X-Ray Photoelectron Spectroscopy: Theory and Practice

X-Ray Photoelectron Spectroscopy: Theory and Practice PHYS-581 (Fall 2010)

Contact Information for EMS in RRC-East Alan Nicholls, PhD Director of Research Service Facility - Electron Microscopy Research Resources Center-East 845 West Taylor Street SES Building, Room 110 Email: nicholls@uic.edu Office: (312) 996-1227 Ke-Bin Low, PhD Senior Research Specialist - Electron Microscopy Research Resources Center-East 845 West Taylor Street SES Building, Room 112 Email: kebinlow@uic.edu Office: (312) 355-2087

Outline of Lecture (1) Background (2) Vacuum 101 (3) Analytical Capabilities (4) Instrumentation (5) Spectrum Simulation (6) Summary

Background 1921 1981 Photoelectric effect discovered by Albert Einstein Nobel Prize Photoemission as an analytical tool demonstrated by Kai Siegbahn (Electron Spectroscopy for Chemical Analysis ESCA) Nobel Prize

Background specimen

Background

Background j = l ± s j: Total angular momentum l: Orbital angular momentum s: Spin angular momentum Quantum Numbers n l s j 1 0 ± 1/2 1/2 2 0 ± 1/2 1/2 2 1 + 1/2 3/2 2 1-1/2 1/2 3 0 ± 1/2 1/2 3 1 + 1/2 3/2 3 1-1/2 1/2 3 2 + 1/2 5/2 3 2-1/2 3/2 Spectroscopist Notation nl j 1s 1/2 2s 1/2 2p 3/2 2p 1/2 3s 1/2 3p 3/2 3p 1/2 3d 5/2 3d 3/2

Background XPS probes core-levels Binding energies in the range of 10 10 3 ev Kinetic energies of similar magnitudes when Al-Kα or Mg- Kα radiation is used Electrons with such low KE easily scattered (REMEMBER THIS) B in d in g E n e r g y (e 1400 1300 1200 1100 1000 Binding Energy vs. Atomic Number for XPS n=1 n=2 n=3 n=4 900 800 700 600 500 400 300 200 100 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Atomic Number 1s 2s 2p3 2p1 3s 3p3 3p1 3d5 3d3 4s 4p3 4p1 4d5 4d3 4f7 4f5 5s 5p3 5p1 5d5 5d3 5f7 5f5 6s 6p3 6p1

Background Universal Curve shows that photoelectrons with KE in the 10 10 3 ev range have inelastic mean-free-paths (IMFPs) from 1 3.5 nm IMFP depends on: (1) Material (atomic #, density) (2) Kinetic energy 1000 100 IMFP, λ (nm) 10 1 Universal IMFP vs. KE Curve Kinetic Energy (ev)

Background 95% of all photoelectrons detected are generated within 3λ of the surface = Sampling Depth (65% within 1λ). 3λ is used as the benchmark definition for Sampling Depth in XPS. So the sampling depth for XPS is typically 3-10 nm Surface-Sensitive! Instrument must be run under ultra-high vacuum! I s = I o exp (-d / λcosθ)

Vacuum 101 Pressure (Torr) Average distance gas molecules between collisions λ Time for formation of a monolayer of gas Number of gas molecules per litre Atmospheric Pressure 760 0.000066 3.3ns 2.5x10 22 Medium Vacuum 1 0.066 3.3 µsec 2.5x10 19 10-1 660 µm 33 µsec 2.5x10 18 10-3 66 mm 3.3 msec 2.5x10 16 High Vacuum 10-6 66 m 3.3 sec 2.5x10 13 Ultra-high Vacuum 10-9 66 km 55 mins 2.5x10 10 10-10 660 km 550 mins 2.5x10 9 VISCOUS FLOW - When diameter of tube > 100λ, gas molecules more likely to bump into each other. Molecules in general move towards lower pressure end of tube. Unlikely to get backstreaming. MOLECULAR FLOW - When diameter of tube is < λ, gas molecules are more likely to collide with the tube wall than each other. There is free movement of molecules in either direction, the numbers directly related to ratio of pressures at each end of tube. At high vacuum this ratio is likely to be close to 1. Backstreaming a concern. TRANSITIONAL FLOW - Intermediate between Viscous and Molecular.

Vacuum 101 So for an 60mm diameter tube VISCOUS > 0.1 Torr > TRANSITIONAL > 1 mtorr > MOLECULAR OTHER PUMPS PUMPS Turbo DIFFUSION ION ROTARY Molecular PUMP Pump Used Used from from 10 1Pa Pa to to 10 10-8 -7 Pa Pa Used from 10-1 Pa to 10-9 Pa Cryosorption - oil free, capture, Atmosphere Extremely Used Historically from 10 atmosphere down to 0.1Pa Gas molecules -1 high Pa most speed widely (10,000rpm) that are pumped used high are mechanical vacuum Problems pump, pumps with really corrosive typically a vapor of with Diaphragm - oil free, transfer, Atmosphere trapped to 1Pa inside pump by the gettering jet pump. magnetic condensable levitation Pumping speed gases bearings virtually (H 2 O) for EM action of the sputtered Ti - constant limited Claw pump - dry, transfer, Atmosphere use. to 10Pa lifetime. below Potential 10-1 Pa source of oil contamination Molecular Drag - dry, transfer, 10Pa to 10 of Works Absolute -6 Major vacuum Pa efficiently problem freedom system in - Backstreaming; if Molecular from pressure oil in Flow line to region contamination minimised system - needs is not by using kept to be with in backed. Sublimation - oil free, capture, 10-1 Pa to 10-9 Pa no a viscous low moving vapour flow parts. pressure regime. No backstreaming Ideal for high oil. of oil when REMEMBER -- No Pump exerts a force that drags vacuum or pulls systems gas but are operating not BAD At well or NEWS near at suited atmospheric full speed. molecules to it. Pumping is purely diffusion of -gas on do systems not molecules let air pressure that into are a an from high pressure to low pressure cycled diffusion oil Major regions mist frequently concern is pump! ejected - preventing to through atmosphere. outlet physical valve. damage Must vent to pump! outside or through a filter.

Analytical Capabilities of XPS (1) Identify elements/compounds (except H and He) (2) Determine oxidation states (e.g. Ti 3+ or Ti 4+ ) (3) Identify types of chemical bonds (e.g. Si-O or Si-C) (4) Semi-quantitative analysis (10-15% error) (5) Determine adsorbate/film thickness (6) Highly surface-sensitive (3 10 nm from the surface) Detection limit 0.1 to 1 at% Ultra-high vacuum required!!! Minimize/delay surface reactions and contaminations

Analytical Capabilities of XPS Survey spectrum for element identification 25000 Survey Spectra of 061010 20000 X-Ray Source: Monochromatic Al K-alpha (1486.6eV) @ 15kV/10mA Scan Parameters: Pass-Energy 80eV, Step 0.5eV, Dwell 200msec, 5 sweeps, Charge Neutralization On Chamber Vacuum: ~ 2E-10 Torr (Base); ~ 7E-10 Torr (During Acquisition) O 1s Intensity (cps) 15000 Na 1s 10000 O KLL Auger F 1s Na KLL Auger Cl 2s P 2p Al 2p 5000 Ti 2p N 1s C 1s Cl 2p P 2s Al 2s Na 2s 0 1200 1100 1000 900 800 700 600 500 400 300 200 100 0 Binding Energy (ev)

Analytical Capabilities of XPS 1400 Binding Energy vs. Atomic Number for XPS 1300 1200 1100 1000 Binding Energy (e 900 800 700 600 500 400 300 200 100 0 C Na 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Cl Ti Atomic Number 1s 2s 2p3 2p1 3s 3p3 3p1 3d5 3d3 4s 4p3 4p1 4d5 4d3 4f7 4f5 5s 5p3 5p1 5d5 5d3 5f7 5f5 6s 6p3 6p1

Analytical Capabilities of XPS XPS spectra show characteristic "stepped" background. Due to inelastic processes (extrinsic losses) from deep in bulk. Electrons deeper in surface loose energy and emerge with reduced KE, apparent background increase at higher BE

Analytical Capabilities of XPS Typical Features and Artifacts of Core-Level Peaks Can get multiple peaks from core levels must be aware of where they come from in order to carry out chemical analysis (not all may be present) 1. Spin orbit splitting leads to additional peaks (no splitting for s, splitting for p,d,f etc.) 2. Additional peaks due to, for example, chemical shifts and oxidation states 3. Ghost peaks at lower binding energies (achromatic X-ray only) no useful info! 4. Shake up/ off peaks at higher binding energies (result of energy being transferred from the ejected photoelectron electron to a valence electron). 5. Plasmon loss peaks (due to electron excitations) 6. Photon-induced Auger peaks 7. Effects of charging of non conductive specimens

Analytical Capabilities of XPS No spin-orbital splitting for s Spin-orbital splitting for p, d, f O 1s 537.0 534.0 531.0 528.0 525.0 Binding Energy (ev)

Analytical Capabilities of XPS Ti 2p 1/2 and 2p 3/2 chemical shift for Ti and Ti4+. Charge withdrawn Ti Ti4+ so 2p orbital relaxes to higher BE

Analytical Capabilities of XPS Peak 0: C-C / C=C Peak 1: C-OH Peak 4: π π* Peak 2: C=O Peak 3: COOH

Analytical Capabilities of XPS Schematic of Ghost and Shake-up peaks 600 500 Monochromatic Achromatic 400 300 Shake-up Peak Ghost Peak 200 100 Main peak 0 293.8 293.4 293 292.6 292.2 291.8 291.4 291 290.6 290.2 289.8 289.4 289 288.6 288.2 287.8 287.4 287 286.6 286.2 285.8 285.4 285 284.6 284.2 283.8 283.4 283 282.6 282.2 281.8 281.4 281 280.6 280.2 279.8 279.4 279 278.6 278.2 277.8 277.4 277 276.6 276.2 Binding Energy Kinetic Energy

Analytical Capabilities of XPS Electrical insulators cannot dissipate charge generated by photoemission Process. Surface picks up excess positive charge - all peaks shift to higher BE Can be reduced by exposing surface to neutralizing flux of low energy electrons - "flood gun" or "neutralizer. BUT must have good reference peak.

Analytical Capabilities of XPS I s = I o exp (-d / λcosθ) I s : Intensity at surface I o : Intensity from infinitely-thick sample Beer-Lambert relationship (Numerical expression Describing the photoelectron Intensity generated from a material) d: depth λ: Inelastic mean-freepath (IMFP) θ: Spectrometer takeoff angle

Analytical Capabilities of XPS Using the Beer-Lambert expression to estimate film thickness I Film I = Film d 1 exp λ Film I Substrate I = Substrate exp λ d Substrate, Film SiO 2 Surface Layer Si Substrate

Analytical Capabilities of XPS When λ, θ and all the respective intensities are known, film thickness can be determined by taking the ratio of I film to I substrate and solve for d Si d oxide I I Si SiO 2 cosθ ln I SiO2 I Si = λsio 2 + 1 SiO 2

Analytical Capabilities of XPS Semi-Quantitative Analysis Photoelectron intensity from a homogeneous material is also dependent on instrumental factors, and can be alternatively-described by I = JCσζTλ J: X-ray flux C: Concentration of the element-of-interest σ: Ionization cross-section ζ: Spectrometer angular acceptance T: Spectrometer transmission function λ: IMFP of the element

Analytical Capabilities of XPS C = I/(JσζTλ) = I/JF F is termed the atomic sensitivity factor: - incorporates all the terms associated with the spectrometer and material - empirically-determined by XPS manufacturer - values are normalized against Fluorine

Analytical Capabilities of XPS Atomic fraction of an element (A) in a multi-component material (ABCD ) can be estimated using the following, Atomic % A = (I A /F A ) / Σ (I n /F n ) Quantification using this expression is valid only if: (1) Material is homogeneous, (2) Material surface is smooth and flat.

Instrumentation Charge Neutralizer (built into lenses) Ion Gun

Instrumentation Kratos Axis-165 XPS system in RRC-East

Instrumentation Total resolution (i.e. peak s full-width-half-maximum) of instrument is convolution of: (1) X-ray energy spread, (2) Spectrometer broadening, and (3) Intrinsic line-width of the element-of-interest. Total FWHM = {FWHM x-ray2 + FWHM spectrometer2 + FWHM intrinsic2 } 1/2 100 % Intensity 50 % Total FWHM Energy

Instrumentation Binding energy shifts due to different chemical states or bonding configurations can be subtle (1eV or less) for certain elements. Say, if achievable total FWHM of a peak is 2 ev, the XPS instrument will not be able to resolve 2 peaks that are separated only by 0.5eV!!! Total FWHM can only be decreased by minimizing the FWHM of: (1) X-ray (2) Spectrometer * Intrinsic line-widths is a non-controllable term!

Instrumentation Ideal Candidates for X-ray source: Al- and Mg-Kα (1) High energy - allow wide scan range (2) Narrow energy spread - improve resolution

Instrumentation Twin-Anode Achromatic X-ray source: Bombard metallic anode with 10-25kV electrons with ~10mA of current to generate X-rays. Can generate high X-ray flux producing high signal BUT specimen may be damaged by heat generated by the X-rays and continuum radiation and source emits X-ray satellites (additional weak lines at lower binding energies) Simple, relatively inexpensive High flux (10 10-10 12 photons s -1 ) Beam size ~ 1cm

Instrumentation Monochromatic X-Ray source: Diffraction from bent SiO 2 crystal focusing primary λ at specimen. Other λ 's focused at different points in space (filtered). Always use Al Kα which is diffracted from quartz (no equivalent crystal for Mg Kα) Beam size ~ 1 cm to 50 mm Electron Spectrometer Quartz Crystal Eliminates satellites peaks simpler spectra Decreases FWHM of X-ray energy Flux decreases at least an order of magnitude leading to less damage, improves S/N (no X-ray continuum) but lower signal Sample X-ray Source Rowland Circle More complicated and expensive

Instrumentation Most common type of electrostatic deflection-type analyzer: Concentric Hemispherical Analyzer (CHA) or spherical sector analyzer Energy resolution dependant on radius. Capable of collecting photoelectrons of larger angular distribution. Photoelectrons of a specific energy are focused by the lens at the slit of the spectrometer. Lens also controls sampling area. Photoelectrons travel through a circular path and exit into a series of channeltrons (electron multipliers).

Instrumentation An electron of kinetic energy ev = V o will travel a circular orbit through hemispheres at radius R o Since R o, R 1 and R 2 are fixed, in principle changing V 1 and V 2 will allow electrons of different KE to be detected. Negative potential on two hemispheres V 2 > V 1 Potential of mean path, R o through analyzer is V o = (V 1 R 1 +V 2 R 2 )/2R o

Instrumentation Single-Channel Electron Multipliers (Channeltron) - Hollow glass tube with semiconducting layer on inner surface - Electrons/ions entering Channeltron produce secondary electrons (SEs) - Avalanche effect as SEs accelerate down the tube under HV - Signal amplified by ~ 10 6 to 10 8

Instrumentation But E/E o = S/2R o E = (S/2R o )E o Peak FWHM Spectrometer Term (constant) Photoelectron KE Spectrometer broadening (i.e. FWHM spectrometer ) is a function of photoelectron KE entering spectrometer i.e. Resolution is non-uniform across XPS energy spectrum!

Instrumentation To circumvent the problem (1) Hemisphere potentials are fixed to allow electrons with a fixed KE (the pass energy, E p ) to reach the electron detectors (Channeltrons); (2) Electrostatic lens before the slit decelerate photoelectrons of a particular KE to E p ; (3) Magnetic lens focus these electrons with E p at the slit, so that only electrons with that pass energy are allowed to enter the slit. Scanning of the energy-scale is achieved by varying the decelerating potential on the electrostatic lens instead of the hemisphere potentials Fixed energy-resolution across the energy-scale!!! Something to Ponder Should an infinitesimally-small pass energy yield an extremely mild spectrometer broadening???

Instrumentation Most XPS systems also come equipped with an ion gun. Purpose: (1) Removes surface contaminants (usually oxides and hydrocarbons); (2) Allows depth-profiling study. How does it clean a surface? (1) Ionizes an inert gas (usually Ar) (2) Focuses and accelerates the ions towards the specimen surface (3) Rasters the ion beam across the surface (4) Ions impart energy to surface contaminants Sputtering them away!

Instrumentation Example: Successful removal of native oxide from CdTe Surface Atomic Fraction of Cd Atomic Fraction of Te Cd: Te Ratio As-received 0.65 0.35 1.86 10 mins Sputter 0.51 0.49 1.04 20 mins Sputter 0.55 0.45 1.22

Instrumentation Potential Problems Associated with Sputter-Cleaning: Surface roughening; Alter oxidation states of some elements; Selective etching of surface due to different sputtering rates in a multi-component material.

Instrumentation Example: Oxidation-state changed in Pd after sputtering 500 450 400 Pd 3d (As-received) Pd 3d (10 min Sputter) PdO 3d5/2 Pd 3d3/2 Orbitals Elemental Pd 3d5/2 Intensity (CPS) 350 300 250 200 150 100 352 350 348 346 344 342 340 338 336 334 332 330 328 326 Binding Energy (ev)

Spectrum Simulation (Peak-Fitting) Simulation of XPS spectra (i.e. peak fitting) to match experimentallyobserved spectra. Purpose: (1) Background noise subtraction to reveal true peak intensities; (2) Precise determination of peak position; (3) Deconvolute spectra into individual components when 2 or more peaks are in close proximity.

Spectrum Simulation (Peak-Fitting) Ce 3d3/2 (3+) s Ce 3d3/2 (4+) s Ce 3d5/2 (3+) s Ce 3d5/2 (4+) s

Spectrum Simulation (Peak-Fitting) Select Background Model Background Quality? Good Poor Modify Peak Parameters Adding Synthetic Peak Option 2 Option 1 Providing Initial-Guess For Peak Parameters: (1) Line position (2) Area (3) FWHM (4) Gaussian-Lorentzian Ratio (5) Asymetry (6) S.O.S. Software Fitting Fit Quality? Poor Good End of Fitting

Spectrum Simulation (Peak-Fitting) Types of Background: Linear (1) Linear (2) Shirley (typically-used) (3) Tougaard Shirley Tougaard

Spectrum Simulation (Peak-Fitting) Types of Line-Shapes: (1) Gaussian Function Describes the measurement process (e.g. instrumental response, X-ray line-shape, Doppler and thermal broadening) (2) Lorentzian Function Describes lifetime broadening (intrinsic line-width) XPS peaks can usually be described by varying G-L ratio

Spectrum Simulation (Peak-Fitting) Peaks associated with pure metals may tend to be asymmetic Need to introduce a Tail Modifier term to the G-L functions

Spectrum Simulation (Peak-Fitting) XPSPEAK version 4.1 is a free windows-based XPS peak fitting program written by Prof. Raymund Kwok (Chemistry, CHUK). Installation file downloadable at: http://www.uksaf.org/xpspeak41.zip You will need this software to complete the XPS assignments! A peak-fitting procedure will be demonstrated using XPSPEAK during the lab session.

Summary X-ray Photon Spectroscopy (XPS) (Originally called ESCA) is a surfacesensitive technique that probes the chemical properties of the top ~10 nm of a solid surface that must be UHV-compatible no wet specimens! It is the most versatile and quantifiable of all the surface chemical analysis methods: elemental ID, oxidation-states, chemical bonds, film thickness, depth-profiling, semi-quantitative analysis Limitations: (1) does not detect H or He; (2) Radiation damage possible (worse for achromatic sources); (3) Charge neutralization needed for insulating material; (4) Chemical analysis can be limited to functional groups and in some cases chemical shifts are not resolvable.

Electron Microscopy Service @ UIC JEM-1220 JEM-3010 JEM-2010F HB601UX JSM-6320F AXIS-165 XPS VT-SPM Ramascope 2000 S-3000N