Scanning Electron Microscopy
Volume 1986 Number 2 Article 3
7-16-1986
Ion-Induced Auger Emission from Solid Targets
Josette Mischler Université Paul Sabatier et Institut National des Sciences Appliquées
Nicole Benazeth Université Paul Sabatier et Institut National des Sciences Appliquées
Follow this and additional works at: https://digitalcommons.usu.edu/electron
Part of the Biology Commons
Recommended Citation Mischler, Josette and Benazeth, Nicole (1986) "Ion-Induced Auger Emission from Solid Targets," Scanning Electron Microscopy: Vol. 1986 : No. 2 , Article 3. Available at: https://digitalcommons.usu.edu/electron/vol1986/iss2/3
This Article is brought to you for free and open access by the Western Dairy Center at DigitalCommons@USU. It has been accepted for inclusion in Scanning Electron Microscopy by an authorized administrator of DigitalCommons@USU. For more information, please contact [email protected]. SCANNING ELECTRON MICROSCOPY /1986/11 (Pages 351-368) 0586-5581/86$1.00+0S SEM Inc., AMF O'Hare (Chicago), IL 60666-0507 USA
ION-INDUCED AUGER EMISSION FROM SOLID TARGETS Josette MISCHLERand Nicole BENAZETH Laboratoire de Physique des Solides, Associe au C.N.R.S. Universite Paul Sabatier et Institut National des Sciences Appliquees 118, Route de Narbonne - 31062 TOULOUSECedex (France)
(Received for publication February 28, 1986: revised paper received July 16, 1986)
Abstract Introduction We present a review of the Auger emission Impact of heavy ions on surfaces gives rise induced from light elements (Mg, Al, Si) bombarded to a variety of collision events leading to ejec by ions of intermediate energy (1 keV - 200 keV). tion of secondary or reflected ions, sputtered The different physical phenomena at the origin of atoms, electrons and photons. In principle each Auger emissions are outlined, in particular the sort of particle carries information about the mechanism of molecular excitation responsible for surface and can lead to a method of surface analy the production of inner-shell vacancies in colli sis : secondary ion mass spectroscopy (SIMS), Ru sions between two complex particles and the pro therford backscattering spectroscopy (RBS), proton cesses of Auger decay and electron transport induced X-Ray emission (PIXE). In practice, SIMS in the solid. Auger spectra partially consist of is the most frequently used technique. A different L23VVelectrons corresponding to decays in the type of study of energy distribution of electrons bulk; this part is similar to that observed from ejected from solids by photon and electron impact electron-induced Auger spectra. Superimposed on has produced many useful spectroscopies [Auger this broad structure appear different narrow lines electron spectroscopy (AES), X Ray or U.V. photo due to de-excitations from excited sputtered atoms. electron spectroscopies (XPS and UPS)] for charac These atomic-like lines are assigned to different terization of the electronic, chemical and geome L23MMor ITL23)2MMJAuger transitions. The depen trical properties of surfaces. In this paper, we dence of the width and shift of these lines on dif shall discuss the present understanding of Auger ferent parameters (e.g., ionic energy, emission electron spectroscopy induced by heavy ions from polar angle) is interpreted by Doppler effect. On solid targets and investigate the information such the other hand, the experimentally determined a study can bring to ion-surface interactions. L23VVand L23MMAuger yields are compared to values Under heavy ion impact, at intermediate ener calculated by computer simulations of collision gies E (in the keV to hundred keV range) electron cascades (from EVOLVEand MARLOWEcodes). Lastly emission is, primarily, due to ionizing colli~ons the azimuthal and polar angular distributions of undergone by primary ions, energetic recoil atoms both L23VVand L23MMAuger electrons are analysed and fast electrons. On the energy distribu by taking into account the important role of ion tion curve these "true" secondary electrons make induced surface topography. up the intense peak centred at low energy (about 2 eV),is followed by a steadily decreasing tail. In addition, in some cases, electrons of specific energy appear on the spectrum as structures super imposed on the continuous energy tail. These stru~ tures are due to Auger transitions involving a two electron process and leading to decay of the inner shell vacancies created in either the target atom and/or the incident ion. This ion-induced Auger emission was observed for the first time by Snoek et al. (1965). Since 1975, several research wor KEYWORDS : Ion-solid interactions, Auger emission, kers took up this line of investigation. The most Ion-induced surface topography, Surface. studied elements are Mg, Al and Si which give un der ion bombardment L Auger spectra with unexpec Address for correspondence : ted features. Fig. 1 gives an example of the energy distri J. MISCHLER, Laboratoire de Physique bution of electrons emitted by bombarding an Al des Sol ides, Associe au C.N.R.S. - Universite target with different ions. The Al Lz3 Auger spec Paul Sabatier, 118 Route de Narbonne trum appears between 35 and 80 eV. With proton 31062 Toulouse Cedex bombardment, it is essentially composed of a broad FRANCE Phone N° 61.55.68.21 structure assigned to an L23VVtransition invol-
351 J. Mischler, N. Benazeth
ving two electrons belonging to the valence band ions can transfer a sufficient amount of kinetic (V) which is similar to that observed in electron energy to the target atoms to displace them from aluminium collisions. In addition, under heavy ion their lattice site (primary knock-on atom). If impact, the spectra obtained are mainly composed the energy transferred to the primary knock-on of three narrow peaks, superimposed on the broad atom is high enough, it can displace and ionize structure ; their amplitude depends on the ion another target atom. These secondary recoiling mass. This spectrum in no way looks like that in atoms may in turn displace further atoms resulting duced by binary collisions of 50 keV Al+ striking in a displacement cascade. A surface atom becomes Ar gas which exhibits a great number of lines, as sputtered if the energy transferred to it has a shown in Fig. 2. component normal to the surface which is larger Because the ion-induced Auger spectra of these than the surface binding energy. light elements look like neither the spectra ob So, from medium energy heavy ion-solid target tained in electron (or proton) -solid target col interactions, asymmetric collisions between the lisions nor those induced in binary gas collisions, incident ion and the target atom, and symmetric their interpretation has been the subject of con collisions between two target atoms can occur. troversy. Although the narrow lines are now inter Moreover, but with a lower probability symmetric preted by most authors as the decay of excited collisions between an implanted ion and an inci sputtered atoms, an Auger model does not yet exist dent ion can also take place. which totally explains all the features of these Production of inner-shell vacancies in ion-atom Auger spectra. collisions (Kessel and Fastrup, 1973) In the next section, we will outline the dif Either of two extreme approximations are gene ferent mechanisms which intervene in the ion-indu rally used to deal with the creation of ced Auger spectra. Then, after a brief survey of inner-shell vacancies in ion-atom coll is ions : the the experimental techniques used for the electron direct Coulomb interaction and molecular excita spectroscopy studies, we will review the more re tion (Garcia et al., 1973). cent results obtained and the interpretations pro The former is valid when the projectile is a posed by the different authors. high-velocity point charge (e.g., H+) which, by direct Coulomb interaction with the electrons of Different physical processes the target may ionize an inner-shell . To solve involved in ion-induced Auger emission and this problem different models have been proposed computer simulations such as : the plane wave Born approximation (P.W. B.A.) and the Binary Encounter approximation(B.[A.). Collisions in solids The molecular excitation applies to collisions When a sufficiently high-energy particle pe between two complex atomic systems at a velocity netrates a crystal, it interacts with the atomic high enough to form a quasi-molecule. During this nuclei, core electrons and conduction electrons collision, the internuclear separation, R, varies (Lehmann, 1977). The projectile loses its initial from infinity to approximately zero and back to kinetic energy along its path by successive colli infinity. When R is large. the atoms do not inter sions and suffers many successive deflections from act and may be referred to as separated atoms. When its initial direction. It describes a trajectory R approaches zero, the electron shells see a sin which ends where the kinetic energy has dropped to gle nucleus called the united atom, the atomic practically zero and the particle comes to rest. number of which is the sum of the atomic numbers, This is true for electron, proton and heavier ion Z1 and Zz, of the colliding particles. Electronic projectiles. However, an electron has, in contrast transitions occur from an initially-filled to a heavy ion, a constant charge and a very small molecular orbital to another empty or only par mass. These features give rise to differences in tially-filled orbital if the approach of both nu the penetration behaviour. A fast electron loses clei is close enough to allow an interaction bet practically all of its energy in ionizing colli ween these two molecular orbitals. Such a process sions and only a fraction of its energy is lost has been extensively studied with gaseous targets. (e.g., 10-5 for electrons of 3 MeVin Cu) in elas Diabatic correlation diagrams, drawn using the one tic collisions with atomic nuclei. Energy loss of electron molecular orbital approximation, provide projectiles heavier than electrons is due to nu the variation of the energy of the one-electron clear collisions (i.e., elastic or quasi-elasTic states with the distance between the two nuclei. collisions with a target atom seen as a whole en By means of such diagrams, Fano and Lichten (1965) tity) and to electronic coll is ions (i.e., inelas were able to explain the inner-shell promotions tic collisions with the electrons of the atoms of observed in Ne+-Ne and Ar+-Ar collisions. The pro the solid and with conduction electrons in metals). motion of one or two L23 electrons in Ar+-Ar colli In the low velocity regime, nuclear collisions do sions (Fig. 3) is attributed to the diabatic 4f ou minate whereas at high velocities the stopping is molecular orbital (M.O.) which carries two 2p elec governed by electronic collisions. trons of Ar through a series of closely spaced M.O. This paper deals with the interactions between crossings in such a way, that, after the collision, proton and heavier ion projectiles (in the energy the electrons are likely to find themselves left range 1 keV to 200 keV) and solid targets. For a in highly excited orbitals of argon. So, one or proton, such kinetic energies correspond to velo two L23 vacancies appear in the separated argon cities of about 108 cm/s or more and the stopping ions after the collision. The Fano and Lichten ru power is essentially due to electronic collisions. les are extended for drawing diabatic M.O. corre For heavier iogs (10 < Z1 < 54) the velocities va lations to asymmetric systems (Barat and Lichten, ry from 6 x 10 to 108 cm/s. They are slowed down 1972). In the absence of extensive M.O. calculations by both nuclear and electronic collisions. These simple M.O. correlation diagrams have proven to be
352 Ion-induced Auger emission from solid valuable for the discussion of inner-shell electron promotion. In the simplest model, the atomic orbi tals of separated atoms are connected with straight lines to the atomic-orbitals of the united atom (Fig. 4). Although the diabatic correlation dia grams which were constructed by using the Barat Lichten rules are qualitatively consistent with 0.1 the available data, it should be noted that some 0.2 crossings not shown on one-electron correlation diagrams may well be found to exist when more exact computer simulations become available orwhen N-electron total potential energy curves are used. 1.0c:.::::::__ __ ..c:..::--=--+-1 2
:j 0 w
500 ls 1sG lO00L__L___J'---'---'--'--..,___.__..____, .01 .02 .05 .1 .2 .5 10 2 5 10 Kr R(a.u)- Ar+ Ar
Fig. 3 - Diabatic correlation diagram Ar.,.-Ar (from L1chten, 1967).
Xe'lOkeV E 10 20 30 40 50 70 80 E(e.V)
Fia. 1 - Al Lz3 Auger electron distribution obtai ned from a polycrystalline Al target bombarded by different ions (from Benazeth et al., 1978). 45 3d
x103 10
8 Zp Ar• Zs Ar• 6 Fig. 4 - Diabatic M.0. diagram Ar+-Al (from Barat Fig. 2 - Al Lz3 Auger electron distribution ob tained in 50 keV Al+-Ar gas collisions (from Dahl anc!LTchten, 1972). et al . , 1976). 353 J. Mischler, N. Benazeth When the target is a solid, the M.O. model is 1984). So it can be seen that the decay occurring valid if the collisions between incident and lat inside the solid can contribute to Auger emission tice atoms can be treated as isolated, two-body only if the electron is ejected from a distance events and if only the electronic inner-shells z less than two or three times the mean free path are considered so that the electronic states can Ainel . With the two assumptions : (i) excitation be localized. Moreover, in a solid, the projecti is isotropic, and (ii) the inner-shell vacancy le can yield an important part of its kinetic cross section a is constant over a distance z of energy to a target atom (see previous section) a few electron mean free paths, a very simple which in turn can behave as a projectile. So, one expression of the number of electrons, N(e),ejec important problem is the determination of the re ted at the polar angle e and the total number of lative efficiency of the asymmetric and symmetric electrons, 0~, can be obtained : collisions according to the ion and target para Ne( ) dn = -,r::-no A- case dn = N case dn ( 2) meters (e.g., energy, mass, incidence angle of 41T me 1 . 0 the ion, nature of the target). and : Ji,;i=;nxoxA. l (3) Auger decay process (Chattarji, 1976). 4 me . When the particle is ionized 1n an inner-shell W, n is the atomic concentration. the system tends to reorganize itself so as to at From polycrystalline samples, the flux atte tain a state of lower energy through either of two nuation, due to inelastic collisions, is the only processes. First a transition may occur in which transport effect which is observed. the inner-shell is filled by an electron from an From single-crystal targets, the analysed outer shell Y, the excess energy appearing as ra- electron wave lengths (5 to 2 ft- for electrons~1ith diation (X-ray). In the second process, the inner an energy between 10 and 100 eV), being of the shell vacancy is still filled by an electron from same order of magnitude as the crystal lattice para level Y, but the released energy is imparted by meters, multiple scattering effects (coherent elas Coulomb interaction to another electron of the sa tic process) also have to be taken into account. me atom belonging to a shell Z which is ejected But some experimental studies on the azimuthal with a kinetic energy of about : angular distributions of electron or ion-induced Ewvz = Ew - Ey - E'z (1 l Al L Auger electrons showed that this phenomenon where E'zis the ionization potential for shell Z leads to small variations in the emitted Auger with respect to an atom which is already ionized electron intensity (Mischler et al., 1979 - Negre in shell W. Such a radiationless emission is known et al., 1985). as the Auger effect and the corresponding Auger Computer simulations transition is labelled as WYZ(Fig. 5). - Calculations of the origin, and decay of Ar+- In a solid target, if an inner-shell vacancy bombardment-induced 2p inner-shell excitation in W is created on the first bound electronic level, Al and Si were first performed by Andreadis et al. the two electrons which participate in the Auger (1983). The Monte Carlo code EVOLVEwas used to si transition belong to the valence band and the tran mulate both the excitation and decay processes in sition is termed WVV.Such transitions give rise volved. These calculations, based on the collisional to an Auger peak, the profile of which is roughly excitation analysis of Vrakking and Kroes (1979) the convolution square of the density of states assume that at low ionic energies the 2p excitations (for example, Al L23 VVobserved under electron only arise from symmetrical collisions. or proton bombardment). Moreover if the projectile Recently, computer simulation results were is a heavy particle, the excited target atoms may obtained by Hou et al. (1986) and Benazeth et al. stay located in their normal lattice position or (1985) with the MARLOWEprogram (version 11) by be displaced from it during the collisions produ taking into account both symmetrical and asymme cing inner-shell vacancies. If the lifetime of the trical excitation collisions and also the surface inner-shell vacancy is high enough, some moving Auger neutralization process. In these calculations, excited atom will be able to decay outside the so the actual crystalline structure is taken into lid. So Auger decay will occur partly inside and account and the Moliere approximation to the Thomas partly outside the target. For Auger transitions Fermi potential is used with Firsov s.creening taking place outside the solid, the outer electro lengths. In aluminium, a 2p inner-shell ionization nic shells of the emitting atom are bound levels takes place when atomic collisions occur between and the width r of the corresponding Auger line is moving atoms involving a distance of closest ap related to the lifetime, of the vacancy by rela proach smaller than a threshold value, re, An ad tionship r =~/,.As the emitting atom is moving, ditional energy parameter, E;, is introduced to Auger lines are broadened by the Doppler effect. cut off trajectories of unexcited moving atoms Electron transport (Ganachaud, 1977). that are unable to produce a collision with an - For Auger electrons excited inside the target, apsis smaller than re, A Monte-Carlo method is used the transport effects are due to two processes : to describe Auger electron emission and escape . - inelastic collisions : electron-electron It uses two parameters : the inelastic electron mean collisions and collective excitations (surface and free path Ainel. for an elec~ron in a~uminium and bulk plasmons) undergone by an electron before it the lifetime, of a 2p hole man Al 10n. In order reaches the surface ; to localize the Auger decay, a clock is attached - elastic processes (coherent scattering). to each ionized atom, starting when ionization Generally, only inelastic events are taken takes place. The probability of an ion decaying into account. This is done by the introd~ction of along its flight segment between collisions i and a mean free path A which is about 2 - 5 A for an i + 1 is given by : e 1ectron energy between 10 and 200 eV (Powell , 354 Ion-induced Auger emission from solid been demonstrated to be the only ones val id. There E fore a systematic study of their influence is still required. Experimental techniques Electron spectroscopy utilizing ionic bombard Ez Ey ment requires the association of both a particle accelerator and a system for the energetic and/or angular analysis of secondary electrons. A wide variety of ion guns are used, from low energy (< 10 keV) up to Van de Graaf accelerators which supply energies up to several MeV. We shall not develop this part of the experimental arrangement and shall only describe the main electron spectro meters used for surface analysis. Essentially two families can be distinguished, electrostatic or Fig. 5 - Schematic energy representation of WYZ magnetic deflection spectrometers and retarding 7\ugerdecay. A deep hole of energy Ew is filled potential spectrometers(Roy and Carette, 1977). by an electron coming from a higher state of ener Deflection spectrometers gy Ey (or Ez), the excess energy thus released Whenelectron spectroscopy appeared, the first is used to excite an electron from Ez (or Ey) to instruments used were magnetic devices. However, E. magnetic instruments are now used much less since various types of electrostatic spectrometers which Pem = exp (- ti / T) - exp (- ti+1 / T) (4) are easier to construct and to use have been de The probability of the electron reaching the veloped. They can be easily protected from stray electronic surface is assumed to be dependent only magnetic fields by mumetal shields. So, we will on the inelastic mean free path Ainel. only present some electrostatic deflection spec Pex=exp(-o/Ainel.l (5) trometers. where o is the distance along the emission direc 127° Cylindrical Deflector Analyzer (C.D.A.) tion between the site of decay and the electronic (Fig. 6) - The basic theory of this analyser was surface located at some distance, E, from the atomic surface. An electron reaching the surface will escape if it is able to overcome a surface energy barrier, Vs. ion /lectrostatic selector Ions may be ejected from the surface without beam having decayed in the bulk provided they overcome a surface binding energy EB. It is well known that an ion flying close to a metallic surface may be neutralized by an Auger process involving surface conduction electrons. This process is discussed by Hagstrum. The survi val probability for an ion at an infinite distan ce from the surface reference plane is given by : P(v_J_)= exp (-A/ av1_) (6) where VJ_ is the component of the ion velocity nor mal to the surface,A is the transition rate of an ion in the surface reference plane and a is re lated to the rate of fall off of the bulk wave Fig. 6 - Analysis system with 127° cylindrical function outside the solid. electrostatic spectrometer for angular-resolved The neutralization of sputtered ions by an energy distributions of electrons. L23VVprocess was estimated by equation (6). The value estimated by Garett et al. (1984) was used presented by Hughes and Rojansky (1929). The ca for the A/a ratio. libration equation which relates the pass energy, The parameters involved in target description, E0 , to the potential difference, 6V, between the the cascade model and the Auger emission have gene electrodes is . [ R _ rally accepted standard values. 21 1 E = e 6V 2 lg KJJ (7) From this model, for a given set of parame 0 ters, different results have been obtained : first, where R2 and R1 are the radii of the outer and the emission distribution from the effective elec inner eTectrodes respectively. The resolution is tronic surface both for electrons escaping from linked to geometrical parameters such as semi the bulk and for those emitted by free sputtered angular apertures and slit widths at the entrance atoms ; secondly, the emission yield by surface and exit. In practice, resolution is also limited Auger neutralization of excited sputtered atoms ; by the effects of electron reflections and space and thirdly the relative efficiency of the symme charge. The performance may be improved bv usinq trical and asymmetrical collisions. It must be grid electrodes, backed by a collector plates to noted however that the set of model parameters reduce these effects. Under these conditions in chosen to predict Auger emission yields has not combination with a low pass energy filter, a re- 355 J. Mischler, N. Benazeth solution of better than 0. 1 eV can be obtained. research workers are Ma, Al and Si which give, in One advantage of this analyser is its simple geo heavy ion collisions, an Auger spectrum with unex metry which makes its construction easy. A rectan pected features. Such typical spectra are shown in gular section beam acceptance is quite suitable Figs. 9,10, and 11. The three tar(]ets were bombarded for some applications. by different ions of several energies. The spec Spherical Deflector Analyzer (S.D.A.) - The tra exhibit a narrow main Auger peak (I) and, on basic principle of this analyzer was first inves the low energy side of this line, two weaker struc tigated by Purcell (1938 ) who described E by tures (II and III). In each case, the energy dis 0 tance between lines I and II is the same as the e6V (8) distance between II and III (4, 7 and 10 eV for '2,- ~ Mg, Al and Si respectively). On the high energy 1 R2 side of peak I, a less intense structure appears where 6V, R1, R2 are used as previously defined. (IV). Its amplitude strongly depends on the nature The basic resolution is similar to that of the and the energy of the incident ion. These sharp C.D.A. but its focusing properties are much better. peaks are labelled L23MMatomic-like lines [or Several kinds of S.D.A. are used : S.D.A. with (Lz3 )zMM]. They are superimposed on a broad struc 180°, 158°, 135°, 90° spherical radius which exhi ture called L21vv. bit different properties. The resolution of a A detailea analysis of the Auger spectra was S.D.A. can also be improved by pre-retardation of performed by first subtracting the continuum back electrons in a good electron optic system. With ground (the true secondary electron spectrum) and pre-retardation, values as low as 20 mVfor elec then by separating the contribution of the Lz3VV trons emitted at 2,000 eV have been obtained transition from that of Lz3MM(Fig. 10). (Kuyatt and Plummer, 1972). The evaluation of the continuum background Cylindrical Mirror Analyzer (Fig. 7) - This differs according to the workers. For example, apparatus 1s now the most widely used in the field Whaley and Thomas (1984) used an empirical formula of surface analysis. The full 360° symmetry around given by Sickafus (1977) although it is not suit the cylinder axis is utilized to improve transmis able for aluminium. We prefer to estimate the sion and sensitivity. The basic theory of this continuum background in the Auger energy range analyzer was first given by Zashkvar et al. ( 1966) using the following expression : and later by Sar-el ( 1967). But the general study of second order focusing was computed and tabula N(E) = ~ ak E exp [- ; EIJ (9) ted by Risley (1972). The commercially manufactu k=1 red analyzers combine a high sensitivity and a and to match it to the experimental background by narrow energy window. They offer good resolution : using six fitting points outside the Auger spec 6E / E ~ 0.7 % for a semi-angular aperture of 6°. trum (Benazeth et al., 1985). For all these electrostatic analyzers, the After subtracting the Auger spectrum, the derivative of the energy spectra can also be ob second step is the evaluation of the Lz3VVstruc tained by a modulation voltage added to the poten ture. Authors generally agree that the L23VVspec tial of one of the two electrodes. trum under ionic bombardment differs only slightly Hemispherical retardinq potential spectrometer from that obtained for the same target under pro Electron energy analysis may also be performed ton or electron irradiation. An objection can be by a retarding potential system 1vith electronic raised that under heavy ion impact only a and synchronous detection (Fig. 8). The basic prin part of the decay inside the solid occurs in par ciple of this spectrometer is that of a high pass ticles located on an atomic site ; the other part filter in which only particles having an ener originates from moving recoil atoms. But, even gy higher than the potential barrier, generally for fast recoils (E ~ 1000 eV), the velocity is applied between two grids in front of the collec much lower than the Fermi velocity. They should tor, can be detected. A small sinusoidal modula behave as stationary atoms whose valence electrons tion voltage is superimposed on the retarding po belong to the valence band of the solid. So, the tential. The tuning of the lock-in amplifier to Lz3VVcomponent of the ionic-induced Auger spec the frequency of the modulating voltage gives the trum should be identical to the Auger spectrum energy distribution N(E), while the tuning to the observed under electron impact. Indeed in this double frequency provides its derivative dN(E)/dE. case the Auger transitions only occur inside the This type of analyzer has high sensitivity since target. To model this L23vv structure, we record its geometry allows high transmission, but only an electron-induced Auger spectrum obtained in medium resolution. The main limitations of the re the same experimental set up, and we adjust its solution come (Taylor, 1969) from field penetra magnitude to fit the ion-induced Auger spectrum. tion and trajectory deviation at the grids. But After subtracting the L23VVspectrum, the a great advantage of this analyzer is that absolu atomic-like transitions are obtained. With a Mg te electron yields can be obtained directly. target, only four peaks can be seen but the Al and Si targets exhibit additional features (Viel Results et al., 1976 - Hou et al, 1986 - Whaley and Thonias, 1984). Auger emission due to the target atom The double character of these spectra (bulk Although Auger electrons characteristic of and vacuum) can be partly attributed to the life the target atom have been observed for many ele times (McGuire, 1971) of the inner-shell vacancies ments, compounds and alloys (Benazeth et al .,1978- (93, 13.4, 3.7 x 10-14 s for Mg, Al, Si) which Viel et al., 1976), the materials studied by most allow some atoms, ionized in the bulk, to escape 356 Ion-induced Auger emission from solid l N(E) b ,50 /; 'l. Fig. 7 - Cross section of a cylindrical mirror a analyzer in its general geometry. A full 360° sym metry around the axis is possible. Usually both the source O and image I are on the axis and 80 42.3° for second-order focusing (from Roy and ...... Carette, 1977). 0 20 40 60 Fig. 10 - a - Secondary electron energy distri bution obtained from 40 keV Ar+-Al collisions, --- true secondary electron distribution ; b - Al L23 Auger spectrum, L23MMspectrum, --- Cz3VVspectrum. dN(E) dE Fig. 8 - A retarding potential analyzer. 1 - Dia x30 phragm ; 2 - collector ; 3 - grids ; 4 - target ; 5 - control for polar incidence angle variation ; 6 - oven. V x100 x10 II x30 0 20 40 60 E (eV) 0 20 40 60 80 100 120 E(eV) Fig. 9 - Secondary electron energy distribution Fig. 11 - Secondary electron energy distribution 4 4 obtained in 40 keV Ar - Mg collisions. Mg Lz 3 observed in 40 keV Ar - Si collisions. Si L23 Auger spectrum can be seen between 30 - 60 eV. Auger spectrum can be seen between 40 - 110 eV. In upper part of the figure is the derivative form of the energy distribution. 357 J. Mischler, N. Benazeth from the solid before decaying. In heavier ele (obtained from different projectiles) can be sum ments, such as Ti, Ni, Cu, which give rise to ion marized in a universal diagram using the maximum induced M23 Auger spectra only exhibiting a bulk energy transfer between the projectile and the structure (M23VV),the lifetimes of the M23 vaca~ target atom for scaling. So, the main parameter cies are shorter:typically 10-15s (Mc Guire, 1972). would be the energy of the fast recoil atom rather Identification of the atomic-like lines - than the incident particle energy. The authors Various authors, cons1der1ng that the L23vv struc concluded that, in the explored energy range (5¥eV ture near the electron-induced Auger spectrum is 30 keV), the inner-shell excitation was mostly due well-known, first qualitatively studied the atanic to symmetric collisions (T-T). Another study like Auger spectrum. Each line was assigned to an (Hennequin et al., 1984) of alloys (Fe-Al) based atomic transition in comparing its energy posi on a theoretical work of Jayes (1972) proposed tion either to calculated energies from different that the Auger yields are proportional to the models (Whaley and Thomas, 1984; Metz et al., square of the concentration of the studied ele 1980) or to lines observed from gaseous collisions ment if only T-T collisions are efficient and pro (Dahl et al., 1976). All the lines are due to the portional to the concentration if asymmetric P-T decay of a single (I, II, Ill, V) or double (IV) collisions predominate . From the_ dependence 2p vacancy in the inner-shell of the target atom, of the intensity of the main Al Auger peak but correspond to different initial and final sta on the Al concentration studied for different tes of the emitting particle. Some calculated and ionic energies, it appears that at 5 keV only experimental Auger line energies determined by Al-Al collisions are efficient but when the different authors are given in Table l. Other Auger incident energy increases asymmetric collisions energy values can be found in the papers of Metz (P-T) play a larger role (cO % at 20 keV). Recen et al. (1980), Ferraris et al. (1986), Matthew tly, in our group, computer simulations on the ( 1983). relative efficiency of the two types of col The peak identification proposed for the lisions were undertaken for an Ar+-Al combina three elements is very similar. However the com tion in the ionic energy range 5 keV-70 keV. The plexity of the Auger spectrum increases from Mg first calculation showed that the results to Si due to the increase of the number of elec strongly depend on the threshold distances trons in the outer shells of the atom. The most of ionizing collisions for both symmetric intense line (I) is attributed to an s-p transi and asymmetric collisions. tion in which the 2p excited initial state is neu \i-Jidth and shape of the atomic_-like Auger tral. For Si and Al, line II is also assigned to peak - Since normal lifetime broadening r is an s-p transition but from an ionized particle. negligible (r = ~/T, where Tis l1fetime) the Since this line is much weaker than line I, it width of the experimental peaks is due to the may be experimentally deduced that 2p excited Doppler effect, instrumental broadening and pos sputtered particles are predominantly neutral. sibly overlapping of two or more Auger transi Line IV which originates from a double ionized 2p tions. From experimental results, a direct esti level is attributed to an s-p transition. The ad mate of the Doppler effect cannot be made. More ditional peak V situated on the high energy side over, a comparison between the widths observed of line I, which only appears on Al and Si spec by different authors is difficult, since the ex tra at 68 eV and 93 eV respectively (Whaley and perimental conditions (energy resolution, inci Thomas, 1984; Ferraris et al, 1986; Hou et al., dence angle, solid angle and direction in which 1986), listed by Whaley and Thomas (1984), as a the electrons are collected) are different. How p-p transition from a neutral excited initial st1r ever, some authors estimated the contribution of te cannot be observed in Mg spectra where the ex the Doppler effect by studying the dependence of cited atom has only one 3p electron. Since line V the width and the shape of the lines on the pro is weaker than line I, it could be deduced that jectile energy. These results show that the width the s-p transition probability is higher than the increases in the ion energy range 2 keV - 20 keV p-p one. But calculations of the probabilities of (Saiki and Tanaka, 1984; Wittmack, 1979b; Bara the different transitions remain to be done. The giola et al., 1982) and remains constant beyond other additional weak structures which appear in this value (Whaley and Thomas, 1984; Benazeth the spectra will not be detailed here. et al., 1986). It was also shown that the A second question which has been raised con broadening is asymmetric, and occurs towards the cerns the nature of the collision that produces high energy side of the peak (Baragiola et al., the 2p hole. Do Auger electrons come from colli 1982). As both the emitting particles and the sions between a projectile and a target atom Auger electrons are moving in all directions, it (asymmetric collision P-T) or between a fast re is more instructive to study the Doppler effect coiling ion and a target atom at its site : sym on angular-resolved Auger spectra and as a func metric collisions T-T? Most of the time both ty tion of the electron polar emission angle. In pes of collisions co-exist but their relative effi this domain, recent results were obtained by ciency depends on several collision parameters and Pepper et al. (1986). Under oblique incidence at in particular for a given ion-target combination, low ion energy (1 - 5 keV) in certain cases, the on the incident ion energy. From experimental re authors showed an unexpected shape of the main sults it seems that the contribution of T-T colli atomic-like Al line. For example, when electrons sions dominates at low ion energy. Let us mention are analyzed at grazing angle in an azimuthal an for example the experimental work of Wittmack (1979) gle opposite to that of the ion beam, the Al line and Hennequin et al. ( 1984). The former showed exhibits a split shape which is not observed at that the intensities of the main Si Auger line normal incidence. This experimental observation 358 Ion-Induced Auger emission from solid TRANSIT IONS ------CALCULATEDEXPERIMENTAL PEAKS from to VALUES(eV) VALUES(eV) Mgo 2p5 3s2 3p --,. 2p6 3s sp 44.4 (a) 44.5 (c) 43.0 (b) I 2 6 Mgo 2p5 3s 3p -+ 2p 3s ( 2s) sp 41. 18 (b) 44.5 (f) 3 Mg+ 2p5 (3s 3p P) --+ 2p6 sp 38. 6 (a) 40.5 (c) Mg+ 2p5 (3s 3p 1p) -+ 2p6 sp 41.4 (a) I I (b) 2 6 39.0 Mg Mgo 2p5 3s 3p -+ 2p 3p (2P) ss 36.75 (b) Mg+ 2p5 3s2 --,. 2p6 ss 35. 1 (a) 36 (c) I I I Mg+ 2p5 3s2 --,. 2p6 ( 1S) ss 32. 13 (b) 34.0 (b) 2 Mg+ 2p4 ( 10) 3s 3p -+ 2p5 3s sp 53.9 (a) 55.5 (c) 2 54 (b) IV Mg+ 2p4 3s 3p -+ 2p5 3s Sp 56 (b) 55.5 (f) Al 0 2p5 3s2 3p2 2p6 3s 3p sp 63.6 (a) 63.5 (c) 62.2(b) - I 0 2 2 Al 2p5 3s 3p -+ 2p6 3s 3p (3P) sp 61. 74 (b) 63.5 (f) 62.9(g) 2 Al+ 2p5 3s 3p -+ 2p6 3s sp 57.8 (a) 56.5 (c) 0 2 2 Al 2p5 3s 3p -+ 2p6 3p2 (3P) ss 54.71 (b) I I 56.2 (b) Al+ 2p5 3s2 3p --,. 2p6 3s ( 2s) sp 57.7 (b) 2 6 Al Al+ 2p5 3s 3p -+ 2p 3p ss 51. 13 (a) 49.5 (c) ') I I I Al+ 2p5 3s'- 3p -+ 2p6 3p (2P) ss 49.0 (b) 49.0 (b) 1 Al+ 2p4 ( o) 3s2 3p2 --+ 2p5 3s 3p sp 73.7 (a) 76.0 (c) 76 (b) IV 2 2 5 Al+ 2p4 3s 3p -+ 2p 3s 3p (3P) Sp 75.3 (b) 75.5 (f) 0 2 2 Al 2p5 3s 3p2 -+ 2p6 3s ( 1s) pp 66.37 (b) 68 (c) 67.8(b) V 2 85 (c) 86 (b) Sia 5 3 --+ 2p6 3p2 (20) 2p 3s 3p 3s Sp 83.30 (b) 85.0 (d) 85.5(e) I Si+ 5 2 2 6 3 75 (c) 76.0(b) 2p 3s 3p -+ 2p 3s 3p ( P) sp 75.47 (b) 75.0 ( d) 75 (e) I I 5 2 2 65 (c) 66 (b) Si Si+ -+ 2p6 3p2 (3P) 2p 3s 3p ss 65.34 (b) 65.1 (d) 65 (e) I I I 4 2 3 5 107.D (b) 105 ( c) Si+ 2p 3s 3p ---+ 2p 3s 3p2 sp 103.0 (b) IV 106 (f) 107 (e) 5 2 3 6 2 Sia 2p 3s 3p --+ 2p 3s 3p (2P) pp 9 3. 15 (b) 93 .0 (b) 92.4(d) V Table 1 - Calculated and experimental energies of L23MMpeaks for Mg, Al and Si. (a) Dahl et al .(1976) (b) Whaley and Thomas (1984) (c) Viel et al. (1976) (d) Saiki and Tanaka (1982) (e) Wittmack(1979a) (f) Hennequin et al. (1984) (g) Baragiola (1982). 359 J. Mischler, N. Benazeth 15 was attributed to the Doppler effect resulting from the lifetime of the 2p hole is shorter (5 x io- s a flux of high energy excited atoms moving in this for Al) than the standard values (1.34 x 10-1_ s direction. on free atoms : Mc Guire, 1971 ; 2 x 10-14 s in In our group, we are now performing a study sol id : Citrin et al., 1979). of the Doppler effect, observed on the angular re Recently, absolute L23vv and L23MMAuger yields solved Auger spectra, in bombarding a polycrystal were directly measured from the experimental Auger line Al target, under normal incidence, by 40 keV spectra obtained in Ar+-Al collisions with a retar Ar+ ions. The experimental set up is derived from ding potential spectrometer (Fig. 8) where all the the hemispherical retarding field energy analyzer emitted electrons were analysed. The dependence of (Mischler et al., 1986) : the collector is cut in the Auger yields on both incidence angle (0 - 60°) to 27 receiving elements isolated from each other. and ionic energy (5 - 60 keV) was studied for po Each element collects the electrons emitted with lycrystal and single crystal Al targets (Hou et al., a polar angle e for all azimuthal angles¢. By com 1986). Comparison with computer simulation results paring the Auger spectra obtained on each part of from the MARLOWEprogram described above gave the collector, it is shown that the atomic-like line reasonable agreement (Fig. 13). In this cal remains symmetric but moves towards high energy culation, both symmetrical and asymmetrical col when the emission polar angle (relative to the nor lisions and the Auger surface neutralization pro mal at the surface) decreases. A shift of 0.7 eV cess are taken into account. Furthermore, a stan is observed when e varies from 80° to 0° (Fig. dard value for the 2p hole lifetime was used. The;e 12). Furthermore, the width increases with the po results indicate that L23VVand L23MMAuger yields lar angle: 1.1 eV at normal incidence and 1.5 eV increase more rapidly with the incidence angle than fore= 80°. These results may be explained by the the "true" secondary electron yield. From single Doppler effect. Under normal incidence, the angu crystals ion channeling and shadowing effects were lar distributions of the emitting atoms are symme also observed. tric relative to the direction of incidence. If it Angular distributions : azimuthal profiles - is assumed that all these atoms move with a mean In our research group, a study of angular distri velocity vi, the electrons emitted perpendicularly butions of Auger yield has been ongoing for some to the surface will have a velocity, in the labora years (Mischler et al., 1979 - Mischler et al, tory frame, between v0 (velocity of the electrons 1983 - Negre et al., 1985) . In order to investigate in the reference frame) when they a re emitted from single crystal effects, the azimuthal profiles of atoms moving parallel to the surface and v0 + Vi Al Lz3VVand L23MMelectrons emitted from (100), when they are emitted from atoms moving normally to (110), (111) crystalline surfaces under normal in the surface. The electrons ejected parallel to the cidence ion bombardment have been studied. surface will have a velocity between v0 - Vi and The analysis system was a 127° electrostatic v0 + Vi when the emitting atom moves in the opposi selector, with an aperture of 2° (Fig. 6). The azi te or the same direction as the analysed electron muthal distributions from polycrystal targets are in respectively. So, in this case, the Doppler effect dependent of the azimuthal angle¢ ; but from sin does not affect the position of the peak. gle crystals, they exhibit a weak modulation (less Calculations of the shape and the shift of than 10 %) which has the periodicity of the crys the Auger peak induced by the Doppler effect were tal lattice and is dependent on the ion beam para carried out for comparison with experimental re meters and electron energy (Fig. 14). sults observed in angular-resolved Auger spectra. An experimental study of the azimuthal angu For these estimations we need to know both the lar distribution N(¢), as a function of the energy energy and the angular distributions N(E*, et) of and nature of the incident ions (Mischler et al., emitting sputtered particles. We assume that this 1983) and of the emission polar angle e showed tllat function is the product of a function of the ener two phenomena participate in the observed anisotro gy, EZ, and a function of the polar angle, er. The py : transport effect and preferential emission pro energy distributions were obtained by computer si cess. mulation cascades (Benazeth et al., 1985), but so For_xe+_and_Kr+_ions (M2 / M1 > 1 ).Whatever far, the angular distribution cannot be reasonably the polar emission angle, the shape of the Lz3Vv estimated because of the statistical uncertainties azimuthal distribution curves agree reasonably with of the computer results. Since the target is bom profiles calculated from a model (Negre et al., 1985) barded normally, a cosine function was chosen for itself derived from a photo-emission model (Abraham the angular distributions. From these calculations Ibrahim et al., 1978) and taking into account both the energy shift of the peak was 0.5 eV when e va Auger decay and electron transport (that is to say ried from 10° to 90° and the width was 0.5 and 1. 1 multiple scattering effects owing to the wave length eV at normal and grazing ejection respectively. of the Auger electrons). Anisotropy rates of a few This is less than the measured widths in which ex percent were observed (Fig. 14). On the other hand, perimental broadening intervenes. These results the L23MMAuger electrons although emitted from ex can be seen in Fig. 12. cited sputtered atoms, exhibit a non zero anisotro Auger yields - The dependence of Si (or Al) py. However, it must be noted that this anisotropy atomic-like L23MMand band-like L23VVAuger yields (less than or equal to 1 %) is much lower than the on the angle of incidence of the Ar+ beam was exa L vv one ; it could be due to an Auger decay of mined by Andreadis et al. (1983) for low ion energy part23 of the electrons outside the solid but in the (<5 keV). They showed that their experimental mea vicinity of the surface where the effects of the surements were in reasonable agreement with the va crystal still exist. lues obtained by cascade computer simulations assu For_Ar+_and_Ne+_ions (M2 / M1 > 1). When the ming that : (il the 2p excitation arises only from polar emission angle is far from a close-packed di symmetrical co1lisions Si-Si (or Al-Al), ana (ii) rection, the observed anisotropy is, for the Lz3VV 360 Ion-induced Auger emission from solid E (eV) 0 0 0 63,5t=------a 63 llE(eV) e b 0 0 coo o o o th. Auger 0 0 0 60° 80° s Fig. 12 - Dependence of : (a) energy position E, (6) width l'IE, of the mean Lz:iMMAuger peak, on ejec tion polar angle e. Circle : experimental values. Solid line : calculated curves.- The Al target was bombarded normally by 40 keV Ar•. 3 EXPERIMENT o SIMULRTIDN 2. 5 40keV Ar• /Al (poly) • lil D 4.5% 1 0 D Cl' D D D 0 S 'IJ D ()° 20" 40° 60" ~ Fig. 14 - Azimuthal profiles of 66 eV L VV 0 10 20 30 40 so 60 Auger electrons obtained from a (111) Al 2~arget INCIDENCE ANGLE (DEG I bombarded by Xe+, Kr•, Ar• and Ne•. Calculated curves from a model taking into account both Auger Fig. 13 - Calculated dependence of Al L23 Auger decay and transport (Th. Auger) and only transport yields pMMon the incident beam angle, for poly (Th. cont.) (8 = 40°). crystalline Al target bombarded by 40 keV Ar• ions. electrons, in agreement with the proposed trans sion of electrons occurs from fast excited atoms port mo~el. But, when the polar emission angle is sputtered along the close-packed directions. It near a [110] close-packed direction (situated at was shown (Senger, private communication) that in cj> = 0° in Fig. 14), maxima of the emitted current high velocity collisions, electrons are preferen are observed on both Lz3VV, and Lz3MMangular pro tially emitted in the direction of the emitting files. The anisotropy rates (a few percent) are particle. But, so far, we have not been able to of the same order of magnitude for the two types justify this hypothesis through lack of data on of electrons. Furthermore it was shown that the excited sputtered atoms (in particular angular and anisotropy increased as the ion energy decreased. energy distributions). Since this effect also appears when the Auger Polar angular distributions : topography ef decay occurs in the vacuum, it could be due, not fects - It has previously been shown that the to a transport phenomenon, but to an anisotropic polar angular distributions, N(e), expected for excitation process. For example, preferential emis- electrons isotropically excited, follow a cosine 361 J. Mischler, N. Benazeth ~ '~\.,~;i ~~ti,.i.,~~~"~' 10~ I Fig. 15 - SEMmicrograph of polycrystalline Al Fig. 16 - SEMmicrograph of (111) Al bombarded bombarded with 25 keV Xe+ ions to a total dose of w1!~ 25 keV xe+ ions to a total dose of 5.1018 5.1018 cm-2. cm . e 80° 60° 40° 80° 60° 40° :::j ::, cu ro a iD iD V, V, 0 0 ~ ~ ::,;: > z1------~==,------==-----c=-~..;;:::::c'----=- z> z 0 .5 0 case Fig. 17 - Experimental polar angular distribution 80° 60° 20°0° NyvTcos 0) of 63.5 eV Lz3Vv electrons ; compari ::, son with calculated curve (solid line). ro b 362 Ion-induced Auger emission from solid topographies essentially composed of repetitive other hand, at high values of 8, the number of elec depressed and faceted structures, in agreement trons is decreased because a large part of emitted with many works on the modification of the surfa electrons are blocked by the relief. So, the in ce by ion irradiation (Chadderton, 1979 - Kelly fluence of topography on the number of analysed and Auciello, 1980 - Carter et al., 1983 - Auciello L23MMelectrons is important when the electrons are 1984 - Kelly, 1984 - Carter et al., 1985). An ex collected only in an angular sector. It would be ample of intergranular effects, resulting from less when the all-8 electrons are taken into accrunt the presence or different crystallites, is shown because the increase for the low values of 8 is in Fig. 15. The variation in sputtering yield for partly balanced by the decrease for high values of different crystal orientations gives rise to the 8. creation of grains of various elevations across Such experimental studies will be performed at the surface, and the development of cliffs at grain oblique incidence because calculations we recently boundaries. An example of intragranular features, performed have shown that, in this case, the yields which are structures occurring on a single crystal- and angular distributions of electrons emitted in 1 ite in polycrystalline target or on a single crys side the target are also modified. tal, can be seen in Fig. 16. In this figure, ob Auger emission from the projectile tained from a (111) Al single crystal surface, re When the pr0Ject1le has a complex electronic petitive features composed of three equivalent structure it may also be ionized in its inner elec slightly tilted systems of facets, situated at tronic shel 1s during asymmetric coll is ions between 120° to each-other are observed : this symmetrical projectile and target atom and/or symmetric colli arrangement is compatible with the symmetry of this sions between incident and previously implanted face. projectile. Most of the time, it is difficult to Some models were set up to schematize the determine the relative efficiency of symmetric and main surface structures observed such as bowls, asymmetric collisions. Indeed threshold energies ripples and terraces. From these models, yields and cross-sections for inner-shell vacancy forma and angular distributions of electrons emitted in tion are different in these two cases. Moreover, side and outside such structured surfaces were cal other numerous parameters intervene such as projec culated. To calculate the number of analysed elec tile implantation in the first layers of the solid, trons in these two cases we took into account the cross section of the inner-shell excited projectile influence of the local incidence and emission an backscattering, and the variation of these physical qles (i and a respectively) different from the cor features with incident angle and ionic energy. So, responding experimental angles (Band 8), and the only the experimental results obtained from the blocking of a part of emitted electrons by the re same ion-target combination with an identical in lief. cident angle and incident energy should be compared. Two mains assumptions were made in our calcu Results were obtained for Ar (Legg et al., lations. First, it was assumed that the electrons 1980: Wittmack,1979a). Ne (Benazeth et al., 1980; generated inside or outside the crystal come from Zampieri and Baragiola, 1984; Pepper and Aron, the vicinity of ion impact. Second the electron 1986; Pepper, 1986) and Na (Benazeth et al., 1980) source was assumed isotropic. in collisions of their ionized gases with different Under normal incidence, experimental distri solid specimens. butions obtained for L23VVelectrons, whatever the The Ar L Auger spectrum was observed by dif irradiation conditions, turn out to closely follow ferent authors from Ar+ bombarded Mg, Al, Si, Ti a cosine law which is that expected from a flat and Fe targets. In all these experiments, a main surface (Fig. 17). Nevertheless, the total number Ar L Auger peak appeared at about 215 eV (the ener of L23vv electrons emitted is higher from a textu gy position varied from 211 to 217 eV according to red surface than from a flat one. Such results are the workers) assigned to L23VVand L23M23M23 tran in good agreement with angular distributions and sitions. This spectrum also exhibits a structure yields calculated from models for electrons excited at lower energy (about 195 eV)sometimes interpre inside the crystal. ted as bulk plasmon energy loss (Viaris de Lesegno Figs. 18 a and b present, for the same expe and Hennequin, 1979). Moreover, from Si and Ti an rimental conditions, polar angular distributions additional peak at 230 eV attributed to a transi of L23MMAuger electrons emitted under Ar+ and xe+ tion from a double ionized Ar L23 shell was obser bombardment. With argon ions, for which the surfa ved. The interpretation of the Ar L Auger spectrum ce is only weakly perturbed, the N(cos 8) curve cor depends on the nature of the target atom. In Ar+- responds to an isotropic distribution law, expec Ti (or Fe) and Ar+- Si coll is ions, most of the Ar L ted for electrons generated outside the crystal Auger spectrum was attributed to asymmetric colli from a flat surface. Similar results were publi sions. This interpretation is based on several ar shed by Saiki et al. (1984). With xe+ ions, for guments. First, experimentally, the increase in the which the surface topography is strongly modified Ar L line width with an increasing ionic energy was by ion irradiation, the experimental angular dis observed. The broadening is consistent with Doppler tribution rapidly decreases as the value of the broadening during Auger emission from a fast par polar angle 8 increases. This result is in reaso ticle (Fig. 19). Moreover, Viaris de Lesegno et. al. nable agreement with the curve calculated in the (1979) showed that the abrupt increase in the Ar L case of relief in the form of various tilted ter Auger intensity above 6 keV is of the same type as races for electrons emitted outside the solid. More the variation in the cross section for Ar 2p vacancy precisely it appears, from the model, that at 8 = 0~ creation in Ar+ - Ti collisions. These authors also the number of electrons collected from a structured suggested that in such collisions the argon L Auger surface can be increased by about 100 % compared to emission is mostly due to asymmetric Ar+ ➔ Ti col the number collected from a flat surface. On the lisions where one or two vacancies are created in 363 J. Mischler, N. Benazeth the 2p level of the argon atom because after an assigned to 2p5 3s2 + 2p6 transition and three asymmetric collision the vacancies are preferably satellites of lower amplitude centred at 28 eV, created in the lighter partner (Fano and Lichten, 32 ev and 36. 7 eV. In the ionic energy range 6 keV - 1965). The latter argument is not concrete eviden 60 keV, the main peak width remains constant. Such ce. In effect, from a calculation of M.O., in the a sharp line is inconsistent with an Auger emis near symmetric collision system Si + Ar, Schneider sion from a fast particle. So this NaL23 Auger et al. ( 1983) showed that two different mechanisms spectrum was attributed to an emission from an im give rise to L shell excitation. One contributes planted Na atom at the surface excited in a symme to L shell excitation in the heavier colli tric collision with an incident Na ion. But, a mo sion partner and the other in the lighter del similar to that for Ne Auger emission did not partner. interpret the spectral dependences of the Auger From Ar+- Mgand Ar+- Al collisions, the pro yield on different collision parameters (incidence bability of creating a 2p hole in Ar from asymme angle, ionic energy,target temperature). Such a tric collision is very low. So, only the rare col study will be carried on in our group. lisions Ar+ implanted Ar can lead to 2p excitation of the argon. This explains the very low Ar Auger Conclusion intensity observed from such ion-target system and the disappearance of the high energy Ar Auger As we showed, ion-induced Auger emission from line: (L23)2 M23M23. In Ar - Ar collisions, the Mg, Al and Si exhibits both bulk (L23vv) and vacu two vacancies would be shared between the two col um (L23MM)components. liding atoms and practically no Ar atom would have Tne L23VVcomponent is considered by most 2p vacancies after such a collision (Viaris de Le authors to be identical to that excited by elec segno and Hennequin, 1979). tron (or proton) impact. However, this transition A Ne L23 Auger spectrum obtained in Ne+ - Mg might be slightly modified because part of the de and Ne+ - Al collisions was also studied (Benazeth cay occurs in moving atoms and in a perturbed area et al., 1980; Zampieri and Baragiola, 1984; (by ionic bombardment). Pepper and Aron, 1986 ; Pepper, 1986). This spec Although, at present, authors generally agree trum situated between 20 eV and 34 eV essentially with a decay from sputtered atoms to interpret the displays two peaks (at about 22 and 25 eV), the po atomic-like transitions L23MMand that the diffe sitions of which slightly differ according to the rent lines are assigned to Auger transitions (by authors (Fig. 20). They were assigned to : comparing the observed energies to predicted values), 1s2 2s2 2p4 (3P) [3s 3p (3P)]( 3P)( 3D) an Auger model does not yet exist which entirely 2 2 5 explains the composition of the spectra and the + 1s zs 2p (peak at 22 eV) line profiles. In particular the differences obser and 2 2 4 3 1 ved between the Auger spectra obtained from solid 1s 2s 2p ( P) f3s 4s (3s)] ( P) targets and those from gas phase, and the role of + 1s2 2s2 2p5 (peak at 25 eV). the field of the surface on the lineshape have to In the ionic range from 7 keV to 20 keV, be studied. Benazeth et al. ( 1980) showed that the width and To advance in this domain, computer simula the position of the structures were independent tions of collision cascades could be a useful tool. of the ionic energy. This seemed inconsistent with In particular, such calculations allow the deter an Auger emission from an incident ion because in mination of different parameters hardly available this case one should observe Doppler broadening. with the experimental measurements (for example So, in these collisions, they attributed most of energy and angular distribution of excited sput the Ne Auger emission to a decay on an implanted tered atoms ; relative efficiency of the asymme Ne. A model allowing the calculation of Auger yield tric and symmetric collisions in different ener from implanted particles as a function of energy gy ranges and for different ion-target atom com and incidence angle of the ion beam and of the binations).But to have confidence in these calcu target temperature was set up. This model takes lations, the different collision cascade parame into account the various parameters which govern ters have to be known without ambiguity. From the implantation profile in the target. Experimen MARLOWEcode, the crystalline effects can be stu tal results were in reasonable agreement with cal died, and recent calculations performed in our culated Auger yields. However, in a lower ionic group showed a significant anisotropy on the Al energy range (0.5 - 5 keV) and for angular resol L23 Auger yields excited by Ar+ impact, due to ved Auger emission, some authors (Zampieri and cnanneling and shadowing effects. Such a work will Baragiola, 1984; Pepper and Aron, 1986) attributed be developed. the Ne L23 Auger spectrum to the decay of back Experimentally only a few results were ob scatterea Ne ions, excited in asymmetric collisions tained from single crystal targets. They showed, Ne+- Al (or Mg). An increase of the peak widths in agreement with the computer simulations, the was observed when the energy increased from 0.5 influence of the crystal lattice on Auger yields. to 2 keV and then seemed to saturate. This spec Such an influence was also observed on the azimu tral dependence was interpreted in terms of Doppler thal angular distributions of bulk and vacuum shift given to the Auger velocity by the excited components of the Auger emission : a preferential atom ejected into the vacuum. ejection of the electron along the close-packed In collisions between Na+ and different solid directions appears on the angular profiles. This targets (Mg, Al, Be), Na L23 Auger spectrum was study must be carried on because, so far, only observed ( Benazeth et al . , 1930). This spectrum hypothesis can be put forward to explain this phe is composed of a narrow[full width at half maxi nomenon but a theoretical model remains to be set mum(F.W.H.M.) 0.7 eV] intense peak at 25.7 eV up. 364 Ion-induced Auger emission from solid of bulk and atomic decays that make them more com plex than electron-induced Auger spectra (Thomas, Ar•- Si(Ar) ) 6keV 1984). Secondly, in the general case, where both ,~ :r" (,.14) ,•' I I I symmetric and asymmetric collisions participate in the creation of inner-shell vacancies the Auger yield is a poorly-known function of the concentra I tion. Moreover, ionic bombardment poses certain problems (some of which also exist in SIMS)in par ticular the composition of the surface is modified 9keV by the ionic bombardment ; on the other hand, the (,.JS) Auger yields may be greatly modified by ion-indu ced surface topography ; lastly it is a destruc tive method. However !AES could be used for light elements which have high Auger yields and which, 150 200 250 150 200 250 3(X) on the contrary, are difficult to be detected by ELECTRON ENERGY (eVJ other techniques ; then, it would allow the direct determination of the depth profiles by the erosion Fig. 19 - Comparison of Ar L Auger spectra at dif of the surface by sputtering. ferent ionic energies (Wittmack, 1979a). References N(E dN(EI Ne Abraham-Ibrahim S, Caroli B, Caroli C, Roulet B. dE (1978). Deep-hole interferences in photon-induced X 25 Auger emission from solids. Phys. Rev. B 18, 6702- 6712. - Andreadis TD, Fine J, Matthew JAD. (1983). Auger electron emission from the decay of collisionally excited atoms sputtered from Al and Si. Nucl. Inst. Meth. 209/210, 495-502. Auciello 0. (1984). Historical overview of ion-in duced morphological modification of surfaces, in: Ion bombardment modification of surfaces.a. Auciel lo, R. Kelly (eds), Elsevier, 1-25. Banouni M, Mischler J, Negre M, Benazeth N. (1985). Surface topography effects on energy-resolved po lar angular distributions of ion-induced seconda ry electrons. Surf. Sci. l~i, L 720-L729. Baragiola RA. (1982). Principles and applications of ion-induced Auger electron emission from solids. Rad. Effects,§__!_, 47-72. Baragiola RA, Alonzo EV, Raiti HJL. (1982). Ion induced Auger-electron emission from aluminium. Phys. Rev. A..?..?_,1969-1976. Barat M, Lichten W. (1972). Extension of the elec 0 20 40 E(eV) tron-promotion model to asymmetric atomic colli sions. Phys. Rev. A~, 211-229. Benazeth C, Benazeth N, Viel L. (1978). Spectres Fig. 20 - Energy distributions of secondary elec Auger et sections efficaces d'ionisation de la trons emitted by 8 keV Ne+ - Mg. couche L23 du magnesium et de l 'aluminium sous im pact de protons et d'ions lourds (10-100 keV). New results concerning the influence of the Surf. Sci. 78, 625-647. ion-modified surface topography on the polar an gular distributions of the Auger electrons empha Benazeth N, Leonard C, Benazeth C, Viel L, Negre size the necessity to control and take into ac M. (1980). Emission d'electrons Auger caracteris count this topography in the analysis of ion-in tiques du neon et du sodium, observee lors de col duced Auger emission results. lisions de ces ions avec differentes cibles sol ides. Better resolution of spectrometers should Surf. Sci. '}]__,171-190. bring more information and give more confidence Benazeth C, Benazeth N, Hou M. (1985).Absolute Au to the interpretations proposed for some experi ger yields in Ar+-Al collisions : experiments and mental results (in particular Doppler effect). computer simulations. Surf. Sci . .!2_1_,L137-L143. Although the fundamental study of ion-indu ced Auger emission (!AES) is worth pursuing for Benazeth C, Hou M, Mayoral C, Benazeth N. (1986). several reasons, this spectrometry, according to Argon-induced Al Auger electron emission : struc most authors, does not appear, at this time, to ture effects and Doppler broadening. Symposiumon be a tool for surface analysis. First, Auger spec sputtering - June 86 - Spitz a.d. Donau-Vienna, tra induced by noble gas ion impact are a mixture Austria. Nucl. Inst. Meth. ~-, in press. 365 J. Mischler, N. Benazeth Carter G, Nobes MJ, Whitton JL. (1985). Sputtering in: Case studies in atomic physics. M.R.C. McDowell, induced topography development on f.c.c. metals. E.W. McDaniel (eds), North-Holland, Vol. 3, n°3, Appl. Phys. A38, 77-95. 139-213. Carter G, Narunsek B, Whitton JL. (1983). Heavy ion Kuyatt CE, Plummer EW. (1972). Field emission de sputtering induced surface topography development, flection energy analyser. Rev. Sci. Instr. 43, in: Topics in Applied Physics, Vol. 52. Springer 108-111. Verlag, Berlin Heidelberg, New-York, Tokyo, 231- Legg KO, Metz WA, Thomas EW. (1980). Auger spectra 269. induced by ion impact on metals. Nucl. Inst. Meth. Chadderton LT.(1979). The stability of cones and 170, 561-564. pyramids on sputtered surfaces of copper. Rad. Lehmann C. ( 1977). Interaction of radiation with Effects Lett., 43, 91-95. solids and elementary defect production, North Chattarji D. (1976). The theory of Auger transi Holland Publishing Company, S. Amelinckx, R.Gevers, tions. Acad. Press, London, 13-29. J. Nihoul (eds), 105-137 and 197-252. Citrin PH, Wertheim GK, Schluter M. (1979). One Lichten W. (1967). Molecular wave functions and electron and many-body effects in X-ray absorp inelastic atomic collisions. Phys. Rev. A~. tion and emission edges of Li, Na, Mg, and Al me 131-142. tals. Phys. Rev. B~_Q_,3067-3114. Matthew JAD. (1983). Auger spectroscopy of solid Dahl P, Rodbro M, Hermann G, Fastrup B, Rudd ME. surfaces : Electron versus ion excitation. Physi (1976). Auger spectroscopy on heavy-ion-atom col ca Scriptn, T6, 79-87. lisions II. Na+, Mg+, Al+, r+, s+, c1+-Ar colli Mc Guire EJ. ( 1971). Atomic L-shel l Coster-Kronig sions. J. Phys. 82_, 1581-1599. and radiative rates and fluorescence yields for Fano U, Lichten W. (1965). Interpretation of Ar+ Na-Th. Phys. Rev. A}_, 587-594. Ar collisions at 50 keV. Phys. Rev. Letters. 14, Mc Guire EJ. (1972). Atomic M-shell Coster-Kronig, 627-629. - Auger and radiative rates, and fluorescence yields De Ferrariis L, Grizzi 0, Zampieri GE, Alonso EV, for Ca-Th. Phys. Rev. A~, 1043-1047. Baragiola RA. (1986). Observation of a new atomic Metz WA,Legg KO, Thomas EW. (1980). Auger spec like peak in the ion-induced Auger spectrum of Si. tra induced by 100 keV Ar+ impact on Be, Al, and Surf. Sci.~, L175-L180. Si. J. Appl. Phys.~, 2888-2893. Ganachaud JP. (1977). Thesis. Nantes, France. Mischler J, Negre M, Benazeth N, Spanjaard D, Garcia JD, Fortner RJ, Kavanagh TM. (1973). Inner Gaubert G, Aberdam D. (1979). Experimental and shell vacancy production in ion-atom collisions. theoretical study of the angular resolved Auger Rev. Mod. Phys. 45 , 111-1 77. emission L23 of a single crystal target of alumi nium excited by ion impact. Surf. Sci. 82, 453- Garett RF, McDonald RJ, O'Connor J. (1984). A de 460. - termination of the ionization probability for alu minium secondary ion emission. Surf. Sci. 138, Mischler J, Negre M, Benazeth N. (1983). Evidence 432-448. - for electronic preferential ejection process in energy-resolved angular distributions of ion-in Hennequin JF, Inglebert RL, Viaris de Lesegno P. duced secondary electrons emitted from Al single (1984). Secondary ions and Auger electron emission crystals. Rad. Effects 2.Q_, 117-129. from Ar+-ion-sputtered Fe-Al Alloys. Surf. Sci. .1iQ_,197-206. Mischler J, Banouni M, Benazeth C, Negre M, Bena zeth N. (1986). Surface topography effects on ener Hou M, Benazeth C, Benazeth N, Mayoral C. ( 1986). gy-resolved polar angular distributions of elec Incidence angle and ion energy dependences of ab trons induced in heavy ion-Al collisions : expe solute Al L23 Auger yields in Ar+-Al collisions : riments and models. Rad. Effects, 87, 255-266. experiments and computer simulations. Nucl. Inst. and Meth. B..!l_,645-651. Negre M, Mischler ,l, Benazeth N. (1985). Azimuthal angular distributions of Xe+ and Kr+ ion-induced Hughes AL, Rojansky V. (1929). On the analysis of Al L23 Auger electrons. Surf. Sci. 2.~' 436-450. electronic velocities by electrostatic means. Phys. Rev. 34, 284-290. Pepper SV.(1986). Binary collision model for neon Auger spectra from neon ion bombardment of the Joyes P. (1972). Influence of asymmetrical corre aluminium surface. Surf. Sci. 169, 39-56. lations in the secondary emission of solid com pounds. J. Phys. C~; 2192-2199. Pepper SV, Aron PR. (1986). Angle-resolved Auger electron spectra induced by neon ion impact of Kelly R.(1984). The varieties of surface altera aluminium. Surf. Sci. ~~, 14-38. tion : structural, topographical, electronic and compositional, in: Ion bombardment modification of Powell CJ. (1984). Inelastic mean free paths and surfaces. 0 Auciello, R. Kelly (eds.), Elsevier, attenuation lengths of low-energy electrons in 79-126. solids. Scanning Electron Microsc. 1984 ; IV : 1649-1664. Kelly R, Auciello 0. (1980). On the origin of py ramids and cones on ion bombarded copper surfaces. Purcell EM. (1938). The focusing of charged parti Surf. Sci. J__QQ_,135-153. cles by a spherical condenser. Phys. Rev. 54, 818-826. Kessel QC, Fastrup B. (1973). The production of inner-shell vacancies in heavy ion-atom collisions, 366 Ion-induced Auqer emission from solid Risley JS. (1972). Design parameters for the cy Discussion with Reviewers lindrical mirror energy analyser. Rev. Sci. Instr. 43, 95-103. C. Boiziau : The comparison of the Figs. 9, Roy D, Carette JD. (1977). Design of electron 10 and 11 suggests that a correlation exists bet spectrometers for surface analysis, in: Electron ween the low energy electron emission and the in spectroscopy for surface analysis. H. Ibach (ed), tensity of the atomic-like lines. Is it exact? Springer-Verlag, NY, 13-58. And, if it is, is it possible to explain this phe nomenon? Saiki K, Tanaka S. (1982). Ion-induced Auger elec Authors : Collision events taking place near tron emission from Si surface. Japanese Journal of the surface are responsible for the emission of Applied Physics,~. L 529-L531. "true" secondary and atomic-1 ike Auger electrons. Saiki K, Tanaka S. (1984). Ion-excited Auger ~ec Only violent collisions, in the first layers of the solid between heavy particles (ion-atom or tron emission from Mg, Al and Si surfaces. Nucl. recoil atom-atom) can lead to L23MMAuger emission Instr. Meth. B2, 512-515. because first the 2p hole creation needs a distan Sar-El HZ. (1967). Cylindrical capacitor as an ce of closest approach smaller than a threshold analyser. I. Non relativistic part. Rev. Sci. value re of about 0.4 ~ and second the excited Instr. 38, 1210-1216. atoms have to pass through the surface before de Schneider D, Nolte G, Wille U, Stolterfoht N. caying. About the "true" secondary electrons, a direct transfer of the ion energy to the electrons (1983). Total cross sections for L-shell Auger of the solid and the collisions between fast elec electron production and vacancy production in slow trons and free target electrons can participate Si-Ar collisions. Phys. Rev. A~~. 161-165. in the excitation process as well .Some of these Sickafus EN. (1977). Linearized secondary-electron may come from target depth more than the inelastic cascades from the surface of metals.I. Clean sur mean free path if they were excited with an ener faces of homogeneous specimens. Phys. Rev. B]i, gy high enough to pass through the surface, after 1436-1447. undergoing several inelastic collisions. Snoek C, Geballe R, Van der WegWF, Rol PK, Bierman The secondary emission coefficient values y DJ. (1965). Structure in the electron energy spec of Al and Mg, obtained in the same ionic bombard trum from multi-keV Ar+-Au and Ar+-cu collisions. ment conditions, are comparable because the ato- Physica ll_, 1553-1556. mic masses not much different so leading to the sur face coll is ion events roughly identical. But the Taylor NJ. (1969). Resolution and sensitivity con Mg L23 MMAuger yield PMMis higher than the Al siderations of an Auger electron spectrometer ba one, although the 2p inner-shell ionization cross sed on display LEEDoptics.Rev. Sci. Instr. 40, sections are almost equal, because the 1ife time 792-804. - of the 2p hole in Mg is longer and so more exci- Thomas EW. (1984). Ion bombardment induced photon ted atoms will pass through the surface before de and Auger emission for surface analysis. Vacuum caying. 34, 1031-1044. On the other hand, any factor which increases the surface collision events such as the increase Viaris de Lesegno P, Hennequin JF. (1979). Ar+ and of the incidence angle or the ionic energy, will e- excited argon Auger electron transmission from lead to an increase of bothy and PMM·However, Mg, Al, Si, GaP and Ti surfaces. Surf. Sci. 80, some differences may be noted in the variation 656-662. laws of each type of electrons. For example, we Viel L, Benazeth C, Benazeth N. (1976).Spectres showed, in some experiments, that PMMyields in Auger sous bombardement d'ions Ar+ de 60 keV de crease more rapidly with the incidence angle than quelques elements purs et de composes binaires. y. For Auger electron, a complementary process Surf. Sci. 54, 635-646. occurs which explains this result : when the in cidence angle increases, the number of excited Vrakking JJ, Kroes A. (1979). Ion-induced Auger atoms displaced from their atomic site with a mo electron emission of Mg, Al and Si as a function mentum towards the surface is increased. of ion energy. Surf. Sci. 84, 153-163. Whaley R, Thomas EW. (1984). Auger spectra induced S.M. Durbin : It is clear from this review that by Ne+ and Ar+ impact on Mg, Al and Si. J. Appl. the fundamental interactions in IAES are not yet Phys. 56, 1505-1513. fully understood. You point out that IAES at pre sent may not be a useful probe of surfaces, but Wittmack K. (1979a). Projectile-energy dependence that there are materials for which IAES is well and line shape of Ar-L Auger spectra from argon suited. Under what type of conditions would IAES bombarded silicon. Phys. Lett. 74A, 197-200. be useful for materials characterization? Wittmack K. (1979b). Characteristics of ion-exci Authors : So far, ion-induced Auger spectros- ted silicon L-shell auger spectra. Surface Sci. copy 1s not of general value for surface analysis. 85, 69-76. However, used simultaneously with other methods it can probe the solid surface with different Zampieri GE, Baragiola RA. (1984). Ion-induced quality of information. IAES could be useful for Auger emission from solids : Correlation between characterization of elements with a high inner Auger energies and work functions. Phys. Rev. B 29 shell vacancy cross section and a low fluores 1480-1482. _, cence yield. These conditions are fullfilled for Zashkvar VV, Korsunskii MI, Kosmachev OS. (1966). weakly bound electronic shells of light elements. Magnetic spectrometer with spatially varying field. It can also be pointed out that ion-induced Auger Sov. Phys.- Techn. Phys. --1..!_,96-99. electron spectroscopy in combination with surface 367 J. Mischler, N. Benazeth channeling can be used to identify the surface atoms and to locate the position of both the ab- sorbate and substract atoms (Schuster and Varelas, 1983). Additional Reference Schuster M, Varelas C. (1983). Surface analysis by spectroscopy of Auger electron induced by sur face channeling ions. Surf. Sci . .!.l..'.!_,195-222. 368