C. Uniform CoSi2 started forming at 560°C and the CoSi2 remained uniform at higher temperatures. The presence of a thicker (100Å) Ti barrier lowered the effective concentration of Co at the growth interface to such an extent that CoSi2 started to form as first phase after annealing for 30 min at 600°C. At 700 and 800°C non-uniform CoSi2 formed. For Fe-silicide formation the use of 50Å and 100Å Cr barriers, as well as CrSi2 barriers, delivered very similar results. There was no change in the normal Fe-silicide phase formation sequence, as non-uniform FeSi was the first phase to form at 500°C and thereafter FeSi2 started to form at 600°C. At 700°C the use of Cr barriers resulted in the complete formation of FeSi2 of greater uniformity than was formed in the Si-Fe binary system without the presence of a diffusion barrier. In this study dynamic real-time RBS has been used for the first time to prove without any doubt that diffusion barrier layers can be used to bring about “phase skipping”. These results have been interpreted in terms of the Effective Heat of Formation (EHF) model and are good examples of concentration controlled phase selection (CCPS). In general it was found that the thicker the diffusion barrier layer, the higher the temperature of silicide formation. Furthermore, silicide formation was generally found to be more uniform at higher annealing temperatures and when thinner diffusion barrier layers were used. === AFRIKAANSE OPSOMMING: Die vorming van Ni-, Co- en Fe-silisiedes deur verskillende diffusie sper-tussenlagies is ondersoek. Die diffusie sperlagies onder beskouing was Ta, Ti en Cr. In sommige gevalle is die invloed van die dikte van die sperlagie en van ‘n deklagie ook ondersoek. Die dun-film strukture is voorberei op enkelkristal Si-substrate d.m.v. Elektronbundel Vakuum Deposisie. Die monsters is in vakuum uitgegloei vir tye wat wissel van 10 tot 60 minute by temperature wat wissel van 340 - 800°C en die karakterisering van die monsters is uitgevoer d.m.v. konvensionele RBS, dinamiese RBS, kanaliserings RBS en X-straal diffraksie (XRD). Die gebruik van a dun (20Å) Ta sperlagie in die Ni-Si sisteem het reaksie verhoed selfs na ‘n uitgloei van 10 min. by 400°C, maar RBS resultate het getoon dat uniforme NiSi skielik gevorm het as eerste fase na ‘n 15 min. uitgloei by 400°C. XRD sowel as dinamiese RBS metings het hierdie abrupte formasie van NiSi in plaas van die normale eerste fase Ni2Si bevestig. Volgens die Effektiewe Hitte van Formasie (EHF) model toon dit dat die diffusie sperlagie die effektiewe konsentrasie van die Ni-atome verlaag tot ‘n waarde waar die effektiewe hitte van formasie van NiSi meer negatief is as dié van Ni2Si en sodoende word die eerste fase formasie van NiSi termodinamies bevoordeel. Die dikte uniformiteit van die eerste fase NiSi wat deur die dun Ta sperlaag gevorm het, het verbeter met uitgloei by hoër temperature. ‘n Dikker (100Å) Ta sperlaag het ook Ni diffusie vertraag en nie-uniforme, eerste fase NiSi het eers by 500°C begin vorm. Die uniformiteit van hierde NiSi het ook verbeter met toename in temperatuur, maar die gebruik van die 20Å Ta sperlagie het meer uniforme eerste fase NiSi produseer in die 400 tot 700°C temperatuur gebied. Die gebruik van ‘n dun Cr (30-50Å) sperlagie het ook by 400°C die vorming van hoofsaaklik NiSi opgelewer, maar XRD spektra het Ni2Si ook aangedui. Die uniformiteit van NiSi het verbeter by hoër temparatuur uitgloeiings. Soortgelyke resultate is verkry van monsters met ‘n dikker (100Å) Cr sperlaag by laer temperature, d.w.s die vorming van NiSi as eerste fase by 400°C, maar die eerste fase NiSi wat by 500 tot 700°C gevorm het, was nie-uniform. In die geval van Ti-sperlagies was die dikker (100Å) Ti minder effektief as die dunner Ti sperlagies vir die verkryging van uniforme eerste fase NiSi in die 500 tot 700°C temperatuur gebied. Die gebruik van ‘n dun (30-50Å) Ti sperlagie het ‘n mengsel van Ni2Si en NiSi as eerste reaksie gevorm by 400°C, maar ‘n 10 min. uitgloei by 500°C het uniforme NiSi gevorm, soos bevestig is deur RBS en XRD metings. Die uniformiteit van die NiSi het verbeter met toename in uitgloei-temperatuur tot by 700°C. In die geval van die dikker Ti sperlagie het geen reaksie by 400°C plaasgevind nie en nie-uniforme eerste fase NiSi het by 500°C gevorm. Al drie dun sperlagies het NiSi2 gevorm by temperature van 750°C en hoër, maar die dun Ti sperlaag het die mees uniforme di-silisied gevorm. Die NiSi2 wat deur al drie die dikker sperlae by 800°C gevorm het, was nie-uniform. Die gebruik van ‘n dun (10-30Å) Ta diffusie sperlagie het Co-silisied formasie voorkom tot by 560°C. Die effektiewe Co-konsentrasie by die groei-intervlak is verlaag, derhalwe word die gewone eerste fase formasie van Co2Si by 450°C oorgeslaan. By 560°C het ‘n mengsel van CoSi en CoSi2 gevorm, soos bevestig deur XRD. Die CoSi2 wat by 640°C gevorm het (‘n hoër formasie temperatuur as sonder ‘n sperlagie) se dikte was redelik uniform, maar XRD metings het getoon dat daar ook CoSi teenwoordig was. Die gebruik van dikker (100Å) Ta sperlae het die diffusie van Co-atome vertraag tot by temperature so hoog as 600°C. Uitgloei by 700°C het CoSi2 sowel as CoSi gevorm en by 800°C het nie-uniforme CoSi2 gevorm. Die toevoeging van ‘n Ta deklagie (van verskillende diktes) in samehang met ‘n 30Å Ta diffusie sperlagie het nie Co-silisied formasie wesentlik beïnvloed nie. Die gebruik van dun (10-30Å) Ti sperlagies het gelei tot die oorslaan van die Co2Si voorloper fase en die vorming van redelik uniforme eerste fase CoSi by 520°C. Uniforme CoSi2 het by 560°C begin vorm en by hoër temperature was die CoSi2 steeds uniform. Die teenwoordigheid van ‘n dikker (100Å) Ti sperlagie het die effektiewe konsentrasie van Co by die groei-intervlak so verlaag dat CoSi2 as eerste fase begin vorm het na ‘n 30 min. uitgloei by 600°C. By 700 en 800°C het nie-uniforme CoSi2 gevorm. Vir Fe-silisied formasie het die gebruik van 50Å en 100Å Cr sperlae, sowel as CrSi2 sperlae, soortgelyke resultate opgelewer. Daar was geen verandering in die gewone Fe-silisied fase formasie volgorde nie, want nie-uniforme FeSi was die eerste fase wat by 500°C gevorm het en daarna het FeSi2 begin vorm by 600°C. Uitgloei by 700°C deur Cr sperlagies het gelei tot die volledige formasie van FeSi2 wat meer uniform was as dié wat in die Fe-Si binêre sisteem gevorm het sonder ‘n diffusie sperlagie. In hierdie studie is dinamiese intydse RBS vir die eerste keer gebruik om bo enige twyfel te bewys dat diffusie sperlae gebruik kan word om die “oorslaan” van fases te bewerkstellig. Hierdie resultate is interpreteer in terme van die Effektiewe Hitte van Formasie (EHF) model en is goeie voorbeelde van konsentrasie-gekontroleerde fase seleksie. In die algemeen is bevind dat hoe dikker die diffusie sperlagie, hoe hoër die temperatuur van silisied-formasie. Bowendien was silisied-formasie oor die algemeen meer uniform by hoër uitgloei-temperature en met die gebruik van dunner diffusie sperlagies.