EE 245: Introduction to MEMSLecture 7: Process Modules IIICTN 9/17/09Copyright © 2009 Regents of the University of CaliforniaEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 29Silicon Crystal Orientation• Silicon has the basic diamond structureª Two merged FCC cells offset by (a/4) in x, y, and z axesª From right:# available bonds/cm2<111># available bonds/cm2<110># available bonds/cm2<100>xyza[100](100) planeIncreasing@ coordinate (1,0,0)<100> plane ⊥to this vectorxyza][110](110) planexyza][111]e(111) planecoordinate (1,1,0) defines vector<110> plane ⊥to this vector@ coordinate (1,1,1), <111> plane ⊥ to resulting vectorEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 30Anisotropic Wet EtchingAnisotropic etches also available for single crystal Si:ª Orientation-dependent etching: <111>-plane more densely packed than <100>-planeSlower E.R.Faster E.R.…in some solventsOne such solvent: KOH + isopropyl alcohol(e.g., 23.4 wt% KOH, 13.3 wt% isopropyl alcohol, 63 wt% H2O)E.R.<100>= 100 x E.R.<111>EE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 31Anisotropic Wet Etching (cont.)Can get the following:(on a <100> - wafer)Si54.7°<111> <100>SiO2(on a <110> - wafer)Quite anisotropic!Si<110> <111>SiO2EE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 32Wet Etching SiO2SiO2+ 6HF H2+ SiF6+ 2H2OGenerally used to clear out residual oxides from contactsnative oxidecan get this just by exposing Si to air 1-2nm-thickProblem: Contact hole is so thin that surface tensions don’t allow the HF to get into the contactGenerally the case for VLSI circuitsoxideHFbubblent300nmSolution: add a surfactant (e.g., Triton X) to the BHF before the contact clear etch1. Improves the ability of HF to wet the surface (hence, get into the contact)2. Suppresses the formation of etch by-products, which otherwise can block further reaction if by-products get caught in the contactEE 245: Introduction to MEMSLecture 7: Process Modules IIICTN 9/17/09Copyright © 2009 Regents of the University of CaliforniaEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 33More Wet Etch Chemistries• Wet etching silicon nitrideª Use hot phosphoric acid: 85% phosphoric acid @ 180oCª Etch rate ~ 10 nm/min (quite slow)ª Problem: PR lifted during such etchingª Solution: use SiO2as an etch mask (E.R. ~2.5 nm/min)( A hassle → dry etch processes more common than wet• Wet etchining aluminumª Typical etch solution composition:ª Problem: H2gas bubbles adhere firmlly to the surface →delay the etch → need a 10-50% overetch timeª Solution: mechanical agitation, periodic removal of wafers from etching solution80% phoshporic acid, 5% nitric acid, 5% acetic acid, 10% water(HNO3)(CH3COOH) (H2O)(H2PO4)(1) Forms Al2O3(aluminum oxide)(2) Dissolves the Al2O3EE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 34Wet Etch Rates (f/ K. Williams)EE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 35Film Etch Chemistries• For some popular films:EE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 36Dry EtchingEE 245: Introduction to MEMSLecture 7: Process Modules IIICTN 9/17/09Copyright © 2009 Regents of the University of CaliforniaEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 37Dry EtchingAll based upon plasma processes.~+++++++++waferE-fieldRF (also, could be μwave)Develop (-) bias)Plasma (partially ionized gas composed of ions, e-’s, and highly reactive neutral species)(+) ions generated by inelastic collisions with energetic e-1’sGet avalanche effect because more e-1’s come out as each ion is generated.Develops (+) charge to compensate for∴(+) ions will be accelerated to the wafer• Physical sputtering• Plasma etching• Reactive ion etchingEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 38Physical Sputtering (Ion Milling)• Bombard substrate w/ energetic ions → etching via physical momentum transfer• Give ions energy and directionality using E-fields• Highly directional → very anisotropicPR PRfilmSiionsplasmaSteep vertical wallEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 39Problems With Ion MillingPR etched down to herePRPRfilmSiPR1. PR or other masking material etched at almost the same rate as the film to be etched → very poor selectivity!2. Ejected species not inherently volatile → get redeposition → non-uniform etch → grass!• Because of these problems, ion milling is not used often (very rare)Once through the film, the etch will start barreling through the SiEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 40Plasma Etching• Plasma (gas glow discharge) creates reactive species that chemically react w/ the film in question• Result: much better selectivity, but get an isotropic etchPlasma Etching Mechanism:1. Reactive species generated in a plasma.2. Reactive species diffuse to the surface of material to be etched.3. Species adsorbed on the surface.4. Chemical reaction.5. By-product desorbed from surface.6. Desorbed species diffuse into the bulk of the gasMOST IMPORTANT STEP! (determines whether plasma etching is possible or not.)PR PRFilm to be etchedSi123456plasmaEE 245: Introduction to MEMSLecture 7: Process Modules IIICTN 9/17/09Copyright © 2009 Regents of the University of CaliforniaEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 41Ex: Polysilicon Etching w/ CF4and O2• F° is the dominant reactant → but it can’t be given a direction → thus, get isotropic etch!CF4CF4++ CF3++ CF2++ CF++ F++ F0+ CF2++ …Neutral radical (highly reactive!)SiCF6, SiF4both volatile dry etching is possible.∴e-+ CF4→ CF3+ F + e-SiplasmaPRpolySiF0F0SiF4isotropiccomponentEE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 42Ex: Polysilicon Etching w/ CF4and O2PRpolySiF0F0SiF4isotropiccomponent• Problems:1. Isotropic etching2. Formation of polymer because of C in CF4ª Solution: add O2to remove the polymer (but note that this reduces the selectivity, Spoly/PR)• Solution:ª Use Reactive Ion Etching (RIE)EE C245: Introduction to MEMS Design LecM 4 C. Nguyen 8/20/09 43Reactive Ion Etching (RIE)• Use ion bombardment to aid and enhance reactive etching in a particular directionª Result: directional, anisotropic etching!• RIE is somewhat of a misnomerª It’s not ions that react … rather, it’s still the neutral species that dominate
View Full Document
Unlocking...