PRq~155ING

&.Lithographicmodelingspeedsthin-film-headdevelopment

Chris A. Mack, FINLE Technologies; Gary E. Flores,Tencor Instruments; Warren W. Flack, UltratechSteppc:r; and Elizabeth Tai, Read-Rite Corporation

II hin-tllm heads are producedusing planer rechniques rharare quire similar ro rhose usedin manufacturing inregrared

circuirs [11. From a lirhographic standpoim, the thickphororesisrs (up ro 40 mm) and large aspecr ratios(up [() 6: I) of rhin-film heads pose a considerablecha]]enge, which will mosr likely increase in rhe years~lhead as drive makers arrempr ro boost headperformance by narrowing rrack widths and polebraps.The [arrer will allow more birs ro be srored perunir lengrh of track. while rhe former will increaserrack densiry [21.

There's a carch, though. A5 individual transitionregions shrink in size, more copper coils are required soheads can read rhe magneric dara embedded in rhemedia. More coils increase the srrucrural heighr of thehead, which will require thicker layers of phororesisrsto process the thin-film head [1,3].

Indeed, Immy of the crirical issues in thin-film-head

r:lbricarion direcrly involve the lirhographic process.'I'he dimensions of the copper coils and poles musr bei..:areh.lllycomrolled since rhey can affecr head perform-~II1Ce.However, the large ropography of a typical head!\:quires thick phororesist rllms and long exposurelill1e" resulting in a loss of line-widrh comrol [4].~tecp slopes in a thick phororesisr film also reduce crir-ical dimension comrol. And the high reflecriviryof theplated meral film can affecr critical dimension comrolbv rrjaacrina a srandina wave.. = '" '"

In the tace of these rribularions, lithographic simula-tion may prove ro be the wisesr course. There are sev-eral software programs thar simulare phororesist chem-isrrv under varying oprical and exposure condirions ropredicr [he line width and shape of rhe resisr parrerns.'['he semiconducror indlmry, for example, uses litho-graphic modeling ro rrim developmem time and rohelp berrer understand complex problems. For exam-

ple. broadband i-line lithography [5] and deep UVexcim<;:rlithography [6] have been simulared using a

software program called PROLITH/2. These semi-conducror processes involve subsranrially smallergeomerries and rhinner phororesisr films rhanrhin-film-head processes. However, rhe phororesisr

aspect ratio (i.e., height ro line widrh) for rhin-filmheads is acrually larger than the aspecr rario used inchip making, suggesring thar rhe lirhographicchallenges are jUStas great for thin-film heads. Clearly,the thin-film-head indusrry can benefit from processmodeling as much as the semiconducror indusrry.

For more than a dozen years, lithographic model-ing has been used ro simulare the processes of imageformation, exposure, and developmem. When prop-erly used, simularion can accurarely predict rhe

resulting phororesisr profiles for a large range of pho-

aMODHIH~H{UHiHI

PR!Dlm IH! B!HAVIOR

Of THICIPHOiORfSISI\

um feR MHH

Jnlli.fiii'li~,m

INPUTPARAMETERSFORLITHOGRAPHICSIMULATION

~~

.Imaging tool:Wavelength=420.0 run

Bandwidth= 50.0 run

NumericoIa~~,~.21:Reductionratio= 1.0'

Imageflare= 0.00

Aberrations:None

Partialcoherence=='0.85

LineWidth;'4.00~m;'

Pitch= 8.001UR

Mask bias= O.O}!lll

. FocoIpasilion =.' ,3.00 IUR

CEliOI' Top ARC:

Notused

.Intermediatelayers:None

~e:'SiIiam

Exposureenergy= .500.0MIlan 2

Resist system:rPositive

1hiOOiess=~10:000~m ;

AbSOiptionpOrlimeter;A";03511IUR

;;~~'~t~;B"; O.0201/~mExposUreratecOO$t.C==0.012 an2jMJ

RefrOctive: Uldex = 1.670

". Jeve10pmentmodel:OriginalMack

Maxdeveloprate=80.0 nm/s

MIDdeveloprate= 0.3nm/s

Threshold,M;"OJ

Seledivilrparameter,n=. 2.10

Relativesurfacerate=. 0.010

Inhibitiondepfu= 1.00IUR

PESdiffusionlength= 70,0nm

Developmenttime = 60.0See

May/June1996DATASTORAGEm

Figure 1, /VItrioureel eIevelopmem rates versw reuniVi' PAC concmtrationsJOr HoechJ't Cehnese AZ 4000 (!.eft)

(liid Shipt<.I' S7R (nght) phowresists lit fiLm thicknesses of (a) 211m, (b) 5 11m, and (,j 10 Jim, Development

/ille partzmeters detemzined through I'Igressiol/ tlIltILysis tlre Listed in the Uppl~' right comer of each i(mph,

rolirhographic processes, including rhoseused tor rhin-film-head manufacruring,

Lirhography modeling began in rhe1970s when researchers ar IBM firsr

anempred ro characrerize rhe behavior ofphororesisrs. Since rhar rime, much efforrhas gone inro undersranding rhe physicsand chemisrry of phorolirhography.

'laday, lirhographic simularion uses weB-resred phvsical models. The simularion canbe' perr()rmed on a deskrop PC or in aworksrarion using commercial sofrware.Simularion programs require rhe inpur ofIirhograph ic paramerers such as maskTl';![Un:siL.e. numerical aperrure of rhesrepper lens, rhe rype of resisr used,e'xposure dose, and developmenr rime.\X/irh proper inpur paramerers, simularionprograms can accurarely predicr rhe resisrline widrh, side-wall angle, loss. depth ofr()(us. :U1dorher perrinenr metrics associat-ed wi(h (he limographic process.

Through lirhographic modeling phorot:ng,inens can perform many "experi-mc:ms" quickly and ar very low cosr, usingonly a few resr wafers. Indeed. as rhin-film head lirhography becomes morecomplc:x. sending processing COS(s

-~jDATA STORAGE May/June 1996

(hrough rhe roof~ modeling will galI1 anirresistible appeal.

One of rhe keys w simulation is rheavailabiliry of accurare inpur paramerers.A rypical ser of inpur paramerers for alimographic simulation program is lisred inthe rable. Many of rhe paramerers ei(herare known (e.g., exposure wavelengrh) orare easily obrainable (e.g., stepper parrialcoherence). For best resulrs, resisr/

developer properries must be known.

DISSOLUTION BEHAVIOR

Phororesisr dissolurion has been extensively

srudied for rhin-film phororesisr applica-rions [7]. However, how weB exisring min-rIlm models apply ro the behavior of rhickphororesisr films is unclear. Thus. rhe firsrsrep is ro measure me dissolurion behaviorof me resisrsat differenr micknesses. If me

developmenr model adequarely describesrhe dissolurion behavior of the resist, the

fulllirhographic simularor can be used andcompared wirh acruallirhographic resulrs.

Two commercial phororesis( producrs,Hoechsr Celanese'sAZ 4000 and Shipley'sSTR 1000, were examined for meir devel-

opmenr rare behavior and phororesist pro-

flies through focus. Barh marerials arespecifically designed for thick phororesis(

applicarions. 4620, a high-viscosiry formu-larion of rhe AZ series, was used ro srudy5- and IO-~m-mick regimes, while P411awas used in 2-~m-rhick processes. Ahigh-viscosiry formularion of rhe STRseries, 1075 (44% solids), was used ro

srudy a 1O-~m-rhick regime, while ocherShipley formularions (1045 and XP90 190)were used ro smdy 5-~m and 2-~m-rhickregimes, respecrively.

An Ulrrarech Srepper model 1500 wasused in all experimenrs. Projecrion op(ics,based on a Wynne-Dyson-Hershel IXlens, provided broadband illuminarion ofthe g and h mercury lines (including (hecominuum from 390 ro 450 nm). The

numerical aperrure was ser ar 0.24 tor alltesrs; parcial coherence was fixed ar 0.85.The Hoechst Celanese phororesisr washeared (i.e., "sofr baked") in a convecrionoven ar 105°C for 45 minures, while the

Shipley phororesisr was heared on a horplate ro 100°C for 90 seconds.

The developmenr rare of me tWo pho-roresisrs was measured ar three differenr

rhicknesses (2, 5, and 10 /-lm), using s(an-dard open frame. dose-ro-clear methods(conrrasr curves). A range of exposures(50 ro 650 MJ/cm2, in incremenrs of15 M]) were used in all resrs. Oevelop-menr times were ser berween 1 and

4 minures for 2-~m-rhick phororesisrfilms; berween 2 and 8 minures for

5-~m-rhick films; and berween 3 and 12minures for 1O-~m-rhick films.

Phororesist developmenr rares werethen calculared for e:lch exposure energy.Based on rhese rares, an effecrive averagedevelopmenr rare was esrablished for agiven film thickness and exposure dose.A corresponding average phoroacrivecompound (PAC) concenrration wasderermined for rhe various exposure ener-gies and film rhicknesses. Phororesisrdevelopmenr rate versus relarive PAC isshown in Figure I ror borh AZ 4000 andShipley STR 1000 phororesisrs. A devel-opmenr rare model was rhen fir ro rhisexperimenral data [8], viz:

RLzte = Rma,.,(l- e -EC)" + Rm;n

where Rmaxis rhe maximum developmenrrare of fully exposed phororesisr; Rm;n' meunexposed developmenr rate; n, me devel-oper selectiviry; E, me exposure dose; andC. rhe effecrive phororesisr rare consranr.

AZ4000photoresists ShipleySTRphotoresists50 50

II) 4110

.f .1

XP9019040 2.mlilm 2-mfilm30 Rmax= 30 nml s

30 \R..ax=63nm/s

R..in= 0.5 nm/s R",;"=0.5nm/s20 ""1 20 n=4.410 ]0

0 00.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8

- 50 - 50" b) 4620 " b) 1045! 40 5.mfilm ! 40 5-mfilm'" '"

\"§ 30 Rmax= 60 nm/s "§ 30 Rmax= 55nm/s-= Rmin= 2nm/s -= R..in= 3 nm/sE 20 n=2.5 E 20 n=5.0a.. a..c

1010 >'" '"Q 0

Q 00.2 0.4 1 0.2 0.4 0.6 0.8

50 50c) 4620 c) 1075

40 10-mfilm 40 ]O-mfilm

30 Rmax= 30 nml s 30 Rmax= 75 nm/s

20Rmin= 1 nm/s 20 L

R..in= 2 nm/sn= 1.20 n=4.5

10 10

0 0-'"

0.2 0.4 0.6 0.8 I 0.2 0.4 0.6 0.8Relative PACconcentration Relative PACconcentration

a) AZ4620simuiation

Figure 2. Theefftaoffocw on the resistprofile shape under a 10-11m-thick coating o(Hoechst Celanese

..17 )(,20 photoresist. Simu/med profiles tire to the left o(actua/ resistprofiles.

The rerm e'FCis approximarely equivalenr[0 the relative PAC concenrration; hence,

this equation can be ht ro the experimemalresul tS ro determine developmenr rateparameters. The maximum developmemrate, Rm<L"for me AZ 4000 £1lT1ilywas onthe order of 30 ro 60 nm/s; for the STR

]000 family, it was slighdy higher, aroundGOro 75 nm/s.

One notable difference between the

rwo phororesist families is their dissolu-[ion selecriviry,or n value, which is relatedw [he phororesist conrrasr. The STR

] 000 phowresisr exhibits n values of

approximately 4.5 ro 5.0, while theA7 4000 phororesist has n values of

].2 ro 2.5. The higher the n value, thegreater the exposure and focus latirude.

A scanning electron microscope wasused ro evaluate a parrern of lines andspaces in me phororesisrs as a function offocal depth. The results indicate that theSTR 1075 photOresist provides bettercomrol over a larger focal range than meAZ 4620 photOresist. These results werealso predicted in the simulation srudies,which pinpoimed the relative importanceof the developer selectivity, n, in litho-graphic performance. All in all, the

simulation results closely matched the

experimemal results over a wide range offocal depths (see Figure 2). ..;..

.;;-t-:',k' ~ DATA STORAGE~ May/June ]996

REFERENCES

1. Oshiki, M. and Hamasaki, S., "Thin-film

head rechnology," hyitsu Sci. Tech.}., vol.26 (4), Pl'. 353-363 (Fehruarv ]')91).

2. Kryder, M. H., "Dara-sroragc rcchnologies

ror advanced compuring," ScientificAmerican, vol. 257 (4), pp. ] ]7-125(Ocrober ] 987),

3, Ohdoi, Y. er. aI., "A 48-rurn rhin-film head

for ]50 MB/in2 recording," IEEE 7im1Jtlc-

dons on Magnetics, vol. 26 (5), pp. ]67]-1673,

4, Gau, ]. S" "Phorolirhography ror jnregraredthin-film read/wrire heads," Optical/Laser

Microlithography If, Proc., SPIE vol. 1088,

pp, 504-514 (1989).5. Flores, G., Flack, W., D"ycr, L., "Lirho-

graphic perrormance of a new generarion

i-line oprical sysrem: A compararive analysis,"

Optical/Laser Lithography VI, Proc" SPIE vol.]927 (1993).

6, Mack, C, er. at, "Modeling and characreriza-rion or a 0,5-llm deep ulrravioler process,"

}. Vae. Sri. Techno!, B 9 (6), pp, 3143-3]49 ]991),

7, Mack, C, "New kineric model ror resisr

dissolurion," I Elecrrochem. Soc., ]39 (4),pp, L35-L37 (1992),

8, Trefonas, P., Mack, c., "Exposure doseoprimizarion for a posirive resisr comainingpoly-runcrional phoroacrive com pound,"Advances in Resist TechnoLogyand ProcesJ'ingVII/, Proc,, SP]E voL 1466, pp. 270-282(1991).

Chris A. Mack is president and chief tech-nical officerat FlNLE Technologzesand an

adjunct fizculty at the University of TexasatAustin. He earned a BS in physics, chem-istry, electrical engineering, and chemical

engineeringfrom Rose-Hulman Institute ofTechnology and an MS in electrical engi-neeringfrom the University of Maryland.

Gary E. Flores is manager of'inspection

equipment at Tencor instruments, Hereceiveda BS in chemical engineeringfrom

the Universityof Califtrnia at Berkeleyandan MS from the University of Califtrnia atSanta Barbara,

warren w: Flack is the applications man-

agerjOr Ultratech Steppa He earned a BSand an MS in chemical engineeringfrom

the Georgia institute of Technologyand aPhD, alsoin chemical engineering,from the

University of Califtrnia at Berkeley.

Elizabeth Tai is a staff R&D engineer

at Read-Rite Corporation. She received

a BS in molecul£trbi%g)! fom New Vork

University.