The binding energy measured in XPS reflects the total energy required to remove a core electron from an atom to the Fermi level. That energy is sensitive to the local electrostatic potential at the nucleus, which is determined by:

• the nuclear charge
• the shielding provided by valence electrons
• the charge donated to or withdrawn from the atom by its bonding partners

Bond order modulates all three of the latter terms. As bond order increases between carbon and oxygen, for example, the oxygen draws progressively more electron density away from carbon. The reduced valence electron density at the carbon nucleus means less shielding of the core electrons, raising their binding energy. This is the chemical shift — and bond order is one of its most systematic predictors.

The relationship can be expressed through the charge potential model:

E_B = k·q + V + E_ref

where q is the atomic charge and V is the Madelung-like potential from surrounding atoms. Higher bond order to electronegative neighbours increases q positively, shifting the binding energy to higher values. The model is approximate but directionally reliable and useful for rationalising spectra from organic and inorganic systems alike.

Carbon is the element where bond order effects in XPS are most routinely encountered and most practically important. The C 1s binding energy scales predictably with the oxidation state of carbon, which in organic chemistry maps closely to bond order with oxygen:

Each step of approximately 1.5 eV per unit increase in bond order to oxygen is a useful working rule, though inductive effects from neighbouring groups and the specific molecular environment will modulate the exact shift. The incremental nature of these shifts is why peak fitting of complex polymers or functionalised surfaces can resolve four or five chemically distinct carbon environments in a single C 1s envelope.

In transition metal oxides, bond order considerations underpin the interpretation of both the metal core levels and the O 1s spectrum. As metal oxidation state increases — which corresponds to increasing effective bond order in the metal–oxygen interaction — the metal binding energy shifts to higher values due to reduced d-electron shielding.

For a series such as W(0) → W(IV) → W(VI), each step represents an increase in the number of W–O bonds per tungsten centre, and increases the binding energy. The O 1s spectrum simultaneously shows sensitivity to bond order: lattice oxygen in a fully coordinated oxide environment (high bond order, high local symmetry) appears at lower binding energy (~530 eV) than hydroxyl oxygen (single bond, ~531.5 eV) or adsorbed water (~532.5 eV).

This hierarchy is directly interpretable through bond order: the greater the formal bond order between oxygen and its bonding partner, the greater the back-donation of electron density from oxygen into the bonding interaction and, consequently, the lower the O 1s binding energy.

Aromatic and conjugated systems present cases where bond order is a non-integer and delocalisation must be accounted for. In benzene, for example, the C–C bond order is 1.5 rather than alternating 1 and 2. The C 1s spectrum reflects this through a narrow, symmetric peak at approximately 284.5 eV — characteristic of equivalent aromatic carbons with identical bond orders and equivalent electrostatic environments.

Departures from uniform delocalisation, such as functionalisation at one position on an aromatic ring, break the equivalence. The perturbed carbons shift according to the electron-withdrawing or electron-donating character of the substituent, which can be rationalised through the inductive and resonance contributions to effective bond order with the substituent.

In nitrogen-containing heterocycles — pyridine, pyrrole, pyrimidine — the N 1s binding energy reflects the bond order to adjacent atoms and the degree to which the nitrogen lone pair participates in the aromatic π system. Pyrrolic nitrogen (lone pair delocalised into the ring, effective N–C bond order > 1) appears at lower binding energy than pyridinic nitrogen (~398.5 eV vs ~400 eV), a distinction routinely used in characterising nitrogen-doped carbon materials and electrocatalysts.

Bond order has a secondary but diagnostically important influence on satellite structure in XPS. π→π* shake-up satellites, most familiar in the Cu 2p spectrum of Cu²⁺ compounds and in aromatic carbon systems, arise from the simultaneous excitation of a valence electron from a bonding π orbital to an antibonding π* orbital during the photoemission process. Their presence and intensity depend on the existence of accessible π states — which requires bonds of order two or greater, or aromatic delocalisation.

In purely single-bonded systems (bond order = 1 throughout), π→π* satellites are absent. Their appearance in a spectrum is therefore direct evidence of multiple bonding or conjugation. In transition metal compounds, the intensity of multiplet-split and satellite features scales with the number of unpaired d electrons, which itself reflects the effective metal–ligand bond order and the degree of covalency in the metal–oxygen or metal–nitrogen interaction.