University of Groningen. Hollow-atom probing of surfaces Limburg, Johannes

Transcription

1 University of Groningen Hollow-atom probing of surfaces Limburg, Johannes IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 1996 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Limburg, J. (1996). Hollow-atom probing of surfaces Groningen: s.n. Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date:

2 3 Experiment All experiments discussed in this thesis have been carried out between September 1992 and March 1996, at the KVI Atomic Physics facility in Groningen. This facility is located in the \atomic physics hall" (bronzaal) about 25 meters long and 8 meters wide. At the time of writing of this thesis (May-August 1996) ve experimental set-ups were connected to a central beamline fed by a new home built 14 GHz ECR ion source (ECRIS4). The ion-surface work presented in this thesis has been measured in Sir (an acronym for SURface PHYsics set-up), located at the end of the beamline (see gure 3.1). 3.1 ECR ion source In 1982, the KVI acquired its rst electron cyclotron resonance ion source (ECRIS1) from the CEN-Grenoble laboratory. This source { a copy of the MINIMAFIOS source developed by Geller and Jacquot 61 {was used for about ten years, both in conjunction with the old KVI cyclotron and as a source for the atomic physics set-ups. In 1990 a new source (ECRIS2) was designed and built for the atomic physics group at the KVI. This source stayed in operation until the spring of Both ECRIS1 and ECRIS2 were operated at 10 GHz radiofrequency (RF) power. The need for higher currents of multiply charged ions led to the development of a new, 14 GHz ECRIS3 source in This source is used as injector of heavy ions in the new superconducting AGOR cyclotron. A copy 21

3 22 Experiment Atlas Cheops Ecris 4 selection magnet quadrupoles quadrupoles 45 o 45 o 45 o 110 o selection selection magnet selection magnet post accelerator magnet Sirφ Agora Elcid Figure 3.1: Atomic physics hall (bronzaal). of this source, ECRIS4, was installed in the atomic physics source room in the summer of 1995 replacing ECRIS2. Most of the Auger spectra presented in this thesis have been obtained with beams from the old ECRIS2 source. Part of the work presented in chapters 9 and 11 has been performed using the new ECRIS4 source. The lay-out of ECRIS4 is shown in gure 3.2. ECRIS4 (and 3) were designed to ensure a high quality magnetic trap conguration 62 which is necessary for optimal operation (see the next section). In these sources, large mirror ratio's are obtained by using a B max ' 1T hexapole and coils having comparable magnetic elds on their axes. The rst ionization stage and biased disk, present in ECRIS2, 63 have been left out in the design of ECRIS3 and 4. The 14 GHz RF power is coaxially injected, together with the process gasses. Ions are extracted using a movable puller lens which can be put on a negative voltage to ensure proper focusing of low charge state or low energy beams. The main dierences between ECRIS4 and 3 are the shorter hexapole of ECRIS4 (260 vs. 190 mm respectively) and the smaller radius of the hexapole (73 mm vs. 65 mm respectively). Principle of ECR operation In an ECR source, highly charged ions are produced stepwise by electron impact ionization. During bombardment by electrons ions or atoms are subsequently ionized leading to charges q =1;2;3; :::;q max. The yield of highly charged ions is limited by aspects such as neutralization by electron

4 3.1 ECR ion source 23 Figure 3.2: Drawing of the ECRIS4 source. capture, the increasing amount of impact energy needed to ionize higher and higher and the connement time of the ions in the trap. Electron capture mainly stems from ion-neutral collisions and can be minimized by achieving low background pressure inside the source. Typical background pressures in the KVI sources are of the order of 10 5 Pa. To achieve the stepwise ionization needed to obtain highly charged ions, a wide range of electron energies is needed. To singly ionize an atom, an energy of about 1 a.u. is needed. But to strip the last electron o an ion of nuclear charge Z at least an energy of Z 2 =2 is needed. Moreover, the ionization cross sections peak around impact energies about 2 times the ionization energy. So, to obtain for instance fully stripped oxygen ions, electrons in the energy range 1 { 100 a.u. (' 25 { 2500 ev) are needed. In an ECR source, these electrons are produced inside a magnetic trap. This trap, a radial hexapole magnet and two axial magnet coils, consists of magnetic elds increasing in all directions. Electrons moving inside the trap will gyrate around the magnetic eld lines. The revolution frequency! c of an electron gyrating in a eld B is given by! c =2eB=m e. Moreover, electrons gyrating in the direction x of an increasing eld experience a repulsive magnetic force perpendicular to their radial axis. Depending on the and on the electron velocity v x, the movement can be reversed, eectively reecting the electrons 1. This way electrons are conned inside the trap and a sucient electron density can be obtained. High energy electrons are produced by applying a radiofrequency eld. When the RF frequency equals the revolution frequency! c \electron cyclotron resonance" occurs: the electrons are resonantly accelerated by the RF eld. By choosing a proper combination of magnetic eld strength (of the order of 1 T) and RF frequency (2.5 { 14 GHz), a closed ECR surface { along which the reso- 1 The reection coecient is determined by the mirror ratio Bmax=Bmin.

5 24 Experiment nance condition is fullled { can be obtained inside the vacuum chamber. Measurements of radiation produced in such a connement have shown the existence of electrons with energies up to 600 kev. 64 Ion extraction The ions produced in the ECR-plasma are extracted by putting the complete trap to high voltage. Ions reaching the extraction ange are accelerated towards the puller lens. This way, ions of virtually all charge states are extracted. A m=q selection magnet mounted behind the source is used to select and guide the desired charge state to the set-ups. Output After many optimizations, especially after mounting a biased disk, ECRIS2 typically produced some particle A ofhydrogenic ions. For O 7+ a maximum intensity ofabout9eawas achieved. The improved design of ECRIS3 resulted in a factor 10 increase in highly charged ion yields. This source produces typically 50 A ofo 7+ and 10 A O 8+ ( 62 ). At the moment of writing ECRIS4 is still subject to optimization. Up to now, the source has produced maximum currents of about 25-50% of those obtained with ECRIS3. The reason for this can probably be found in the very narrow and strong hexapole used in this source. This hexapole has strong fringing elds near its end which isvery close to the ion extraction ange. These elds probably have a strong defocussing eect on the exiting ions, especially for higher ion charges. At this moment, simulations of the extraction eld conguration are performed in order to get a grip on these problems. 3.2 Sir A photograph of the Sir set-up is shown in gure 3.3. The set-up consists of a 300 mm diameter UHV -metal collision chamber 2, a target holder, a deceleration lens system, an electrostatic analyzer (ESA) 3, and three timeof-ight tubes 4. The collision chamber is hooked up to the main beam line via a dierentially pumped line and a 45 switching magnet. The collision chamber is kept at a pressure less than Paby a 400 l/s triode getter pump and a titanium sublimation pump. The top and bottom anges of the chamber are magnetically shielded by 2 mm thick -metal, mounted inside 2 Manufactured by Vacuum Generators, Hastings (GB). 3 All manufacturedby the Lab. voor Technische Natuurkunde, University of Groningen 4 Manufactured at the KVI.

6 3.2 Sir 25 Figure 3.3: Side-view of the Sir set-up. The vacuumchamber, the beam line connecting the chamber to the 45 magnet and the three time of ight tubes can be recognized.

7 26 Experiment Figure 3.4: Schematic of the Sir set-up showing the entrance diaphraghms, the chopper and sweeper system, the deceleration lens system, the electrostatic analyzer (ESA), the sputter ion source and on of the three time of ight tubes. the chamber. The residual magnetic elds in the chamber are of the order of a few T Collimator and deceleration system Aschematic overview of the interior of the apparatus is shown in gure 3.4. The primary ion beam enters the apparatus via a set of diaphragms D 0, D 1, D 2 and D 3 with diameters given in the inset of the gure. The acceptance of the collimator is 5.5 mm mrad leading to an overall transmission of the ECR beam of about 0.5%. The overall transmission depends somewhat on the mass/charge ratio of the beam. For large M=q, the transmission tends to be somewhat worse. For the beams used in the experiments in this thesis, the currents entering the collision chamber are typically on the order of a few tens particle na. Up to diaphragm D 3, the beam line is on ground potential. In between diaphragms D 3 and D 4, the ion beam can be decelerated, in principle down to zero energy. This is achieved by connecting the complete set-up, including the getter and sublimation pumps and part of the electronics, directly to the ECR ion source potential. The actual deceleration takes place in a four element electrostatic lens system (designed by Siebe de Zwart 65 ), mounted between diaphragms 3 and 4. The lens system has almost 100% transmis-

8 3.2 Sir 27 sion for deceleration factors up to 100, while the beam spots on the target are only slightly larger than for the undecelerated beams. For even larger deceleration factors the beam diverges more strongly. The energy of the projectiles entering the collision chamber is now determined by applying an oset potential V to the set-up in series with the source voltage V ecr. This way the beam energy is completely determined by V and possible uctuations in V ecr are canceled. The energy of the beam E p is now simply given by its charge state times the sum of the oset voltage and the plasma potential inside the source: E p = q(v + V plasma ) (3.1) The energy oset arising from the plasma potential can be measured in two ways: directly by the electrostatic analyzer at 0 with respect to the beam and indirectly by measuring the target current versus positive target bias voltage. Both methods have some drawbacks. The electrostatic analyzer is equipped with a channeltron to detect ions and electrons. To avoid overload of the channeltron during a direct measurement of the beam energy, the intensity of the beam has to be reduced drastically 5. This can be done either by reducing the amount ofrfpower applied to the source or by detuning the beam line quadrupole magnets. Either way a sample having dierent properties than the normal beam may be selected to measure the plasma potential. In the second method, the determination of the plasma potential by measuring the target current while biasing the target, the impinging beam may strongly be defocussed. However, for the \old" ECRIS2 source a plasma potential of about 15 2Vwas found, while for the new ECRIS4 potentials between V were measured. All projectile energies listed in this thesis have been corrected for the appropriate plasma potential. The (undecelerated) beam current is measured in the Faraday cup labeled FC1 (see gure 3.5). Faraday cup FC2 registers the ion current after deceleration by the lens system. The target manipulator oers four degrees of freedom for target movement. The incident angle can eectively be varied between 0 and 90 and the target can be rotated over an angle between 180 and Moreover, the tilt angle between target normal and the rotation axis can be changed and the target can be moved horizontally into or out of the beam. This movement, the sway, is actually a rotation around an axis 257 mm above the target center.

9 28 Experiment Figure 3.5: Lens system, target manipulator and ESA. The manipulator has four degrees of freedom labeled tilt,, and sway. The beam intensity can be measured by Faraday cups FC1 and FC2. The ESA observation angle can be varied between 0 and Electrostatic analyzer All electron and ion energy spectra presented in this thesis are measured using the 180 spherical electrostatic analyzer sketched in gure 3.5. Since in the works of De Zwart 65 and Folkerts 66 a thorough description of the ESA is given, here only the main features of the spectrometer will be discussed. The ESA observation angle ranges between 0 and 140 with respect to the beam. The radius of the central trajectory of the ESA is 50 mm, the inner and outer spheres have radii of r 1 = 48 and r 2 = 52 mm respectively. The exit slit measures 0:5 1:9 mm 2. The entrance diaphragms D 5 =1:9 mm and D 6 =0:4mm diameter yield an acceptance angle of 2: sr at the center of the target. The energy resolution of the ESA is 0.5% FWHM, therefore the total acceptance of the analyzer is 1: E (sr ev). The uncertainty in the acceptance is estimated to be 15%. Electrons having energies higher than 20 ev can be measured with nearly 100% eciency. For energies lower than 20 ev, the residual magnetic eld distorts the transmission leading to a loss of detection eciency. 5 Typically intensities of a few pa are used.

10 3.2 Sir 29 Pass energy In a measurement, the voltages (V 1 and V 2 ) applied to the inner and outer spheres of the ESA are varied such that the potential V r on the central trajectory is zero. For two centered spheres the eld E r between inner and outer hemisphere is given by: E r = (V 2 V 1 )r 2 r 1 r 2 (r 2 r 1 ) ; (3.2) with r in between r 2 and r 1. V r is obtained by integration of E r with the boundary conditions V r1 = V 1 and V r2 = V 2 : V r = (V 2 V 1 )r 1 r 2 r(r 2 r 1 ) + V 2r 2 V 1 r 1 r 2 r 1 ; (3.3) Taking V r = 0 on the central trajectory (r =(r 2 +r 1 )=2) gives the ratio of the voltages to be applied to the two spheres: V 2 V 1 = r 1 r 2 : (3.4) The pass energy E 0 of a projectile with a charge state q on the central trajectory is now given by: E 0 = q(v 2 V 1 ) r 2 =r 1 r 1 =r 2 (3.5) in which the factor F =1=(r 2 =r 1 r 1 =r 2 ) (ev/v), the spectrometer proportionality factor gives the ratio between applied voltage and energy of the detected particle. Inserting the radii gives V 2 =V 1 = 0:923 and F =6:24. Our ESA is only partly spherical therefore its V 2 =V 1 and F factors deviate slightly from the ideal values. Calibration of the spectrometer 66 gave V 2 =V 1 = 0:935 and F =6:42 respectively. Absolute yield The conversion from raw ESA counts, measured at an energy E in a time interval t to the absolute number N of emitted particles per ion is given 2 = 19 counts q 1:6 10 P e( 0 ) 1: E I t t F I F g particles ion sr ev (3.6) Where q is the charge state of the projectile. Calculation of the absolute intensity involves three correction factors P e ( 0 );F I and F g :

11 ) s t i n u. b r a ( y t i s n e t n I 30 Experiment Emission probability. The factor P e reects the emission probability of a particle under an angle 0 = with respect to the target plane. Auger electrons are for instance emitted more or less isotropically whereas specularly scattered ions have a strongly peaked emission probability Target current. The measured target current is the sum of the impinging ion current and the secondary particles (electrons/ions) leaving the target. Measuring the beam current in Faraday cup FC2 gives the true particle current within 10% uncertainty. F I is given by the ratio of FC2 current and target current Displacement (sway) (mm) Figure 3.6: ESA signal versus the target displacement (sway) for a 150 ev O 7+ beam measured at electron energy E = 480 ev. Incident and observation angles were 15 and 90 respectively. Geometry correction. The geometry correction factor F g reects the amount ofoverlap between beam spot and analyzer spot on the target. For grazingly incident beams observed at large, the target surface covered by the beam can be larger than the part observed by the ESA. An estimate of the fraction seen by the ESA involves a measurement ofthebeam prole. The beam prole is measured using a grazingly incident beam and the ESA set at a large angle of observation (say,90 ). Projected on the target surface and assuming a circular beam, the beam and analyzer spots are ellipses. The analyzer spot can now be shifted through the beam spot by moving the target (adjusting the sway) through the beam. This way, parts of the beam spot are sampled by the ESA leading to an intensity prole. The resulting prole of which an example is shown in gure consists of a convolution of the actual beam density prole and the analyzer prole. For a full calculation of F g the reader is referred to appendix A of De Zwart's thesis Time of ight system The TOF system consists of three time of ight tubes, mounted at angles of 10, 30 and 140 degrees with respect to the beam (gure 3.4), and a chopper {sweeper system mounted in front of the deceleration lenses.

12 3.2 Sir 31 Acceleration tube Channeltron D 8 G 1 G 2 72% transmission D a D b D c D d D e D f 8mm 6mm 2mm Figure 3.7: Drawing of one of the time of ight tubes. Indicatedare the diameters of the inner diaphragms and the lengths of the inner and outer tubes. Chopper { sweeper system Chopped beams are produced by a chopper-sweeper system installed between diaphragms D 2 and D 3. This system, sketched in gure 3.8 consists of two pairs of parallel deection plates of size (a b =2035 mm 2 ), separated by d = 3 mm. One of the chopper plates is grounded while on the other one a block pulse voltage is applied. At both zero crossings of the applied voltage, the beam is swept over the diaphragm D 3 located 251 mm behind the center of the chopper plates. The positive ank of the chopper pulse has a much longer rise time than the negative ank. To ensure proper time resolution, pulses from the positive ank can be deected by the sweeper plates. The slope of the negative chopper ank can manually be adjusted between 20 ns and 20 s. The time spread of the beam pulses produced by the chopper { sweeper system is discussed in appendix B. Time-of-ight tubes The lay-out of the time-of-ight tubes is sketched in gure 3.7. Each tube consists of a stainless steel outer tube carrying an isolated inner tube of 421 mm length. The inner tube is equipped with 6 diaphragms, the last of which is 2 mm in diameter. After this diaphragm a channeltron 6 detector is mounted. The sensitivity of the channeltron for ion detection is 100% for ions with energies exceeding 1 kev. 67 The ight distance from the center of the target to the channeltron head is 682 mm. The corresponding acceptance angle is tof =6:810 6 sr. 6 Manufactured by Galileo, type 4839.

13 32 Experiment 3mm 11mm 20 mm Ion beam 35 mm 20 mm 3mm 7mm D2=1mm V sweeper V chopper Time Figure 3.8: Schematic view of the chopper-sweeper system. Focussing The inner tube can be used to separate ions and neutral particles. By oating this tube on a negative or positive voltage, ions are accelerated or decelerated towards the detector. A homogeneous eld between tube and entrance is ensured by two parallel metal grids having a total transmission of about 72%. Still the tube has a focusing eect on the accelerated charged particles. Due to the acceleration, ions emitted in a solid angle 0 are projected into a smaller solid angle tof. This leads to an overestimate of the ion contribution in the measured ions/neutrals ratio. The ratio 0 = tof for which ions are projected on D 8 is a function of projectile energy and charge E; q and tube acceleration voltage V t : s p 1+qV t =E = tof 250 p 1+qV t =E (3.7) The true ion/neutral ratio is found by a division of the measured ratio by this number. Foratypical qv t =E =0:3 used in the experiments, 0 = amounts up to In the experiments the total correction factor K(qV t =E) is determined using ion/neutral intensity ratios I =I 0 measured for dierent V t. Extrapolation of these to V t = 0 gives the true I =I 0 ratio.

14 i t n r A d e r e t t a c S l i o c e r -l A li o c e r - O l i o c e r - H 3.3 Targets ) s u. b r a ( s t n u o C Contaminated Al Clean Al Velocity (arb. units) Figure 3.9: Flight time spectra of 10 kev Ar projectiles scattered o a dirty (top) and clean (bottom) Al(110) surface. Indicated are the recoil Al, O and H peaks. 3.3 Targets The majority of experiments presented in this thesis are performed on three types of single crystal targets: Al(110), Si(100), and LiF(100). A few of the Auger spectra shown in chapter 4 stem from tungsten and nickel targets which are described in more detail in Folkerts' thesis. 66 Aluminum has been used as target material because it is the best known physical example of a free electron gas target. The Al(110) crystal has FCC structure with lattice constant a =7:6, Fermi sphere radius r s =2,Fermi energy E F =11:7 ev and work function W =4:2 ev. The target is cleaned in situ using 3 kev to 10 kev Ar sputtering at incident angles between 5 and 15 degrees and subsequent annealing to about 700 K. Cleanliness is monitored using the time of ight tube mounted at 30. An example of 10 kev Ar time of ight spectra taken on a contaminated and a clean Al target respectively, are shown in gure 3.9. The Al(110) target is regarded clean for oxygen contaminations less than 1%. The azimuthal direction of the Al(110) target can be found by measuring the azimuth-dependent yield of scattered projectiles for a low energy (a few kev) He beam grazingly incident onthe target. The scattered projectile yield shows maxima for angles which agree with low indexed { close packed { crystal directions. The azimuth angle can now be calibrated by comparing the measured distribution with a simulated distribution, which can be obtained using the marlowe program 68 (see e.g. C. Bos 69 ). The silicon Si(100) crystal is boron (p) doped, the doping being of the

15 34 Experiment order of 0.1 to 0.01 ppm. The majority carriers of this target are holes. The crystal has diamond structure. Its work function is 5.3 ev. From the atomic point of view, the Si and Al targets are very similar, having the same core level structure and binding energies. Because of the positive doping the electron densities are dramatically dierent. The cleaning procedure used for the Si target is roughly the same as for Al. The LiF(100) crystal is ionic, built up of Li + ions surrounded by six F ions, and vice versa. The unit cell has lattice constant a = 7:6. The valence band of the crystal is formed by the least bound lled F (2p) level, at ' 12 ev. At elevated temperatures of about 700 K the LiF crystal conducts ions (but conserves its insulating properties) allowing ion beams to reach the target without charging up macroscopically. Heating up the target to this temperature has a welcome side-eect: at 700 K a large fraction of the contamination is removed. On top of the temperature induced cleaning, the LiF target was sputter cleaned using 3 kev Ne beams at =5 incidence before each experimental run.