An XAS scan from low energy to high energy looks like a flat line, then a step, then a wobble. The usual split is:

  • XANES (X-ray absorption near-edge structure): the pre-edge, the edge, and roughly the first 50 eV above it.
  • EXAFS (extended X-ray absorption fine structure): the decaying oscillations from about 50 eV above the edge out to several hundred electronvolts.

In soft X-ray and surface-science papers, the near-edge region is often called NEXAFS (near-edge X-ray absorption fine structure). The names differ; the region is the same. Soft X-ray groups tend to say NEXAFS, hard X-ray groups tend to say XANES.

Inside a beamline at DESY / HASYLAB. The monochromator and experimental chamber turn a continuous spectrum into a scanned photon energy. Image: Wikimedia Commons.

XANES / NEXAFS: valence, symmetry, empty orbitals

Near the edge, the initial state is a core level (1s for a K-edge, 2p for an L-edge) and the final state is an unoccupied state above the Fermi level. The spectrum therefore reports:

  • Oxidation state / valence. The edge often shifts with valence. That shift is the chemical shift.
  • Coordination symmetry. Some pre-edge peaks are allowed only for particular symmetries. On some transition-metal K-edges, the pre-edge is used to discuss tetrahedral versus octahedral sites.
  • Orbital character. Light-element K-edges show π* and σ* resonances. Graphite, carbonate, and carbonyl carbon spectra can differ by a whole fingerprint.

For 3d metals, a soft X-ray L-edge is often more chemically direct than a hard X-ray K-edge. The L-edge is 2p → 3d, so it lands on the d orbitals that matter for valence, with rich multiplet structure and sensitivity to high-spin / low-spin character and charge transfer. The cost is that the theory is harder: a fit is not just an edge shift.

The useful order is: compare reference spectra first, then talk about fitting or calculation. Metal foils, oxides, and known complexes are the ruler. Guessing a valence from a peak, with no ruler, easily turns surface contamination or charge neutralisation into a false conclusion.

The absorption-edge family. XANES / NEXAFS sits on the edge; EXAFS is the longer oscillation above it. Image: Wikimedia Commons.

EXAFS: turning oscillations into bond lengths

As the photon energy rises further, the core electron is promoted into the continuum. The outgoing photoelectron is scattered by neighbouring atoms and interferes with the outgoing wave. Constructive and destructive interference, as a function of energy, is the EXAFS oscillation.

After the standard treatment — edge subtraction, normalisation, $k$-weighting, and a Fourier transform — the oscillation can be read for:

  • the distance to neighbouring atoms (bond length, often to about 0.01 Å when the data allow it);
  • the coordination number and the Debye–Waller factor (disorder);
  • sometimes the identity of the scattering atom.

Soft X-ray EXAFS is not the common case. The energy window is short, the background and self-absorption are more awkward, and the statistics are often worse than a hard X-ray K-edge. Many SXAS papers stop at NEXAFS, where valence and fingerprint are already enough. Hard X-ray EXAFS is the routine tool for coordination chemistry and in-situ catalysis.

One spectrum, two questions

XANES / NEXAFS EXAFS
Energy window pre-edge to about 50 eV past the edge tens to hundreds of eV above the edge
Main physics transitions into unoccupied bound or quasi-bound states photoelectron scattering from neighbours
Usual answers valence, symmetry, orbital fingerprint bond length, coordination number, disorder
In the soft X-ray very common, especially C/N/O K-edges and metal L-edges limited window; use it carefully

Experimentally they are two slices of one scan, not two instruments. In a report they are better kept apart: use the near edge to say what the chemical state is, and the extended region, if it is there, to say how far the neighbours are.

Two mistakes are worth blocking early.

  1. Do not read a XANES pre-edge height as a bond length.
  2. Do not treat an uncorrected Fourier-transform peak as a physical distance. In $R$-space the apparent peak is usually about 0.3–0.5 Å shorter than the true bond length. The distance that belongs in a paper is the one after a phase-shift correction from a scattering-path fit, for example FEFF in Athena / Artemis.

The next note walks through a typical soft X-ray experiment: source, monochromator, TEY / TFY, and normalisation.

How to cite:

Tianlu Pang. XANES and EXAFS: two ways to read one absorption spectrum[EB/OL]. PANG Blog, 2026. https://blog.pang.work/en/2026/08/19/xanes-and-exafs/