Roland A. - PVD for microelectronics (779636), страница 74
Текст из файла (страница 74)
11.7). For example. in Afthe preferred emission is along the ( 1 10) crystallographic direction. Thedistribution of grain orientation within a polycrystalline metal target thenaffects the overall angular e~nissiondistribution of the sputtered atomsfrom the tareet. Generally. the thermomechanical process of target manufacturing imparts a preferred alignment to the grains, or "texture," whichintluences film uniformity.
Figure 11.8 shows calculated and measured angular distributions of sputtered A1 from ( 100) and ( 1 10) single crystal targets. Figure 11.9 presents the calculated film uniformity on a 200-mmwafer from these single-crystal A1 targets and from polycrystalline A1 withtexture ranging from strongly ( 100) to strongly ( 1 10). The simulation suggests that a predominantly random spatial distribution of crystallites witha very slight (100) texture produces the best thickness uniformity. Highlyrandomized orientation offers another potential benefit.
Namely, as a target with a preferred orientation erodes during sputtering. its surface topography evolves to expose a variety of new crystal planes at different areasover the surface (e.g., on the sides of the erosion grooves). This can resultin an erosion-dependent film uniformity. Experimental work on both AI alIoys and Ti confirms that a randomly oriented target texture is preferred foroptimum blanket film uniformity and consistency through target life.Grain Size Target grain size affects the deposited film in several ways.When the grain size is large (> 1 mm) and the target-to-wafer spacing isSPUTTERING TARGETS393(111)TargetackedionSputteredAtom(110)~ 0ncident,.w,OO....,, o ,- '- {' - - .b., o ,I,"~.o-9.9990t_90 ~,.9."- ! .:,,,I.,.'9.' " ,b9 i:.Q9 ..', ~.p e"Q9999.~".:.9e#i099".~.9,999..' X"'~e~99,090*.o9999o.o-"'''~176ooo~~"...
. . . . . . . . . . . . . . . . . . . . . ..t9oio~-'~,..--'"0..................................P(e) (arb. units)190 ~FIG. 11.7 Angular distribution of sputtered atoms from single-crystal (100) Al displaying lobes associated with close-packing directions [11.9].small ( ~ 5 cm), local differences in each grain's angular emission distribution can create local regions on the wafer that have different depositionrates, which affects global uniformity. This problem can be solved by reducing target grain size below 500/~m or using much larger source-to-substrate spacing.
This is supported by data such as that in Fig. 11.10 (fromref. 11.7), which shows sheet resistance uniformity of A1-0.5%Cu filmsfrom A1 alloy targets of different average grain size. Grain size can also influence particle generation (as illustrated in Fig.
11.11) for reactive sputtering of TiN from a 4N5 purity Ti target [11.7]. Figure 11.11 shows that,,. . . . . . . . : . . . . .~- - ~ 1 7.6 1 7, -6 - . , ,.. ,-'".~...-:..9"";"'-""-.. ~9~176..,. ~:.!-""'.""-'"""-,Dgp.,,:."-',.~,i:.~":9 -"....-........-~176:" ......."-.... -...'...-.":......-........:9............9...... --..:"6~ 1 7~ 6... .......--....._....d:9:-:,~176...,,""-~".,.."~9..--- .... .-.-.." "."".".:" ~ ~176 9:.'......"'"'-.>.-'.i~.. . . . ~ .
."-.~-. ....e-...-....".'-..".."~...-........'..",..........~..'..."....-". ..... :-~............',,-..-. ........ ......... -,,HG. 11.8 Calculated and measured angular distribution of sputtered A1 atoms from single-crystal (100) and (110) AI targets and similarly oriented polycrystalline AI targets [11.9].SPUTTERING TARGETSL. . . ."~" 7 - - - - - - - . ~.~. 6i,,_,",,..................'"',,Spacing' "'.!' .'"" .....
'l- i~"" 44 mm...........\o5' ~,D4.......,-- 3, ,395--~.....- "-"'56 mm--I--optimum.....'0I-.o~1,,0.......~-~, ,,,,,"Crystallographic(100)TextureSpectrum"qpq) O 0 ,B....,,,•(110)FIG. 11.9 Calculated Al film uniformity versus target-to-substrate spacing for single-crystal (100)and (110) AI targets and polycrystalline targets with texture ranging from strongly (100) to strongly(110) [11.9].submicron particle levels (size range = 0.3-0.5 ~m) could be greatly reduced in this case by going to fine and even ultrafine grain targets with average grain size ~ 10 ~m, and that a high-performance target can maintainparticle levels of ~ 0 .
0 l / c m 2 through life.Second-Phase Microstructure Both the resistivity and reflectivity ofA1 alloy films can be influenced by the amount of second-phase precipitation (e.g., A|zCu ) in the target [l 1.10, l l . l l ] . It is believed that the emission of second-phase molecular species (e.g., Si 2, AlzCu) from the targetcreates second-phase nucleation sites in the PVD film, which can influencehow stress is relieved. For example, in AI-Si-Cu films at elevated temperatures ( > 400~second-phase precipitates can suppress hillock formation in favor of the growth of single-crystal "whiskers."Copper Segregation Segregation of copper in macroscopic regions ofthe target is also to be avoided since this can cause microarcing, nodules,or film segregation effects such as variation in electromigration resistanceor dry etch rate across the wafer.
Both of these effects depend on the localconcentration of Cu in the A1Cu alloy film.R. POWELL AND S. M. ROSSNAGEL3962.521.5=%1 sigmaUniformity0.50#1 #2 #3 #4 #5Target NumberTa,,,rget#, l, ,,TargetFace Mid-Radius c~!er263120I>300>3002>30069362726946653561I Avenge>2286569FIG. 11.10 Sheet resistance uniformity for AI-Cu alloy targets depends on average grain size (datafrom ref.
11.7).11.8 Particle GenerationA major source of particles in a well-designed PVD module can often bethe target itself. A general issue is that the nonuniform erosion can lead tomaterial being sputtered from heavily eroding areas of the target and backonto more lightly eroding a r e a s - typically near the center. Build-up ofsputtered material can then lead to flaking. A related issue is the use of aTi target exclusively for reactive TiN d e p o s i t i o n - for example, dedicating a PVD module for deposition of a TiN antireflection coating (ARC)layer. TiN has a high compressive stress, and over the course of time athick layer of TiN will build up on the shields that can spall and produceao,~eI N!,L o q l u e q l ssoals aox~o I s e q qO!A~pUeS N!M!,L oql p u g N!,L ol u o ! s o q-pc p o o g s e q ! i "spIo!qs poleOO-N!,L oql u o ao,(e I !,L e l ! s o d o p ol s e g OX!lOeO.~l n o q l ! ~ p o a o l l n d s st.
log.ml oql q o ! q ~ u! g u ! l s e d !,L POlleO-os t u a o j a o d ~ilgO-!po!.md ol s! u o ! l e n l ! s s!ql q l ! ~ g u ! I e o p j o ,~eh~ ou 0 "SOlO!laed p u g SO~leLI"(D~I~ 'uetul!O "d jo ,~sm-moo) iool aalsnlo ClAd tunnoeA-q~!q e pue la~.rel !,L zouettuojaod-q~!q e ~u!sn ,~q OJ!l lo~agl q~no.lqlzma/SOla!lyed I0"0 ~ jo SlOAOI le pou!mu!etu o.re aojeA~ tutu-00 E e uo StUlg N!.L u! ,q!suop Ola!lJed(q) '.[L"I 1] mr/0t~ > ozls u[ea~ ,,~pue '(mr/00I >) su!e.l~ oug 'sulea~ osyeoa ql!n~lo~yel LLe tuoaj .mjem tutu-0g I e oluo pol!sodop StUlg NLL u! SOla.u.redtur/-~;'0-C0 jo ,q!suzcl (e) II "II "Did(q)(Jq'MN) euut;eJl'l l e w000;~009t00800~ L'"00bo00"0:go'oI~80"0~90"0m~lt~0"0OL'O~k'OI~I.'0wo "bs/selo.qJed 8 0 0 LifO = ueelN:leBJel wn!um,!l ~O q~MeoueuJJoped elo!ped N!IgL'Ob~~8L'O0~'0(~)sJejeM000909L8.009~~4.-vv|IP ....... .
. . . . . . ~. . . .wvv m~~/. .v. . . . . . . . . . . .v. . . . .~. . . .". . . .9. .. .9. . . . . . . . . . . . . . . . . /,, ................ ~...0~.00"0o~oO'lz'OI. /. . . . . . . . . . . . . . . . . . . . . 9. . .V....09~ k_...........................................~~r--.....!tir . .
. .a. . .r.......................909"0V.........."0_.;~08"0...........................................~O0"L ~". . . . . . . . . . . . . . . . . . . . . . .[ .........../>..L". /. ..............................................l!ioO-HOO9 O8"L),eSJei u!eJE)-eUL-Im9~oOb'L.O9"L...............................................................................L6s),e6Jel u!eJo-esJeoo 9] ......- 00"~SIHD~IY.I. DNIH~I.I-I.iqdSR.
POWELL AND S. M. ROSSNAGEL398alone. By using the Ti as a stress-reducing paste, additional TiN depositions can be carried out without shield cleaning. In a production environment, pasting might occur after every 100 wafers or so. Conversely, if thesputter chamber is equipped with a mechanical shutter that can be positioned between the target and wafer, pasting for just a few seconds mightbe done after each wafer while the shutter is closed. In either mode, the cumulative effect of pasting on overall wafer throughput needs to be considered, since the practice can reduce useful kW-hr of Ti target life by20-30%.Other issues of particle generation relate to target quality and tend tobe materials-specific.
For example, areas of low density (microvoids)produced in Ti targets during target manufacturing can trap gases (seeFig. 11.12). As these gases are released during target erosion, they cancause local high-pressure regions that induce electrical arcing from thetarget to the plasma with related particle generation. Also, since the gasesin the voids are at relatively high pressure compared to the mTorr processambient, a microburst of released gas can ballistically launch a piece oftarget material that eventually impacts the wafer surface. Refractorymetal targets such as TiW that have not been properly burned-in also havea tendency to grow dendrites or cones on the target surface (Fig.
11.13).These cones continue to grow until they reach a critical height above thetarget surface at which arcing will occur.Contamination, on the target surface or included in the bulk, will alsotend to arc and flake. For example, a relatively high density of A1203 inclusions in AI targets has been correlated with arcing events at the targetFIG. 11.12SEM of a microvoid in a Ti target (from ref. 11.12).SPUTTERING TARGETS399TiW Sputter Target DendriteFIG. I1.13SEM of dendritic growth on TiW film (from ref. 11.13).surface. Because the inclusions are insulating, they charge up due to ionbombardment of the target, producing local electric fields that can exceedthe dielectric strength of the insulator. An arc is then initiated by the"flashover" that occurs when dielectric breakdown is reached. The arc inturn can produce localized melting and explosive emission of AI dropletsfrom the target and onto the wafer.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
