CA2160234A1 - Electromigration resistant metallization structures for microcircuit interconnections with rf-reactively sputtered titanium tungsten and gold - Google Patents

Abstract

Two metallization schemes of PtSi/TiW/TiW(N)/Au (Type I) and PtSi/TiW/TiW(N)/TiW/Au (Type II) and associated process are described for microcircuit interconnections. The metallization schemes and process are capable of ICinterconnections with a metal-pitch as small as 1.5 .mu.m, or even smaller. The metallization schemes are reliable for continuous high temperature and high current operations.

Description

WO 95122838 i P~ S~OOIS2 21~Q234 FT FCTROMIGRATION RESISTANT ME~ALLIZAIION STRUCTURES
FOR ~ICROCIRCUIT INTERCONNECIIONS
WlTH RF-REACIIVE~Y ~U l l~KED TITANIUM
- I~J(i~l~J AND GOLD

SBACKG~l~EJND OF'1 H~ rNVEN~lON
1) Field of the InventiaL
The p~esent i}vention r~lates, in gene~al, to m~ tion schPmes for mi~ ;uil iDte~nnloc1loll~ of electronic devices and, more parricularly, to PtSi/lrlWtTiW~N)/Au (Type I) and PtSi/TiWtTiW(N)t~lW/Au 10 ~ype ~) gold ~.~u~ ~nc of ~Pmi~nn~ t~tnr devices and :~C~ tf'd ~l.,ces~s of m.o~lli~t;Qn 2) Di~c~ ion of Related Art.
~ ~ent~l, high p~rU~ ,n~n~ and a higher degree of ;..lt~g.,.l;on of 5~mjo~n~Uctor deYic~s re~e r~duction of overall ~limenci~mc~ in~ lrling the 15 ~i;",Pn~;nn~ of metal leade~s and contact hole ~i~m~ptprs (inn3u~in~ via t~ t t~ s ~l~ctinn m ~imP~n~ion~ cau~s i~ic~s in chip o~ ;n~
t~ dlll~5 and cuIrent d~ti~S in the me~al stTipes and int~ es bel~n silicon and the metal con~lll~ in the contact holes.
- At the samc time, a decrease m - contact hole 20 conse~u~lly inCl~S its aspect ra~o ~l.e., st~p height/~ m~t~), and hence WOl~S the s~p cove~age of the metals. A smaller step coverage in the contact holes causes filr~her inCl~lS~ of the cuIr~t density in the contact holes.
~ Iigh power tra~L~)ls, in particular, gencldl~ large ~u~ and develop heat. The high curr~t density and high ~ op~ion~ of the 2~ de~ices rEsult in ~ather fast d~ ti'~n of an ~1..." ". .., alloy (CO~ nly i~CI~in~ c~ryer and silicon) llased ih~ ~nn~l;nnc due to an eles~ ti-)n effe~t. The de~llvlll;~tion~ caDd by mo~ r ~ of ele~ v~s~ causes fo""~tion of voids and an i~uylion in ~e metal con~u~ors.

WO 95n2838 ` ~ PCI`/SE9S/00152 2~6~234 Gold is a good conductor and is more than seven times larger in atomic weight than ~lnminllm Gold based intclcon.~ ;onc are very resistant to the electromigration effect. However, because of the Au-Si e~te~tic (--370C), diffusion barriers are needed which should possess some e-ss~
5 ~lopel ~ies, such as:
long-term high te.ll~,dture stability, good electromigT~tion reCict~nce with a high current density, good adhesion to both inc~ tin~ films and gold cnn-luctor, a good conductor, low contact reCict~nc~ to doped (n+ and p+) silicon, good step coverage in contact holes, non-partic~ tion, and a wide ~l~SS l~itude even for small ~1;,.. <;o,.c (~ l~m).
Gold based inler~onnP~tionc can usually be formed in two ways.
First, a lift~ff technique based on evaporation of ~iffiUci~ n b~Tiers and goId 15 can be used, e.g., a lift-offof Ti/Pt/Au metals. See~nd, a s~ul~ ;.lg technique combined with plasma etched gold or ele~;tloplal~ gold cr~n~lu~orC can be used.
One of the ma3or advantages of the lift-off technique is that it rather easy to form metal-pitches below 2 ~m. A sucr~ccfill Lift-off technique l~U~._S an elec~on-beam C~uld~Ol with a large ~ ce between metal 20 sources and wafers. For ;~ ;.nc~, for 6n ~ metp- ~rafers the fliC~nce must belarger than 145 cm m order to obtain an inr~d~nt angle deviation of the evaporated metals less than 3 degre~s from the normal to the vafer S1~A~
Consequently, the size of the e~ apo~oi will be huge and the waste of precious metals will be ve~y large. F~ ..U~re, a l~ui~ ~t on the photoresist 25 profiIe is quite high: a thick (> 2.5~Lm) and ~f II~AIII profile having negatively sloped sidewalIs (~90) as shown in Figure lc. The l~lu~que yields ve~y low step coverage and con~icttontly proves very ser,siLi~e to particles with re~ect to the reliability of the diffilsion ba~
The ~ull~ g and electroplating t~chni.lues co...hit-rd with 30 plasma etching is easily feasible with common 4 ~ nl and have no f~iffi~l~lty WO 9S/22838 , PCI/SE9S/OOlS2 with a large wafer tli~m~t~r (> 6") in contrast to a lift~ff techu~ique.
However, a drawbac~ with this technique is the difficulty in chrinking a metal-pitch below 4 ~m due to some senous process b~mers, such as:
- photoresist loss in an electroplating bath, geometry d~pendPnc~ of plating speed and ~iffi~ ty in plating fine lines together with large size geometry, too rough Au~o~ductor surfaces after a wet-etch of the s~u~
Au-film used for electroplatMg and ~ole~;lion of diffusion b~rri~rs, and etch undercuts at the diffusion b~rrito~.
Even by a plasma etch technique it is ~1iffi-` l1t to elimin~t~ some undercuts be~u~ the plated gold layer must act as an etch mask instead of a photO~al.
Gold is much softer than diffusion b~.l;P ~, and hence it is .1imr.1lt to apply a strong physical apu~ g effect in a plasma etch ~roc~a~
At the same time the gold etch mask can not offer any side passivation with o,~c ~l~llle,a like a photoresist etch mas~ The.erol~ it is very to Su~pl~a ~e etch undercut for the microcircuit ih~ orn~hr~n~
In the past !;~ t~-~ or/and nitri~ ~;t~.-;.. -t.. -~t. n, herein ~no~ as TiW and TiW(N) (which do not stand for a stoichiometri-co,n~osiLion but for a ~pseud~alloy" of TiXWlX o~ d from a c~thode eo",~,ilion of X = 0.30 (eorresponding to Ti = 10 wt9~)), have been utili7~ The ac~ual co",~ailion of the ~uu~.~d TiW- and rlW(N)-films vanes by ~,~ .;n~ p~r~m~tprs~ and accordingly varies the film p~u~lies as a d;rru~;on barrier and adhesion layers, etc.
Previously, ~e dif~sion b~rriP~ of TiW and TiW(N) have been used in the mPt~lli7~tion systems of Si/TiW/Al or PtSi/l~iW/Al as IC--;On~ or Al(bondpad)/TiW(N)-TiW/Au (wire or bump) for p~r~Ps.
J.A. Cllnn;n~ et al., ~co~ùaion Re~;~l;.n-~ of Seve~al Tntc~t~l-circuit lupt~ 7atio~ Systems,~ IEEE Tr~ ions on Reliability, VoL R-l9, No. 4, November 1970, pp.182-187, dicc~c~s a mp~ni7~tinn WO 9S122838 `~ PCI/SE9S/OOlS2 21~Q23~ -system of Al(bondpad)/TiW/Au(wire), and R.S. Nowicki et al., ~Studies of the TiW/Au mPt~lli7~tion on ~ Thin Solid Films, 53 (1978), pp. 195-205, ~iicCuccPs a m~Pt~lli7~tion system of Al(bQn~ Tiw(N)/Au. The Nowic~i et al. article in~ dP5 obs~ ~lions about ~l~ e ;nlf ~ ;ng of TiW
5 and Au at 300C for 6 l~ t ~s whe~n the TiW-film was relatively pure. They also observed that a reactively *Jull~ film of Tl~,O) illl~JlU~ the diffusion ba~Tier plopelly by orders of m~ni~de Nltrogen and o~cygen in the TiW(N,O)-film was --42 and --8 at.% (atomic ~l), r~Li-~ely. The o~cygen atoms in the TiW(N,O)-film were ~ ;oll~lly 10 involved from residual gases in the sputter ch~ r. The TiW(N,O)-film even had minim~l in~inC-c stress, otherwise a TiW(O)-film with O--4 at.% had tensile stress.
Two patents to R.K. Sharma et al. ~ sP the use of a double-layer diLru~on barrier of TiW(~)/TiW in a mpt~ ti~)n system of Al(bondpad)tTlW(N)(1.5-3.0kA)/TiW(0.5-l.OkA)/Au(2-7 kA) for TAB (ta~e alUll)lll~led bonding) or wire bonding ap~li~tionC See U.S. Patent Nos.
4,880,708 and 4,927,505 dated Nov. 14, 1989 and May æ, lsso, ~
Prpfipl-hlp ~ intervals are g;ven in the 1~ r~s The TiW(N) film was ob~ilined by ~ul~ulg with a gas ~ ul~ concicting of at least 30 vol9~ N2 20 in Ar. There L no illr~ AI;on about the ~putter t~ ique.
The ma3or reason to in~ de Ti into W is to i.~ ove the A~lh~
~lu~.ly to SiO2-layer bec~,~ strong rl-o bonds could be ~ 1~ at the int~rfA~ The II~A~;IIIIIIII olubility of rl into W i~c only about 10 at.96 at 600C. Any e~cess Ti will be ~ us~ically di~l~ibut~d in a "pseudo-alloy~
25 of ~lW, and even located at g~ain boundAri~s and i,...Ç--~c The svbatAI~llAl improvc"le.ll of the diffusion barrier pl~ could be a~ ~ by n~ifl-Atinn or/and n~ Ation of the Ti-atoms during ~lt. ~ ;np of TiW~athode (~r~et).
The form~tion of stable TiO2 and TiN at grain b~uad~ particularly se~ems to cffecdvely slow down ih~ nL grain boundary ~.rr..Q~ However, one of 30 major dld~lJaC]~ of the o~id~tion and nitridAtion of the ~l-atoms is the loss or WO 95n2838 i P~ k9S/OOlS2 2~602~

reduc~ion in the ability of fo~ Ti-O bonds at the interface between SiO2 and TiW.
For the p L~ge appli~tionc~ ~unnin~h~m et al. and Nowicki et al. have been suc~ r~l in using TiW(N) directly on top of Al-bon~
5 bc~n~, of good ~dh~or bcl~xn Al and W which can form i..tr....~ niC
alloys with each other. Furthe~ ul~, one could easily apply strong sputter etch on the Al-surface in ord~ to c~i...;n~lto. A1203 and on the passivation film. The s~ong sputter etch could roughen the s~ s, and hence could .onh~n~-R
~he5ion In any way, some lateral ~ on losses of a few mic,ul~l~te~ in 10 an etch due to weak ~lh~nn bcl~cen ~lW(N) and the passivation film of SiO2 (~SG, BPSG glasses) or Si~ can not infl~l~nce reliability be~ause usually the diffusion barrier covers l~early 10 ~m over the passivation film.
However, the sihl~tion is quite dirf~le,lt in the miclueil~;uiL
intelconn~tionC where a metal-pitch can be as small as 1.5 ~m. The ~h~ion 15 strength bet~.~n a SiO2-layer with contact holes (< 1 ~Lm) and a diffusion barner has a decisive ;- n~ ~ ~ on the size of the etch undercut and rP~ ity~
In this case the coveragc of the diffusion barrier over the contact hole edge (Fig. le) may be a few tenths of a micron where one-tenth micron is very t. The appl;~ of a sputter etch is very limited due to vesy shallow 20 depths of j!~.cl;on~ for ~F ~igh ~u~ncy) power tr~nsi~tors and high speed de~rices in gen~t DeI~ing ct al., ~Reliability of High Te~ dLu~ t~ldl~d Circuits,~ E~E/rnh -- ~n>n~1 Reliability Physics Sv~ osiu~ll Proc.. 1984, pp.3~36, ~ s a ~ ti~-n system of PtSi/TiW(250 A)/TiW(N)a lcA)/TiW(250 A)/AU(5 }A)/TiW(500 A) for IC~ er~onn~tit n The ar~cle .~.fn~ nS a lift-offt~hn;~ applied to 3" wafers. There are no de~ails about the lift~ff tc:chni~lue in tbe paper. However, there are some serious feasibility problems in IC-pru-Ju~ For i~ ;.n~, it is very lifficl-lt to use a pho~l~st or some polymer ~Ih --c for the lift~ff due to too large heat develo~-~ent 30 :~tSOc .,.led with c~ ;on and co~(~Pn~tion of the W-metal (boiling point =

WO 95122~38 , ``. PCI~/SE9SIOOlS2 ~60234 5660C) because the pattem will be deformed by heat and stress. Fur~h~n~re, even ~ith 3" wafers the stress level was so high that Au~nductors formed hillocks and voids. The high stress level also will limit to increase the wafer size. An input of N2-gas dunng the eV~I~Q~tinn of the Ti- and W- metals will 5 cause tliffir~llty to lift-off the metals due to step coverage g~ ~P- -t~ by collicionc with the N2-mol~-lrs Most of all, it is almost illl~ssil~le to reproduce the film quality because during the reactive evaporation the melted Ti- and W-metals in the crucibles will react vith the N2-gas.
Acc~ g to the present inventor's ezpf ;~-~ce as well as that 10 l~ d in the Dening et al. ar~cle, the ~rlhesion s~ lh bclv~n SiO2 and TiW(N) vas not s lffiriPntly strong to withc~n~ the s~ess causecd by thPrm~l mi.~ t~hin~ b~ the deposition t~ of TiW(N) may be in a Iange of 100 350C and a diLrc,cnce of TOE (thlorrn~l c~ffiriPnt of e~p~ncion) can be 4-8 ppm/C. This weak ~hP~ion caused lln~rr~t~hly large etch Lnder~;u~
15 and o~ent;...~-s left a small caviq at a sidewall of the contact holes. The form~tinn of ~e cavity s-P~i~ucly caused rç1i~hility problem because Au-diffusion into the bulk-Si as well as Si diffusion into the Au-layer was observed after ~nnP~ling at 420C for 30 min. The ~ h~.~;c... seems to be sur~ce diffusion.
K.A. Lol~Len et al., in U.S. Patent No. 5,1?3,449 dated Dec.
22, 1992 PntitlP~ "~P~lli~tion Process", flicrlocP a InP~lli~tion process for a 5~h~--C of ~lW(0.2~.8 kA)/TiW(N)(2-5 kA)/TiW(0.2-0.8 kA)/Au(5-20 kA)/TiW(1-4 kA, as an etch mask for Au-p~ ..;n~ and being etched away later) for a~pli~ti~ n~ of microdevice i~l~On~P~tiQnc M~jor ~lvcess sk~ps are 25 based on ~u~ P and plasma etching techniques. However, the ~lucess l~tit~ (feasibility) for~ ";;~a the fine line Au~n~uct~ (1-1.25 ~Lm width or less) seems to be low be~ e the plasma etching of a thick Au-layer is based on pure physical ~l~u~ by argon and o~ygen gases. Consequently, an ~NO3-boil is l~C~ n~ed after the Au~tch in order to clean the etch l~ldUe5 30 of Au, Ti and W-metals. However, the nitlic acid boil can easily attack ~e WO gsn2838 ~ PCT/SE9SIOOlS2 21~0~4 TiW-layer unless the TiW-surface had already been o~cidiæd.
The o~ tion of the TiW-surface at the same time causes an o~ tion of the etch residues of Ti- and W-metals from the eroded etch mas~.
Consequently, the clP~nin~ plo~lule is in~rfs~/;~te and risky. The o~ tion 5 step can ~otr.~t;~lly cause an ~hP~;nn failure b~L-.~n the Au- and TlW-layers, e~lly with fine line Au~on~n~o s .~cn et al. in U.S. Patent No. 5,173,449 desc2ibe in detail the ~mPt~pr settings of a sputter system (~C~03, M~tPri~l Res~.,h Corp., N.Y.), P~e~lly about N2-gas purge ~l~l~ after a reactive s~,u~r~ ;n~ of 10 the TiW(N)-film. However, there is no deseli~lion about the applied sputter technique which is of decisive ~ll~ Ldl~ce for the reactive ~l~ul~ g of the TiW(N)-film. They use a N2/Ar-gas ll~i;~Lul~ of which the N2 conrPn~ti~ n is in the range of 14-38 vol~o. The mPhl~ tinn scheme ~l~yos~ by the Lol~ zcn et al. patent is not ~ullwll with respect to contact r~Cichncp and 15 high ~ stability be~ bottom layer of TiW has direct contact with the Si Suball~.

SUMMARY OF l~ INV~NIION
The present invention p~vides a good ohmic contact and stability by forming a thin layer of PtSi in co,.~ It is well known that Pt~;i and PtSi2 20 reduce contact re~ic~nr~ However, it was not known that the PtSi-layer could act as a 5~hili7~ of a TiW based dirru~ol~ barrier. The s~bili7in~ "~f,~t~ m was either a 1~ ~yl~c or ".feh~ni~l c~ like a stress buffer, and the e~cact m~h~nism iS not eS~hli~h~ yet. It is also we~l known that TiW-film is quite stress sensitive and in~n~ stress is tensile. A direct contact be~
25 a TiW-film and bulk-Si (~+ and n+ doped) was not stable at a high tClllpCldlul~ (420 C).
The present ~wllion provides a very stable diffusion ba~rier by ua~ing a nitn~led TiW, TiW(N), o~l~incd by an RF-~ a~u~ g technique.
A co.~ ,C study of the diffusion b~ ~ulL~d by dilr~.~nt ~echniques WO 9S122838 `~ PCI~/SE9SIOOIS2 2 ~6Q23~ -was ~ru~ ed by SlMS (secQnd~ry ions mass a~;llull.eter) and SEM
(Sc~nning electron mic~seope). The stability of the Tl~(N)-films obtained by DC-m~ .un and R~-m~n~otron s~ f ;llg is much infenor to the RF-~llU~ films (Figs. 2 and 5).
SUpprionty of the RF-reactive ~ has also improved adh sion strength and ste~ cove~age as well as lcd~ particle ~ .l;on C~ler.lly the DC-I~n~ ~o~ technique is e~clusively used in IC-mPt~lli~tion because of a higher sputter ~ate.
The present invention provides good ~r1hPQ~n ~l-.~n SiO2 and TiW layers and ~,tl.~ the diffusion ba~ier and a Au-layer. The invention offers two lcinds of interf~ s which are dP~-ndent on co.~ of residual gases of H20 and 02 in a sputter rh_---hf . For a high vacuum beKer than 10-' Torr, an intf ~G~ of TiW(N~Au ~ype I) is ~efL.~ble for c~ n~d~ high rPli~bi1ity.
However, for a moderae orrather poor vacuurn cQ~ iL;o.~ an intPrfac~ ûf T
Au ~Type II) is ~l~f~ble. The latter il,l~oves slighdy the ~ lhP~ n ~
but has a shorter liff ~ due to form~ti- n ûf an intL. ~nf~l~llie alloy ~h. ~C~ the Au- and Ti-metals.
The ~lhff;on SL~ is depP-n~if~nt on o~cygen and nlL~o~cen lf rL~ in the TiW(N,O)-film. Au-peeling was mosdy caused by a thin rlo2-film at the intP.~r~e which crr~;livcly ~ ed Au ~ ,h~l;on ~ JU~h the o~ide into the diLru~on ba~Tier. The Dening et al. article obs~ved that the ~hPcinn ~ 1 between I~W(N,O) and Au was wea~er than the inl~
b.,.~,.~ TiW and Au. In their paper there is no ~ n~ of the reason.
HoweYer, the weaker ^~lhe~i~m might largely depend on a rather high o~cygen c~ntPnt (--6%, by AES, Auger Elcctron s~oscop~) just at the intP~r~ and uncontroll_d ni~geA contcnt (which might be too high) in the rlW(N,O)-film due to a lift~ff t~ ~hn; l ~
The: ~hP~ L~ ~I..~n SiO2 and bottom glue layer of TiW-film ha an ih~ ;..n.,~ on the etch ~nd~.;~ll and the deYice 30 ~li~hility. Weak ~h~Q~ easily creates a small cavity (c~asse) in the contact wo ssn2s3s Pcr/SEsS/ools2 2 ~ 3 2 3~l hole due to te~ dllllC mic ~ rl~ g dur~ng alloying. As the result, surface diffusions of the Au-atoms into the Si-sub~ t~, and the Si-atoms into the Au-con~uctors will be g~dtly f~l`ilit~t~
- The ~lcS~L invention also offers corrosion resistant met~lli7~tion 5 s~h~ ~..PS for the microdevice i"L~.~n.l~tionc Thus, the ~ t invention ~lates to se~co~ u~r devices, especially to gold mL~onn~;~;oll~ of silicon based devices. The invention relates to deYices for high current and high ~ c oper~tinns with good ~li~hility and a loQg lifetime. The invention dil~:tly relates to two types of 10 mul~layer diffusion b~-.;o.s bcl..~n gold and silicon b~ce of the gold-silicon eu~ti. at--370C.
This invention relates to mul~laya diffilsion bArriP~s with low contact recic~Ance to the silicon a~sl.At~ and with good A~lh~n to inc-llA*n~
layers of SiO2 ~ncl~i~ely, doped glasses of PSG and BPSG) and Si3N4 and to 15 the gold cQndl~ctors This inven*on also relates to mul*~ayer diK~on barIiers which cause smaller undercut in a plasma etch.
This i,l~llion fi~er r~latcs to a ~ e sputter technique which is sui~ble to obtain most r~ t;~Al ylOp~l;~ of a di~usion bamer; for 20 c~A...plP long-terrn high b IIl~ Allu~ stability, good Al1h~ n, less partir,l-l-Atinn, better step cove~age in contact holes and wide process la~i~de even with a metal pitch of 1.5 ~m, or smaller.
This invention -A~fliti~n~lly relates to an ea~ly feasible and cost , mPtAlli7Ation ylY)C~ it~hle for mass ~lu~l;on of fine geol..eL
25 gold COfidU~

BRI~ DESCRIPIION OP DRAWINGS
llle present invention will now be ~e~ ~ ;hed by reference to the following d~ gs.
Figs. la-lg illl-~te the f~hri~ti~ s~u~ccs of the u~ d~,.ice WO 9S122838 ~ CI/SE95100152 2~ 0 23~ -ul~r~nnections .
Fig. 2 shows a S~S' analysis of the RF reactively ~yuu~d Type I m~t~lli7~ n scheme after ~nn~linE at 420C for 280 hours (in N2).
The mul~layers are: PtSi(200 A)/TiW(250 A)/TiW(N~(1.8 kA)/Au(l kA).
Fig. 3 shows a SIMS analysis of the DC-m~ on reactively ~yuU~d diffusion barrier ~nnP~l~ at 420aC for 30 min. (in N2). The multilayers are: PtSi(200 A)/TiW(N)(3 kA)/l~lW(l kA)/Au(10 kA).
Fig. 4 shows a SIMS analysis of the R~;-m~g...,l..,n ~ ly *~ul~led Type II mP~lli7~tion scheme after ~nnP~lin~ at 420C for 30 min. ~n N2). The multilayers are: PtSi(200 A)/TiW(250 A)/TiW(N)(1.8 kA)/TiW(120 A)/AU(12 kA).
Fig. S shows a SrMS analysis of the RF rea~vely ~ulh~d Type II m~t~lli7~ n st~h~m~o after ~nnP~inp at 420C for 90 hours (in N2).
The multilayers are: PtSi~200 A)/TiW(250 A)/TiW(N~(l kA)/TiW(250 15 A)/AU(1 kA)-DETA~ ~ l l DESCRIPIION OF T~ INVENTION
Two mPt~lli7~tion seh~--Ps of PtSi/l~ W(N)/Au (TypeI) and PtSi/TiW/TiW(N)/l~lW/Au (Type Il) and ~ ~ process are ~ &d for mi~.*,ir.;uil ~nnP~I;o~ c The m-~hlli7~tion srh~ S and plOC~ are 20 capable of IC-ihl~-"nnP~hnn~ with a metal-pitch as small as 1.5 ~m, or even smaller. The Ine~lli7~tion s~h- ..P~ are reliable for co..~ ous high t~
and high current ope~tirmc In a~ldanc~ with the present invention, the disadv~nl;~g~s of TiW/Au, TiW(N)/Au and TiW(N)/TiW/Au types of m~t~lli7~tir~n S~ p~ s for 25 mic~odevice i~ ~nnp~l;ons are illl~ )V~d w~th r~t to ~rlhP~i~n and high ~lll~dlul~ long-term rPli~hility as ~ slnn ~a-~ The pl~nt Ul~ i pro~rides the poscil~ility of ~3u~ g the whole m~l1i~tinn system to the range of 1 ~m line width. The present invention has no or in~igJ~ nt f~ili1y problems. At the same time, ~e present invention is one of ~e most cost WO 95/22838 i P~ISE9SIOOlS2 -21 ~23~

effe~tive met~lli~tio~ plOCC55~S.

T~e present invention consists of two met~lli7~tion schPmes and an associated plo~ss:

Type I: PtSi(0.1-0.4kA)/TiW(0.2-O.SkA)/TiW(N)(1.~2.0kA)/Au(0.3-1.0 kA)/Au(7-15 kA, electroplated);

Type II: Pt~i(0. 1-0.4 kA)/TiW(0.2-0.5 kA)/TiW(N)(1.0-2.0 lcA)/TiW(0. 1-0.5 kA)/Au(0.3-1.0 k~)/Au(7-15 kA, e~ l~o~l~ted).

The Type I is c~ncid~ored better for a high va~;uulll ( ~ 10-7 Torr) 10 condition in a ~putter chamber, while the Type II is better for a moderate or rather poor vaclmm ( > 1~ Torr) system. For both cases the N;~ content in Ar is 5-10 vol~o for the reactive sl)u~ of the TiW(N)-film. For the better ohmic contact and stability of the multilayers of ?/~lh~ n and dif~usion barrier, a thin layer (10~400 A) of PtSi is formed in t_e contact holes. The invention in~ des a depoQt a pure Pt-metal (app~ A~ly half of the PtSi ll"~ "~c~) by ~ltlf~ P (~l~f~dbly in the same sputter mA~hine for the rest of the metals) or e~rd~laLing at ;~n o~tide-free s~ee of the Si-~lb~l.A~. A thin (< 20 A) native SiOz-film is s~lffi~nt to make the whole ~ tinn ~)lOCCSS unc~tain with re~pect to the film th;~ llf ~S The fonnAti~n of the PtSi-layer is p~,~dbly pe~rul~ed in an inert ~Ao~hf l~ of N2 by ~nnP-ting at 600~20C for 30 min. The su~p}us Pt-metal is preferably ;,~ pod away in a co~ .l;on~l ..A~nf ~ with a ~L~ e of HNO3 ~ 3HCl sQltltioTl.

Before the ~ull ;ng of the first glue layer of I~W-film, a mild sputter etch is Feferably pc~ru~ ed by an RF-sputter l~i~ue in order to 25 f li,,,;,.~ some ~ive SiO2 ~i well as ~;2O at the s ~. G-~r 5 of the PtSi and SiO2-film, ~ ~. Wlthout the sputter ctch it is very ~liffi~llt to P.li~";n water and lllo~ ,r of OH-groups which are h~d~g~.~ bonded to the SiO2 s~ One of the major re~wns for failure to adhere to the SiO2 surface is the hyd~.cn ba}ded surface wa~er and too high residual gas con~ .AI;on~ of 30 H20 and 2 iD ~e sput~er ch~s.

wo 9Sn21~38 `~ . ~ PCI/SE9S/OOlS2 216Q23~

A presputter with the TiW-target is prefe~bly for ~some .~ ul~ s (2-5) in order to clean the target s~l~c~ and to getter the residual gases of H20 and 2 in the sputter c~
A thin layer aoo-500 A) of TiW-fiLrn is ~uU~od and used as S a glue layer. The punty and thirl~nPcs of the TiW-film is i~ nt. A high content of o~cygen and nill~g~ in the film causes a weak ~lhPcion to the SiO2-film. A too thick TiW-film is not preferable ber~l~ce there is a ~ndPncy to form some si~ ps by taking the Si-atoms from the PtSi-layer (Figs. 3-5).
~lthout the PtSi-layer the s~bility of the TiW-film was quite poor. Rather fast 10 diffusion of the Ti- and W-atoms into the Si-aubsl~lf' was obse.~
A r~tively ~uu~l~ TiW(N)-film (1.0-2.0 kA) is obtained by an RF-~ul~ technique with a gas ~ e of N2 and Ar. The mi~cing ~atio ~N2~or.rf~n~tit-n in Ar = 5-10 vol%) is ~epPndent on an RF-power density (W/cm~), yl~ul~ (1~15 mTorr), N2-gas flow rate and wafer t~
15 which de~ ne the rh~mi~l reaction rate of l~hv~,en with the Ti-and W-atoms.
A high RF-power density (> 1.5 Wtcm2) is p.~fc ,~hle in order to obtain a ~ el~ o~tygen free film of TiW(N) which is needed to get good ~h~ion with the Au-layer in the Type I ~LIU~:tUIC. This RF-l~h~ y 20 ~u~ d film of TiW(N) f.- ~ ;onc as a IEal dirrusion bamer. The con~ol of the nihogcn and o~ygen ~...;..tr~.l;nn~lly ;~ .,de~ fmm the residual gases) cl~nt~ t~ in the film is very ~~ for b~l~nr-in~ the ylu~hes of tliffi~cit~n bamer, ~Ah~on~ sheet ~ n~x and s~css. I~ne higher co.~r,~.l,,,l;l~nc Of nitrogen and o~ygen in the I~W(N) film are, the better r1iffi~cirm b~
be. Ho~ , such films have too weak ~h~cion to ~e Au-layer and have too high resisti~ity. The ~Esistivity of the ~lW(N)-film ic typically more than three ~mes of a rather pure ~W-film. Ac 1- .I;nl~d herein else ~." the effect of ~lh~ion los_ iS ~r~m~ti~lly due to the o~ide fo. ~ I;nn of ~lO2 at ~e su~ce of the TiW(N)-film.
The choice of an ade~lualc sputter technique has also an ~l~t WO 95/22838 I PCTlSE9S/OOlS2 21~0234 effect on the quali~ of the TiW(N) film. The present invention reveals some ~cse~ 1 difr~,nces of the TiW(N) films obtained by an RF- and m~gnptron ~cW~;~t~i (Dc-m~ ~on and RF-m~ un) sputter techniques. The - au~ ~ ;o~iL~ of the RF-a~uLl~d TiN(N) fil n :~nn~ ~i at 420C ~or 280 hours 5 (in N2) is demo~ l~ in a SIMS analysis (Fig. 2). The SIMS analysis also in~ir~tPS a good adhpc~o~ b.,L .~n Au and TiW(N)-layers. The SIMS analyses of the DC-m~gnP~on ~ul~d and RF~ un s~uLL~d diffusion bal~;e.s mnP~lP~ at 420C for 30 min (in N2) are shown in Figs. 3 and 4, ~ ly.
Both DC-.~ ~n -l~un and RF-m~nPtron *)ull ;n~ techniques 10 produce quite lln~t~hle diffusion b~ .; s. These m~gnPtron ~ t~ reactive ~u~l- ;--g techniques cause uneven chemi~l reactions of niLlUge~l against the wafers and TiW-target (c~thod~P) because the m~nP-trons generate a very intense plasma just un~Pmp~th the m~nfl-ùllS. The serious cons~ence is uneven nitnd~tion of the TiW(N)-film at the wafers. This intense and uneven 15 nitri-1~tion of the target causes particles. Ayy~ ly nobody has hitherto clearly pointed out t-h-ese kinds of negdlive effects of the m~pnP,~n associated reactive a~vll~ ;ng of the TiW(N)-film. Ho..~le, a well lalown negative effect of less step coverage due to form~tioll of cavilies in the iarget lm~ h the m~gne~rons cdn be illlylu-ltd by a latest techs~ique of rota~ng or SC~nning 20 ~ .ons. However, the uneven nitrid~tion can not completely be avoided even by the advanced techniques.
The present invention reveals the illl~l~nce of a proper thi~ ,ss of a pure TiW-layer under the Au-layer in the Type II m~ ti~n in case this glue layer is ne~is-~ y. The in~bility of a rela~ively pure 25 TiW-film as a difruh~n bamer b~ Si and Au has been r~,~ol~d (in the Nowicki et al. ar~cle). A new breakdown ~C ~ m was ~ in~ by the role of Au-Ti i.,t~ alloy. The glue layer of Tl~J loses stability at a high t~ ~Ipr ;~ 400C) and starts to form the Au-Ti intP~rn~t~llic alloy after the Ti-metal's u~w~ diffusion into the Au-layer (Figs. 3 and 5). The 30 amount of the Au-~l in~ llic alloy decides the level of tensile stress on the WO 9S122838 - ` PCI~/SE9S/OOlS2 21~0234 diffusion barrier at high L~ pe~dl~s. The tensile stress on the diffilsion bamer f~ tzt~s the Au~iffusion into the bulk-Si and Si-diffusion into the Au-layer.
This is one of the major breakdown ."~h~n;~...c of the Type S barrier. Th~ e it is i~ lant to choose a thin layer of TiW (e.g., 100 500 A). After the ~epo~;l;s)n of the TiW(N)-film, it is illl~lt~t to ~ ,.r,i.
ni~u~en saurce~C both from the sputter ~hZ~.he~ and the TiW-~arget (r~thode) surface by p- ~f~lll~ng, c.g., a careful purge in the ~h~mb~ and a p~c~ulhr for some (2-5) ...in~ 5 with the TiW-target before a~ ;r~ ûf the glue TiW-layer 10 in the Type I~ scheme b~- ~ the ~3rget surf~ce is ~itr~-lod d~g a~ull ;.~g of the TiW(N)-film.
Af~ ~1~pocitinr ~ of the TiW(N) for the Type I or the last glue layer of TiW-film for the Type II, a th~n layer (0.3-1.0 kA) of Au is a,uul~d in situ (without brea~ing the vacuum) at a high sputter rate in order to obtain 15 a pure Au-fdm and to enh~nce ~r~hecion to the underlying layer. Before Spu~ ;nf~ of t-h-e thin Au-layer it is preferable to p~ rl .. a short (0.5-1 min) IUU~ with the Au-targe~ The thin Au-layer wor3~ as a ~loLee~ion film against o~hon and ch~-mi~l attac3c tl~ the TiW- and Tl~(N)-fdms. It also worl:s as a condl~chn~ and in;l;~;~ film for even ele.;l~ ;.l;n~ of the Au-20 c~nductors (7-15 ~A).
As previously ~ic~-cc~ a~out the breakdown ~,Prh~.~icm of the di~sion b~m~s, the Type I ~igs. lb and 2) m~~ on s~-h~ o., which has no tPndenry to form any Au-Ti ~~ alloy, is more stable than tne Type II (Fig. 1c). Fig. 5 shows a S~S analysis of the l~pe ~ diffucinn barrier after 25 ~nn~lin~ at 420C for 90 hour-c ~n N2). Upward di~usion of the Ti-me~al and dvwllwald diffilsion of ~e ~l-, W- and Pt-me~als into the bulk-Si are clear.
The PtSi-layer is dissolved after co...~ g ~vith the Ti- and W-atoms for Si-atoms. The ~ Ani~m of ~e Ti-, W- and Pt-atoms' m-,ie..,~
into the bulk-Si looks like a i~ l;on of c~ ~ However, e~ven the Type 30 ~ barrier ea~ly fulfills n~u~wll~nls of high ~ ..~..l~., and high current WO 9S122838 ' PCI~/SE95/OOlS2 216023~

ope~ions. The HF power t~nCictQr~s couId withc~n-1 thP~rn~l k!~-linE for 40 hour.s at 420C without any loss of yield. The result col-~nd to a Iifetime of more than 200 years with an o~ t~ c of 130C and Ea=0.9 - eV (Ea = ac~vation energy in the ~ e~ on, A*e cp(-EaM'), where 5 A and k (Polt7m~nn) are COf.~ c and T = ~h50l~ lr ,.I...~), ~ t~f~rrin~ to Fig. la-lg, a mPthod embodying the ~
invention i_ ill~trated. Fig. la ill~ t~s a si-subC~t~ 1 which norrn~l1y iS
covered with an inc~ ;ng layer of SiO2 or layes~ of Si3N4/SiO2 or SiO2/Si3N4/SiO2 films 2. The inC~ tor film 2 i_ opened by a pIa ma or wet or 10 a combination of wetldry (Fl~cm~) etch in order to form con~:t holes. Slightly sloped con~act holes _ay be p~f~ hle for a better st~p coverage of the ~iffi~70n ba~ier (see below). Some dopants (n- and ~t~pes) may be imrl~nted or dc~osiled and driven-at high t~ -lr ,~ 1000C) into the Si-~ob~
1 through the contact holes. A thin Pt-film (100 200 A) may be ~uL~cd or 15 c~ ~d aftcr d~tion of a thin SiO2-filrn formed in the con~ct holes.
The de~ tion is ~ r.. , .~ by ;.. ~.. ~.n~ the Si-,u~ 5 1 in a diluted (1-2%) HF-water sollltinn.
A PtSi-layer 3 is formed by ~nnP~lin~ at 600 620C for 30 min in an inert ~I",o5phr..c of N2. An eventual e~oess Pt-metal is a~ ~ ;r~d away in20 a regal s~ hon (ENO3+3HCl).
Fig. lb and Fig. lc illl-~t~ the prEsent invention of the Type I and Type Il m~ nn 51~t.. ~ $, ~ 1y. A thin layer of a Tl~-film aoo 500 A) 4 is ~uU~d after a careful pum~down and a mild sputter etch ~ith an R~-a~vlh ~ mode. This Tl~V-layer 4 wor~s as a glue layer to the 25 in~ film(s) 2.
The ~h~.~n str~ ngth is de~d~nt of thep~ity of the -ll~-layer 4 which is in tu~n dc~ i by the ~sidual gas l~el of H20 and O2 in the aputter cl~l~, a~ulh --~ rate and hyLo~ bonded water to the surf~ce o,f SiO2 2. A ~uu~n level better than 1~' To~r and a high s~ ate are 30 prefe~ble in order to ..-; ;-.. ~r, an o~cygen cOntP~lt in the Tl~-film 4. Ho..~.~, WO 9S122838 ~ P~ ihgS/OOlS2 ~1 6(323~ -without the mild sput~er etch it is ~iffi~..lt to obtain an optimum ~hP~;on beeause a monolayer of H20 is chPmi~lly ~ly~u~l-bond) bonded to the SiO2 surface 2.
- The mild sputter etch also ~nh~n~5 the ~h~l n by ph~ai~dlly S rou~h~nin~ the s~ r~s 2 and 3. It is also very ~llyO~ku~t to take a ~ ;on not to ~m~f~e the PtSi-layer 3 too much. A thi~l~nl~c~ loss of--50 A seems to be acceptable for most cases. In ~ ition to the above .~.e~s, a ~cS~uL~
of the I~ rget (c~th~e) for 2-5 ...;~ is prPfe~hlP in order to get~er the residual gases of H20 and 02 in the ch~...b r and to clean the tar~et. Strong 10 ~hPsinn b~ ~. the in~ ting film(s) 2 and the Tl~ glue layer 4 h S an rSy~ l mP~ninf~ in order to - n; ~ f an etch and to p~EvPnt surface tliffil~inn of Au 9 into the siIicon svl~al~ p 1 bf~ the stress caused by t~.ll~.
mi~m~t~h can form a cavity bc~n the contact hole and the Tl~-~ayer 4.
The real di~on bamer of the TiW(N)-film (1.~2.0 lcA) 5 is L5 s~u~ lly s~?ulL~f;i by an RF-reactive spu~l~ ;ng lechni.luc. An RF-power density more than 1.5 W/cm2 and a N2 con~r.l.,.l;nn of S-10 vol~o in the Ar-gas flow are pre~hle~ A gas p~ of 10 mTorr is a colllpl~l~se for step cove~age, stress and sputter ~ate. The st~ el~Y, iS depCn~f-nt of the conta~t hole's aspect ratio, side angle of conta~t holes, sputter p~ ,~ .. t~ ~ (e~pesially 20 p~SaUl-,) and the al~u~ technique.
There is a clear !~ f ~ y that a TiW(N)-film produces less step -cove~age and g~ rS more par~cles than a ~W-film. However, these ~egative effects of the Fh~~ are very small vith an RF-r~tive ~J(~
te~hnique co~p~cd with a m~tron ~ ~ r~tive ,~u~t~ hnique 25 (DC and RF). Mos~ of all, the stability of the RF-reactivdy ~ulL~ film is much supenor to a fi~ d by a m~ Ton ~c~ ~t d ~active ay~llh ;.,~, A thin film C300 1000 A) of Au 7 (Fig. lb) is ~4~ f~ Atly ~u~tL.C~i after ~JUl~g the Nrgas and a short (0.5-1 min) pl~ull~ of the Au-target. It is i...~u~ku~L to ~ r~ e Au-~l~ lh ;.~g wiLhuul br~ng ~UUlU
30 a*er the I~W(N)-film. Othe~wLce it G~uses a~h~on loss or Au-pee~ng due to WO 9S/22838 PCI/SE9S/OOlS2 2~234 TiO2-form~hnn at the surface of the TiW(N)-film. It is preferable to use a high RF-power density in order to obtain better ~th~cion even though the film thic~lrne5$ is s;mall.
pl f.~ hl~ to the Type ~ m~lli7~hon s~ (Fig. lc), a thin glue layer (100 500 A) of TlW 6 is ~yulLL~d after a carefill pu~lg of the N~-gas in the chamber and a y~yuU~f of the T;W-target for 2-5 Ill .l~ules t)~a~
the target su~e is nih~-~ du~ing ~yull ~ of the I~ -film. S~u~ .ng of the Au-film 7 is identi~l with the Type L
Fig. ld ilt..~ a (ler .1~ t) photoresist pattern 8 over the Au-10 film 7. The phuLul~l pat~crn for Au~o~ Gm be obl~ ~ b~raconvtontinn~l ~ptical lill.oE;l~h~ yl~cess. It is pr.~ lP to create a r~P .t~ t (angle > 90) or vertical pholul~l profi7e 8 in order to get a better step coverage of a fin~l passivation film 10 over the pla~ Au 9. For a very small n of a me~al-pi~ch (1.5~ ~m), a co~h;nA~;~n of an image reversaI
photol~l (AZ5218E or AZ5214E, ~oe~hct) and an i-line (wavelength = 365 nm) or g-line (436 nm) wafer s~er ~ Y can ea~ly create a half n~icron ~ c with a snm~nt1y thick resist (> 1.5 ~m) with the ~ , nt profile.
EYen a standard positive pholo~l, which can ~ e a 20 ver~cal resist p~file, wor~s vesy well for most cases. After a hard baking 5C/30 mm. with a stlff;~Pnt air or N2-flow) of the r~st p ~ an O2-plasma des~lm should be applied in order to clean the Au-surface 7. This plasma des~tm is ~ t be~ a dirty Au-sur~ will not start to be pla~ed.
ig. le ill..~ t,r c Cl~h.Jp~ Au 9 with ph.~o.~ c;~l p~l....c 8.
The Au plating bath is not based on an AuCN c~mplP~ which causes n si_t loss etc., but on a ~AuSO3 comple~. The la~ pla~ng bath with a proper pH-~ralue (7.2-9.5) is quite gentle against the photo~ .-c The bath also co~.l;.;nC a leYeller of which pl~n~- ;,, l;on effect i_ shown in the Au-profile 9.
30 An Au-thi~ n~ of 0.7-1.5 ~Lm is snm~nt for the most IC-a~li~ti~nc wo ssn2s3s ` P(,~ gS/OOlS2 2~23~

- However, the AU-~hit~1~n~c can be easily inaeased to the ~ange of Au-bumps m) for TAB (tape ~l~k~ i bnn-iin~) appli~tion~ with a thic~er resis~
After plating the Au cQn~llctor 9 the photoresist 8 is stripped in ,qr~ton~, and the uafers are rinsed in plo~nol and D~-water (d~-Q~ i) in 5 sequence. After the phok,.c~ L~ g an 02-plasma descum should be . rc~ l again in order to obtain a clean Au~ face before a wet etch of the ~in Au-film 7.
Fig. lf ~ S~ f s the profiles of the plated Au~on-luctor 9 and the wet etched thin film of Au 7 and the plasma etched glue layer of Tl~ 6 10 (only for the Type Il) and the difEi~sion ba~rier l(ayer of TiW(N) S and bottom glue layer of l~W 4. The plasma etch of the laye~s 6, 5 and 4 can be ~,lu,ed by using SF~gas in a plasma mode (RF-anode cou~lill~) or RIE
(rea~tive ion etch, RF~thode coupling) mode d~y etcher. It is preff~h~
apply an RIE mode dry etcher in order to ~ ;n; I;~e the etch undercuts. It is not advisable to apply a wet etch for the layers 6, 5 and 4 by a H2O2-s~uti- n The wet e~ch causes the undercuts and is too slow against the ~lW(N)-layer 5.
After the etch of the layers 7, (6), 5 and 4, an ~nn~linE is olluecl at 420C for 30 min. m a N2-ahnosphfre in or~ to get better ohmic ct:tnt~c't and to -nh~n~ some ;~ of the metals for better 20 ~rlh.~ n - Fig. lg illl~ f s the final profiles of the Au~2~ u~vto( 9 and a PESiNX ~plasma c lhan~ ~licon ni~ide, ~8 lcA) ~ lion film 10. The ~ ion film 10 p~Ol~:b the Au~n~t~ r 9 against both ~ ni~l and rhlomi~l (c~ ù5iOn) ~ r5, ThiS illtL~u~n~.eO~ n scherne is p~rh~lly O~O~l free wi~uul a p~v~ion film be~ nn~lin~ in the N2-~tmos~hpre and an air e~ u~ f~ thin p~ule~lion filrns of TiN and 1~lO2 a~
all ~secl ~ ,s of the I~W and I~W(N)-film~s 6, 5 and 4. The choice of the PESiNx-film also d~ n~s on its better th~nal CQ~ u~l;vil,y than SiO~. For high power ~nc~CtQ~, ~ ~liv-e heat ~ ading i~s a very i~l ~u~l~ for 30 ~i~

WO 95/22838 , ~ P~ gslools2 216~23'1~

The b~n~r~ opening in the PEsiNx-film 10 is ~ rO ..~Pd by a co~ r,l;o~l plasmaetchbyanetchgasof Nl 3 ora~ eofCF~ ~ 2(--iO
vol~0) after a no~...;L;.~l litho~l~phy step. After the bon~lr~ls o~g the - photoresist is ~iy~d in ~r~tol-ç or some other sol~L~.
Fig. 2 shows a S~S analysis of the R~-reactively S~UL~.
~imlcion ba~rier af~r ~nn~ling at 420C for 280 hours in N2. The multilayers of a Type I m~ tinn SC~ r are:
P~Si(200 A)/T~ A)./I~lW~(1.8 kA)/Au(l kA).
The d u~on barrier is s~ll very stable af~Ler a severe thermal s~s test, and adhesion ~cl~n Au- and TiW(N~-~yers is good b~.l~ Au-dowl~w~ls dif~LLcion into the TiW(N)-layer is clearly seen. The ohmic con~t be~ the TiW- and PtSi-layers is also good. The PtSi-layer is also inta~t be~use the upper-signal poC~iti~n~ of the Si- and Pt-ions are clearly "l~os~. The N2~nt~nt in Ar-gas is 10 vol% for the TiW(N)-film ~ ul h . 1"~
Flg. 3 shows a SIMS analysis of the DC m~n~tron ~
~uu~ed diffusion barrier ~nnP~ at 420C for 30 min. in N2. The multilayers of the m~71i7~tinn srh~-mç are:
PtSiaOO A)/TiW(N)(:3 kA)/TiW(l kA)/Au(10 kA).
The .I;~ . barrier is l~Li~ very llniC~blP- Upward ~liffilcion of the ~l- and Si-atoms in~co the Au-~yes as we31 as duw~
diffusion of the rl-, W-, Au- and Pt-atoms into the bulk-Si are ~ery clear.
Di~ ;o~l of the PtSi-layer i aIso clear. The N2~on~nt in Ar gas is 30 vol%. The in~hility of the ~li~us,on bamer is an ~U~lU~ ~
25 sputter t~ue for the I~W~)-film and too thick glue layer (1 kA) which causes the f~ Al;nn of an Au-Ti ;-~ i(' alloy.
Fg. 4 shows a SD!~S an~lysis of the RF-~IIA~
ayu~d ~li~u~on ba~ier ~nn~l~ at 420~C for 30 min. in N2. The mul~layers of the md~lli7~t~on s~hPme are:
PtSi(200 A)~w~so A)rr~ .s ~A)~w(l20 A)/AU(12 lcA).

WO 95/22838 . -" PCI~/SE9S/OOlS2 The dif~ion ba~ier is relatively unstable. Upwar~ 1l. fr. I~,on of the Si-atoms into the Au-layer as well as downward di~usion of the 1~1-, W-, AU- and Pt-atoms into the bulk-Si is ve~y clear. Huwe~ e uywaL~
dif~sion of the Ti-atoms observed in Fig. 3 is not visible in this case.
S D~ tion of the PtSi-3ayer is very clear. The N2 contPnt in Ar gas is 10 vol~0. The stability of the f1.rru~ ba~ier is slightly be~er th~n the case of Fig. 3. The inS~hility iS caused by an ih~y~O~ reactive sputt~ t~hai~ue for the Tl~(N)-film even though a l`~rye II configma~ion is used.
Fig. 5 shows a SlMS analysis of the RF-~ vely ~yuLh~d diffasion bamer ~nnP~led at 420C for 90 hours in N2. The multilayers of a Type II mP~lli7~tion scheme are:
PtSi(200 A)/I~W(æSO A)/I~W(N)(1 kA)/Tl~(250 A)/AU(1 kA).
The lifrus,on bamer b~O...P~ in~hte after a seYere ~
streSc test. Uywdlds~t;rr~l~ o~ of the Ti- and W-atom_ as well as dowuw~s--15 diffi~sion of t_e Ti-, W- and Pt-atoms are clear. Dc~ nn of the PtSi-layer has started, butis not severe. I~U..~I;nn of an Au-Ti ;~ " - hlliC alloy is visl~le in the uu~OS~u~ (SEM). Howe~ er, the di~sion ba~ier is q~te stable for a thPrm~l stres up to 40 hours at 420C without any yieId los_ of ~F
power ~ IO~ eYen though ll"~L"~ of the TiW(N)-film L OI~ly 1 kA.
As ~e.~ofi~ ~ above, an adequa~e choic~ of a film config~l~tion, thi~nP~c and sputter technique verified long-~n rPIi~hl~ity~
good ~h~cion to SiO2, Si3N~ and gold layer and low contæt re~n~ to both p~(boron) and n+(~hG~ho~ uuS, a~s~c or ~ntimnny) doped silicon a~Jl~l .,.t.~, For the sequ~Lial d~ n-~ of the I~W-and I~ films an appli~tinn of an RF-~ technique instead of mostly used DC-~ h~Jn played an cs5~ 1 role with re~e~t to the long-term r~ hility, good ~h~l~n, good st~p covesage of contact holes and les parhr~ tinn The sta~ility of the RF-~ ~ ~uU~ IIW(N~ film as a d;rr"~;n.~ ba~ i much ~ o~ to the ~ ol~ ul by ~ un ~ eaL~e ~ ; -P. ~DC-m~gn~o~Ton and RF~ ù.~ ne wo 9S/22838 pcrlsE9slools2 superionty is d~monc~t~ by S~S (s~on~ y ions mass a~l~u,lleter), SEM
(sc~nning elect~on mi~scu~e3 analyses and the t; f , - ~c of ~: power tr~ncictQrS
The m~ tti7atinn sch-~m~c of the Type I and Il are sup~or to some known s~h~ s of Si/TiW(N3/TiW/Au, Sill~ W(N3/TiW/Au or SilTl~lTiW(N3/Au with ~pect to long~ m reliability and contact r cic~n~
However, the Type I is ~ fif,.AhlP when a good ~uu~ (<1~ T~) is obt~in~hl~ in tne sputter ~ . The Type I is sUpPnnr to the Type ~ wi~h r~ to the long-~n high ~lll~~ e stabiIity, pl~a cimpli~ity and 10 process l~tihlde by less esch tmderc~t.
HoweYer, the Type Ir avoids Au-peeling when the contf .1~ of recidual gases of H20 and (~2 in the sputter c~ ..b ~ is rather high due to a poor pu~u~g system. The thin ~W layer b~l~n the TiW(N) and Au layers in the Type II ~nh~r~ces the ~th~m by Au-i,~te.diffu~on. HoweYer, the pure I~W-15 film tmd~F~Il. the thic~c Au-layer catlseâ the forT~ nn of an Au-Ti n~ tlic alloy which ~ lf 5 the breakdowns of the Type II.
The Type I and Type II with~nd at least a th~ l s~ess of 420~C1280 hours and 420C140 hours, l~;li~dy, without any loss of yield of HF power ~ ~ a EYen the shortest lifeliule of the Type Il co~ ds to more than 200 years ~tion with a de~rice le~dl~, at 130C.
- The fP~cibility and cost Pffi~eney of the mPt~lli7~ti~n ploce~,~s are ~r~npnt by ~plying we~ established techniques of ~ull .r.~,, gold "plaLillg and plasma ~tling.
It will be ~c;~ by those slcilled in the art that the ~pe~fi~
~ des~ ed her~n are mer~ly for t~ s of r ~ n~ and tha~
many ~ may be used without depar~ng ~om the spirit and scope of the lion. Ac~ dillgly, it is intPnliP~ that ~r~tinr~ as will oceur to t`hose s~lled in the art based ~ the rti~loSllres herein be -~ tud~i in the sc~e of the claims which follow.

Claims (48)

Claims:

1. A method of fabricating an interconnection on a semiconductor device, said method comprising the steps of:
forming an insulating layer on a silicon substrate;
forming a contact hole in said insulating layer to expose a portion of said substrate;
forming a PtSi layer in said contact hole on said substrate;
forming a TiW layer in said contact hole on said PtSi layer, forming a TiW(N) layer in said contact hole on said TiW layer;
and forming a Au layer in said contact hole on said TiW(N) layer.

2. A method in accordance with claim 1, wherein said PtSi layer reduces contact resistance and stabilizes an interface between said substrate and said TiW layer, which acts as a glue layer.

3. A method in accordance with claim 1, further comprising the steps of:
a mild sputter etch, after formation of said PtSi layer and before formation of said TiW layer, to enhance adhesion between said insulating layer and said TiW layer, said mild sputter etch further removing SiO2 on said PtSi layer.

4. A method in accordance with claim 1, further comprising the step of:
pre-sputtering, after formation of the PtSi layer and before formation of said TiW layer, to reduce residual gases of H2O and O2 in the sputter chamber and from the surface of a TiW-target.

5. A method in accordance with claim 1, wherein said TiW layer is a pure TiW-film.

6. A method in accordance with claim 1, further comprising the step of:
eliminating nitrogen from a TiW-target surface and sputter chamber by a combination of the pre-sputter and purging, after formation of the TiW(N) layer and before sputtering of said thin film of Au layer.

7. A method in accordance with claim 1, wherein said step of forming said Au layer further comprises the steps of:
forming a thin film of Au by sputtering in said contact hole on said TiW(N) layer, and forming a plated layer of Au in said contact hole on said thin film of Au.

8. A method in accordance with claim 7, further comprising the step of:
etching said thin film of Au in a solution.

9. A method in accordance with claim 7, further comprising the step of:
etching said TiW(N) layer, which acts as a diffusion barrier, and said TiW layer, which acts as a glue layer, by a plasma etch with said plated Au layer acting as an etch mask.

10. A method in accordance with claim 9, wherein said etching step is performed in situ.

11. A method in accordance with claim 1, further comprising the step annealing said Au layer to obtain an optimal contact and adhesion strength by interdiffusion of the gold into said TiW(N) layer.

12. A method in accordance with claim 1, wherein said interconnection is a bondpad.

13. A method in accordance with claim 1, further comprising the step of:
forming a PESiNx passivation film on said Au layer.

14. A method in accordance with claim 1, wherein thickness ranges are:
PtSi(0.1-0.4k.ANG.)/TiW(0.2-0.5k.ANG.)/TiW(N)(1.0-2.0k.ANG.)/Au(0.3-1.0 K.ANG.)/Au(7-15k.ANG.,plated).

15. A method in accordance with claim 1, wherein RF-power density for sputtering of said TiW(N) layer and said TiW layer are of a same order to avoid internal stress.

16. A method in accordance with claim 1, wherein RF-power density for sputtering of said TiW(N) layer and said TiW layer are larger than 1.5 W/cm2.

17. A method of fabricating an interconnection on a semiconductor device, said method comprising the steps of:
forming an insulating layer on a silicon substrate;
forming a contact hole in said insulating layer to expose a portion of said substrate;
forming a ptSi layer in said contact hole on said substrate;

forming a first TiW layer in said contact hole on said PtSi layer;
forming a TiW(N) layer in said contact hole on said first TiW
layer, forming a second TiW layer in said contact hole on said TiW(N) layer; and forming a Au layer in said contact hole on said second TiW
layer.

18. A method in accordance with claim 17, wherein said PtSi layer reduces contact resistance and stabilizes an interface between said substrate and said first TiW layer, which acts as a glue layer.

19. A method in accordance with claim 17, further comprising the step of:
a mild sputter etch, after formation of said PtSi layer and before formation of said TiW layer, to enhance adhesion between said insulating layer and said TiW layer, said mild sputter etch further eliminating SiO2 on said PtSilayer.

20. A method in accordance with claim 17, further comprising the step of:
pre-sputtering, after formation of the PtSi layer and before formation of said first TiW layer, to reduce residual gases of H2O and O2 in thesputter chamber and from the surface of a TiW-target.

21. A method in accordance with claim 17, wherein said first TiW
layer is a pure TiW-film

22. A method in accordance with claim 17, further comprising the step of:

eliminating nitrogen from a TiW-target surface and sputter chamber by a combination of the pre-sputter and purging, after formation of the TiW(N) layer and before formation of said second TiW layer.

23. A method in accordance with claim III, wherein said step of forming said Au layer further comprises the steps of:
forming a thin film of Au by sputtering in said contact hole on said second TiW layer; and forming a plated layer of Au in said contact hole on said thin film of Au.

24. A method in accordance with claim 23, further comprising the step of:
etching said thin film of Au in a solution.

25. A method in accordance with claim 23, further comprising the step of etching said TiW(N) layer, which acts as a diffusion barrier, and said first and second TiW layers, which act as glue layers, by a plasma etch with said plated Au layer acting as an etch mask.

26. A method in accordance with claim 25, wherein said etching step is performed in situ.

27. A method in accordance with claim 17, further comprising the step of:
annealing said Au layer to obtain an optimal contact and adhesion strength by interdiffusion of the gold into said second TiW layer.

28. A method in accordance with claim 17, wherein said interconnection is a bondpad.

29. A method in accordance with claim 17, further comprising the step of:
forming a PESiNx passivation film on said Au layer.

30. A method in accordance with claim 17, wherein thickness ranges are:
PtSi(0.1-0.4 k.ANG.)/TiW(0.2-0.5 k.ANG., first)/TiW(N)(1.0-2.0 k.ANG.)/TiW(0.1-0.5k.ANG., second)/Au(0.3-1.0 k.ANG.)/Au(7-15 k.ANG., plated).

31. A method in accordance with claim 17, wherein RF-power density for sputtering of said TiW(N) layer and said first and second TiW
layers are of a same order to avoid internal stress.

32. A method in accordance with claim 17, wherein RF-power density for sputtering of said TiW(N) layer and said TiW layer are larger than 1.5 W/cm2.

33. An interconnection on a semiconductor device, said interconnection comprising:
an insulating layer on a silicon substrate;
a contact hole in said insulating layer to expose a portion of said substrate;
a PtSi layer in said contact hole on said substrate;
a TiW layer in said contact hole on said PtSi layer;
a TiW(N) layer in said contact hole on said TiW layer; and an Au layer in said contact hole on said TiW(N) layer.

34. An interconnection in accordance with claim 33, wherein said PtSi layer reduces contact resistance and stabilizes an interface between said-substrate and said TiW layer, which acts as a glue layer.

35. An interconnection in accordance with claim 33, wherein said TiW layer is a pure TiW-film.

36. An interconnection in accordance with claim 33, wherein said Au layer further comprises:
a thin film of Au in said contact hole on said TiW(N) layer; and a plated layer of Au on said thin film of Au.

37. An interconnection in accordance with claim 33, wherein said interconnection is a bondpad.

38. An interconnection in accordance with claim 33, further comprising:
a PESiNX passivation film on said Au layer.

39. An interconnection in accordance with claim 33, wherein thickness ranges are:
PtSi(0.1-0.4k.ANG.)/TiW(0.2-0.5k.ANG.)/TiW(N)(1.0-2.k.ANG.)/Au(0.3-1.0 k.ANG.).

40. An interconnection on a semiconductor device, said interconnection comprising:
an insulating layer on a silicon substrate;
a contact hole in said insulating layer exposing a portion of said a PtSi layer in said contact hole on said substrate;
a first TiW layer in said contact hole on said PtSi layer, a TiW(N) layer in said contact hole on said first TiW layer;

a second TiW layer in said contact hole on said TiW(N) layer;
and an Au layer in said contact hole on said second TiW layer.

41. An interconnection in accordance with claim 40, wherein said PtSi layer reduces contact resistance and stabilizes an interface between said -substrate and said first TiW layer, which acts as a glue layer.

42. An interconnection in accordance with claim 40, wherein said first TiW layer is a pure TiW-film.

43. An interconnection in accordance with claim 40, wherein said Au layer further comprises:
a thin film of Au in said contact hole on said second TiW layer, and a plated layer of Au on said thin film of Au.

44. An interconnection in accordance with claim 40, wherein said interconnection is bondpad opening.

45. An interconnection in accordance with claim 40, further comprising:
a PESiNx passivation film on said Au layer.

46. An interconnection in accordance with claim 40, wherein thickness ranges are:
PtSi(0.1-0.4 k.ANG.)/TiW(0.2-0.5 k.ANG., first)/TiW(N)(1.0-2.0 k.ANG.)/TiW(0.1-0.5 k.ANG., second)/Au(0.3-1.0 k.ANG.).

47. An interconnection in accordance with claim 33, wherein thickness ranges are:
PtSi(0.1-0.4.k.ANG.)/TiW(0.2-0.5k.ANG.)/TiW(N)(1.0-2.0k.ANG.)/Au(7-15 k.ANG., plated).

48. An interconnection in accordance with claim 40, wherein thickness ranges are:
PtSi(0.1-0.4 k.ANG.)TiW(0.2-0.5 k.ANG., first)/TiW(N)(1.0-2.0 k.ANG.)/TiW(0.1-0.5 k.ANG., second)/Au(7-15 k.ANG., plated).