Coulomb Explosion: Strip the Electrons and the Molecule Tears Itself Apart

Removing the glue faster than the nuclei can react

A molecule is a set of positively charged nuclei that would violently repel one another if the shared valence electrons were not sitting between them, cancelling most of that repulsion and adding an attractive exchange term. Chemistry works because the electrons stay. A Coulomb explosion is the deliberate violation of that arrangement: an ultrashort, ultra-intense pulse takes away a large fraction of the electrons in a time much shorter than a bond vibration, and the nuclear frame is suddenly left holding an enormous amount of unscreened electrostatic potential energy.

Two very different pulses can do it. The first is an optical strong-field pulse, typically 800 nm light from a chirped-pulse-amplified titanium:sapphire laser, focused to 10¹⁴–10¹⁶ W/cm². The second is a hard-X-ray pulse from a free-electron laser such as LCLS at SLAC, SACLA in Japan, or the European XFEL, which does not pull on the valence electrons at all but drills holes in the inner shells and lets Auger–Meitner cascades do the stripping.

Either way the end state is the same and is best described as a vertical transition in the Franck–Condon sense: the electronic configuration changes drastically while the heavy nuclei, thousands of times more massive, have not yet had time to move. What sits in the focal volume for a moment is a frozen sculpture of the molecule made entirely of like charges — and then it lets go.

Why a laser can outmuscle a chemical bond

The intensity thresholds are not arbitrary. A valence electron in an atom experiences a binding field of order the atomic unit, e/(4πε₀a₀²) ≈ 5.1×10¹¹ V/m. A laser field of intensity I has peak amplitude E = √(2I/cε₀), which at 10¹⁵ W/cm² gives about 9×10¹⁰ V/m — roughly a fifth of the binding field, and that is already enough. The laser field does not have to overpower the Coulomb well; it only has to tilt it. Adding a linear potential −eEz to the atomic well creates a finite barrier on the down-field side, and the electron tunnels out through it.

Whether this is tunnelling or multiphoton absorption is decided by the Keldysh parameter γ = √(I_p/2U_p), where the ponderomotive energy is U_p[eV] ≈ 9.33×10⁻¹⁴ I[W/cm²] λ²[µm²]. For N₂ (I_p = 15.6 eV) at 800 nm, γ ≈ 1.1 at 10¹⁴ W/cm² but falls to ≈ 0.36 at 10¹⁵ W/cm² — squarely in the tunnelling regime, where ionization happens in a burst each half cycle, in a fraction of the 2.7 fs optical period of 800 nm light. Push a little harder and the barrier disappears altogether: the barrier-suppression intensity I_BSI ≈ 4×10⁹ I_p⁴/Z² W/cm² is only 1.4×10¹⁴ W/cm² for hydrogen, and each successive charge state simply requires more intensity.

The X-ray route works on completely different grounds. At LCLS, 8.3 keV photons sit above iodine's L edges (L₃ = 4.56 keV) but far below its K edge (33.2 keV), so each absorbed photon ejects a 2p electron. The resulting deep hole refills by Auger–Meitner emission in well under a femtosecond, ejecting further electrons, and the cascade repeats as more photons arrive. In the 2017 Nature experiment by Artem Rudenko, Daniel Rolles and co-workers, CH₃I absorbed enough photons in a ~30 fs pulse to reach a total molecular charge of order +50. The striking result was that the molecule charged up more than an isolated iodine atom would have: the highly charged iodine acted as a sink, pulling electrons off the methyl group and reopening its own valence shell for further ionization — the behaviour that earned the experiment its nickname, the molecular black hole.

The governing relation: 14.4 electronvolt-ångströms

Once the electrons are gone, the arithmetic is elementary electrostatics. For two point charges the potential energy is V = q₁q₂ e²/(4πε₀R), and in the natural molecular units e²/(4πε₀) = 14.4 eV·Å. Two singly charged ions one ångström apart therefore hold 14.4 eV — some fifty times a typical hydrogen bond, and about four times the C–C bond energy they just replaced. For a polyatomic you simply sum over pairs: V = Σ_i<j q_iq_j × 14.4/R_ij eV.

Essentially all of that becomes kinetic energy release (KER), and momentum conservation splits it inversely with mass: for a two-body break-up E₁/E₂ = m₂/m₁, so the light fragment takes almost everything. H₂ at its equilibrium bond length of 0.74 Å stores 19.5 eV; each proton leaves with ~9.7 eV, which is a speed of 43 km/s, or 0.43 Å per femtosecond. N₂ at 1.098 Å stores 13.1 eV in the (1,1) channel, and each N⁺ would depart at roughly 9.5 km/s.

The timescales follow from the same numbers. Integrating μ d²R/dt² = e²/(4πε₀R²) for N⁺ + N⁺ gives an initial acceleration near 1.6×10¹⁸ m/s² and a bond-length doubling time of roughly 13 fs — shorter than N₂'s 14.1 fs vibrational period (2359 cm⁻¹) and utterly negligible against its ~8.4 ps rotational period. For H₂ the reduced mass is fourteen times smaller and the bond doubles in about 2 fs. This ordering of timescales — ionization faster than vibration, vibration faster than rotation — is the whole basis of the technique.

It is worth being clear about what this is not. Ten electronvolts corresponds to 116,000 K if you convert it to a temperature, but nothing here is thermal. The energy is stored in a single, perfectly ordered electrostatic configuration and is released as directed momentum along the original bond axes. There is no heat bath, no equilibration, and no Boltzmann distribution to speak of.

How it is measured: foils, reaction microscopes and free-electron lasers

The first systematic exploitation of the effect used no laser at all. In 1989, Zeev Vager and Ron Naaman of the Weizmann Institute, working with Elliot Kanter at Argonne National Laboratory, reported in Science that firing MeV-energy molecular ion beams through free-standing carbon foils only about 100 Å thick produced clean, invertible explosions. A 1 MeV/u ion crosses that foil in roughly 0.7 fs, so the bonding electrons are simply scraped off before the nuclei can respond. The fragments then coast metres downstream to a position-sensitive detector, and their arrival positions and times give the full momentum vector of every piece.

The modern gas-phase workhorse is the COLTRIMS reaction microscope (cold-target recoil-ion momentum spectroscopy), developed through the 1990s by Horst Schmidt-Böcking, Reinhard Dörner, Joachim Ullrich and Lew Cocke. A supersonic jet delivers cold, dilute molecules to a focus; weak electric and magnetic fields guide every ion and electron produced by a single molecule onto delay-line detectors with near-4π acceptance; and time-of-flight plus hit position yields all three momentum components of each fragment in coincidence. Because the events are recorded one molecule at a time, the momentum sum can be used as a filter: only events whose vectors add to zero are kept, which rejects false coincidences almost perfectly.

At the intensity extreme, XFELs pair the same detectors with 10¹⁹–10²⁰ W/cm² X-ray pulses. And in clusters, Todd Ditmire and colleagues showed in Nature in 1997 that exploding xenon clusters throw off ions with energies up to ~1 MeV; in 1999 they turned deuterium clusters into a tabletop fusion source, the exploding D⁺ ions colliding hard enough to produce 2.45 MeV DD-fusion neutrons.

Coulomb explosion imaging: reading the molecule out of its own debris

The payoff is structural. If the nuclei really do not move while they are being ionized, then the asymptotic momentum vectors are a one-to-one map of the starting geometry, and you can integrate Newton's equations backwards under a pure point-charge potential to recover the bond lengths and angles. This is the axial-recoil approximation, and its validity is quantitative: a ~15 fs explosion against an 8.4 ps rotational period means the molecule turns by only about a hundredth of a radian — well under a degree — on the way out.

Two experiments show what the method can do that diffraction cannot. In 2013, Martin Pitzer and co-workers in the Dörner group published in Science the direct determination of the absolute handedness of a single chiral molecule, CHBrClF, from the sign of the triple product of three fragment momenta — a stereochemical measurement made on individual molecules in the gas phase. In 2014, Ibrahim and colleagues used Coulomb explosion imaging to watch a proton migrate across acetylene, C₂H₂, as it isomerized towards vinylidene, capturing a hydrogen atom moving on a femtosecond clock.

The resolution has a hard floor set by quantum mechanics rather than by the apparatus. Even a perfectly cold molecule has zero-point motion: for N₂ the root-mean-square bond-length spread is √(ħ/2μω) ≈ 0.03 Å. No inversion, however careful, can localize the nuclei better than the wavefunction they occupy, so measured geometry distributions are always at least this broad — a feature, not a bug, since the width itself carries information about the vibrational state.

Where the simple picture breaks, and what it is confused with

The clean point-charge model systematically overestimates the kinetic energy release, and the reason has a name. Charge-resonance-enhanced ionization (CREI), predicted independently in 1995 by Tao Zuo and André Bandrauk and by Tamar Seideman, Misha Ivanov and Paul Corkum, notes that a stretching molecular ion becomes easier, not harder, to ionize. At a critical internuclear distance — typically a few times the equilibrium bond length, and 7–10 atomic units for H₂⁺, whose equilibrium separation is 2 a.u. — the inner barrier between the two nuclear wells rises above the field-suppressed outer barrier, so an electron localized in the up-field well finds itself sitting above the exit and ionizes with near-unit probability. The consequence is that the highest charge states are created at stretched geometries, and by V ∝ 1/R the observed KER falls short of the equilibrium estimate. This is precisely why N₂ → N⁺ + N⁺ is measured at about 7–10 eV instead of the vertical 13.1 eV. Residual bound electrons that screen the nuclear charges, and non-instantaneous charge-up during long pulses, push in the same direction.

Two look-alikes deserve separating. The first is the Rayleigh instability of a charged liquid droplet, worked out by Lord Rayleigh in 1882: when the surface charge satisfies q² > 64π²ε₀γR³, electrostatic repulsion beats surface tension and the droplet fissions. Electrospray practitioners routinely call this a Coulomb explosion, and the name is fair, but the physics is macroscopic hydrodynamics on microsecond timescales, not bare nuclei on femtosecond ones. The second is the long-running debate over ion tracks in solids, where a swift heavy ion leaves a damage trail: whether the lattice is destroyed by a Coulomb explosion of the ionized column or by a thermal spike depends on how fast conduction electrons neutralize the charge, and in metals they do so far too quickly for any explosion to occur.

Open questions

The technique is being pushed hard, and the frontier is mostly about how far the point-charge inversion can be trusted. How large a molecule can be imaged? The number of pairwise terms grows as N², coincidence rates fall, and correctly assigning every fragment becomes combinatorially hard; molecules of around ten atoms are now within reach, but twenty-atom molecules are not.

Other open threads: how much charge transfer happens between fragments during the first few femtoseconds of the explosion, which corrupts the assumed constant charges; whether machine-learning inversions trained on classical trajectory simulations can beat analytic back-integration; whether Coulomb explosion imaging can be triggered at a chosen delay with attosecond precision to make genuine molecular movies of transition states rather than of reactants and products; and how the XFEL charge-rearrangement enhancement scales with molecular size, which matters directly for setting the radiation-damage limits of single-particle X-ray imaging of proteins.

Four genuine routes to a Coulomb explosion — and one impostor that shares the name

Route	How the electrons leave	Charge reached and timescale	What it is used for
800 nm femtosecond laser, 10¹⁴–10¹⁶ W/cm²	Tunnel ionization twice per optical cycle; the ~9×10¹⁰ V/m peak field bends the binding potential into a barrier that leaks	About +1 to +10 per atom within a 10–40 fs pulse	Coulomb explosion imaging of small molecules; strong-field and attosecond science
Hard-X-ray XFEL (LCLS, 8.3 keV)	Inner-shell photoionization — 8.3 keV is above iodine's L₃ edge at 4.56 keV — followed by Auger–Meitner cascades, with charge pulled in from the neighbouring atoms	Total charge of order +50 on CH₃I in a ~30 fs pulse	Extreme-charge chemistry; setting the radiation-damage limit for single-particle X-ray imaging
MeV molecular-ion beam through a ~100 Å carbon foil	Bonding electrons are simply swept off during the ~0.5–1 fs foil transit	Near-bare nuclei, stripped in well under a femtosecond	The original Coulomb explosion imaging (Vager, Naaman and Kanter, Argonne, 1989)
Rare-gas or deuterium cluster in an intense pulse	Outer ionization drives electrons off the cluster as a whole, leaving a positively charged ball of ions	keV to MeV ion energies from clusters of 10³–10⁵ atoms	Tabletop deuterium–deuterium fusion neutrons (Ditmire et al., 1999)
IMPOSTOR: a charged droplet past the Rayleigh limit	Nothing is stripped — excess like charges simply accumulate on a liquid surface until repulsion beats surface tension	Microseconds to milliseconds, limited by viscosity and surface tension	Electrospray ionization and droplet fission — same phrase, entirely different physics