Resonance-enhanced above-threshold ionization of ammonia under strong fields Authors: Ignacio M. Casasús, María E. Corrales, Arlind Sherifi, Sonia Marggi Poullain, Luis Bañares Published: 2026 ABSTRACT Molecules exposed to intense femtosecond laser fields display a complex interplay of multiphoton ionization, resonances, and tunneling dynamics. Although above-threshold ionization (ATI) has been extensively characterized in atoms, its resonance-enhanced counterpart in polyatomic molecules has remained elusive. Here we investigate the strong-field ionization of ammonia (NH3) by 800 nm femtosecond pulses coupled with photoelectron velocity map imaging, exploiting its distinctive vibrational fingerprint in the umbrella bending mode. We uncover a weak Stark-induced Freeman resonance whose ATI replicas are shown to dominate over the parent channel. A strong resonance-enhanced above-threshold ionization (REATI) phenomenon is thus demonstrated here in a polyatomic molecule. Moreover, our results connect non-resonant and resonant multiphoton ionization, Freeman resonances, ATI, and REATI, into a unified roadmap, establishing ammonia as a benchmark system for exploring vibrationally resolved strong-field ionization. Beyond unveiling a distinct regime of molecular ionization, our findings highlight how the vibrational structure governs strong-field phenomena in complex molecules. Introduction High-Intensity femtosecond lasers offer a unique window into how atoms and molecules absorb and redistribute energy under extreme conditions1,2. Ionization dynamics as a function of the laser field intensity are governed by different phenomena, some of them extensively investigated. At the lowest field intensities, perturbative light-matter interactions dominate, leading to multiphoton ionization (MPI)3,4 and resonance-enhanced MPI (REMPI)5, in which several photons are absorbed to reach the ionization continuum, either directly or via resonant intermediate states. As the field intensity increases, additional regimes emerge: electrons may absorb extra photons beyond the ionization threshold in above-threshold ionization (ATI)6,7, while intermediate states can be transiently brought into multiphoton resonance by the ac-Stark shift, giving rise to the so-called Freeman resonances within defined intensity windows8–15. At still higher intensities, ionization proceeds predominantly through tunneling in the field-distorted Coulomb potential, followed by laser-driven rescattering of the liberated electron in the parent ionic field1,16,17. These rescattering dynamics give rise to high-energy ATI (HATI) ladders as well as to high-harmonic generation (HHG)18, marking the onset of strongly non-perturbative behavior. Under intense femtosecond laser pulses, the Keldysh parameter γ1 is commonly used to assess whether ionization proceeds predominantly within a perturbative (multiphoton, γ ≫ 1) or a nonperturbative (tunneling, γ ≪ 1) picture, with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma =\sqrt{{{\rm{I}}}_{{\rm{P}}}/2{{\rm{U}}}_{{\rm{P}}}}$$\end{document}γ=IP/2UP, where IP is the ionization potential of the system and UP is the ponderomotive energy associated with the laser field. In atomic systems, this parameter provides a convenient classification of strong-field regimes, linking the laser intensity and wavelength to the dominant ionization mechanism. However, this description implicitly assumes that ionization involves a limited set of well-separated electronic states, such that internal structure plays a minor role once the field parameters are fixed. In molecular systems, additional nuclear degrees of freedom introduce dense manifolds of rovibrational states that can participate in the ionization process. As a consequence, molecular strong-field ionization at a given value of γ may involve multiple competing pathways, as demonstrated by channel-resolved coincident studies19,20. This is particularly relevant in the intermediate regime preceding fully developed tunneling, where resonance-enhanced pathways such as Freeman resonances can persist over broad intensity ranges rather than appearing as isolated features. How these resonance-assisted channels evolve, reorganize, or dominate the ionization dynamics within dense rovibronic manifolds remains less systematically explored. Ammonia (NH3) is particularly suited to disentangle resonance-enhanced ionization dynamics in molecular systems. Excitation or ionization from the nitrogen lone-pair electrons, nN, transforms the pyramidal ground state geometry (C3v) into a planar one (D3h) characterizing both Rydberg and ionic states. This large geometrical change gives rise to unusually long vibrational progressions in the umbrella bending mode (ν2, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\nu }_{2}^{+}$$\end{document}ν2+)21. Figure 1 shows the vacuum ultraviolet (VUV) absorption spectrum of NH3 (adapted from22,23) along with the corresponding He I photoelectron spectrum associated to the formation of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\rm{NH}}}_{3}^{+}$$\end{document}NH3+ (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}$$\end{document}X~A2″2) (adapted from24). In Fig. 1, both panels reveal these long umbrella-mode progressions that directly connect electronic excitation with nuclear motion. The different bands observed in Fig. 1a are attributed to Rydberg states from one-photon excitation of the nitrogen lone-pair electrons, nN (3a1 in C3v). These states have been extensively assigned in the literature, labeled from \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{A}$$\end{document}A~ up to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{H}$$\end{document}H~22,24, all converging to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\rm{NH}}}_{3}^{+}$$\end{document}NH3+ (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}$$\end{document}X~A2″2). The detailed spectroscopic analysis of ammonia has also benefited from resonance-enhanced multiphoton ionization (REMPI) experiments25–29. The series of Rydberg states extends to high energies and overlaps with the ionization continuum, opening the possibility of autoionization processes and long-lived resonances that can strongly interact with the driving laser field.Fig. 1VUV-photoabsorption and photoelectron spectra of ammonia.a High-resolution VUV photoabsorption spectrum of NH3. In black, the 5.5–10.8 eV photon energy range using the data from ref. 22; in gray, the 10.0–11.7 eV range from ref. 23. Both are normalized to match intensities. The umbrella mode ν2 = n progression (n = 0–14) is indicated for the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{A}$$\end{document}A~ band. b He I photoelectron spectrum of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}$$\end{document}X~A2″2 ground state of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\rm{NH}}}_{3}^{+}$$\end{document}NH3+ with data from ref. 24, showing the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\nu }_{2}^{+}=n$$\end{document}ν2+=n progression. Red gaussian markers indicate the energy of several 800 nm multiphoton absorption (full-width-at-half-maximum, FWHM = 35 nm). VUV-photoabsorption and photoelectron spectra of ammonia. a High-resolution VUV photoabsorption spectrum of NH3. In black, the 5.5–10.8 eV photon energy range using the data from ref. 22; in gray, the 10.0–11.7 eV range from ref. 23. Both are normalized to match intensities. The umbrella mode ν2 = n progression (n = 0–14) is indicated for the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{A}$$\end{document}A~ band. b He I photoelectron spectrum of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}$$\end{document}X~A2″2 ground state of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\rm{NH}}}_{3}^{+}$$\end{document}NH3+ with data from ref. 24, showing the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\nu }_{2}^{+}=n$$\end{document}ν2+=n progression. Red gaussian markers indicate the energy of several 800 nm multiphoton absorption (full-width-at-half-maximum, FWHM = 35 nm). Here, we experimentally investigate the near-infrared (NIR) strong-field ionization of ammonia. The present results show that for a considerable range of field intensities, ATI replicas of a resonance-enhanced channel dominate over the parent signal, a weak Stark-induced Freeman resonance. This phenomenon labeled here resonance-enhanced above-threshold ionization (REATI) contrasts with previous experiments in formic acid30, where high-order ATI spectra (HATI) were enhanced at specific laser intensities. From a theoretical point of view, calculations have demonstrated that resonant pathways can enhance and modulate ATI structures. For example, Li et al.31 have reported theoretical enhancements in the HATI spectra of Ar, Xe and N2 when high-lying excited states become accessible. Similarly, calculations by Rodríguez et al.32 have shown that tuning the extreme ultraviolet (XUV) energy in and out of resonance produces modulations in the relative strength of ATI signals. In addition, Shu et al.33 have theoretically described resonance assisted ATI in the polar system NaK, concluding that the strong-field dressing of the states favors the emission to higher ATI orders. In this work, photoelectron velocity map imaging (VMI) is employed to investigate the strong-field ionization of ammonia by intense 800 nm femtosecond pulses. The photoelectron VMI images are Abel-inverted (rBASEX) to extract kinetic energy and angular distributions34,35. By scanning the laser intensity up to ~45 TW ⋅ cm−2, we trace the evolution of ionization pathways from non-resonant and resonant MPI, through Freeman resonances, ATI and REATI regimes, before tunneling ionization dominates. Our findings show that the dominance of ATI replicas is favored by the presence of long-lived field-dressed resonances, which extend their interaction time with the laser field and enhance the probability of additional photon absorption. This unveils a mechanism for resonance-assisted strong-field ionization in molecules. Results The external field intensity was gradually increased by adjusting neutral density filters, driving ammonia through a sequence of strong-field ionization regimes. Figure 2 presents three representative Abel-inverted photoelectron images (PEIs) from 800 nm ionization, obtained at low, intermediate, and high intensities. These images provide a visual overview of the evolving dynamics (from left to right panels of Fig. 2): from a first landscape of resonant pathways with rich structures, to a new series of resonant signals that overtake the previous one, followed eventually by interferometric patterns as tunneling ionization takes over.Fig. 2Photoelectron images from the ionization of ammonia at selected intensities.Selected Abel-inverted photoelectron images from photoionization of NH3 with femtosecond pulses centered at 800 nm. Approximate laser intensities at the interaction region with the molecular beam, a 12.0, b 28.7 and c 43.3 TW ⋅ cm−2, are indicated in the bottom-right corner of each image. Photoelectron images from the ionization of ammonia at selected intensities. Selected Abel-inverted photoelectron images from photoionization of NH3 with femtosecond pulses centered at 800 nm. Approximate laser intensities at the interaction region with the molecular beam, a 12.0, b 28.7 and c 43.3 TW ⋅ cm−2, are indicated in the bottom-right corner of each image. The photoelectron kinetic energy distributions (PEKDs) are obtained by angular integration of each image as a function of the laser intensity. For an n-photon ionization process with additional s above-threshold photons, the expected electron kinetic energy (KE) is 1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rm{KE}}=(n+s)\hslash \omega -\left[| {{\rm{I}}}_{{\rm{P}}}| +{{\rm{U}}}_{{\rm{P}}}+{{\rm{E}}}_{{\rm{int}}}^{+}\right],$$\end{document}KE=(n+s)ℏω−∣IP∣+UP+Eint+, where ℏω is the photon energy, ∣IP∣ the ionization potential, UP the ponderomotive potential, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\rm{E}}}_{{\rm{int}}}^{+}$$\end{document}Eint+ the internal energy of the cation. This relation enables a self-calibration of the intensity scale, since UP grows linearly with the laser intensity by definition [UP (eV) = 0.0933 ⋅ I (W ⋅ cm−2) ⋅ λ2 (m2)]17,36–38. Using this approach, we can determine the maximum intensity in our measurements to be I0 = 43.3 ± 0.4 TW ⋅ cm−2. Additional details on the self-calibration methodology can be found in the Supplementary Note 1 (including Suppl. Figs. 1 and 2 and Supplementary Table 1). Figure 3 shows the evolution of the PEKDs across the full intensity scan. Panel (a) presents the self-normalized PEKDs as a function of the laser intensity, dominated by signals that remain nearly constant in kinetic energy, a clear signature of resonant pathways. Panel (b) displays the derivative of these distributions with respect to intensity36–38, which removes stationary resonant signals and reveals the shifting non-resonant MPI contributions. Dashed lines correspond to the predictions from Eq. (1), confirming that the calibration consistently reproduces the expected slopes for non-resonant MPI and ATI channels. Furthermore, non-resonant signals also appear to be broadened, suggesting underlying vibrational excitation, consistent with reported results from He I single-photon ionization24.Fig. 3Intensity-dependent photoelectron kinetic energy distributions (PEKDs).a 2D mesh-grid false-colormap constructed with the self-normalized PEKDs, showing their evolution with increasing external field intensity. b 2D mesh-grid diverging-color map, showing the evolution of non-resonant MPI signals after derivation with respect to the external intensity. Diagonal dashed lines are plotted following Eq. (1) for different total (n+s) photon channels (indicated in white squares next to the lines), with I0 = 43.3 TW ⋅ cm−2 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$| {{\rm{I}}}_{{\rm{P}}}| +{{\rm{E}}}_{{\rm{int}}}^{+}\approx$$\end{document}∣IP∣+Eint+≈ 10.5 eV. Intensity-dependent photoelectron kinetic energy distributions (PEKDs). a 2D mesh-grid false-colormap constructed with the self-normalized PEKDs, showing their evolution with increasing external field intensity. b 2D mesh-grid diverging-color map, showing the evolution of non-resonant MPI signals after derivation with respect to the external intensity. Diagonal dashed lines are plotted following Eq. (1) for different total (n+s) photon channels (indicated in white squares next to the lines), with I0 = 43.3 TW ⋅ cm−2 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$| {{\rm{I}}}_{{\rm{P}}}| +{{\rm{E}}}_{{\rm{int}}}^{+}\approx$$\end{document}∣IP∣+Eint+≈ 10.5 eV. Figure 4 contains a general overview of the resonant landscape, from the unperturbed regime to the tunneling one, containing all recorded PEKDs between ≈ 6 and 45 TW ⋅ cm−2. The spectra are dominated by vertical features that do not vary in kinetic energy. To denote them, we use lowercase Roman numerals i-v for the sub-ATI regime (KE ≤ ℏω), adding a superscript for every extra absorbed photon in the ATI replicas at +ℏω energy (i1, i2, i3, etc).Fig. 4Resonances and photoionization regimes in NH3.Self-normalized PEKDs of NH3 at different external field intensities (marked as tick-labels in the y-axis). Vertical dashed lines indicate resonant peaks at 0.33, 0.71, 0.99, 1.15, and 1.26 eV (i-v), together with their ATI replicas marked with numerical superscripts (i1, i1, i2, etc.). Resonances and photoionization regimes in NH3. Self-normalized PEKDs of NH3 at different external field intensities (marked as tick-labels in the y-axis). Vertical dashed lines indicate resonant peaks at 0.33, 0.71, 0.99, 1.15, and 1.26 eV (i-v), together with their ATI replicas marked with numerical superscripts (i1, i1, i2, etc.). The PEKD measured at the lowest intensity already contains all five resonant signals in the sub-ATI regime, with peaks denoted as i-v. As discussed in more detail in the next section, these resonances can be assigned to multiphoton excitation into Rydberg states converging to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${{\rm{NH}}}_{3}^{+}$$\end{document}NH3+, likely involving vibrational excitation in the umbrella mode in agreement with the absorption and ionization spectra shown in Fig. 1. To complement the energy-resolved analysis, we also extract photoelectron angular distributions (PEADs) through radial integration of the PEIs. PEADs provide additional fingerprints of the MPI pathways, since they directly depend on both the photon order and the symmetry of the intermediate resonant states37. They can be expressed as an expansion in spherical harmonics39: 2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$I(\theta )\approx \mathop{\sum }\limits_{n=0}^{n=N}{\beta }_{2n}\cdot {Y}_{2n}(\theta )=\frac{\sigma }{4\pi }\mathop{\sum }\limits_{n=0}^{n=N}{\beta }_{2n}\cdot {P}_{2n}(\cos \theta )$$\end{document}I(θ)≈∑n=0n=Nβ2n⋅Y2n(θ)=σ4π∑n=0n=Nβ2n⋅P2n(cosθ) where β2n are asymmetry related parameters, θ is the angle between the direction of the emitted photoelectron and the laser polarization vector, and P2n are Legendre polynomials of degree l (Pl). All the distributions are characterized by a favored photoelectron emission parallel to the laser polarization, as the strong-field multiphoton ionization leads to a selection of molecules oriented along the polarization axis40. The Supplementary Note 2 (and the figure and tables within) contains the β2n (2n = 2, 4, …16) directly extracted from the rBASEX inversion algorithm as a function of both kinetic energy and laser intensity. It also includes tables where these parameters have been computed for a range in PE kinetic energy associated to a given resonant contribution. Figures 5a and 6a show 2D maps (kinetic energy vs. θ) of the Abel-inverted PE images recorded at 12 and 24 TW ⋅ cm−2, respectively. Additionally, PEADs for selected resonant contributions are depicted in Figs. 5b and 6b. For signals ii-v, the last significant asymmetry parameter is β10 with strong contributions of all previous β2n. This is consistent with a dominant maximum angular momentum l = 5 together with a strong mixed character from intermediate resonances (see Fig. 5). Signal i also extends to β10 but differs in the sign of β4 and β6, giving rise to a distinct double-lobe structure (Fig. 6).Fig. 5Photoelectron angular distributions at a relatively-low laser intensity.a Abel-inverted photoelectron image of NH3 at 12.0 TW ⋅ cm−2 shown as 2D maps in polar coordinates. b (right subplots) Representative photoelectron angular distributions (PEADs) extracted for the resonant peaks (ii-iv) and one ATI replica (iii1).Fig. 6Photoelectron angular distributions at a moderately-high laser intensity.a Abel-inverted photoelectron image of NH3 at 24.0 TW ⋅ cm−2 shown as 2D maps in polar coordinates. b Representative photoelectron angular distributions (PEADs) extracted for the sub-ATI resonant peak (i) and its ATI replicas (i1, i2, i3). Photoelectron angular distributions at a relatively-low laser intensity. a Abel-inverted photoelectron image of NH3 at 12.0 TW ⋅ cm−2 shown as 2D maps in polar coordinates. b (right subplots) Representative photoelectron angular distributions (PEADs) extracted for the resonant peaks (ii-iv) and one ATI replica (iii1). Photoelectron angular distributions at a moderately-high laser intensity. a Abel-inverted photoelectron image of NH3 at 24.0 TW ⋅ cm−2 shown as 2D maps in polar coordinates. b Representative photoelectron angular distributions (PEADs) extracted for the sub-ATI resonant peak (i) and its ATI replicas (i1, i2, i3). The relative intensities of these resonant features are highly intensity-dependent. Several regions of interest can be identified. In the low-to-moderate intensity window (10–20 TW ⋅ cm−2), signals ii-v dominate, while channel i and its ATI replicas are depleted. At 12.0 TW ⋅ cm−2 (Fig. 5), PEADs for sub-ATI peaks ii-iv exhibit a four-lobe pattern characteristic of l ≃ 5. Their ATI counterparts, such as iii1, shift into five-lobe structures as an extra photon is absorbed. In the moderate-to-high intensity regime (20–35 TW ⋅ cm−2), resonance i reappears together with a series of ATI replicas (i1, i2, i3), whose intensities increase with field strength (Fig. 4). Their angular distributions (Fig. 6) differ sharply from ii-v: the feature i shows a double-lobe, and although the number of lobes increases for its ATI replicas, their shape is clearly modified. Notably, the ATI replicas exceed the signal intensity of the parent resonance, constituting the operational signature of the REATI behavior introduced above. Discussion Ammonia offers an unusually favorable landscape for resonance-assisted ionization because its long umbrella-mode progression creates near-continuous opportunities for multiphoton resonance with 800 nm light. A total of five main resonant channels have been found, labeled i-v (see Fig. 4). Their relative strength is modulated with the laser field intensity, but their position in kinetic energy does not vary. Eqn. (1) describes the expected energy of a photoelectron in a non-resonant pathway, where only the ionization potential experiences the ponderomotive shift. Resonant pathways are characterized by intermediate Rydberg states that, in contrast to the ground state, become modulated by the ponderomotive potential up to a similar extent. Therefore, the momentum splitting between the ejected photoelectron and the molecular cation becomes constant since they are always spaced by the same energy difference. This can be described by the following equation: 3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\rm{KE}}=(m)\hslash \omega +{{\rm{E}}}_{{\rm{R}}}-| {{\rm{I}}}_{{\rm{P}}}| +\Delta {{\rm{E}}}_{{\rm{int}}},$$\end{document}KE=(m)ℏω+ER−∣IP∣+ΔEint, where m is the number of photons connecting the Rydberg and ionic states, ER is the energy of the intermediate Rydberg state, and ΔEint is the difference in vibrational excitation between the Rydberg and the cation states. The five resonant channels have been assigned taking into account the symmetry and the g or u parity conditions for dipole-allowed transitions. Based on the Franck-Condon principle and the geometric similarities between the Rydberg states and the ground state of the cation, ΔEint ≈ 0 is expected to be favored.28Fig. 7 schematically illustrates the resonant pathways associated with channels (i-v). It is plotted using literature values41–45 while no intensity-dependent Stark shifts or laser dressing effects are included. In this field-free picture, we assign resonant signals i − v to the following pathways: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(i)\,\widetilde{X}\mkern5mu{}^{1}\mkern-4muA_{1}^{{\prime} }(2{p}_{z})\mathop{\to }\limits^{7h\nu }7p(e)\to \widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}+{\rm{PE}}\,(0.33\,{\rm{eV}},\Delta {{\rm{E}}}_{{\rm{int}}}\approx 3\,{\nu }_{2})$$\end{document}(i)X~A1′1(2pz)→7hν7p(e)→X~A2″2+PE(0.33eV,ΔEint≈3ν2)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(ii)\,\widetilde{X}\mkern5mu{}^{1}\mkern-4muA_{1}^{{\prime} }(2{p}_{z})\mathop{\to }\limits^{7h\nu }\widetilde{E}(d)\mathop{\to }\limits^{1h\nu }{\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}}+{\rm{PE}}\,(0.71\,{\rm{eV}},\Delta {{\rm{E}}}_{{\rm{int}}}\approx 0)$$\end{document}(ii)X~A1′1(2pz)→7hνE~(d)→1hνX~A2″2+PE(0.71eV,ΔEint≈0)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(iii)\,\widetilde{X}\mkern5mu{}^{1}\mkern-4muA_{1}^{{\prime} }(2{p}_{z})\mathop{\to }\limits^{7h\nu }\widetilde{F}(d)\mathop{\to }\limits^{1h\nu }{\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}}+{\rm{PE}}\,(0.99\,{\rm{eV}},\Delta {{\rm{E}}}_{{\rm{int}}}\approx 0)$$\end{document}(iii)X~A1′1(2pz)→7hνF~(d)→1hνX~A2″2+PE(0.99eV,ΔEint≈0)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(iv)\,\widetilde{X}\mkern5mu{}^{1}\mkern-4muA_{1}^{{\prime} }(2{p}_{z})\mathop{\to }\limits^{7h\nu }\widetilde{G}(s)\mathop{\to }\limits^{1h\nu }{\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}}+{\rm{PE}}\,(1.15\,{\rm{eV}},\Delta {{\rm{E}}}_{{\rm{int}}}\approx 0)$$\end{document}(iv)X~A1′1(2pz)→7hνG~(s)→1hνX~A2″2+PE(1.15eV,ΔEint≈0)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(v)\,\widetilde{X}\mkern5mu{}^{1}\mkern-4muA_{1}^{{\prime} }(2{p}_{z})\mathop{\to }\limits^{7h\nu }\widetilde{G}(s)\mathop{\to }\limits^{1h\nu }{\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}}+{\rm{PE}}\,(1.26\,{\rm{eV}},\Delta {{\rm{E}}}_{{\rm{int}}}\approx 1\,{\nu }_{2})$$\end{document}(v)X~A1′1(2pz)→7hνG~(s)→1hνX~A2″2+PE(1.26eV,ΔEint≈1ν2) where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{1}\mkern-4muA_{1}^{{\prime} }(2{p}_{z})$$\end{document}X~A1′1(2pz) and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}$$\end{document}X~A2″2 are the ground electronic state of neutral and cationic NH3, respectively. As observed, channels (ii) to (iv) can be assigned to similar (7+1) pathways involving different electronic intermediate states, and channel v is characterized by an intermediate Rydberg state more vibrationally excited than the final cation state. In contrast, channel (i) is attributed to a 7-photon absorption associated with a 3a1 → 7p(e) transition, followed by autoionization.Fig. 7Resonant pathways in ammonia photoionization.Field-free schematic of the resonance-assisted ionization pathways (i-v) in NH3via multiphoton ionization with 800 nm femtosecond pulses. Neutral Rydberg states (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{E}$$\end{document}E~, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{F}$$\end{document}F~, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{G}$$\end{document}G~, 7p(e)) converging to the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}$$\end{document}X~A2″2 ionic ground state are shown together with multiphoton excitation steps and electron emission into the continuum. Intensity-dependent Stark shifts or laser dressing are not included in the scheme. Photon absorption and electron emission are indicated by upward and downward arrows, respectively. Resonant pathways in ammonia photoionization. Field-free schematic of the resonance-assisted ionization pathways (i-v) in NH3via multiphoton ionization with 800 nm femtosecond pulses. Neutral Rydberg states (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{E}$$\end{document}E~, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{F}$$\end{document}F~, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{G}$$\end{document}G~, 7p(e)) converging to the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\widetilde{X}\mkern5mu{}^{2}\mkern-4muA_{2}^{{\prime\prime}}$$\end{document}X~A2″2 ionic ground state are shown together with multiphoton excitation steps and electron emission into the continuum. Intensity-dependent Stark shifts or laser dressing are not included in the scheme. Photon absorption and electron emission are indicated by upward and downward arrows, respectively. The assignment of channel i to excitation into the 7p(e) Rydberg state is further supported by previous vacuum-ultraviolet and photoionization studies of ammonia, which have shown that members of the Rydberg series extend well above the ionization threshold and can decay through autoionization pathways41–45. Moreover, Locht et al.42 described cases in which autoionization in ammonia deviates from the so-called propensity rule, which establishes that only autoionization processes with Δν = − 1 are expected, observing autoionization processes with up to Δν2 = − 9. In particular, excitation to the 7p(e) state has been experimentally observed to autoionize into the ground cationic state with Δν2 = − 3 42, a ΔEint that closely matches the kinetic energy of resonance (i) at 0.33 eV. Within the performed intensity scan, just a variation of only 2 TW ⋅ cm−2 causes a ponderomotive shift of 0.119 eV, comparable to the energy of one vibrational quantum of the umbrella mode ν2. Over the full intensity range explored, a change of 25 TW ⋅ cm−2 corresponds to a shift of approximately ≈ 1.50 eV, comparable to the photon energy of the driving field. The total ponderomotive shift at the end of the scan is approximately 2.7 eV, yet low-intensity resonances (ii, iii) remain significantly prominent. Focal-volume averaging (FVA) implies that, even when the on-axis peak intensity is off resonance, lower-intensity shells within the Gaussian focus profile can still efficiently drive REMPI processes15,36. We estimate that, in our experimental conditions, the Rayleigh length zR is approximately 0.63 mm. Considering that the ratio between the maximum and the minimum laser peak intensities is ≈ 7.5, all intensities within the scan window are covered in a range of 2.65 ⋅ zR = 1.76 mm, which still constitutes a substantial fraction of the interaction volume within the vacuum chamber. Together with the long ν2 progression characterizing ammonia (Fig. 1), this explains why low-intensity resonances persist up to the appearance of tunneling ionization signatures. The Stark shift is responsible for channel closing and channel switching effects, which are largely reported in atomic systems like Xe8,9, H11, Ar13 and Kr14, and in molecules17,36–38. The channel closing effect can be understood in terms of Stark shifting of the ionization potential. As the ponderomotive potential increases the ionization barrier, a given number of photon are not sufficient to ionize at a certain threshold laser intensity. The channel closing effect appears mainly for non-resonant MPI pathways, as depicted in Fig. 3. The channel switching effect is in contrast explained in terms of Stark shifting of intermediate Rydberg states. At a given threshold intensity, they are brought into resonance with a number of photons leading to strong contributions called Freeman resonances. In ammonia, both mechanisms are observed when increasing intensity as the PE spectra undergoes a gradual but substantial reorganization. The transition is smooth thanks to the long vibrational progressions and the FVA effect. By mapping energy- and angle-resolved yields versus UP(I), one can identify when specific Rydberg ladders come in (and out of) multiphoton resonance. This effect is illustrated in Fig. 8, where different numbers of laser photons become progressively resonant with Stark-shifted states and effectively cross the ionization potential.Fig. 8Channel closing and channel switching effects in the photoionization of ammonia.a Field-dressed displacement of the He I photoelectron vertical binding energy (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\widetilde{X}}^{2}{A}_{2}^{{\prime\prime} }$$\end{document}X~2A2″) versus laser intensity (adapted from ref. 24). b Field-dressed displacement of the VUV absorption manifold (Rydberg progression) versus laser intensity (spectra adapted from refs. 22,23). Energies are plotted against intensity (TW ⋅ cm−2). Vertical black arrows and their labels indicate nℏω (800 nm) photon sums. The map visualizes how increasing intensity (via the ponderomotive shift UP) moves Rydberg levels and the ionic threshold, explaining the reorganization of resonant pathways observed in Figs. 4–6. In particular, around ~ 10 TW ⋅ cm−2, the shift brings parts of the Rydberg manifold into multiphoton resonance with an alternative even-odd integer photon number; and around ~ 25 TW ⋅ cm−2 (denoted by the energy position of vertical arrows), the eight-photon channel crosses the vertical binding energy, while intermediate photons are in resonance with the centroid of absorption bands, reorganizing the accessible pathways and redistributing spectral weights. Channel closing and channel switching effects in the photoionization of ammonia. a Field-dressed displacement of the He I photoelectron vertical binding energy (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\widetilde{X}}^{2}{A}_{2}^{{\prime\prime} }$$\end{document}X~2A2″) versus laser intensity (adapted from ref. 24). b Field-dressed displacement of the VUV absorption manifold (Rydberg progression) versus laser intensity (spectra adapted from refs. 22,23). Energies are plotted against intensity (TW ⋅ cm−2). Vertical black arrows and their labels indicate nℏω (800 nm) photon sums. The map visualizes how increasing intensity (via the ponderomotive shift UP) moves Rydberg levels and the ionic threshold, explaining the reorganization of resonant pathways observed in Figs. 4–6. In particular, around ~ 10 TW ⋅ cm−2, the shift brings parts of the Rydberg manifold into multiphoton resonance with an alternative even-odd integer photon number; and around ~ 25 TW ⋅ cm−2 (denoted by the energy position of vertical arrows), the eight-photon channel crosses the vertical binding energy, while intermediate photons are in resonance with the centroid of absorption bands, reorganizing the accessible pathways and redistributing spectral weights. In Fig. 8a, the channel closing effects are visible through the Stark-shifting of the ionization potential. When considering the adiabatic biding energy (10.186 eV, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\nu }_{2}^{+}=0$$\end{document}ν2+=0), the 7ℏω channel closes at 11.12 TW ⋅ cm−2, and the 8ℏω closes at 37.08 TW ⋅ cm−2. When considering the vertical binding energy (10.91 eV, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\nu }_{2}^{+}=6$$\end{document}ν2+=6), the 7ℏω channel is always closed, and the 8ℏω channel closes at 24.95 TW ⋅ cm−2. The channel switching effect, illustrated in Fig. 8b, is however particularly relevant to explain the modulation of the observed resonant pathways. Based on the experimental results from Fig. 4, we define four different intensity windows where signals undergo substantial reorganization. The first window corresponds to the unperturbed regime, where all i − v signals are present, defining the starting point at the lowest intensities. A small ponderomotive shift of ~0.6 eV at 10 TW ⋅ cm−2 changes the landscape: signal i and its ATI series fade out of the spectra, leading to the dominance of low-to-moderate intensity resonances ii-v. In the second intensity window, from 10 to 20 TW ⋅ cm−2, high-lying Rydberg states lose the 6ℏω resonance and become resonant with 7ℏω laser photons (Fig. 8). Parity selection rules highly favor the change from even to uneven number of photons, enhancing their relative strength. At moderate external fields, small ac-Stark shifts perturb vibrational accessibility and redistribute Franck-Condon overlap, inverting their relative strengths in the PE spectra. Signal v decreases in intensity, while iii and iv become dominant (Fig. 4). In the third intensity window, for intensities above 20 TW ⋅ cm−2, the shift in energy is comparable to the photon energy, UP(I) ≈ ℏω, bringing back the previously depleted signal i as a Stark-induced Freeman resonance, now correlating with a strong ATI series. The redistribution is parity- and angular-momentum selective: even and odd photon orders preferentially access states of different orbital character. This is why, when approaching the 7-photon channel closure (10 TW ⋅ cm−2), signals i and its replicas do not survive the channel-switching mechanism; whereas all resonances are still present after nominal closure of the 8-photons channel at 40 TW ⋅ cm−2. At the highest intensities, instead of observing an additional channel-switching mechanism, tunneling and re-scattering start dominating the picture, masking any resonance assisted pathway. The relevance of the tunneling regime in our experiments at high intensities can be evaluated based on the estimated Keldysh parameter1: 4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\gamma =\sqrt{\frac{{{\rm{I}}}_{{\rm{P}}}}{2{{\rm{U}}}_{{\rm{P}}}}},$$\end{document}γ=IP2UP, where, according to the definition, the lower the value of the Keldysh parameter, the more important the contribution of tunneling ionization is. It approximately shifts from γ(10 TW ⋅ cm−2) = 2.92 to γ (35 TW ⋅ cm−2) = 1.56. This last value justifies the tunneling ionization observed at the highest intensities. In non-resonant strong-field MPI processes, the angular momentum l acquired by the ejected photoelectron can be directly related to the minimum number of photons needed to reach the continuum from the initial neutral ground state and to its orbital quantum number. In REMPI processes, considering favorable parity conditions, this momentum can be additionally related to the extra number of photons necessary to connect the last intermediate resonant Rydberg state and the continuum17,36,37. Besides, the dominant angular quantum number l of these photoelectrons can be extracted from fitting the PEADs to squared Legendre polynomials \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${({P}_{l}(\cos \theta ))}^{2}$$\end{document}(Pl(cosθ))2 of order l or, in a simpler approach, just by the number of local minima found between 0-180o angle of emission. Figure 9 shows a comparison between measured PEADs and the expected \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${({P}_{l}(\cos \theta ))}^{2}$$\end{document}(Pl(cosθ))2 distribution according to the number of local minima.Fig. 9Measured versus expected photoelectron angular distribution.Comparison between the PEADs extracted for the resonant peaks i, iii and iii1 (top part of every subplot, from 0 to 180o) and their expected \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${({P}_{l}(\cos \theta ))}^{2}$$\end{document}(Pl(cosθ))2 distribution according to the number of observed lobes (bottom part of every subplot, from 180 to 360o). Measured versus expected photoelectron angular distribution. Comparison between the PEADs extracted for the resonant peaks i, iii and iii1 (top part of every subplot, from 0 to 180o) and their expected \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${({P}_{l}(\cos \theta ))}^{2}$$\end{document}(Pl(cosθ))2 distribution according to the number of observed lobes (bottom part of every subplot, from 180 to 360o). Resonant channels ii to v exhibit similar angular momentum patterns, with four-lobes or five local minima that are associated with a dominant l ≊ 5 (see Figs. 5 and 9). The extracted β2n parameters (SI-2) show indeed that the last significant asymmetry parameter is β10 with strong contributions of all previous β2n. These l values are compatible with an 8-photon MPI from the nitrogen lone-pair, nN, which has l = 1, as the angular parity of the photoelectron is therefore preserved for an even-photon MPI process. However, the high angular momentum, l ≊ 5, of the photoelectrons reflected by the PEADs are not in agreement with the assigned intermediate Rydberg states neither with any excited state lying in this energy region22. In opposition to the field-free picture, the strong-field dressing of the states not only shifts their energy, but is observed to also cause an appreciable l-character mixing of nearby Rydberg members, which would explain the complex angular structure observed, and the structural similarity on the family of ii − v patterns. We note, however, that interference patterns from tunneling ionization lying in the background cannot be decoupled from the observed angular distributions, which also affects the extracted β2n parameters. The assignment proposed here for channel (i) is additionally supported by the huge shift in angular momentum from signals ii − v, with four-lobe structures, towards signal i, with only two lobes. This points towards the presence of a resonance with a low-angular momentum state while no extra photons are absorbed, which is consistent with autoionization from the 7p(e) Rydberg state. However, despite having the expected number of lobes for a \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${P}_{3}{\left.(\cos \theta )\right)}^{2}$$\end{document}P3(cosθ)2 distribution, their position does not match (see Fig. 9). This could reflect a high mixing between nearby laser-dressed vibronic states. The PEADs are hence showing that the intermediate step in the ionization process can not be considered as a single well-defined quantum state. A quantitative description of this vibronic mixing would require dedicated theoretical calculations beyond the scope of the present work. Finally, the PEADs for the REATI pathways in as well as iin − vn are characterized by a n increase of visible lobes and thus of angular momentum as expected when absorbing extra photons. Figure 5 shows this effect for signals iin − vn, while Fig. 6 does it for signals in. The present results show an intensity window between 20 and 35 TW ⋅ cm−2 where resonant channels can be understood in terms of Stark-induced Freeman resonances. A weak resonance emerges at ≈ 0.33 eV together with its ATI replicas separated by ℏω. For an increasing intensity, the replicas outgrow the parent line, which is almost depleted, and eventually dominate the picture, defined here as a REATI phenomenon. Previous studies have demonstrated that the intensity ATI structures can be modulated via external field-strengths in both atoms and molecules due to resonance-enhanced ionization pathways30–33. Experimentally, Wang. et al.30 describe modulations in the high-order ATI spectra (HATI) of formic acid at specific laser intensities. Theoretically, Li et al.31 have reported enhancements in the HATI spectra of Ar, Xe and N2 when high-lying excited states become accessible. Calculations by Rodríguez et al.32 have also shown that tuning the extreme ultraviolet (XUV) energy in and out of resonance produces modulations in the relative strength of ATI signals. Finally, resonance assisted pathways33 were theoretically explored in the polar system NaK, where strong light-matter coupling produces Autler-Townes-split field-dressed intermediates, in which recycling through the dressed state biases emission into higher ATI orders. However, these works mainly point towards high-order ATI (HATI) reweighting, situations in which the plateau of emitted replicas can be enhanced or suppressed as a function of the intensity. Here, the observed inversion occurs already between the Freeman parent and its first few ATI orders, and does not manifest as an intensity modulation of high order replicas30,31. In the final step of a MPI process, several competing ionization pathways are available under strong-field conditions. Direct photoionization into the continuum or further photon absorption into ATI replicas are the main ones. However, in NH3, promotion to a high-lying Rydberg state that extends far into the continuum becomes possible. Therefore, autoionization through vibronic coupling with the ionic manifold, dissociative ionization or predissociation into neutral fragments have to be considered too. The relative importance of these processes is imposed by their characteristic timescales and by the degree to which the intermediate state remains coupled to the driving laser field. The presence of REATI therefore implies that population initially promoted into a resonant intermediate state is not immediately depleted by the fastest available decay channels, but instead remains coupled to the field long enough for additional photon absorption to become competitive. Otherwise, autoionization or direct ionization would dominate, and ATI replicas would remain weaker than the parent resonance, as observed for conventional Freeman resonances. We explain the origin of this preference towards further photon absorption in NH3 based on two main experimental observations. First, channel (i) is spectroscopically assigned to excitation into the 7p(e) state followed by autoionization, consistent with its low kinetic energy and with previous photoionization studies. Field-free autoionization lifetimes of high-lying Rydberg states in small molecules are typically longer than an optical cycle and can be comparable to or exceed the pulse duration, enabling further photon absorption during the pulse. In the case of NH3, these high-lying Rydberg states have long vibrational progressions with very narrow bandwidths (below 15 cm−1), which would indicate lifetimes between hundreds of femtoseconds up to picoseconds41,43. Although these are field-free lifetimes, which may differ under laser dressing due to continuum coupling and dynamic Stark effects, they nevertheless indicate a sufficiently long effective residence time within the laser-dressed manifold. Second, their PEADs reveal strong deviations from the pattern expected for a well-defined intermediate state, pointing to substantial mixing of the electronic character. Under the effect of the laser field, the vibronic structure of the molecule is modified by ac-Stark shifts and laser dressing. As a result, the intermediate step in the ionization process cannot be considered as an isolated Rydberg state, but rather a laser-dressed manifold of near-resonant vibronic states. Upon strong-field excitation, the molecule can transferred within the manifold of coupled states during the pulse duration, before autoionization or dissociation happens. The unusually long vibrational progressions along the umbrella mode (with narrow bandwidths), combined with strong-field dressing, lead to a redistribution of population among nearby states and to extended lifetimes within the manifold. Under these conditions, the reported results are consistent with a scenario in which autoionization becomes sufficiently slow, depleting the parent Freeman line while leading to a preferred further photon absorption, responsible for the pronounced REATI observed here. This behavior, which seems to be favored by the unusually long umbrella-mode progressions, differs from conventional Freeman resonances, where the enhancement is largely confined to the parent line, and from HATI re-weighting, where intensity-dependent effects primarily modulate the high-energy plateau. Conclusion The strong-field photoionization of ammonia has been thoroughly investigated by means of femtosecond photoelectron velocity map imaging. The photoelectron images were measured for intensities up to 45 TW ⋅ cm−2. The results consist of the corresponding photoelectron kinetic energy as well as angular distributions as a function of the field intensity. Different intensity windows are identified and a roadmap of phenomena including non-resonant and resonant ionization, Freeman resonances, above-threshold ionization (ATI) and tunneling ionization are recovered. The results demonstrate that ammonia exhibits a resonance-enhanced above-threshold ionization (REATI) regime in which the ATI replicas of a Stark-induced Freeman resonance dominate over the parent line. The effect emerges in a well-defined large intensity window (20–35 TW ⋅ cm−2), where field dressing brings parts of the dense Rydberg/ν2 manifold into near-multiphoton resonance. High-lying Rydberg states of ammonia overlap with the continuum and span over long energies, enabling autoionization seeding and longer dwell times. This strongly enhances the probability toward extra-photon absorption, almost depleting the sub-ATI channel and producing strong ATI replicas. Beyond establishing REATI in a polyatomic molecule, the intensity-tagged, angle-resolved approach works as a Stark-scan of field-dressed structures, turning strong-field ionization into a vibrationally specific probe of resonant intermediates. Methods The experiment consists of a femtosecond laser system and a molecular beam setup to record fragment ion or photoelectron images using the velocity map imaging (VMI) technique46. The laser (Spectra Physics) is an amplified Ti:sapphire system delivering 50 fs, 3.5 mJ pulses centered at 800 nm with 1 kHz repetition rate. The laser enters into the reaction chamber and is focused by a 30 cm focal length lens on the gas-phase molecular sample. The sample is inserted into the reaction chamber through a ~ 20 μm diameter pinhole (found in the repeller plate) in the form of an effusive molecular beam of pure NH3, with measured vapor pressures in the interaction chamber of ~ 1 ⋅ 10−6mbar. The electrons generated in the interaction region are extracted perpendicularly to the laser propagation axis and accelerated by the electrostatic lens system working in VMI configuration. At the end of a 40 cm time-of-flight (TOF) tube, charged particles are projected onto a 2D detector composed of two microchannel plates (MCP) in Chevron configuration coupled to a phosphor screen (P47). The TOF tube is equipped with a μ-metal cylindrical shield to avoid interference of the photoelectron trajectories with external magnetic fields. Typical repeller voltages were −4.5 kV, with optimum velocity mapping conditions found for Vextractor/Vrepeller = 0.75. In these conditions all photoelectrons with the same velocity vector are mapped on the same point on the detector. A Peltier-cooled 12 bit complementary metal oxide semiconductor camera (LaVision Imager M-lite 2 CMOS) recorded the phosphorescence emitted by the phosphor screen as a 2D image. The measured raw photoelectron images (PEIs) are Abel inverted to extract the desired information (kinetic energy and angular distributions) using the rBASEX method34 as implemented in the PyAbel Python software package35. The polarization of the laser beam was set parallel to the detector face to provide the cylindrical symmetry required for the procedure of Abel inversion of the ion/photoelectron images. Typical acquisition times for each image are 60 s corresponding to 60000 laser shots. Supplementary information Supplementary Material Supplementary Material