Extended Abstracts of the l99l International Conference on Solid State Devices and Materials, Yokohama, 1991, pp. 17-19

The Influence of Synchrotron X-Ray Damage on Hot-Carrier-InducedDegradation in Subquarter-Micron NMOSFETs

T. Tsuchi.ya, M. Harada, K. Deguchi , and T - Matsuda

NTT LSI Laboratories3-l- Morinosato lYakamlya, Atsugi, Kanagawa 243-OL, JAPAN

Hot-carri-er reliablllty due to residual damag:e in the gateoxide created by synchrotron X-ray lrradiatlon is investlgatedfor subquarter-mLcrometer NMOSFETS. Although lrradiation-1n-duced interface-traps are completely eliminated after 400 oC

post-metallzation- annealing, neutral electron traps partlallyremain. The effect of the residual traps on hot-carrler degra-dation can be negllgible when gate oxides thinner than about 5nm are used. It ls found that there ls no effect of lrradiationdamage on lnterface-trap g:eneratlon during hot-carrler-lnjec-tlon.

1. INTRODUSTIOITSynchrotron radiation (SR) X-ray lithog-

raphy ls pronlslng for fabricating futuredeep-subnicroneter ULSIs. However, there isa great concern over hot-carrier (HC) rell-ability due to residual damage in the gateoxide created by X-ray irradiatlon. Severalprojects have investigated the influence ofthe danage on H0-induced MOSFET degradationtll-t31. These proJects have dealt wlthgreater than half-nlcroneter devices withgate oxides thicke; than-10 rn, and withrelatively snall (LO" mJ/cmz) naxinun irradi-atlon doses. This paper focuses onthe lrradiation and HC-effects forsubquarter-nicroneter NMOSFETs withthlnner. gate oxides und'er., a wlderirradiation range (10-3,000 mJ/cn"l .

2. EXPERITIENTAL PROCEDTIRES

N+ poly-Si-gate NMOSFETs with 3.5-L2.0nn thlck gate oxldes and 80 nn deepsource/d.rain Junctions [4] were used in thisstudy. Gate layers rvere patterned by EBlithography. After gate electrode fabrica-tlon, sanples were annealed at 900 "C for 30ninutes ln Nc atnosphere to elininate"EBdanage. ExceFt f or the gate layer, opt.icalIlthography was used. Irradiation experl-nents were performed in an NTT SOR Facilityt5l. In order to lnvestigate irradlation-lnduced danage and the effect of pos!-metali-zation-anneallng, which is the final anneal-lng through fabrlcation process, directexposure was nade without any reslst layereither before or after the final anneallng lnNt/Hc f ornlng gas at 400 oC f or 30 ninutes.x:rai wavelengths ranged f ron o ,7 -L.2 nn!,.

The naxinun irradiatlon dose (3,000 mJ/emzl

A-1-4

was set to evaluate irradlatlon effects dueto flve- or six-lithography-leveI exposureuslng typical-sensitivity X-ray reslst,assuning a three-level metal process wherehigh-tenperature annealing is not perfornedexcept for the flnal N2/H2 annealing.

3. RESTILTS AIID DISCUSSION3.1. IRMDIATION DAIIAGE

X-ray exposure generates inter-face-traps, posltlve charges andneutral traps in the gate oxlde tOl.Generated interface-traps were evalu-ated by charge punp l ng current I,^r,[7], and they are conpletely eliminafled after the flnal annealing, evenunder a maxlnun dose of 3,000 mJ/cm2,as shown ln Fig. 1. Generated posi-tive charges also disappeared afterthe anneallng. This was conflrned bya conparlson with M0SFET characteris-

lo-7 Tor - 5.0 nm

L = lOpmW = 20pm

AFTEREXPOSURE

lo lo2 lo3

EXPOSURE DOSE (mJ/cmzl

Flg. 1Charge pumplng current vs. exposure dose.

s ro-sCL<,

lo-e

t7

D0SE (mJ/cm2l

3000r00o ;f133'tr

0 tor = 5.0 nm

AFTER EXPOSURE8 ANNEAL

z

= t0-'

reo

ro-'

= ro-2F

ro-l loo lolQrn.r (c/cmzl

Fts. zAV; vs. charges of electrons lnJectedf ro^m substrate QlNl.

tlcs of non-lrradiated devlces.Residual neutral traps were evaluatedby threshold voltage shift A Vr aftersubstrate hot - electron lnJection t8l.AVr due to the trapping of electronsard shown in Fis. 2 as a function oflnjected electron charg:es. AVrlncreases wi th the increase in theexposure dose, and it cannot recoverto the non-irradiated level, evenafter the final anneallng. Thls meansthat neutral traps are not completelyrecovered even after the final anneal-ing. Effects of the gate oxide thick-ness on AVr are shown ln Flg. 3 fornon-iqradiaf,ed devlces, and for 3,00OmJ/cmz-irradlated and subsequently-annealed devlces. InJected electron-charg:es are flxed at 0.5 e/cmz. AVtdramatlcally decreases as the gateoxlde thickness TOX decreases below Inm . Tox wi Il be - a key parameter lnreducin-d- the lnfluence of the residualneutral. traps on HC-degradatlon.

3.2. HOT CARRIER DEGRADATION3.2.L. EFFECTS OF X-RAY DAMAGE

In order to lnvestlg;ate theeffect of X-ray damage on HC-induceddevlce degradatlon 1n subquarter-mlcrometer NMOSFETS, threshold voltageshift A Vt and the increase in I"odependences upon stress gate voltps^eunder fixed stress-draln-voltage weremeasured. The bias-stress was appliedto the devices after exposure. Theresults for O.2 pm NMOSFETs are shownln Flg . 4. Compared wlth non-irradi-ated devices , A Vn increases pemarka-bly wlth incredsed exposure dose,especlally in the hlgher stress-gate-voltag;e range It should be notedthat A I^-, is not dependent on theexposure "dose, whlch means that lrra-diatlon-lnduced damage does not playany role ln lnterface-trap generatlon

lo-3

to-'o2468t012

GATE 0Xl0E THICKilESS (nml

Fle. 3Effect of gate oxlde thlckness on A Vrdue to substrate hot-electron injectionlL"=1O pm.

durlng HC-lnJection. The increase lnAVn ls clearly due to the trapping oftnJ-ecfed-hot-electrons ln the irradla-tlon-lnduced traps.

Here, lt has been reported thatHC-degradatLon 1s signlflcantly 1n-creased ln PMOSFETs due to the en-hanced electron trapping in the oxldeby the resldual traps tgl. Recently,lt was found that HC-degradation modefor quarter-mlcrometer level PMOSFETSls qul te dlfferent from prevlousreports ( 1. e. , effective channel-length reductlon due to trapped elec-trons ) , and the new degradation models caused by lnterface-traps generatedby hot-hole-lnj ectlon t 10 I . Inter-f ace-trap g:eneratlon ls a more lmpor-tant phenomenon than electron trappingfor future PMOSFETS. Therefore, thefact that there 1s no irradiation-lnduced damage effect on lnterface-trap g:eneratlon durlng HC- in j ection lshlshly slgnlflcant.3.2.2 EFFECTS OF ANNEALING AI{D GATE

OXIDE TUICNiTESS ON DEVICE LIFETIMEA V1. and A I.o dependences upo4

stress gate voltag^e for 3,000 mJ/cmo-lrradlated and subsequently-annealed0.2 pn-NMOSFETS are shown in Fig. 5.For comparison, those for non-irradi-ated devlces are also shown in theflgure. Gate oxlde thickness for bothtypes of devices is 5 nm. It can beseen that there are no meaningfuldlf f erences ln A V.n and A I,-r, betweenthe two types of ddvices. The influ-ence of X-ray damage on HC-desradatlonseems to dlsappear after the flnalanneallng. In order to investigatethe effect of the final annealing indetalI, DC device Ilfetimes r wereobtalned as a functlon of substrate

X'IDosE lmJ/cm2l /i

3ooo //\//o \//

\ / / Qlx,l= 0.5 c/cmz

x////t

i I AFrER ExPosuRE

I 'ANNEAL?

l0-l

lo-2

l0-3

td4

AFTER

00SE(mJ/cm2lo0a 100B 400 ro{r lo-{

AFTER EXPOSURE

8 ANiIEAL

STRESS: Vo=3.5V3 hrs.

DOSE (mJ/cm2lo0x 3000

to-8

ldlo

lo-e

lo-l

lo-2

to'

l0s

to8

tde

lo-10

ldrl

voltage.

e(,Att

lo-3

012STRESS GATE VOLTAGE (VI

F1s. 4AVq. and AI^^ vs. stress gate voltage.t"='o-z FEl, T-o-x=s rrm.

current normal Lzed by channel wldthISUn/Weff. t ls deflned &s tlmerEdilirdil-to reach AVr=10 mV. Theresults are shown in FiS. 6 for 0.2pm NMOSFETs with gate oxide thicknessand exposure dose as parameters.Although c is slightly reduced due tothe enhanced electron trapping for 8nm gate oxide ln hlgher exposuredoses, there is littIe irradiatlonef f ect on t f-or 5 nm gate oxlde evenat 3, O0O mJ /cm2 .

Therefore, the influence ofsynchrotron radlatlon' X-ray llthogra-phy on HC-lnduced degradatlon ln O.2pm NMOSFETs 1s consldered negllglblewhen gate oxldes thlnner than about 5nm are used, even when three-levelmetal processing ls assumed.

4. CONCLUSIONSSynchrotron-X- ray- induced lnter -

face-traps are completely ellminatedafter 4OO"C No/Hc anneallng. However,neutral traps" p-artlally remaln evenaf ter the anneallng. The resi-dualdamage effect on HC-degradatlon 1snegllgible for subquarter-mlcrometerNMOSFETs when gate oxldes thlnnerthan about 5 nm are used. It wasfound that there 1s no lrradlationdamage effect on interface-trap gener-atlon durlng HC-iniecti-on. This phe-nomenon will be very lmportant fordeep-submicrometer PMOSFETS where HC-deg:radation is mainly caused by lnter-face-trap g:eneration, not by electrontrapplng in the oxide.

ACIO.IOWLEDGEMET.ITSThe authors would llke to thank

M. Oda and M. Suzuki for thelr supportin lrradlatlon experlments. They also

STRE$i GITE VOLTAGE (VIFls. 5AVr and A Ir.r, vs. stress gatel"=b . e Ffr, T;-X=s nm .

x\

X1\ox

r\\\'\OOSE (mJ/cm2l t.

o0a t00. 1000x 3000

tdz rdr roo ror roa

ls6 /tcfi lpllpalFls.6DC devlce ltfetl-ne tLf-O .2 lr m.

vs . ISUg /V ett .

would llke to thank T. Sakal and K.Izuml for thelr contlnuous encourag:e-ment.

REFERENCESI1l J.Y.-C.Sun, €t &1., 1986 SSDM, 479.I2l C.C.-H.Hsu, €t erl., l-989 IRPS, 189.131 c. Przyrembel, et aI. J. DE PHYSIQUE

(1e88) C4-767.t41 M.Mlyake, et 8I., IEEE Trans. Electron

Devlces, 36(1989) 392.l5I T.Kaneko, et aI., L989 MlcroProcess

Conf., 104.t6l B.Eltan, €t 81., AppI. Phys. Lett.

40(1982) 523.t?l E.H.Nlcolllan and J.R.Brews, MOS Phys.

and Tech. , JOHN WILEY & SONS ( 1982)549.

t8l P.Heremans, et 8I., IEEE Trans.Electron Devices 36 ( l-989 ) 1318.

tgl C.C.-H.Hsu, €t dl., IEEE ElectronDevlce Lett. 19.(1989) 327.

t10I T.Tsuchlya, et aI., L990 SSDM, 29L.

lo{

to3

rot

e,o-F

lr,-Flrll!

J

t9