High resolution ion beam analysis Torgny Gustafsson
Review articles and Books L. C. Feldman, J. W. Mayer and S. T. Picraux, Materials Analysis by Ion Channeling, Academic Press (1982) I. Stensgaard, Surface studies with high-energy ion beams. Reports on Progress in Physics 55 989-1033 (1992). H. Niehus, W. Heiland and E. TagIauer, Low-energy ion scattering at surfaces. Surface Science Reports 17 (1993) 213-303. J.F. van der Veen, Ion beam crystallography of surfaces and interfaces. Surface Science Reports 5 199-288 (1985). R. Hellborg, H. J. Whitlow and Y. Zhang, Ion Beams in Nanoscience and Technology, Springer (2009)
Ion backscattering for materials analysis: 3 energy regimes Low energy ion scattering (~10 kev or less): LEIS Hard to quantify, very surface specific Medium energy (~ 50 200 kev) MEIS Quantitative, somewhat elaborate equipment, surface specific High energy (1 2 MeV) RBS Quantitative, simpler equipment
Advantages of ion beams * Penetrating (can access buried interfaces!) Mass specific Known interaction law (cross sections are known) Excellent depth resolution * Mostly high and medium energy beams
Elementary considerations in ion backscattering (1) Relation between incoming and backscattered ion energies k = the kinematic factor Cross section for backscattering
Elementary considerations in ion backscattering (2) Depth profiling
Structural work using Medium Energy Ion Scattering (MEIS)
MEIS lab at Rutgers XPS ALD IR MEIS NRP XPS 8
Energy Electrostatic ion detector Al 18 O ~20 16 O Depth resolution of ~3 Å near surface Angular resolution 0.2 Mass-sensitive: E = E(M, θ) Quantitative (cross sections are known) Angle 9
Surface structure and dynamics Monte Carlo simulation in the binary-encounter approximation blocking cone shadow cone Δd12 Δd23 Input: individual atomic positions anisotropic rms vibration amplitude Output: R 100 1 N N i 1 Y Y calc exp i i exp Yi 2 Yield (ML) 10 8 6 4 [101] I [011] [211] [011] [110] [011] II III d 12 (%) 5 0-5 -10-15 (a) 0.6000 0.8200 1.040 1.260 1.480 1.700 1.920 2.140 2.360 2.580 2.800 3.020 3.240 3.400 2 50 60 70 80 90 100 110 120 130 Scattering angle (deg) -5 0 5 10 15 d 23 (%)
An (oversimplified) picture of the origin of the oscillatory relaxation
Reconstruction of the (110) surface of Au: Possible structural models consistent with a (1x2) symmetry
Conclusions: Missing row structure Large first layer inwards contraction Buckling in the lower layers, results in charge density smoothing
Another research example: Surface structure of TiC(001) Y. Kido et al, PRB 61 1748 (2000)
Sub-nm Depth Resolution from a Materials Perspective: Who needs monolayer resolution? In this talk, three techniques: Elastic Recoil Detection Analysis (ERDA) Nuclear Resonance Profiling (NRP) Medium Energy Ion Scattering (MEIS) Research examples
Motivation for a lot of work in this field today Gate oxide Source Barrier? Gate >10nm Silicon Drain The SiO 2 /Si system: interface of choice in microelectronics for 40+ years Moore s law says that the gateoxide thickness will soon be too small (~ 1 nm) due to large leakage currents from quantum mechanical tunneling CMOS Gate Structure SiO 2 needs to be replaced by a higher dielectric constant material ( high-k )
Advantages of ion beams Penetrating (can access buried interfaces!) Mass specific Known interaction law (cross sections are known) Excellent depth resolution
Medium Energy Ion Scattering (MEIS) - a low-energy, high resolution version of conventional Rutherford Backscattering (RBS)
A comparison between RBS and MEIS RBS MEIS Ion energy ~ 2 MeV ~ 100 kev Detector resolution ~ 15 kev ~0.15 kev Depth resolution ~ 100 Å ~ 3 Å 2 basic advantages vs. RBS: Often better de/dx, superior detection equipment
Energy Dependence of de/dx for Protons in Silicon Energy (MeV) Maximum of ~ 14 ev/å at ~ 100 kev! This helps, but the greater advantage is the use of better ion detection equipment!
Ion detection equipment Magnetic spectrometers Electrostatic spectrometers Kyoto university (Kimura) Kobelco FOM IBM (Tromp, van der Veen, Saris..) High Voltage Engineering
From the Ion beam analysis laboratory at Kyoto university; note the magnetic spectrometer.
The Kyoto Kobelco very compact MEIS facility Footprint: ~ 2.1 x 1.5 m
TOF New development: 3D-MEIS S. Shimoda and T. Kobayashi 3D-MEIS Pulsed ion beam Scattered (and/or recoiled) particles are detected 2D blocking pattern flight times of scattered (and/or recoiled) particles 3D detector position-sensitive and time-resolving MCP detector x wide solid angle incident beam Ion : He+ Energy : 100 kev Repetition : 500 khz sample periodic atomic structure
Angle from the Si(11 0) plane ( ) (a) Angle from the Si(11 0) plane ( ) 6 4 2 0-2 -4 6 4 2 0-2 -4 (b) Er 18 Si 18 f-1 f-2 f-3 f-4 1 = 67.2 1 = 62.2 1 = 57.2 1 = 52.2 20 f-1 f-2 f-3 f-4 1 = 67.2 1 = 62.2 1 = 57.2 1 = 52.2 20 b-1 b-2 b-3 [011 2] [022 3] [011 1] 22.3 28.8 39.3 22 [331] 22.0 22 Structural analysis of an Er-silicide on Si(111) substrate using 3D-MEIS S. Shimoda and T. Kobayashi 24 24 26 28 30 32 34 Scattered angle 2 ( ) 26 28 30 32 34 Scattered angle 2 ( ) 36 [110] 35.3 Fig.1 3D-MEIS images of the intensities of He particles scattered (a) from Er atoms in the Er-silicide film and (b) from Si atoms in the Si substrate. A A 36 38 38 B B 40 40 42 42 Intensity (counts) 70 60 50 40 30 20 10 0 260 Fig. 2 TOF spectra obtained from the data detected in regions indicated by A and B in Fig.1. a1 (2110) Er 280 a2 300 c B A 320 TOF (ns) 340 Cross section cut by the ErSi 2 (2110) plane Fig. 3 Structural model of the ErSi 2 Si b-1 b-2 b-3 3a=0.665nm 360 b-1 : [0112] _ b-2 : [0223] _ b-3 : [0111] Er Si c=0.403nm
Early High Resolution Work K. Kimura et al. (Kyoto) NIM B99, 472 (1995) Sb on Si(100) with caps of varying thickness; some Sb segregates to the surface
Recent Very High Resolution Work Carstanjen et al. (Stuttgart) (to be published) Note use of N + and N 2+ ions and charge exchange effects
The Stuttgart high resolution ion analyzer
Determining interface strain using monolayer resolution ion scattering and blocking (Moon et al.) Ozone oxide, no strain Thermally grown oxide, significant strain
Spectra and information content Backscattered proton energy spectrum 100 kev p +? ZrO 2 (ZrO 2 ) x (SiO 2 ) y Si(100) depth profile Sensitivity: 10 +12 atoms/cm 2 (Hf, Zr) 10 +14 atoms/cm 2 (C, N) Accuracy for determining total amounts: 5% absolute (Hf, Zr, O), 2% relative 10% absolute (C, N) Depth resolution: (need density) 3 Å near surface 8 Å at depth of 40 Å
Depth resolution and concentration profiling Depth resolution for 100 kev protons (resolution of the spectrometer 150 ev) Stopping power SiO 2 12 ev/å; Si 3 N 4 20 ev/å; Ta 2 O 5 18 ev/å "Near surface" depth resolution 3-5 Å; worse for deeper layers due to energy straggling Layer model: 3 Å H + Layer 1 Layer 2 Layer 3. Layer n 5 raw spectrum layer numbers 3 2 1 4 Energy, kev Areas under each peak corresponds to the concentration of the element in a 3Å slab Peak shapes and positions come from energy loss, energy straggling and instrumental resolution. The sum of the contributions of the different layers describes the depth profile. SUBSTRATE
Oxygen Isotope Experiments: SiO 2 growth mode 1. O 18 uptake at the surface! 2. Growth at the interface 3. O 16 loss at the surface 4. O 16 movement at the interface! Gusev, Lu, Gustafsson, Garfunkel, PRB 52, 1759 (1995)
Schematic model Surface exchange SiO 2 Growth Transition zone, SiO x Si (crystalline) Deal and Grove
Work on high-k films: MEIS spectra of La 2 SiO 5 before and after vacuum anneal yield (a.u.) (1) (1) as deposited film (2) 20 (3) 15 10 5 (2) vacuum anneal at 600 o C (3) vacuum anneal at 800 o C no change in O or Si profiles high-k La-Si-O film minor rearrangement of La upon annealing 0 75 80 85 90 95 proton energy (kev) Annealing up to 800 C in vacuum shows no significant change in MEIS spectra. Surface remains flat by AFM.
La 2 SiO 5 before and after in-air anneal stoichiometry and thickness consistent with other analyses 400 C anneal leads to minor broadening of the La, O and Si distributions 800 C anneal shows significant SiO 2 growth at interface La diffusion towards the Si substrate yield (a.u.) 25 20 15 10 5 (1) as deposited film (2) anneal in air at 400 o C (3) anneal in air at 800 o C int O surf (1) (2) (3) high-k La-Si-O film int Si surf (1) (2) (3) int La surf 0 MO x Si + O 2 MO x SiO 2+ MO x Si 75 80 85 90 95 SiO 2 growth at 800 o C proton energy (kev) La diffusion
Higher temperature annealing high-k La-Si-O film 20 15 (1) as deposited film (2) vacuum anneal at 845 o C (3) vacuum anneal at 860 o C (1) (2) yield (a.u.) 10 5 loss of O at 860 o C surface Si La diffusion (3) 0 75 80 85 90 95 proton energy (kev) The film disintegrates!
6 3 Torr 18 O 2 at 400, 500, 600, 700 o C for 5 min 500 400 as-deposited film 18 O 2 reox. (1 Torr, 500 o C, 5 min) ZrO 2 film re-oxidized in 18 O 2 Zr 5 4 30Å Al 2 O 3 annealed in 3 Torr 18 O 2 400 o C 500 o C 600 o C 700 o C Yield (a.u.) 300 200 16 O 18 O Si x5 yield (a.u.) 3 2 oxygen exchange 16 O 18 O Si interface Al 100 1 0 76 78 80 82 84 86 88 90 92 94 96 Deeper O and Si Proton Energy (kev) No change in Zr profile Surface flat by AFM Significant interfacial SiO 2 growth for ZrO 2, less for Al 2 O 3 Dramatic oxygen exchange: 18 O incorporation and 16 O removal SiO 2 growth rate faster than for O 2 on Si Growth faster under ZrO 2 than Al 2 O 3 SiO 2 Growth (Å) 0 25 20 15 10 5 76 78 80 82 84 86 88 proton energy (kev) 30Å ZrO 2 30Å Al 2 O 3 0 0 200 400 600 800 1000 O 2 Anneal Temp ( o C)
Why GaAs (or Ge)? Potentially great advantages over Si-based devices for both high-speed and high-power applications The electron mobility of GaAs is 5x that in Si Much thinner interfacial oxide
HfO 2 on GaAs: MEIS and TEM comparison TEM * MEIS O Hf No etch a-c HfO 2 GaAs * Grazul & Muller, Cornell Ga-rich oxide O Hf HF etch a-c HfO 2 GaAs Ga-rich oxide TEM and MEIS results are consistent
TEM of Al 2 O 3 on GaAs No etch HF etch no contrast between Al 2 O 3 and Ga x As y O (J. Grazul and D. Muller, Cornell)
0.05 0.04 MEIS data of Al 2 O 3 on GaAs: one-parameter fitting No etch O Interfacial oxide: (Ga 2 O 3 ) 0.37 (Ga 2 O) 0.63 (As 2 O 3 ) 0.17, porous oxide: = 0.5 bulk N (Å -3 ) 0.03 0.02 0.01 Al Ga As 0.00 0 10 20 30 40 50 60 70 Depth (Å) 80 85 90 95 H + Energy (kev) interfacial oxide is much thinner for the HF-etched sample n(ga+as), Å -2 n(al), Å -2 n(o), Å -2 n(o)/n(al) HF etch 0.33 1.48 2.23 1.51 No etch 0.55 1.30 2.22 1.70 rms error 1.00 0.95 0.90 0.85 Etched No etch 0 5 10 15 20 25 30 Interface thickness (Å) N (Å -3 ) 0.05 0.04 0.03 0.02 0.01 O Al HF etch 0.00 0 10 20 30 40 50 60 70 Depth (Å) 80 85 90 95 H + Energy (kev)
Epitaxial Oxides Very versatile materials: Small changes in composition big changes in properties (high-t c!) Combining two different oxides may give multifunctionality and/or entirely new phenomena. 48
Metallic LaAlO 3 /SrTiO 3 (LAO/STO) interface 2D Electron Gas LaAlO 3 (E g = 5.8 ev) by PLD or MBE band insulator SrTiO 3 (E g = 3.2 ev) band insulator Metallic conductivity Ohtomo & Hwang, Nature (2004) Superconductivity Reyren et al, Science (2007) Magnetic order Brinkman et al, Nat. Mat. (2007) T(K) Ohtomo & Hwang, Nature (2004) Reyren et al., Science (2007) Brinkman et al., Nat. Mat. (2007) 49
Origin of metallic state at the interface Polar catastrophe: polar-nonpolar discontinuity Ohtomo &Hwang, Nature 427 (2004) This model assumes an abrupt interface 50
Important points Oxygen vacancies have been shown by many to influence carrier concentrations Conductivity only observed on TiO 2 terminated substrates Cationic interdiffusion Nakagawa et al, Nature 5 (2006), Kalabukhov et al, PRL 103 (2009); Willmott et al, PRL 99 (2007). 51
Samples We have investigated samples from three different labs. None of the samples were made at Rutgers. All made by PLD on TiO 2 terminated substrates. Post-annealed in oxygen. Most data presented today from samples by Prof. H. Hwang, Tokyo (now Stanford). 52
He + channeling spectrum of 4-unit-cell LAO/STO(001) 198.6 KeV He + [001] [111] Tokyo-LO4 La Counts 150 100 50 0 Experiment Simulation-no outdiffusion of Sr and Ti Ti 140 150 160 170 180 x4 He + Energy [kev] Sr 4 u.c. LAO STO (001) rastering beam The measured Sr and Ti peaks fall at significantly higher ion energies than those calculated for a sharp interface: The energies for the Sr and Ti peaks are those of both species present within the outermost u.c. of LAO outdiffusion of Sr and Ti to the LAO film 53
H + channeling spectrum of a 4-unit-cell LAO/STO film 600 500 400 Tokyo-MID4 Experiment Simulation-no outdiffusion of Sr and Ti La Yield 300 200 100 0 Al Sr Ti 85 90 95 H + Energy [kev] 54
H + channeling spectrum of a 4-unit-cell LAO/STO film 500 400 Augsburg Experiment Simulation-no outdiffusion of Sr and Ti La 300 Yield 200 100 Al Ti Sr 0 85 90 95 H + Energy [kev] 55
Angle dependent XPS shows poor agreement with sharp interface model (Chambers et al) 56
Conclusions The LaAlO 3 /SrTiO 3 interface shows evidence for substantial interdiffusion, a useful result for understanding the origin of the many interesting effects shown by this system. More details: Surf. Sci. Rep. 65, 317 (2010) 57