Lecture 5—chemical shift1 Quantization and depth effects, XPS and Auger I.XPS: The Chemical Shift II.Mean free path, overlayer attenuation, etc. III.Auger spectroscopy, final state effects

The XPS Chemical Shift: Shifts in Core level Binding Energies with Chemical State 2 ΔE Chemical Shift In part fromC. Smart, et al., Univ. Hong Kong and UWO

The binding energy is defined as: E b = hv –E k –Φ Where hv= photon energy E k = kinetic energy of the photoelectron Φ = work function of the spectrometer Specifically, the CHEMICAL SHIFT is ΔE b That is the change in E b relative to some chemical standard 3Binding energies and particle size

4 Chemical Shift in Au compounds vs. bulk elemental gold PHI handbook

EFEF E vacuum EBEB hv E kin e- Φ spectrometer Because the electron emitted from the solid has to impact on the analyzer/dectector to be counted, the relationship E kin and E B has to include the work function term of the detector (typically, 4-5 eV): E kin = hv-E B – Φ spectrometer We only need the work function term for the spectrometer, not the sample, because (for a conducting sample) the two Fermi levels are coupled. Obviously, electrically insulating samples present problems (Charging) E vacuum E kin 5

EFEF E vacuum EBEB hv E kin e- Φ spectrometer E vacuum E kin Changes in E B result from : 1.Changes in oxidation state of the atom (initial state effect) 2.Changes in response of the system to the core hole final state: ΔE B = ΔE(in.state) – ΔR + other effects (e.g., band bending) where ΔR = changes in the relaxation response of the system to the final state core hole (see M.K. Bahl, et al., Phys. Rev. B 21 (1980) mainly sometimes

7

8 Primarily an initial state effect

9 ΔE b = kΔq i + ΔV ij V ij often similar in different atoms of same material, so Δv ij is typically negligible

Binding energies and particle size10 ΔE b = kΔq i + ΔV ij Initial state term, often similar for diff. atoms in same molecule In principle, can be obtained from ground state Mulliken Charge Density calculations Valence charge is removed or added to an atom by interaction with surrounding atoms.

Binding energies and particle size11  Chemical shift is dominated by changes in ground state valence charge density:  Changes in valence charge density dominated by nearest-neighbor interactions  Qualitative interpretation on basis of differences in ground state electronegativities

Binding energies and particle size12 C O e- EN = 2.5EN = 3.5 CTi EN = 1.5 e- C C O withdraws valence charge from C: C(1s) shifts to higher BE relative to elemental C (diamond) at eV Elemental C: binding energy = eV Ti donates charge to C, binding energy shifts to smaller values relative to 285 eV

13 Thus, a higher oxidation state (usually) yields a higher binding energy!

Binding energies and particle size14 Electron withdrawing groups shift core levels to higher binding energy

Binding energies and particle size15 Binding energy shifts can be used to follow the course of surface reactions for complex materials: e.g., atomic O /(Pt)NiSi (e.g., Manandhar, et al., Appl. Surf. Sci. 254(2008) 7486 = Ni = Si NiSi (Schematic, not real structure) Bulk Vacuum Atomic O

Binding energies and particle size16 Pauling Electronegativities, Ground State Si = 1.8 O = 3.5 Ni = 1.8 Ni-O or Si-O formation  shift of Ni or Si to higher BE Question: Ni-Si  Ni-Ni. Which way should BE move (think).

17 Si SiO 2 Exposure to atomic O XPS binding energy shifts for Pt-doped NiSi as a function of exposure to atomic O at room temp. (Manadhar, et al., Appl. Surf. Sci. 254 (2008) 7486 SiO 2 peak appears (shift to higher BE) Ni (2p) shifts to lower BE. Why?

PtSi  Pt 1+y Si NiSi  Ni 1+x Si Si transport and oxidation Pt 1+y Si Ni 1+x Si (B) Si transport kinetically inhibited, metal oxidation Si  SiO 2 Pt silicate formation (A) Preferential Si oxidation, Si flux creates metal-rich substrates O + O 2

19 How do we estimate q, Δq? This is usually done with Mulliken atomic charge densities, originally obtained by LCAO methods: Ψ MO = c a Φ a + c b Φ b Φ a(b) atomic orbital on atom a (b)  Ψ  2 = c a c a * Φ a Φ a * + [cross terms] + c b c b * Φ b Φ b * Atomic charge on atom a Atomic charge on atom b Overlap charge

C 2 -B-H C-B-H B-B-H RC-BRC-B Different Boron Environments in orthocarborane derived films (B 10 C 2 H X and B 10 C 2 H X :Y) R c =Ring carbon

Figure 3 B 2 -B CB-B C 2 -B C 2 -B-H C-B-H B-B-H

Binding energies and particle size22 Chemical Shifts: Final Note Calculating ground state atomic charge populations with DFT: Minimal basis sets give best results (LCAO-MO) Such basis sets are not best for lowest energy/geometric optimization

Binding energies and particle size23 Attenuation: Clean surface of a film or single crystal hv e- I = I 0 d film or single crystal with overlayer of thickness d I = I 0 exp(-d/λ) hv Issues: 1.Average coverage 2.Calculating λ 3.Relative vs. Absolute intensities

Binding energies and particle size24 Monolayer Surface coverage = Θ 1 d = d 1 Bilayer Surface coverage = Θ 2 d = d 2 Bare surface Coverage = 1-( Θ 1 +Θ 2 ) We can only measure a total intensity from a macroscopic area of the surface: I = [1-( Θ 1 +Θ 2 )] I 0 + Θ 1 I 0 exp[-d 1 /λ] + Θ 2 1 I 0 exp[-d 2 /λ] = I 0 exp[-d ave /λ]  we can only determine average coverage with XPS!

Binding energies and particle size25 Consider 2 cases: 1.d ave < 1 ML (0< Θ<1) 2.d ave > 1 ML ( Θ> 1) We need to look at the RATIO of I substrate (I B ) and I overlayer (I A ) Why? Absolute intensity of I B can be impacted by: 1.Small changes in sample position 2.Changes in x-ray flux I B /I A will remain constant

Binding energies and particle size26 Calculation of the overlayer coverage First, we need to calculate the IMFP of the electrons of the substrate through the overlayer and the IMFP of the electrons in the overlayer. The formula to calculate the IMFP is (NIST): IMFP=E/E p 2 ([βln(γE)-(C/E)+(D/E2])

Binding energies and particle size27 ElementNvNv ρ ( g cm -3 ) M E (ener gy) E g (Band Gap EpEp βγUCD(E p ) 2 ln(γE)(C/E) (D/E 2 ) [βln(γE)- (C/E)+(D/E 2 ] E p 2 ([βln(γE)- (C/E)+(D/E 2 ]) IMFP=E/E p 2 ([βln(γE) -(C/E)+(D/E2]) Sulfur Co E O in MgO th-- C E C in C th--C Co Thr--C E C in C th--Co E O Thru C Ni Thru C E Mg thru C E Fe thru C E Mg thru MgO E

Terms used in the excel sheet (example Carbon through MgO) ColumnTerm used 1Valence electrons of the element (O) 2Density of the over layer (Carbon) 3Mass of the over layer 4Kinetic Energy of the element(O) After you insert all the four columns, the IMFP is calculated on its own.

Binding energies and particle size29 overlayersubstrateoverlayersubstrate dA(Ini*IcS)B(Ic*IniS)CSi =D6*EXP(-A6/26.36) =E6*(1-EXP(-A6/33.17)) =Area under the curve1915/0.25 =Area under the curve 54544/0.66

Binding energies and particle size30

Take-off angle variations in XPS: Definition θ Take off angle (θ) is the angle between the surface normal and the axis of the analyzer. (Some people use 90-θ) Surface normal θ = 0  normal emission θ=89   grazing emission

Take-off angle variations in XPS: Intensity vs. θ Intensity of a photoemission peak goes as I ~ I cosθ Therefore, intensities of adsorbates and other species are NOT enhanced at grazing emission (large θ)!

Take-off angle variations in XPS: Sampling Depth (d) normal emission (θ = 0) d ~ λ (inelastic mean free path) λ λ λcosθ θ increased take-off angle: d~ λ cosθ (reduced sampling depth)

d~ λ cosθ: Effective sampling depth (d) decreases as θ increases Relative intensities of surface species enhanced relative to those of subsurface: Si SiO 2 λ Si SiO 2 λcosθ SiO 2 Si

In Dragon and other systems: Si SiO 2 Ta sample holders Arrangement of sample holder may cause increased signal from Ta or other extraneous materials. These should be monitored. However, enhancement of SiO 2 relative to Si will remain the same.

Binding energies and particle size36

Binding energies and particle size37 Multiplet Splitting: 1.Valence electrons give rise to different spin states (crystal field, etc.  Cu 2p 3/2 vs. ½ states 2.Formation of a core hole shell yields an unpaired electron left in the shell 3.Coupling between the core electron spin and valence spins gives rise to final states with different total angular momentum.

Binding energies and particle size38 2p 1/2 2p 3/2 Multiplet splitting in Cu

Binding energies and particle size39 Auger Spectroscopy: Final State Effects hv or e- XPS initial State XPS Final State Auger Initial State Auger Final State

Binding energies and particle size40 Kinetic Energy of Auger Electron: This transition is denoted as (KLL) K (1s) L 1 (2s) L 2,3 (2p) e- K (1s) L 1 (2s) L 2,3 (2p) Initial state Final State KE Auger = E K - E L1 – E L2,3 - U eff ~ E K – E L -E L - U eff Note: Auger transitions are broad, and small changes in BE (E L1 vs. E L2,3 ) sometimes don’t matter that much (sloppy notation) What is U eff ? e-detector

Binding energies and particle size41 K (1s) L 1 (2s) L 2,3 (2p) U eff is the coulombic interaction of the final state holes, as screened by the final state response of the system: e.g., Jennison, Kelber and Rye “Auger Final States in Covalent Systems”, Phys. Rev. B. 25 (1982) 1384

Binding energies and particle size42 For a typical metal, the final state holes are often delocalized (completely screened), and U eff ~ 0 eV. However, for adsorbed molecules, or nanoparticles, the holes are constrained in proximity to each other. U eff can be large, as large as 10 eV or more. Nanoparticle, U eff ~ 1/R R Heat in UHV Agglomeration, should see shift in Auger peak as U eff decreases

Binding energies and particle size43 KE(LVV) = E L –E V – E V – U eff as particle size increases, U eff decreases Note shift in Cu(LVV) Auger as nanoparticles on surface agglomerate J. Tong, et al. Appl. Surf. Sci. 187 (2002) 253 Cu/Si:O:C:H

Binding energies and particle size44 Similar effects in Auger KE are seen for agglomeration during Cu deposition at room temp. (Tong et al.) Note corresponding change in Cu(2p 3/2 ) binding energy. Cu(LVV) shift with increasing Cu coverage

Binding energies and particle size45 Auger in derivative vs. integral mode When doing XPS, x-ray excited Auger spectra are acquired along with photoemission lines

Binding energies and particle size46 Auger spectra, though broad, can give information on the chemical state (esp. if the XPS BE shift is small as in Cu(0) vs. Cu(I) Above spectra are presented in the N(E) vs. E mode—or “integral mode”

Binding energies and particle size47  However, in some cases Auger spectroscopy is used simply to monitor surface cleanliness, elemental composition, etc. This often involves using electron stimulated Auger (no photoemission lines).  Auger spectra are typically broad, and on a rising background. Presenting spectra in the differential mode (dN(E)/dE) eliminates the background.  Peak-to-peak height (rather than peak area) is proportional to total signal intensity, and the background issue is eliminated. Except in certain cases, however, (e.g., C(KVV)) most chemical bonding info is lost.

Binding energies and particle size48 Auger (derivative mode) of graphene growth on Co 3 O 4 (111)/Co(0001) (Zhou, et al., JPCM 24 (2012) Homework: explain the data on the right.

Binding energies and particle size49 N(E) KE Peak-to-peak height