Swift Ion Beam Analysis in Nanosciences

By Denis Jalabert, Ian Vickridge and Amal Chabli

Swift ion beam analysis (IBA) of materials and their surfaces has been widely applied to many fields over the last half century, constantly evolving to meet new requirements and to take advantage of developments in particle detection and data treatment.

Today, emerging fields in nanosciences introduce extreme demands to analysis methods at the nanoscale. This book addresses how analysis with swift ion beams is rising to meet such needs. Aimed at early stage researchers and established researchers wishing to understand how IBA can contribute to their analytical requirements in nanosciences, the basics of the interactions of charged particles with matter, as well as the operation of the relevant equipment, are first presented. Many recent examples from nanoscience research are then explored in which the specific analytical capabilities of IBA are emphasized, together with the place of IBA alongside the wealth of other analytical methods.

• Language: English
• Publisher: Wiley
• Release date: Aug 7, 2017
• ISBN: 9781119008675

Book preview

Table of Contents

Cover

Title

Copyright

Preamble: Rutherford and IBA

Introduction

I.1. Interactions with electrons

I.2. Elastic scattering from nuclei

I.3. Nuclear reactions

1 Fundamentals of Ion-solid Interactions with a Focus on the Nanoscale

1.1. General considerations

1.2. Basic physical concepts

1.3. Channeling, shadowing and blocking

1.4. 1D layers: limits to depth resolution

1.5. 2D and 3D objects: aspects of lateral resolution

2 Instruments and Methods

2.1. Instruments

2.2. Methods

3 Applications

3.1. Example of resonances/light element profiling

3.2. Quantitative analysis/heavy element profiling

3.3. Examples of HR-ERD analysis

3.4. Channeling/defect profiling

3.5. Blocking/strain profiling

3.6. 3D MEIS/real space structural analysis

4 The Place of NanoIBA in the Characterization Forest

4.1. Introduction

4.2. Scope of physical and chemical characterization

4.3. Ion-based characterization techniques overview

4.4. Ion-mass-spectroscopy-based characterization techniques versus IBA

4.5. Other characterization techniques versus IBA

4.6. Emerging ion-beam-based techniques

List of Acronyms

Bibliography

Index

End User License Agreement

List of Tables

3 Applications

Table 3.1. Lattice parameters and energy band gap for the wurtzite phase of AlN, GaN and InN compounds

Table 3.2. Areal densities of different oxygen isotopes before and after annealing in ¹⁸O2 atmosphere (P=10−2 Torr of ¹⁸O2) at 490°C for 30 min

4 The Place of NanoIBA in the Characterization Forest

Table 4.1. Overview of the physical and chemical properties qualifying a material

Table 4.2. Typical characterization techniques versus primary and secondary particles as defined in Figure 4.1 (bold acronyms relate to IBA)

Table 4.3. Classification of ion-based characterization techniques (bold short names relate to light ions used as primary particles)

List of Illustrations

Introduction

Figure I.1. Sensitivity of PIXE as a function of atomic number and proton energy for organic samples [JOH 76]

1 Fundamentals of Ion-solid Interactions with a Focus on the Nanoscale

Figure 1.1. Photon wavelength and de Broglie wavelengths of different particles as a function of energy. The top gray area corresponds to the typical distances between atoms within a solid. The bottom gray area corresponds to the nuclear dimensions

Figure 1.2. According to the 2θ deviation of the X-ray beam, the phase shift between photons scattered by adjacent planes causes constructive (left figure) or destructive (right figure) interferences

Figure 1.3. (Left) Maximum intensities of a 10 keV X-ray beam diffracted by atomic planes spaced by 0.2 nm according to Bragg’s law (θ/2θ configuration). (Right) Laue diffraction pattern of a Ge(111) wafer (European Synchrotron Radiation Facility, IF-BM32 beamline)

Figure 1.4. (Left) Maximum intensities of a 200 keV electron beam diffracted by atomic planes spaced by 0.2 nm according to Bragg’s law (θ/2θ configuration). (Right) typical diffraction pattern obtained with a 200 keV electron beam on a silicon crystal [ROU 16]

Figure 1.5. (Left) Maximum intensities of 100 keV proton and helium ion beams diffracted by atomic planes spaced by 0.2 nm according to Bragg’s law (θ/2θ configuration). Diffraction (GIFAD) images recorded with 400 eV He on the c(2 × 2) reconstructed ZnSe(001) surface along the [110] direction [KHE 09]

Figure 1.6. Full width at half maximum of a diffraction peak as a function of the crystallite size

Figure 1.7. Penetration depths of X-rays [SEL 93], electrons [KAN 72] hydrogen ions [AND 77] and helium ions [ZIE 77] in silicon

Figure 1.8. Stopping cross-section of: (top) protons in silicon [AND 77] – (bottom) alphas in silicon [ZIE 77]

Figure 1.9. Penetration depths of helium ions [ZIE 77] and hydrogen ions [AND 77] in silicon

Figure 1.10. Calculated trajectories of protons in the first micron of a silicon sample for two incident energies (left) 100 keV – (right) 2 MeV [AND 77]. The calculation was made using the SRIM software [ZIE 04]

Figure 1.11. Calculated values of straggling for protons and helium ions through different targets (data from [CHU 76])

Figure 1.12. Calculated trajectory of a 100-keV proton in the vicinity of a silicon nucleus with an impact parameter of 100 fm. The diameter of the Si nucleus, about 7.5 fm, and that of the proton (1.7 fm) are not to scale in this figure. The positions of the two nuclei over time are indicated by numbering from 1 to 6 corresponding to time intervals of 1 × 10–4 fsec. In this simulation, the proton comes from the right-hand side of the figure and is scattered upward by the silicon nucleus initially at rest at the position X = Y = 0

Figure 1.13. Variation of the kinematic factor for helium ions as a function of the mass of the target atoms for three different scattering angles (30°, 90° and 165°)

Figure 1.14. Variation of the kinematic factor for helium ions as a function of the scattering angle for four elements (C, Si, Ge and Hf)

Figure 1.15. Scattering cross-sections as a function of the atomic number at a scattering angle of 90° for an incident helium beam of different kinetic energies (0.1, 0.2, 1 and 2 MeV)

Figure 1.16. Scattering cross-section of silicon as a function of the scattering angle for an incident helium beam at different energies (0.1, 0.2, 1 and 2 MeV)

Figure 1.17. Angular dependence of the screening correction factors of the Rutherford cross-section for an incident helium beam scattered by gold at different energies (0.25, 0.5, 1 and 2 MeV). The dotted lines correspond to the correction, independent of the scattering angle, given by L’Ecuyer et al. [LEC 79] (equation [1.14]) and the solid lines correspond to the correction factor, dependent of the scattering angle, given by Andersen et al. [AND 80] (equation [1.15])

Figure 1.18. Schematic representation of the trajectories of channeled protons (the vertical distance scale has been greatly expanded in this figure to allow visualization of trajectories)

Figure 1.19. Critical angle of [001] channeling in silicon as a function of ion beam energy

Figure 1.20. Channeling half-wavelength in the [100] direction of silicon as a function of ion beam energy

Figure 1.21. χmin values for protons and helium ions channeled in silicon along the <110> direction

Figure 1.22. Calculated trajectories of 100 keV protons in the vicinity of a silicon nucleus (the silicon nucleus diameter, about 7.5 fm, is not to scale in this figure and the Coulomb potential used is unscreened)

Figure 1.23. Diagram illustrating the calculation of the shadow cone radius. The incident particle approaches the surface atom with an impact parameter b and scatters at an angle θ, so that it passes the next atom at a distance R

Figure 1.24. Ratio of the shadow cone radius from the screened potential RM to the unscreened potential RC as a function of RC/a (data from [FEL 82])

Figure 1.25. Thomas Fermi screening distance as a function of the atomic number of the target nucleus

Figure 1.26. Experimental values of the surface peak intensity for various crystals expressed in number of atoms per row as a function of the ratio between the thermal vibration amplitude and the radius of the shadow cone. The continuous curve was generated from numerical simulations (data from [FEL 82])

Figure 1.27. Calculated values of the surface peak area for silicon channeled in the [110] direction expressed in the number of atoms per row as a function of the ion beam energy

Figure 1.28. Schematic view of shadowing geometries employed to enhance the scattering yields of the initial atomic layers with respect to deeper layers

Figure 1.29. Schematic view of the shadowing and blocking effects

Figure 1.30. Blocking dip width versus scattered energy for helium ions in silicon along the <110> direction neglecting (straight line) or taking into account (dotted line) the thermal vibration amplitude at 300 K

Figure 1.31. Blocking dip width versus scattered energy for protons in silicon along the <110> direction neglecting (straight line) or taking into account (dotted line) the amplitude of thermal vibration at 300 K

Figure 1.32. Schematic view of atomic displacements due to elastic strain within a crystal

Figure 1.33. Schematic view of the shadowing/blocking geometry used to analyze surface structures (from the Surface Physics Group at the University of York (UK))

Figure 1.34. Theoretical response functions for two point sources resolved according to a) the Rayleigh criterion and b) the Gaussian FWHM criterion. In each case, the single central peak is the response when the two point sources are not separated at all

Figure 1.35. Simulated RBS spectra for nominally resolved Au delta-doped layers in Si, of equal intensity (a and c) and of ten times different intensities (b and d) and with good (a and b) and poor (c and d) statistics

Figure 1.36. Schematic illustration of the particle trajectories in an ion beam of finite emittance

Figure 1.37. Maximum beam intensity, as a function the beam diameter for protons of 10, 100 and 1000 keV, below which the charge space effect can be neglected (considering a distance between the focusing lens and the sample of 1 m)

Figure 1.38. Time to displace 10% of the silicon atoms in the first 10 nm with a 10 nA helium ion beam of 100 keV and 1 MeV respectively according to SRIM calculations [ZIE 04]

Figure 1.39. Schematic illustration of the particle trajectories in a continuous layer (left) and in a set of 3D objects (right)

Figure 1.40. 2D map of scattered ion intensities as a function of the scattering angle and scattered ion energy for normal incidence of 100 keV He+ ions calculated for a 2D layer (top left) or for three nanoparticle shapes: sphere (top right), hemisphere (bottom left) and cylinder (bottom right) [SOR 09]

2 Instruments and Methods

Figure 2.1. Schematic drawings of electrostatic accelerator operation. Left: Van de Graaff principle. Right: voltage multiplier principle. s: ion source. t: accelerator tube. HVT: high voltage terminal. v: voltage multiplying stack. m: belt motor. a: alternator. DC HV: dc belt charge high voltage supply. AC RF: ac high frequency power supply

Figure 2.2. Schematic diagram of an electrostatic analyzer

Figure 2.3. Schematic diagram of a magnetic sector analyzer

Figure 2.4. A typical standard RBS sample analysis chamber. c: collimator; ic: insulated vacuum coupling; ci: current integrator; ift: insulated feedthrough; pa: preamplifier; d: particle detector; sh: sample holder; sm: stepper motor; ca: control and acquisition interface; acq: data acquisition computer; con: control computer

Figure 2.5. Top: typical experimental geometry for standard RBS measurements. Bottom: simulated RBS spectra obtained from a series of SrO thin films of increasing thickness on a Si substrate. For this simulation, a detector resolution of 15 keV was used, and the incident ⁴He+ beam energy is 2.0 MeV

Figure 2.6. Top: typical grazing geometry used to optimize depth resolution in standard RBS. Bottom: Simulated spectra of SrO layers of increasing thickness obtained in the grazing incidence indicated for a beam energy of 0.8 MeV ⁴He+, and a detector resolution of 15 keV

Figure 2.7. Illustration of how angular multiple scattering can degrade effective depth resolution for highly grazing incident beam geometries

Figure 2.8. Schematic diagram of a MEIS experiment

Figure 2.9. MEIS simulated spectra for increasing thicknesses of SrO on Si, obtained with an electrostatic detector typically used in MEIS, with energy resolution ΔE/E of 3 × 10−3, and an incident beam of 400 keV ⁴He+

Figure 2.10. Top: typical experimental geometry for ERDA measurements of hydrogen isotopes in thin films. Bottom: Simulated spectra for SrO films of increasing thickness, charged with equal parts of hydrogen and deuterium, on a Si substrate. For this simulation, the incident ⁴He+ energy was 2.0 MeV, and a 9 µm Mylar film was sufficient to stop the ⁴He+ elastically scattered from the Si and Sr in the sample

Figure 2.11. The cross-section of the ¹⁸O(p,α)¹⁵N nuclear reaction [LOR 79]. The very narrow resonance at 151 keV (magnitude multiplied by 10 so as to be visible on the graph) has been widely used for Narrow Resonance Profiling of ¹⁸O

Figure 2.12. Top: typical experimental geometry used for NRP with the ¹⁸O(p,α)15 N reaction. Note that since the detection energy resolution does not affect depth resolution, a large detector may be used. The poor energy resolution (some tens of keV), the energy straggling in the mylar film, and the large kinematic spread in such an arrangement have no influence on the obtainable depth resolution. Bottom: Simulated NRP excitation curves obtained from the 151 keV resonance in ¹⁸O(p,α)15 N in Sr¹⁸O films of varying thicknesses

3 Applications

Figure 3.1. RBS spectra obtained from a thin SiO2 film on Si with the incident beam channeled along the (110) silicon axis, and detection in grazing (a) and standard (b) geometries [FEL 78]

Figure 3.2. Yield of non-crystalline Si from (110) channeled spectra on SiO2 layers of increasing thickness on Si [FEL 78]. The slope and intercept of the line indicate that in the region of the interface, 2.5 × 10¹⁵ Si atoms cm−2 are either in non-stoichiometric oxide or in a disturbed region of the Si

Figure 3.3. Illustration of the principle of a simple isotopic tracing experiment. Silicon is oxidized first in ¹⁶O2 and then in ¹⁸O2. The position of the ¹⁸O in the oxide gives information about the oxygen exchange, transport and oxide growth processes. In the case illustrated here, oxygen exchange and network diffusion occur without growth in the surface region, whereas oxide growth occurs through interstitial transport of O2 molecules to the SiO2/Si interface, where oxidation of the silicon occurs

Figure 3.4. (a) Measured (points) and calculated (lines) excitation curves of 18O in SiO2/SiC annealed in ¹⁸O2. The calculated excitation curve corresponds to the concentration profiles in (b)

Figure 3.5. (reproduced from Ganem, 2011 [GAN 11], and based on Trimaille, 1994 [TRI 94]). A 20 nm thick thermal oxide of natural isotopic composition is further oxidized (900°C, 5 h, 100 mbar) in pure dry ¹⁸O2. The solid line is the excitation curve calculated for the assumed concentration profile of ¹⁸O shown in the inset

Figure 3.6. Schematic representation of a silicon carbide crystal. The c-axis is vertical in the figure, with the carbon face perpendicular to (0001) and the Si face perpendicular to Mfig-19.jpg by convention. Along the c-axis, the Si and C atoms are separated by just 0.063 nm (and the Si–C bilayers are separated by 0.189 nm)

Figure 3.7. (a) Measured (points) and calculated (lines) NRP excitation curves about the 151 keV resonance in ¹⁸O(p,α)¹⁵N after ¹⁶O/¹⁸O oxidations of Si (open points, dashed lines) and SiC (solid points, full lines). The calculated excitation curves correspond to the concentration profiles in (b)

Figure 3.8. Thermal oxides of different thicknesses are formed on Si and SiC by two different means, so as to be able to vary the oxide thickness and the nature of the SiO2/semiconductor independently. In (a), an oxide of around 50 nm is first grown on all samples and then etched chemically for various times so as to produce a system in which the SiO2/SC interface is identical for all SiC or Si samples, and the overlying thermal oxide thickness varies. In (b), oxide thicknesses of the same range are produced by thermally oxidizing for suitable times, so that in this case the nature of the SiO2/SC interface varies. From [VIC 02]

Figure 3.9. (Top) SEM image of SiC nanocrystals below the SiO2/Si interface after chemical removal of the overlying SiO2. (Bottom) TEM image of a cross-section of one such nanocrystal. The SiC is epitaxial with the Si, but has a lattice parameter about 20% larger than that of Si. The visible periodic contrast is a moiré pattern between the SiC nanocrystal and single-crystal silicon lying beneath it in the TEM cross-section. Images from [PÉC 07]

Figure 3.10. From Deville-Cavellin, 2009 [DEV 09]. (a) Measured (points) and calculated (lines) excitation curves of ¹⁸O in SiO2/Si annealed in ¹³C¹⁸O. The calculated excitation curve corresponds to the concentration profiles in (b). The diffusion and exchange can be divided into three independent processes (see text)

Figure 3.11. Measured amounts of interface ¹³C and ¹⁸O after 90 min annealing of SiO2/Si in ¹³C¹⁸O atmosphere at 1,100°C at various pressures and for various initial oxide thicknesses

Figure 3.12. Examples of non-random RBS spectra. (Top spectrum) the ion beam alignment is too close to an axial channeling direction. (Bottom Spectrum) the ion beam alignment is too close to a planar channeling orientation

Figure 3.13. In-plane lattice parameters, band gaps and corresponding wavelengths for different families of semiconductors. For a color version of this figure, see www.iste.co.uk/jalabert/ionbeam.zip

Figure 3.14. Measured (dots) and simulated (solid line) RBS spectra of an AlGaInN layer deposited on GaN/sapphire at a substrate temperature of 630°C. The surface channels are marked by the corresponding element [MON 03b]

Figure 3.15. Measured and simulated (solid line) RBS spectrum of an Al0.44Ga0.55In0.01N sample grown on GaN/sapphire at a substrate temperature of 650°C. The inset shows details of the indium peak at the sample surface [MON 03b]

Figure 3.16. Illustration of the method used to measure a small amount of material by comparing the measurement and simulation. On the top, the peak area is calculated using the experimental spectrum. On the bottom, the same calculation is done using the simulated spectrum

Figure 3.17. Maximum indium incorporation in the quaternary compound as a function of the Al mole fraction and the substrate temperature [MON 03a]

Figure 3.18. RBS spectrum of a sample containing a Tm-doped GaN layer (104 nm) covered by a Tm-doped AlN layer (180 nm) [AND 05]. The concentration of Tm in GaN is 1.2%. The letters are used to mark the positions of the elements: A is N in AlN, GaN: Tm and AlN:Tm, B is Al in the AlN substrate, C is Al in AlN:Tm, D is Ga in GaN:Tm and E is Tm in GaN:Tm. The arrow with the letter F indicates the position of Tm on the surface of the AlN

Figure 3.19. Normalized growth speeds of GaN layers doped with europium as a function of the concentration of Eu [HOR 04a]. The squares correspond to growth under Ga-rich conditions and the circles to growth under N-rich conditions. The inset shows a typical RBS spectrum corresponding to a multilayer AlN/GaN/AlN/GaN:Eu/AIN

Figure 3.20. RHEED diffraction pattern (left) and AFM image (right) of a plane of GaN quantum dots

Figure 3.21. High-resolution transmission electron microscopy (HRTEM) image of two GaN quantum dots. The top dot is deposited on the AlN surface and the bottom one is embedded in AlN

Figure 3.22. Measured (open circles) and calculated (solid line) RBS spectra of a sample containing two GaN quantum wells of nominal thicknesses 20 Å. The thicknesses of the quantum well buried in AlN and the layer deposited on surface are, respectively, 16.7 and 19.6 Å. The substrate temperature was set at 750 °C

Figure 3.23. Variation of the thinning of GaN quantum wells encapsulated in a matrix of AlN as a function of the nominal thickness of the well. The substrate temperature was set at 750°C for growth of the two layers

Figure 3.24. Variation of the thickness reduction of GaN quantum wells with a nominal thickness of 50 Å buried in AlN as a function of the encapsulating temperature. The substrate temperature was set at 750°C for the growth of GaN wells

Figure 3.25. High-resolution TEM images of two GaN quantum wells of nominal thickness 20 Å, encapsulated in an AlN layer deposited, respectively, at (a) 750 °C and (b) 700 °C as well as the profiles of corresponding inter-planar distances. The thickness of the wells encapsulated at 750 °C varies from 12.5 to 17.5 Å. However, the thickness of the wells encapsulated at 700 °C remains constant and equals 20 Å

Figure 3.26. High-resolution TEM image of a GaN quantum well encapsulated in a matrix of AlN grown at 750 °C. The observed GaN thickness varies from 17.5 ± 1.2 to 12.5 ± 1.2 Å. The lower GaN/AlN interface seems relatively more regular, flat and uniform than the upper AlN/GaN interface

Figure 3.27. RBS spectra measured (open circles) and calculated (solid line) of a sample containing three identical GaN quantum wells of nominal thickness 50 Å covered by an Al flux before encapsulation. Al flux was high enough to completely cover the surface of GaN. The QW1, QW2 and QW3 wells are coated with Al for 1, 10 and 30 min, respectively. The substrate temperature was set at 750°C

Figure 3.28. Measured (open circles) and calculated (solid line) RBS spectra of a sample containing two identical planes of GaN quantum dots, of nominal thickness 18 Å, which are, in one case, buried in AlN and, in the other, deposited on the surface. The substrate temperature was set at 750 °C

Figure 3.29. Variation in the thickness reduction of GaN quantum dot planes deposited on and encapsulated in AlN at 750°C as a function of the amount of GaN nominally deposited

Figure 3.30. HRTEM image of a superlattice of quantum dots GaN/AlN deposited at 730 °C. If the layers of AlN between dot planes are thick enough, a smooth upper AlN interface is observed. However, if the layer is very thin (~20 Å), the roughness induced by the layer of GaN dots is propagated to the upper AlN interface

Figure 3.31. HRTEM image of the wetting layer connecting the GaN islands deposited on AlN and encapsulated in an AlN matrix at 750 °C as well as the corresponding interplanar distance profile. A quantum dot is visible on the right side of the image. The thickness of the wetting layer has been reduced to about 2.5 Å, that is, a single monolayer

Figure 3.32. Change in total free surface area of a plane of GaN quantum dots as well as the respective (0001) and surfaces as a function of the amount of GaN nominally deposited. The surface areas were calculated from AFM data [GOG 03]

Figure 3.33. a) Amount of material contained in the GaN islands as a function of the nominal amount of GaN deposited. The solid line corresponds to the amount of material involved in the formation of the dots, assuming a wetting layer of constant thickness equal to two monolayers. b) Variation of the thickness of the wetting layer of GaN depending on the nominal amount of GaN deposited deduced from curve (a) [GOG 03]

Figure 3.34. Physical structure of an n-type MOSFET: a) perspective view and b) cross-section, showing the characteristic dimensions W and L; the contacts S, G and D and the regions are composed of oxide (dielectric film) and of bulk silicon (substrate) [SED 04]

Figure 3.35. Electron microscopy images showing the evolution of MOSFETs produced by Intel Corporation. a) and b) are the first and second transistor generations strained by a SiGe alloy. c) and d) the first and second transistor generations fabricated with a high-permittivity dielectric barrier (high-κ). e) Implementation of the first transistors in three dimensions (3D). The length L of the channel is displayed on top of each image with the year of manufacture [BOH 11]

Figure 3.36. Experimental MEIS energy spectrum of a HfO2/SiO2/Si (100) stack grown at 430°C. The inset illustrates the scattering geometry: the 101.5 keV He+ incident ion beam is aligned along the [111] axis of Si substrate and the scattered ions along the [211] direction

Figure 3.37. High-resolution transmission electron microscope image of HfO2/SiO2/Si stack deposited at a) 430 °C and b)