Wait, What? X-ray photoelectron spectroscopy can tell you that oxygen is present near a surface — and, with careful interpretation, whether that oxygen is more consistent with an oxide, hydroxide or another chemical environment. Yet the instrument never sees an “oxidation state” directly.
XPS measures electron energies. Chemistry is inferred from how those energies shift, split and acquire satellite structure.
Direct answer
In XPS, a material is irradiated with X-rays of known photon energy. A photon can eject a core electron through the photoelectric effect. The analyser measures the photoelectron kinetic energy, and the instrument converts that to a binding-energy scale. For a solid referenced to the spectrometer Fermi level, the practical relation is commonly written
EB = hν − EK − φspec
where EB is binding energy, hν is photon energy, EK is measured kinetic energy and φspec is the spectrometer work-function term. Core-level energies depend on element and chemical environment. Because photoelectrons lose energy rapidly while travelling through matter, only those generated near the surface escape without inelastic loss, making XPS strongly surface sensitive.
Learning progression
Beginner: X-rays knock electrons out of atoms, and the measured electron energy carries information about where the electron came from.
Secondary Chemistry: different elements have different electron binding energies; electron shells and chemical bonding affect energy.
JC / undergraduate: interpret survey and high-resolution spectra, core-level binding energies, spin–orbit doublets, chemical shifts and approximate surface atomic composition.
Advanced / professional: manage energy referencing, charging, final-state effects, line shapes, satellites, inelastic backgrounds, information depth, sensitivity factors, sputter artefacts and uncertainty before assigning chemical states.
1. XPS begins with the photoelectric effect — but Chemistry owns the interpretation job
The photoelectric effect explains why a sufficiently energetic photon can eject an electron. That underlying physical phenomenon has its own Physics owner. XPS uses it as a chemical measurement mechanism: the question here is how measured electron energies become evidence about composition and chemical state at a surface.
2. Why core electrons identify elements
Core electrons experience the nuclear charge and electron shielding characteristic of each element. Consequently, C 1s, O 1s, Si 2p, Fe 2p and other levels appear in distinctive binding-energy regions. A broad survey spectrum therefore provides an elemental inventory, with important exceptions such as hydrogen and helium, which ordinary XPS does not detect in the same direct core-level way.
3. Why Al Kα appears so often
Modern laboratory XPS commonly uses monochromated aluminium Kα radiation with photon energy about 1486.6 eV. Magnesium Kα at about 1253.6 eV is another classical source. Higher-energy laboratory and synchrotron sources can produce more energetic photoelectrons and increase information depth.
4. Binding energy is an energy difference with a reference
A binding-energy number is meaningless without knowing the energy reference. Conductive samples electrically connected to the instrument can be referenced to a common Fermi level. Insulators can charge under X-ray irradiation, shifting apparent peak positions. That means an apparent 1 eV “chemical shift” may instead be an electrostatic charging shift unless the measurement and referencing are controlled.
5. Chemical shifts are real — but not one-to-one oxidation-state labels
Changing bonding changes electron density and screening around an atom, which can shift the energy required to remove a core electron. More electron-withdrawing environments often move a core-level peak toward higher binding energy, but that rule is not universal enough to assign oxidation state by direction alone.
Observed binding energy contains both initial-state contributions — the electronic structure before ionisation — and final-state contributions — how the material relaxes and screens the core hole after photoemission. Metals, oxides, strongly correlated solids and nanoscale materials can therefore produce counter-intuitive shifts.
6. Spin–orbit splitting is structured evidence
Core levels with orbital angular momentum greater than zero can split through spin–orbit coupling. A p level gives p1/2 and p3/2 components; a d level gives d3/2 and d5/2. The separation and approximate area ratio are constrained by atomic physics. A chemically plausible fit should respect those constraints instead of adding arbitrary peaks to improve residuals.
7. Satellites are not noise by default
Transition-metal compounds can show shake-up, charge-transfer and multiplet features alongside the main photoelectron line. These extra structures contain chemical information. For example, two compounds with similar nominal metal oxidation states may have distinctly different satellite patterns because their electronic configurations and ligand interactions differ.
8. Why XPS is surface sensitive
Electrons travelling through a solid undergo inelastic scattering. At typical XPS kinetic energies, their inelastic mean free paths are often on the nanometre scale. The probability that a photoelectron escapes without an energy-loss event therefore falls approximately exponentially with travel distance.
This gives XPS an information depth commonly of only a few nanometres for standard laboratory conditions. “Surface sensitive” does not mean a mathematically sharp depth boundary: contributions decay continuously with depth and depend on material, kinetic energy and emission angle.
9. Angle changes the effective sampling depth
At more grazing electron-emission angles, photoelectrons travel a longer path through the solid before reaching the surface. Angle-resolved XPS therefore changes the relative weighting of shallower and deeper material without physically sputtering the sample. The interpretation still requires a depth model and sufficiently smooth, well-defined layers.
10. Peak area is not directly atomic percentage
Different core levels have different photoionisation cross sections, analyser transmission and electron attenuation. Quantitative XPS therefore applies relative sensitivity factors or a more complete instrument-response model. A simplified composition estimate is based on corrected intensities:
xi ≈ (Ii/Si) / Σ(Ij/Sj)
where I is an integrated peak intensity and S a sensitivity factor. The result is an estimate for the sampled surface region, not automatically the bulk stoichiometry.
11. Background subtraction encodes an assumption about scattered electrons
Photoelectron peaks sit on a background created largely by inelastic scattering. Shirley and Tougaard-type backgrounds model that contribution differently. Background choice can alter fitted peak areas and minor-component fractions. A fit is therefore not just geometry; it contains a physical assumption about where intensity belongs.
12. Peak fitting can create chemistry that is not there
A mathematically excellent fit with many components is not necessarily a chemically valid fit. Each proposed component should have a reason: known chemical states, required spin–orbit partners, plausible widths and line shapes, corroborating elements, or independent reference data. Overfitting turns noise and background imperfections into fictional species.
13. Charging is one of the most dangerous confounders
Photoemission removes electrons. A poorly conducting sample can become positively charged, shifting peaks toward higher apparent binding energy. If different surface regions charge differently, peaks can broaden or distort. Electron or ion neutralisation can reduce charging, but the remaining reference uncertainty must still be considered.
14. The adventitious-carbon shortcut has limits
It is common to reference an insulating sample by assigning the ubiquitous hydrocarbon C 1s signal a fixed binding energy, often near 284.8 eV. Modern surface-analysis literature has highlighted that this practice can fail because contamination chemistry, sample work function and charging vary. A reference should be justified rather than applied automatically.
15. Sputter depth profiling changes the sample while measuring it
Ion sputtering can expose deeper layers, but bombardment may preferentially remove one element, reduce oxides, mix interfaces or damage organics. A depth profile is therefore a measurement of a sample that is being modified. Cluster-ion sources can reduce damage for some soft materials, but they do not remove the need for controls.
16. Hard X-rays extend the depth window
Hard X-ray photoelectron spectroscopy, HAXPES, uses higher photon energies to produce higher-kinetic-energy electrons with longer escape depths. That makes buried interfaces more accessible, but usually reduces surface selectivity and changes cross sections and instrumental demands. HAXPES is an extension of the measurement window, not “better XPS” for every problem.
17. Ambient-pressure XPS changes the chemical question
Traditional XPS operates under high or ultra-high vacuum because electrons are readily scattered by gas molecules. Differential pumping and specialised analysers now allow near-ambient-pressure measurements, enabling surface chemistry under reactive gases and, in specialised geometries, more realistic catalytic or interfacial conditions. The gain in chemical realism comes with additional scattering and interpretation challenges.
Observation versus inference
Observation: a peak occurs at a calibrated binding energy with a measured width, area, spin–orbit structure and satellite pattern.
Inference: that signal is assigned to an element and chemical environment by comparison with reference data and electronic-structure expectations.
Stronger inference: the exact oxidation state, coordination motif or interfacial reaction mechanism. That may require multiple peaks, complementary spectroscopy, diffraction, electrochemistry or controlled perturbation rather than one binding-energy number.
How we know
- NIST’s XPS database contains tens of thousands of binding-energy, chemical-shift and Auger data records for reference comparison.
- Calibrated elemental standards test the analyser energy scale.
- Known stoichiometric compounds test sensitivity-factor quantification.
- Angle variation and higher photon energies test depth assignments.
- Independent diffraction, Raman, infrared, electrochemical or chemical analysis can test XPS chemical-state assignments.
- Repeated spectra at increasing dose reveal beam-induced changes.
Competing explanations to test
A shifted peak could reflect a different oxidation state, but also differential charging, altered screening, particle size, interface dipoles, a changed reference level or overlapping chemical components. A broadened peak could indicate multiple species, but also charging or poor instrumental resolution. A reduced oxide signal after sputtering could indicate a real depth gradient, or sputter-induced reduction. XPS becomes strong chemistry only when those alternatives are tested.
Misconceptions worth hunting
- “XPS measures oxidation state directly.” It measures electron energies; oxidation state is inferred.
- “Every chemical state has one universal binding energy.” Reference, environment and final-state screening matter.
- “The largest peak means the most abundant element.” Sensitivity factors differ.
- “Surface sensitive means one atomic layer.” The sampled signal extends over a depth distribution.
- “A perfect peak fit proves the species exist.” Overfitting can manufacture components.
- “Sputtering simply reveals the untouched layer below.” It can chemically modify the surface.
- “Charging is only a uniform shift.” Differential charging can broaden and distort spectra.
- “XPS and EDX measure the same depth.” XPS is far more surface weighted under typical conditions.
Transfer checks
- If photon energy rises while a given core-level binding energy is unchanged, should its photoelectron kinetic energy rise? Yes.
- If an insulating sample charges positively, can peaks shift to apparently higher binding energy? Yes.
- If a fitted component has no chemically plausible partner or independent evidence, should a lower residual alone justify it? No.
- If the take-off geometry becomes more surface sensitive, should an ultrathin contamination overlayer contribute relatively more? Yes.
- If sputtering makes a metal oxide look more reduced, is a genuine subsurface reduction the only explanation? No.
Independent reasoning check
Suppose an O 1s peak shifts by 1.2 eV after treatment. Before naming a new oxygen species, write down at least four non-equivalent explanations: chemical-state change, charging/reference shift, final-state screening change and overlapping-component redistribution. Then ask what independent observation would discriminate each one. That is the difference between peak naming and analytical chemistry.
Practical interpretation
Read an XPS result in this order: energy calibration → charging/reference quality → survey composition → high-resolution line constraints → background and fit model → chemical-state alternatives → depth sensitivity → independent corroboration. Starting with “What oxidation state is this peak?” skips the controls that make the answer trustworthy.
Canonical connections
- How Chemistry Works remains the discipline-level Chemistry router.
- Infrared Spectroscopy owns vibrational absorption and dipole-change evidence.
- Raman Spectroscopy owns vibrational scattering and polarizability evidence.
- Mass Spectrometry owns gas-phase ion mass-to-charge analysis.
Research foundations
- IUPAC Gold Book, 5th ed. online (2025): X-ray photoelectron spectroscopy and photoelectron spectroscopy terminology.
- NIST X-ray Photoelectron Spectroscopy Database, SRD 20, for reference binding energies, chemical shifts and Auger data.
- NIST electron inelastic-mean-free-path data and TPP-2M assessments for surface-analysis depth modelling.
- Modern XPS best-practice literature on energy referencing, charging, peak fitting, HAXPES and ambient-pressure XPS.
The quiet ending
The beginner asks, “Which element made this peak?” The developing analytical chemist asks, “Why did the binding energy shift?” The professional asks:
Can I trace the chemical-state claim all the way back to calibrated electron energies, physically constrained line shapes, valid depth assumptions and independent evidence — without mistaking a fitted spectrum for direct sight of the chemistry itself?
