Photon-Stimulated Desorption (PSD)

• Photon-Stimulated Desorption (PSD) is a non-thermal process where photon absorption triggers the ejection of atoms, molecules, or ions via mechanisms like DIET, dissociation, and XESD.
• PSD studies integrate astrochemistry, accelerator vacuum dynamics, and planetary surface physics by linking photon energy, fluence, and surface properties to desorption yields and kinetics.
• Quantitative analyses use time-of-flight distributions, yield scaling, and kinetic modeling to differentiate between thermal and nonthermal desorption pathways and refine predictive models.

Photon-stimulated desorption (PSD) denotes the ejection of atoms, molecules, neutrals, cations, or anions from a surface following photon absorption. In cryosorbed interstellar ice analogs it is defined as a non-thermal ejection process, whereas in surface-science studies of physisorbed noble gases the same experimental framework also resolves a transition between nonthermal and laser-induced thermal desorption regimes (Dupuy et al., 2018, Ikeda et al., 2012). Across VUV, soft X-ray, and EUV photon energies, PSD couples surface binding, electronic excitation, Auger decay, secondary-electron transport, photochemistry, and, in some systems, transient ionic or antibonding states; its quantitative description therefore depends on photon energy, absorbed fluence, coverage, morphology, and the identity of both the absorber and the desorbing species (Basalgète et al., 2022, Devaraj et al., 6 Sep 2025).

1. Definition and physical scope

PSD is studied in at least three major settings. In astrochemistry, it is a critical non-thermal mechanism that returns volatile species from cold grain mantles to the gas phase when thermal desorption is suppressed below $100\ \mathrm{K}$. In accelerator vacuum dynamics, synchrotron radiation impinging on cryogenic beamline walls desorbs residual gases and raises pressure under conditions where non-thermal processes dominate. In planetary surface physics, PSD is invoked for alkali release from regolith-like materials and for the production of suprathermal surface-bound exospheres (Dupuy et al., 2018, Devaraj et al., 6 Sep 2025, Gamborino et al., 2019).

The microscopic meaning of PSD is system dependent. In molecular ices, the desorbing species may be the photon absorber itself, a surface species energized by a subsurface absorber, or a photoproduct formed earlier in the irradiation history. In alkali-bearing solids, UV photons can excite substrate electrons to the conduction band; the adsorbed ion captures an electron, becomes neutral, enters an antibonding repulsive state, and is ejected. In physisorbed noble gases on metals, PSD can proceed either through transient electronic pathways or through photon-induced heating that activates conventional desorption kinetics (Devaraj et al., 6 Sep 2025, Ikeda et al., 2012).

A recurring conceptual point is that PSD is not a single mechanism. The literature explicitly distinguishes direct desorption, indirect desorption induced by electronic transitions (DIET), dissociation-driven channels, kick-out processes, X-ray induced electron-stimulated desorption (XESD), transient negative-ion pathways, and prompt photochemidesorption of newly formed products. This mechanism multiplicity is central to interpreting spectra, yields, and velocity distributions (Dupuy et al., 2018, Dupuy et al., 2021).

2. Mechanistic classes and spectral fingerprints

Direct DIET refers to desorption initiated by single-photon absorption by the adsorbate’s valence states. Its spectral signature is a one-to-one correspondence between the photodesorption spectrum and the absorber’s electronic transitions. Condensed CO is the canonical example: its VUV desorption spectrum is highly structured and peaks between $8$ and $9\ \mathrm{eV}$, tracking the condensed-phase $A$–$X$ vibrational progression. Indirect DIET occurs when absorption by one species drives desorption of another; isotopic layering in CO, and $1\ \mathrm{ML}\ \mathrm{N_2}$ on $20\ \mathrm{ML}\ \mathrm{CO}$, directly show surface-selective desorption with subsurface absorption (Dupuy et al., 2018).

A distinct family of mechanisms is dissociation-driven. In dissociative ices such as $\mathrm{CH_4}$, $\mathrm{CH_3OH}$, and $\mathrm{CO_2}$, fragments can recombine exothermically or transfer momentum to neighbors. For $8$0, the channels explicitly identified include $8$1, releasing about $8$2, and kick-out of $8$3 by energetic H fragments. In $8$4 ice at $8$5, the proposed surface-chemistry sequence is $8$6 followed by $8$7; the flux dependence of CO and $8$8 desorption, together with near-identical evolution cross sections for the two species at fixed flux, supports a common surface-controlled kinetic bottleneck (Hacquard et al., 2024).

In water-rich ices, indirect desorption can be momentum-driven rather than excitonic. A systematic VUV study of Ar, Kr, $8$9, and CO adsorbed on $9\ \mathrm{eV}$0 or $9\ \mathrm{eV}$1 amorphous ice showed that the mass dependence, isotopic effect, and increasing intrinsic efficiency from $9\ \mathrm{eV}$2 to $9\ \mathrm{eV}$3 are explained by collisions between adsorbates and energetic H or D atoms produced by photodissociation of water. The elastic-collision energy-transfer fraction was written as

$9\ \mathrm{eV}$4

with the desorption condition $9\ \mathrm{eV}$5 (Dupuy et al., 2021).

At core levels, PSD is often dominated by Auger-electron cascades. O $9\ \mathrm{eV}$6 excitation of amorphous solid water, and N or O $9\ \mathrm{eV}$7-edge excitation of mixed or layered $9\ \mathrm{eV}$8 and $9\ \mathrm{eV}$9, produce desorption spectra that either follow TEY, implying XESD, or deviate from TEY in ways that identify direct dissociation of the core-excited molecule or of photoproducts such as $A$0. In layered $A$1, irradiation at $A$2 on the $A$3 transition of $A$4 induces $A$5 desorption through an indirect process operating over $A$6–$A$7, explicitly linked to Auger-electron transport and secondary-electron-stimulated desorption (Dupuy et al., 2020, Basalgète et al., 2022).

Metal-supported adsorbates introduce another mechanism class. Xe on $A$8 exhibits one-photon nonthermal desorption at $A$9 that is attributed to transient $X$0 formation and the Antoniewicz neutral-desorption mechanism. Lunar Na release is described differently but analogously in electronic terms: adsorbed $X$1 captures a conduction electron, becomes neutral Na$X$2, and enters a repulsive state because the neutral atom has a larger radius than the ion (Ikeda et al., 2012, Devaraj et al., 6 Sep 2025).

3. Quantification, observables, and diagnostic criteria

PSD experiments are usually reported in terms of yields, rates, fluence dependences, and velocity or time-of-flight distributions. A standard definition is

$X$3

with the rate under illumination

$X$4

and, when absorption is treated explicitly,

$X$5

For constant flux, the coverage evolution may be written as

$X$6

when $X$7 is constant over the relevant interval (Dupuy et al., 2018).

Fluence dependence is mechanistically diagnostic. Linear yield scaling, $X$8, together with fluence-independent translational temperature, is the signature of a one-photon nonthermal channel in Xe/Au. Threshold-like increases and saturation with increasing absorbed fluence indicate laser-induced thermal desorption. In $X$9 ice, $1\ \mathrm{ML}\ \mathrm{N_2}$0 is nearly constant and weakly flux dependent, consistent with a single-photon DIET-like process, whereas $1\ \mathrm{ML}\ \mathrm{N_2}$1 and $1\ \mathrm{ML}\ \mathrm{N_2}$2 grow toward asymptotes and depend strongly on photon flux, showing explicit surface-chemistry control (Ikeda et al., 2012, Hacquard et al., 2024).

Time-of-flight analysis provides a second layer of discrimination. Thermal Xe desorption on $1\ \mathrm{ML}\ \mathrm{N_2}$3 was fitted with Maxwell–Boltzmann forms, including the density-sensitive time-domain expression

$1\ \mathrm{ML}\ \mathrm{N_2}$4

and the flux-space distribution

$1\ \mathrm{ML}\ \mathrm{N_2}$5

At high coverage, post-desorption collisions required a shifted Maxwell–Boltzmann form with stream velocity $1\ \mathrm{ML}\ \mathrm{N_2}$6, diagnosing Knudsen-layer formation rather than a change in the elementary PSD mechanism (Ikeda, 2015).

Velocity distributions on planetary surfaces are non-thermal for the same fundamental reason: PSD and ESD are DIET processes rather than lattice-equilibrated release. For Na, a systematic comparison of Maxwell–Boltzmann, empirical PSD, and Weibull distributions concluded that the measured velocity distribution functions are too narrow for supra-temperature Maxwellian fits, and recommended an offset Weibull form,

$1\ \mathrm{ML}\ \mathrm{N_2}$7

with $1\ \mathrm{ML}\ \mathrm{N_2}$8, $1\ \mathrm{ML}\ \mathrm{N_2}$9, and

$20\ \mathrm{ML}\ \mathrm{CO}$0

This is explicitly a non-thermal release model, even though the surface temperature enters the scale parameter (Gamborino et al., 2019).

For X-ray studies, TEY acts as an absorption proxy. When the desorption spectrum follows TEY, XESD is inferred; when the spectrum instead tracks specific resonances of photoproducts such as $20\ \mathrm{ML}\ \mathrm{CO}$1, direct excitation of those photoproducts is implicated. A compact way to normalize out absorption in mixed or layered ices is

$20\ \mathrm{ML}\ \mathrm{CO}$2

which yields desorption in molecules per eV deposited (Dupuy et al., 2020, Basalgète et al., 2022).

4. Noble gases on metals as a benchmark system

Xe on $20\ \mathrm{ML}\ \mathrm{CO}$3 provides one of the clearest benchmark cases for separating thermal and nonthermal PSD. A Xe monolayer was prepared by dosing about $20\ \mathrm{ML}\ \mathrm{CO}$4 Xe onto clean reconstructed $20\ \mathrm{ML}\ \mathrm{CO}$5 at $20\ \mathrm{ML}\ \mathrm{CO}$6 under UHV, and TOF spectra were measured for $20\ \mathrm{ML}\ \mathrm{CO}$7 and $20\ \mathrm{ML}\ \mathrm{CO}$8 nanosecond laser pulses. Above $20\ \mathrm{ML}\ \mathrm{CO}$9 per pulse, $\mathrm{CH_4}$0 increased with fluence from about $\mathrm{CH_4}$1 toward saturation near $\mathrm{CH_4}$2, and the desorption yield showed a sharp threshold-like increase, quantitatively reproduced by a one-dimensional heat-diffusion model coupled to first-order desorption kinetics. For an impulsive energy deposition, the surface temperature rise was written as

$\mathrm{CH_4}$3

with thermal desorption kinetics

$\mathrm{CH_4}$4

using $\mathrm{CH_4}$5 for Xe/$\mathrm{CH_4}$6 (Ikeda et al., 2012).

At low absorbed fluence, the behavior changes qualitatively. For $\mathrm{CH_4}$7, Xe desorption was observed only at $\mathrm{CH_4}$8. In the lowest-fluence region, $\mathrm{CH_4}$9, $\mathrm{CH_3OH}$0 remained constant at $\mathrm{CH_3OH}$1 while calculated peak surface temperatures were below $\mathrm{CH_3OH}$2, and the yield scaled linearly with fluence, $\mathrm{CH_3OH}$3, corresponding to a nonthermal one-photon PSD cross section $\mathrm{CH_3OH}$4–$\mathrm{CH_3OH}$5. The nonthermal channel was assigned to transient $\mathrm{CH_3OH}$6 formation. The neutral physisorption well was modeled by a Morse potential,

$\mathrm{CH_3OH}$7

with $\mathrm{CH_3OH}$8, $\mathrm{CH_3OH}$9, and $\mathrm{CO_2}$0. Upon electron attachment, the adiabatic potential becomes

$\mathrm{CO_2}$1

and neutralization is taken as

$\mathrm{CO_2}$2

The measured translational temperature and cross section were simultaneously reproduced for $\mathrm{CO_2}$3, implying stabilization of a transient Xe affinity level by image-charge interaction and hybridization with substrate states (Ikeda et al., 2012).

The same Xe/Au system also shows that gas-dynamic effects can overlay the primary desorption physics. At high desorption fluxes and increasing Xe coverage, TOF peaks shift to shorter times and saturate; a shifted Maxwell–Boltzmann fit yields a stream velocity saturating at $\mathrm{CO_2}$4 for $\mathrm{CO_2}$5. This behavior was attributed to post-desorption collisions and Knudsen-layer formation, with $\mathrm{CO_2}$6 and terminal Mach number $\mathrm{CO_2}$7, in agreement with rarefied-gas theory and Monte Carlo simulations (Ikeda, 2015).

5. Cryogenic molecular ices under VUV and X-ray irradiation

Cryogenic ices exhibit strongly molecule-specific PSD spectra and yields. In the VUV, CO photodesorption from a $\mathrm{CO_2}$8 film at $\mathrm{CO_2}$9 varies from about $8$00 down to $8$01 across the spectral range and tracks condensed-phase electronic transitions. $8$02 desorbs strongly only above about $8$03 as a pure ice, but a $8$04 overlayer on $8$05 inherits the $8$06–$8$07 CO signatures through indirect DIET. $8$08 PSD reflects dissociative electronic states, while parent $8$09 yields are very low, below $8$10. In the X-ray range, $8$11 compact amorphous $8$12 at $8$13 shows maximum neutral water yields of about $8$14, much larger than the UV-induced $8$15 maximum of about $8$16; neutral $8$17, $8$18, and $8$19 follow TEY, whereas $8$20 instead follows $8$21 resonances between $8$22 and $8$23 (Dupuy et al., 2018).

Pure $8$24 illustrates the importance of flux and fluence. For $8$25 at $8$26 irradiated at $8$27, the initial $8$28 yield is about $8$29 and shows no clear dependence on photon flux, with $8$30 in the range $8$31–$8$32. By contrast, CO and $8$33 yields rise with fluence and depend strongly on flux: at low fluence in the low-flux regime, CO and $8$34 reach about $8$35 and $8$36, whereas at similar fluence but higher flux they are lower, about $8$37 and $8$38. This establishes a decorrelation between bulk-mediated $8$39 desorption and surface-chemistry-controlled CO and $8$40 desorption (Hacquard et al., 2024).

Broadband UV studies of pure $8$41 at $8$42 reached a complementary conclusion. Directly calibrated QMS measurements gave maximum photodesorption yields of $8$43, $8$44, and $8$45 for CO, $8$46, and $8$47, respectively. The increase of CO and $8$48 desorption with fluence until the CO subsurface abundance maximized at about $8$49 was interpreted as indirect DIET, rather than prompt photochemidesorption, because bulk-formed molecules progressively contributed to desorption as the upper layers were removed (Martín-Doménech et al., 2015).

Pure $8$50 ice at $8$51 exhibits a different partitioning of channels. The parent photodesorption yield,

$8$52

remains approximately constant with fluence, whereas $8$53 and $8$54 photodesorption increase with fluence, indicating indirect energy-transfer mechanisms for the photoproducts. At an interstellar-relevant fluence of about $8$55, the $8$56 photodesorption yield is similar to that of the parent,

$8$57

despite the very small UV absorption cross section of pure $8$58 (Martin-Domenech et al., 2017).

Methanol-containing systems highlight a persistent asymmetry between product desorption and parent desorption. In $8$59 ice analogs irradiated at $8$60, $8$61 formation was observed in the solid, but no photon-induced desorption of methanol was detected. Formaldehyde, by contrast, photochemidesorbs with a representative yield of about $8$62. A separate UV study of pure methanol ice set an upper limit of $8$63 for intact methanol photodesorption, while photoproducts such as CO and $8$64 desorbed efficiently; the negligible methanol desorption was attributed to efficient dissociation of $8$65 by $8$66–$8$67 photons (Martín-Doménech et al., 2016, Cruz-Diaz et al., 2016).

X-ray work extends these findings deeper into the solid. O $8$68 excitation of amorphous solid water at $8$69 and $8$70 releases neutral, cationic, and anionic species, with neutrals dominated by XESD and oxygen-containing ions largely tracing direct excitation of photoproducts such as $8$71. In mixed and layered $8$72 and $8$73 ices, the energy-normalized photodesorption yields cluster around about $8$74–$8$75 and do not depend strongly on photon energy, the photo-absorbing molecule, or its post-Auger state, which is direct evidence that Auger-mediated XESD, rather than an absorber-specific direct process, dominates desorption in these systems (Dupuy et al., 2020, Basalgète et al., 2022).

6. Planetary surfaces, accelerator vacuum, and astrophysical modeling

In accelerator environments, PSD is directly linked to gas load. For synchrotron radiation with $8$76 and $8$77, as in resonant X-ray $8$78 photodesorption, the local desorption rate is

$8$79

for the illuminated spot. The appropriate predictive framework is spectral,

$8$80

because yields vary strongly with photon energy and molecular identity (Dupuy et al., 2018).

Astrochemical rate estimates use the same structure,

$8$81

At $8$82, with $8$83, interstellar-radiation-field-like fluxes of $8$84–$8$85 imply

$8$86

which corresponds to removing one $8$87 monolayer in about $8$88–$8$89 years. The same study explicitly warned, however, that below about $8$90 the $8$91 absorption is weak and actual PSD rates in dense-core environments dominated by Lyman-$8$92 at $8$93 should be much lower (Hacquard et al., 2024).

For nitrogen-bearing ices, the dense-cloud secondary UV flux of about $8$94 gives non-thermal desorption rates of order $8$95–$8$96 for $8$97 and $8$98 when the experimentally derived yields are inserted. For formaldehyde photochemidesorption with $8$99, the same flux gives a rate of about $9\ \mathrm{eV}$00, while a PDR-like photon flux of about $9\ \mathrm{eV}$01 gives about $9\ \mathrm{eV}$02 (Martin-Domenech et al., 2017, Martín-Doménech et al., 2016).

Planetary PSD of alkalis introduces additional model structure. Classic lunar Na models use

$9\ \mathrm{eV}$03

which predicts linearity in photon flux for a given wavelength. Recent simultaneous observations of lunar Na and solar irradiance instead found a non-linear relation,

$9\ \mathrm{eV}$04

with $9\ \mathrm{eV}$05, and showed that EUV photons above about $9\ \mathrm{eV}$06–$9\ \mathrm{eV}$07, especially the $9\ \mathrm{eV}$08–$9\ \mathrm{eV}$09 bands, dominate the response. The same dataset yielded average zenith column density $9\ \mathrm{eV}$10, average characteristic temperature about $9\ \mathrm{eV}$11, and average scale height about $9\ \mathrm{eV}$12, with higher temperatures and scale heights during solar flares (Devaraj et al., 6 Sep 2025).

7. Interpretive issues, common misconceptions, and limitations

A common misconception is that Maxwellian-like TOF spectra necessarily imply thermal desorption. The Xe/Au case shows otherwise: in the nonthermal low-fluence regime, the TOF fits remain Maxwellian-like, yet the translational temperature is independent of fluence and significantly exceeds the calculated surface temperature. Conversely, threshold-like yields and $9\ \mathrm{eV}$13 following the modeled desorption temperature, rather than the absolute maximum surface temperature, are the decisive signatures of LITD in that system (Ikeda et al., 2012).

A second misconception is that PSD yields are generally linear in photon flux. That is true for single-photon channels such as nonthermal Xe desorption at $9\ \mathrm{eV}$14 and the $9\ \mathrm{eV}$15 parent channel at $9\ \mathrm{eV}$16, but it is not true for product desorption controlled by surface chemistry, as demonstrated by the flux dependence of CO and $9\ \mathrm{eV}$17 from $9\ \mathrm{eV}$18 ice, or for lunar Na release in the EUV regime, where a super-linear response is observed (Hacquard et al., 2024, Devaraj et al., 6 Sep 2025).

A third issue concerns the use of absorption proxies. TEY is robust and often proportional to the number of absorbed X-ray photons or Auger decays, but it measures electrons escaping the surface and therefore depends on escape depth and morphology. In water ice, TEY tracking is a strong indication of XESD for neutral desorption, yet ion channels can instead follow photoproduct resonances, especially $9\ \mathrm{eV}$19, and thereby break the TEY correlation (Dupuy et al., 2018, Dupuy et al., 2020).

Composition and irradiation history also matter. Continued X-ray irradiation of $9\ \mathrm{eV}$20 forms $9\ \mathrm{eV}$21 at a few percent abundance, which changes ion-desorption spectra. In $9\ \mathrm{eV}$22 ice, growing CO modifies the $9\ \mathrm{eV}$23 band shape enough that $9\ \mathrm{eV}$24 infrared band areas are unreliable as direct measures of chemistry or desorption without correction. In methanol ices, the repeated failure to detect efficient UV-induced parent desorption, despite clear formation and desorption of irradiation products, means that gas-phase methanol at low temperature remains unresolved by PSD alone (Hacquard et al., 2024, Cruz-Diaz et al., 2016).

Extrapolation beyond the laboratory therefore requires caution. The studies explicitly identify uncertainties associated with QMS calibration, mass transmission, ionization cross sections, flux determination, background subtraction, morphology, porosity, compact versus porous amorphous ice, surface/substrate effects, and evolving composition under irradiation. A plausible implication is that PSD should be modeled as a spectrally resolved, composition-dependent, and time-evolving family of mechanisms rather than as a single yield constant (Dupuy et al., 2018, Basalgète et al., 2022).