Implantation beam angle study for Al implanted in 6H-SiC

zyxwvutsrq zyxwv zyxwvu zyxwv zyxw zyxwvutsrq zyxwvutsrqp E. hilorvan, J. Montsearat, P. Godignon, J. Fernandez, D. Flores, M.L. Localelii** d. Millan and 6.Brszeanu* Centro Nacional de ~ ~ ~ r o @ l ( @ c~~ r ~~ nCampus ~~ ~-~ UAB, ~ 081 ~ 93~Bellaterra, 6 ~ SPAIN , *Faculty Qf Electronics, University Politehnica ~ u c ~ a ~9000, e s ~ S6, , Bucharest, R ~ ~ **CEGELY-ECPA, INSA Lyon, 6$621 Villeurbanne, FRANCE s ~ ~ I A RY In this paper a simulation study of AI implantation into GH-Sic single crystal is presented. Beam of crucial irnpOrt~nCeto achieve good i ~ ~ i a n ~ a tprocess jon control. A first step anafy$is is given and values for optimum ion beam angles are proposed, orientation with respect to crystal axis has been inves~iga~ed. this aspect is ~ o r for Silicon carbide GH-SIC single crystal is known as an exceptional s e ~ i c o n ~ ~ cmaterial specialized devices fabrication in the fields of power and high frequency electronics, o p ~ o e l ~ c ~ ~and onic sensors 1%'. Its possible industrial applications are a ~ ~ o m o ~and ~ v aerospace e e~ectron~~s, power devices for converters, switches or microwave technology. ion i m p l ~ ~ ~ a tt~chnolo~y, ion which has been shown to be a very interesting tool to incorporate doping impur~~ies in Silicon in a conrro~ied manner, begins to be transposed to Sic for device fabrication. Large di~erencesbetween the two materials give rise to many p r o b l e ~for~ Sic ~ e c ~ n ~ which ~ o g ydo not appear in the case of Silicon t ~ c h n o l ~ gDifferent ~. implantation conditions have to be used for Sic in order to achieve semiconductor doping. In the same way, ion implant simulators for c ~ ~ ~ a ~silicon l i n ehave to be to account for the different crystal strudura and composition. modifiedin order In order to study the d j ~ ~ e r easpects n~ of ion ~ ~ p ~ a n t technology a ~ ~ o n for 6H-$iC, a specific ~ i m u l a ~ o ~ has been developed. it uses ~ o n t s c a r ~method o together with semi@mp~r~ca~ and physica~iybased models to calculate the implanted ion t r a j through ~ ~ the ~ ~Sic~single crystal. The slowing down of an incoming ion is evaluated within the ~ r ~ ~ ofe binary ~ o collision ~ k approxj~at~on for nuclear StQpping /2,3/and with advanced models for e ~ e c ~ ~stopping, o n i ~ following the recent work of other research groups /4-61'. In this paper, we present the simulation of one of the numerous problems of ion ~ ~ ~ l ~ nt t ~~ ~c i ~ othe n~ tilt~and~ rotation o ~ angle ~ : o p t ~ ~ i ~ tfar i o nmaximum dopant depth profile uniformity across the implan~~d wafer which ie of ~ r u ~~~ a l ~ for process ~ re~a o o d ~ c ~ ~ i ~ i ~~. n ~ 2. ST A first aspect to be considered is channelling which occurs when a swift ion trawels in a special c ~ t ~ f l o ~dr i aa e~ ~hinside ~~~ ~~ na crystal. A c h a ~ f l ~ion t ~ can ~ d be vie^^ as a particle guided between the walls of a cylinder which generatrices are made by rows of crystal atoms. This case is named "axial c ~ a n n ~ l along ~ ~ na~given " c ~ ~ ~ a i ~ o gdirection. r a ~ h iAn ~ other ~onfigura~ion exists When the ion travels b e ~ e two ~ ncrgrstal planes and is called "planar ~ h ~ n n @ ~~~~i n r ag ~ to ' ~laegiven i ~ r y plane. ~ ~ in a ~ the first case, which is ~ r e d o ~ i the ~ ~moving n ~ , ion has an oscillatory t ~ j ~ c inside t o ~the cylinder. The walk of this channel are somewhat "porrous" for the ion, due to the n Q n ~ n i f o r ~ofi ~the y r~pul§iv~ potential generated by the dispersed positive nucleus of the lattice. Generally speaking, the channelling behaviour of the ion depends on many parameters such as the atomic structure of the cylindsr waNs related to the crystalline structure of the target and the direction of motion of the ion, the ion energy which determines the sensitivity of the ion trajectory on the target nucleus potential, the initial ang/e of incidence of the velocity vector of the ion with respect to the ~ ~ adirection, ~ ~ the e initial l zyxw zyx 393 ~ zyxwvut zyxwvu zyxwvuts zyxw zyxwvuts zyxw zyxwvuts zyxwv zy zyxwvuts Qof the ion ~ in the i plane ~ p e r ~~ e ~ d ~to~ the ~ ~~channel ~ a r axis. A schematic representation of some meters effects is given in Fig d . Further explanation will be found in part 4. The second aspect to be considered is the optimization of the tiltlrotation angles (tilt is the angle of the ion beam with respect to r O O 0 1 ~crystal axis, rotation is measured from €1 1-20> axis). When implanting a wafer for device f ~ ~ ~ i ~ a~ ~ni o~ n~, ofothe r depth m ~ doping ~ ~ profile of the implanted layer has to be taken th ~ ~ ~ ~ t r o s beam ~ a t ~sweep cal ~ ~this problem ~ is critical ~ with ~ respect tonchannelling ~ ion of the ion beam (tilt and rotation) with respect to the e00013 axis of the crystallin@ wafer c ~ during~the sweep. ~ Then, ~ c ~~ ~ n e~bel ~ tiit ~s and n ~rotation sensitive, both angles have to be ~ h ~ s such e n as to m ~ nc~ann~lling i ~ ~ and/or ~ ~to ke constant the channelling Conditions during the w ~ i ~~ ~~ e~ process. a n ~~@ ~ a e r~m ~~nofa~optimum t i ~~n tilthotation angles requires the eval~a~ion of the i ~ n In practice, large tilt angles yieid with respect to both angles for a given i m ~ l a n ~ a tenergy. oided ~ e ~ ofawafer ~ surface ~ e erosion, strang backscattering of incoming ions and mask ~ ~ a d effect. o ~ ~ r n(100) ~ oriented Silicon crystals, 7 9 tilt and 302 rotation are used in standard ~~~~n~ about r ~ n i ~~ o p ~ ~ (Le~worst ~ channelling ~ ~ ~ conditions) ~ a n with low depth s across the wafer. ~ ~ ~ ~ ~ nb ~nh a@ v~o ~off fians ~ ~in the n 6H-SiC ~ ciystal being a consequence of the CrySfal structure and we have ~ ~ ~the~trajectory ~ a of ~ im@ ~ i ~dions n ~ ine a~perfect crystal. This means no thermal f lattice atoms and no defects ~ ~ n ~ ~during a ~ the i o implantation n process simulation. The defect ~ ~ f ~ @wou~d r a tend ~ ~ to o ~ a n d othe ~ icrystal ~ ~ lattice but the minimization of channelling by crystal a ~ o ~ ~ h i iss aad~~ ~e ro@ ~aspect n ~ of ion ~ ~ p ~ a ~ and t a ~will i onot n be treated through this paper. MOr@OV@r, zero degree beam d~Vergencewill be assumed as well as axide free crystal surface. We have chosen a simple pro^^^^^^ to evaluate the channelling yield relative to a given implantation process: ions are ~$~~~~~ to be c h ~ ~ ~and ~ ~l ~ ae ~ d to j the c tail i ~~f the ~ impurity ~ ~ depth profile if they cross a depth limit in the crystal defined as: where R, and sigma(E) are the ~ ~ ~range j and ~ standard c ~ deviation e ~ for random i m p ~ ~ n ~ acondition^ ~Qn at en^^^^ E. This criterion is s ~ ~ @ crude w ~ but a ~represents quite well the atom depth p r ~ f ~ ~ e n ~ ~ ~ rn ~since ~ ~ ~it lis a ~em r~e ~ a ~i ~Q ~~ theI@ dose ~ which ~ ~ penetrates t too deeply into the crystal and thus modify the c o n ~ i g u r ~ ~of~aojunction n farmed by ~ m p ~ a ~ t a t i ~ n . 4. The s ~ ~ s iotf ithe ~ ~j ~h ~ n ~with e respect i ~ ~ nto ~ the tilt and rotation angles can be visualized by plotting, for a given i ~ p l a n ~ ~a ~ ~ a~the ~ ~~ ~~ of c ~h ~~ yn ~ ions ~, (following ~l l ~ ~ the n criterion~described ~ above) ~ versus the ti# and r ~ t a ~ i o angles n on a 30 surface plot. Due to the crystal structure symmetries, we only to r ~the plot far ~ rotation r angles ~ a n from ~ ~OQ~ to 30p. g ~ Far practical ~ purpose, ~ the tilt angle is varied ~ ~ OQ an ~ @ ~ n A 3D plot for 30 keV 27A1+ions implanted into 6H-SiC is proposed in Fig. 2 and shows some interesting aspects of the i ~ ~ l a n process. ~ ~ ~ ~First, o nwe can verify that far a O2 tilt angle, the channelling yield (CY) is the same for any rotation angle used in the simulation. Next, it can be seen that increasing the tilt leads to a d ~ c in CY. ~ This ~ ~is a~consequence @ of incident ions having a tilted velocity vector with respect fo the 4?01flrc ~ s ~ a 6 l ~ ~ axis, ~ a pwhich h ~ c are more likely to Jeave the channel at the beginning of their ~ ~ ~andj suffer @ more ~ random ~ o collisions ~ with target atoms. The main feature of the plot is that a ~ ~ reduction ~ ~of CY ~ appears. ~ aonly ~ when~tilt angle e reaches about Sp, which is a surprising high 394 e zyxwvut zyxwvu zyx zyxwvuts zyxwvu zyxwvuts value. This means that when simulating ion implantation around the <0001> direction, the implanted atom depth profile is not affected until about 5Qtilted imptantation is used. For tilt angles above 5Q,the CY decreases rapidly from 90% to about 50% whichever rotation angle is used, However, an increase of the tilt angle above 15Ggive rise to a special behaviour of The CY: for rotation angles around 30Q,the CY strongly increases eo about 80%. This sharp pic for CY around 2 P tiit and 309 rotation can be explained by the hexagonal structure of 6H-SIC crystal which presents open channels in this ~ i ~ e with ~ ~respect o n to the -=OQQls axis. Even if those channels are discontinuous due to the ~elaf~V@ly complex stacking sequence of GH-SIC, they are effective in channelling implanted ions. An other $~nseqUen$eof this geometrical configuration is the small peak for 2CP ti1t/Og rotation implantation. For rotation a ~ ~ around ~ e sIO9 this ~ h ~effectn is greatly ~ ~ reduced ~ because l ~ it is~more~difficult for implanted ions to enter a channel. They penetrate inside the crystal within a more random direction. As a co~se~uei~$e, one can use tilt angles superior to 10' together with about 109 rotation angle as defined pr@vio~sly, in order lo minimize atom ~ ~ profile p ~nonuniformity h for a given implantation process. Increasing the ~ ~ p l a n energy ~ a ~ will ~ ~not n affect much this optimum conditions since higher energy ions have even lower probability lo be channelled. The energy of the moving ion affects its channelling probabili~ in two ways: increasing the energy will cause the critical angle of channelling to decrease. In other words, for a given incidence angle d ~mpian~ation with respect to a given channel axis, a high energy ion will have a higher transverse energy with respect to the same axis and will have a higher probability to cross the potential barrier formed by the atoms of the channel walls. Even if this ion can be rechannelled in a parallel channel in the crystal, the probability of random scattering after the crossing of the wall is high. On the other hand, for the same incidence angle, if the ion has a low enough energy, it will suffer a "progressive" reflection by the, wall repulsive?potential and will then have an oscillatory motion between the walk of the channel (Fig. 3). Analyticai formulae for critical angle evalluation usually assume an energy dependence in Our simuhtions for cOOQ1=-channels give about E"." behaviour (fig.4). Further investigations have to be carried out. We have inv~stiga~ed the t ~ ~ ~ r o t adependence tio~ of channelled dose during Al implantation into 6H-SIC. The first step of this work leads to ~ ~ ~ tilti of ~102uor more m with respect to e0001> crystal axis and rotation of IOp with respect to 41 1- 2 bcrystal axis. The simulator for GH-SIC developed at the GNM has been usad here and will serve for better ~ n d e ~ s ~ a nand d i f loptimisaticln ~ of the implantation process. w /I/ /2/ /3/ /4/ /6/ NCES H.Morko$ & al. Journal of Applied Physics. 76 (3),August 1994. M.T Robinson PW ai. Physical Rev. B9 (1974) p.51408. J.F Ziegler & al. The stopping and range of ions in solids (faergamon, New York 1985) W.Brandt PW at. Physical Rev. 825 n?3 (1982). Echenique & al. Physical Rev. A33 ne2 (1 986) W. ~ e n g - p ~ n Physical g rev. A52 nQ5(1995) 395 zyxwvut s r qponml k j i h gf e dcbaZY zyxwv zyxwvutsrqponmlk 1 BEAM ANGLETOOHIGH Rp + sigma TARGET ATOMS I/ .".__._ ___ IQN BEAMS channeled Ion5 zyxwvutsrqponmlkjihg zyxwvutsrqponmlk BULK CRYSTAL ION WlMi WSmQNTOO CLOSE TO ATQMC ROW Distance from de crystal surface fig.- 2: definiti~nof the criterion used for the c a i ~ u ~ a o ~f ~the o nchannelling yield (CY). Ions that go beyond the (RP+ sigma) depth limit are sumed to give the CY. 0.0 zyxwvu zyxw ' I - 0.0 25.0 50.0 75.0 100.0 125.0 1 Ion energy (key) Fig.- 4: Critical angle versus energy of implanted ions. The cuwe labeled ucrit.ang-'.5"is the best fit for a hear function of the energy. Resulting Fig.- 3: ~ h ~ ~yield ~ versus @ l tilt ~ and ~ n ~ ~~~ a ~simulated ion energy d e ~ e f l d ~ n of~critical e angle has angles for =AI' ions i ~ p i ~ n into ~ e GH-Sic. d the form erit.ang = k. Em where k is a constant. zyxwvut 39 6