Spectral fringes without subcycles in Schwinger pair production and Dirac materials I. A. Aleksandrov, M. A. Dorodnyi, E. D. Akimkina Abstract Spectral fringes in Schwinger pair creation are usually attributed to structured driving, such as carrier oscillations, pulse trains, or multiple creation events. We show that pronounced fringes can arise even for smooth, carrier-free single-lobe electric-field pulses. Two bell-shaped profiles that are nearly indistinguishable in real time—a Gaussian pulse and a weakly deformed variant—produce qualitatively different longitudinal momentum spectra in the nonadiabatic crossover: the Gaussian spectrum remains smooth, whereas the deformed pulse develops strong fringes as the Keldysh parameter approaches unity. Exact numerical solutions in scalar and spinor QED agree with a semiclassical turning-point analysis and trace the effect to a turning-point dominance transition, where the leading saddle becomes irrelevant and subleading contributions interfere. We demonstrate the same mechanism in a solid-state Schwinger analog described by a gapped two-dimensional Dirac model relevant to epitaxial graphene on SiC, and discuss an energy-resolved pump–probe route to observing the predicted modulation. Introduction. Vacuum pair production in strong electromagnetic fields—the Schwinger mechanism—is a paradigmatic nonperturbative prediction of quantum electrodynamics (QED), connecting tunneling physics to vacuum instability near the critical field scale. Seminal work established the effect for static fields sauter_1931 ; heisenberg_euler ; weisskopf ; schwinger_1951 . Despite its conceptual simplicity, Schwinger pair creation remains a benchmark problem for strong-field quantum theory, with relevance to ultraintense-laser physics and related intense-field settings dipiazza_rmp_2012 ; xie_review_2017 ; gonoskov_2022 ; fedotov_review ; popruzhenko_ufn_2023 . A striking outcome of many studies is the appearance of oscillatory “fringe” patterns in momentum-resolved spectra for structured, time-dependent pulses hebenstreit_prl_2009 ; dumlu_dunne_prl_2010 ; dumlu_dunne_prd_2011_1 ; dumlu_dunne_prd_2011_2 ; akkermans_prl_2012 ; abdukerim_plb_2013 ; kohlfuerst_prd_2013 ; hebenstreit_plb_2014 ; linder_prd_2015 ; aleksandrov_prd_2017 . These fringes admit a natural time-domain multiple-slit interpretation. In particular, semiclassical analyses based on complex-time turning points (or worldline instantons) show that fringes typically require multiple comparable saddle contributions, often generated by subcycle structure, pulse trains, or multi-component fields dumlu_dunne_prl_2010 ; dumlu_dunne_prd_2011_1 ; dumlu_dunne_prd_2011_2 ; akkermans_prl_2012 . We find that a single-lobe, carrier-free pulse can produce pronounced fringes purely from saddle competition, even when the real-time field contains no subcycle structure . We compare two bell-shaped pulses with nearly identical real-time profiles, e(z)=\mathrm{e}^{-z^{2}} and e(z)=\mathrm{e}^{-z^{2}-z^{4}} , and find a sharp change in the longitudinal spectrum in the nonadiabatic crossover. For the Gaussian, the spectra remain smooth and single-peaked, while for the deformed pulse they develop strong oscillatory fringes once the Keldysh parameter \gamma approaches \mathcal{O}(1) . Exact numerical calculations (for both scalar and spinor QED) agree with a semiclassical turning-point (TP) analysis and identify the origin as a TP dominance transition: as \gamma is tuned through unity, the dominant complex-time TP moves rapidly toward i\infty , so the next pair(s) become comparable and robust interference turns on. Crucially, this “interference without subcycles” is controlled by the global analytic continuation of the pulse into complex time rather than by any real-time oscillations. To connect to near-term experiments, we analyze the same mechanism in a gapped two-dimensional Dirac model relevant to graphene. Interband electron–hole generation in Dirac materials under strong in-plane electric fields is widely discussed as a condensed-matter analog of Schwinger pair production (or Landau–Zener tunneling), with both early and more recent theoretical treatments and proposed signatures in transport and (in some settings) radiation allor_prd_2008 ; dora_prb_2010 ; lewkowicz_prb_2011 ; gavrilov_prd_2012 ; akal_prd_2016 ; akal_prd_2019 ; linder_prb_2018 . Epitaxial graphene on SiC provides a mature, wafer-scale platform in which substrate and buffer-layer coupling can break sublattice symmetry and generate an effective Dirac mass (often reported as a gap-like feature near the Dirac point) zhou_nat-mat_2007 ; kim_prl_2008 ; enderlein_njp_2010 ; yu_apl_2013 . The gap is interface-dependent; intercalation can decouple the buffer layer and yield quasi-free-standing (essentially gapless) graphene riedl_prl_2009 , offering a closely related reference system. Epitaxial graphene on SiC is routinely studied with broadband THz-to-mid-IR ultrafast spectroscopy and device-level in-plane driving santos_sci-rep_2016 ; bianco_apl_2015 ; singh_jpd_2024 ; paschke_aqt_2020 . A natural primary readout in this setting is the pump-induced change of the low-energy optical/THz response (transient conductivity/absorption), which is routinely measured in ultrafast spectroscopy on SiC-supported graphene. Nonlinear THz response and emission channels in graphene are well documented hafez_opt-mat_2020 , including THz-driven optical emission interpreted via field-induced e – h generation oladyshkin_prb_2017 . In the parameter regime emphasized below we focus on mid-IR/near-IR driving with duration \tau\sim 10 – 100\penalty\ \text{fs} and peak in-plane fields of order tens of kV/cm, which spans the nonadiabatic crossover for experimentally relevant gaps. We also briefly comment on aligned graphene/hBN as an alternative route to a gapped Dirac spectrum via inversion-symmetry breaking and moiré potentials wallbank_prb_2013_1 ; wallbank_prb_2013_2 ; wallbank_adp_2015 ; jung_prb_2017 ; li_comm-phys_2020 . We establish the effect in QED and then demonstrate its graphene analog. Spectral fringes in setups without subcycles. We start with standard (3+1) QED in units \hbar=c=1 and consider spatially homogeneous, linearly polarized electric fields E(t)=E_{0}e(t/\tau), (1) where e(z) is a carrier-free single-lobe profile with e(0)=1 and \tau sets its duration. The nonadiabaticity is quantified by the Keldysh parameter \gamma\equiv m/(|e|E_{0}\tau) (with m the electron mass and e<0 its charge). We compare two smooth profiles, e(z)=\mathrm{e}^{-z^{2}} and e(z)=\mathrm{e}^{-z^{2}-z^{4}} , which are nearly indistinguishable on the real-time axis. In temporal gauge, A(t)=-\int_{0}^{t}\!E(t^{\prime})dt^{\prime} , we compute the asymptotic particle spectrum n_{\mathbf{p}} by exact mode evolution (see, e.g., Refs. aleksandrov_prd_2017 ; aleksandrov_prd_2016 for numerical details). In the adiabatic regime \gamma\ll 1 , both pulses yield the expected smooth, bell-shaped longitudinal spectrum. In contrast, for \gamma\sim 1 – 2 the deformed pulse produces pronounced oscillations in the longitudinal momentum distribution, whereas the Gaussian spectrum remains smooth (Fig. 1 ). We verify that this change is not a numerical artifact by comparing with a semiclassical WKB/TP construction and by repeating the analysis in both scalar and spinor QED. Figure 1: Momentum distributions of the produced particles as a function of longitudinal momentum p_{\parallel} for p_{\perp}=0 in scalar QED for (a) a Gaussian pulse and (b) a deformed pulse. The numerical results are obtained for a peak field E_{0}=0.1E_{\text{c}} , where E_{\text{c}}=m^{2}/|e| is the critical field strength. Pulse durations are \tau=5m^{-1} (solid), \tau=10m^{-1} (dotted), and \tau=20m^{-1} (dashed), corresponding to \gamma=2.0 , 1.0 , 0.5 , respectively. For clarity, curves are rescaled by the factors indicated in the panels. Semiclassical mechanism. The semiclassical interpretation follows the complex-time TP framework dumlu_dunne_prd_2011_1 ; linder_prd_2015 ; popov_2005 ; oertel_prd_2019 ; taya_jhep_2021 : particle production is controlled by zeros of the effective energy Q_{\mathbf{p}}(t)=\sqrt{\pi_{\perp}^{2}+[p_{\parallel}-eA(t)]^{2}},\quad\pi_{\perp}\equiv\sqrt{m^{2}+p_{\perp}^{2}}, (2) in complex time. Fringes arise when several TPs have comparable weight and interfere. The TPs t_{i} are identified via Q_{\mathbf{p}}(t_{i})=0 , which is equivalent to a(t_{i}/\tau)=\frac{\pm i\pi_{\perp}-p_{\parallel}}{m}\,\gamma, (3) where a(z)=-\int_{0}^{z}\!e(z^{\prime})dz^{\prime} is the dimensionless vector potential. For the Gaussian pulse, one TP remains closest to the real axis for all \gamma , so a single saddle dominates. For the deformed pulse, the analytic continuation along the imaginary axis is qualitatively different: as shown in Fig. 2 , the would-be dominant TP rapidly loses relevance (effectively migrating to i\infty ) near the critical value \gamma_{*}=\int_{0}^{\infty}\!\mathrm{e}^{z^{2}-z^{4}}dz\approx 1.381 (see Supplemental Material SM for a quantitative semiclassical analysis). In the crossover region \gamma\sim\gamma_{*} , several subleading TPs acquire comparable weights and interfere, producing fringes despite the absence of subcycle structure in E(t) . This turning-point dominance transition provides a compact diagnostic—whether \int_{0}^{\infty}E(it)dt converges—indicating when a smooth single-lobe pulse can generate fringed spectra. In other words, the effect is set by the global real-time pulse profile (through its analytic continuation), not by any oscillatory features on the real-time axis. Figure 2: Turning-point trajectories for \mathbf{p}=0 in the first quadrant as \gamma is varied (deformed pulse). For \gamma\ll 1 , the dominant semiclassical contribution comes from the TP on the imaginary axis (red trajectory) that lies closest to the real axis. As \gamma approaches \gamma_{*}\simeq 1.381 , this TP rapidly migrates to i\infty and becomes irrelevant, so that several remaining TPs acquire comparable weights; their interference produces pronounced fringes in the momentum spectrum. For clarity, only the 11 TPs closest to the origin are shown. The 10 initially subleading TPs form degenerate pairs at \gamma=0 and, for \gamma\gg 1 , asymptote to the positions indicated by black circles. The full TP set is symmetric under reflection about the imaginary axis. Inset: critical value \gamma_{*}(a) defined in Eq. ( 5 ) as a function of the deformation parameter a for the pulse family ( 4 ). The TP dominance transition is not restricted to the specific deformation e(z)=\mathrm{e}^{-z^{2}-z^{4}} but persists, for example, for a broad one-parameter family of smooth single-lobe pulses, e_{a}(z)=\mathrm{e}^{-z^{2}-az^{4}},\qquad a>0, (4) which continuously interpolates between a nearly Gaussian envelope and a more rapidly decaying profile along the imaginary axis. In this family the critical value for the disappearance of the would-be leading TP is set by the convergent integral \gamma_{*}(a)\equiv\int\limits_{0}^{\infty}\mathrm{e}^{z^{2}-az^{4}}dz, (5) so that the onset of multi-saddle competition and spectral fringes is expected for \gamma\sim\gamma_{*}(a) (see inset of Fig. 2 ). Varying a therefore provides a simple control knob that shifts the transition and can, e.g., move the fringe-onset threshold into a deeper tunneling regime (smaller \gamma_{*} ), while leaving the real-time waveform single-lobed and carrier-free. Additional numerical examples for a\neq 1 are given in the Supplemental Material SM . Moreover, we have verified numerically that the same TP dominance transition occurs for super-Gaussian single-lobe profiles of the form e(z)=\mathrm{exp}(-z^{4n}) with n=1 , 2 , \ldots , with the corresponding critical values \gamma_{*}\approx 0.91 , 0.94 , \ldots for these cases. Together with the deformation family ( 4 ), this confirms that the mechanism is not tied to a particular analytic form, but is generic across a broad class of smooth carrier-free pulses. We now turn to a solid-state realization, showing the same TP dominance transition in gapped Dirac graphene under experimentally relevant ultrafast driving. Condensed-matter analog in gapped Dirac graphene. To connect the TP dominance transition to near-term experiments, we consider a solid-state Schwinger analog in a two-dimensional massive Dirac system. A minimal model for gapped graphene in a spatially uniform, time-dependent in-plane field is H_{\mathbf{k}}(t)=v_{\text{F}}\,\boldsymbol{\sigma}\!\cdot\!\bigl[\hbar\mathbf{k}-e\mathbf{A}(t)\bigr]+\Delta\sigma_{z}, (6) with instantaneous band energy \varepsilon_{\mathbf{k}}(t)=\sqrt{\Delta^{2}+v_{\text{F}}^{2}|\hbar\mathbf{k}-e\mathbf{A}(t)|^{2}} . As in QED, TPs are given by zeros of the instantaneous gap, \varepsilon_{\mathbf{k}}(t)=0 , so fringes arise from competition among complex saddles. We take v_{\text{F}}=10^{6}\penalty\ \text{m}/\text{s} and label states by wave vector \mathbf{k} (with momentum \hbar\mathbf{k} ). In graphene-like Dirac systems, interband electron–hole generation in a uniform electric field is a well-known condensed-matter analog of Schwinger pair production allor_prd_2008 ; dora_prb_2010 ; gavrilov_prd_2012 ; akal_prd_2016 . Starting from a filled valence band, the post-pulse conduction-band population n_{\mathbf{k}} directly parallels the QED momentum spectrum. We quantify adiabaticity by comparing the intrinsic momentum scale \Delta/v_{\text{F}} to the field impulse |e|E_{0}\tau , which defines the Dirac Keldysh parameter \gamma_{\text{D}}=\frac{\Delta}{|e|v_{\text{F}}E_{0}\tau}\,. (7) We compute n_{\mathbf{k}} by solving the graphene quantum kinetic equations (QKEs) derived from the adiabatic-band (Bogoliubov) system; the full derivation for general and linearly polarized fields are given in the Supplemental Material SM , together with a constant-field benchmark that reproduces the Landau–Zener/Schwinger exponential in the quasistatic limit, n_{\mathbf{k}}\approx\mathrm{exp}(-\pi E_{\text{D}}/E_{0}) , where E_{\text{D}}=\Delta^{2}/(|e|\hbar v_{\text{F}}) is the critical field strength dora_prb_2010 ; gavrilov_prd_2012 ; akal_prd_2016 ; akal_prd_2019 . For gapless graphene ( \Delta=0 ), the QKE system was obtained in Refs. smolyansky_particles_2019 ; smolyansky_particles_2020 . Figure 3 shows that the QED pattern carries over to gapped graphene: for a Gaussian envelope the longitudinal spectra remain smooth, while for the deformed pulse they develop pronounced fringes once \gamma_{\text{D}} approaches unity. For \gamma_{\text{D}}\ll 1 , both envelopes yield spectra whose peak values are close to the constant-field estimate, consistent with a quasistatic nonperturbative production channel. In Fig. 3 we used E_{0}=40\penalty\ \text{kV}/\text{cm} and \Delta=0.1\penalty\ \text{eV} , leading to E_{\text{D}}\approx 152\penalty\ \text{kV}/\text{cm} . In all graphene calculations, the populated wave vectors satisfy |\mathbf{k}|\ll a^{-1}\simeq 4\times 10^{9}\penalty\ \text{m}^{-1} (with lattice constant a=2.46\penalty\ $\mathrm{\SIUnitSymbolAngstrom}$ ), so the low-energy gapped Dirac model remains well justified. Figure 3: Longitudinal momentum distributions of the produced carriers in gapped graphene as a function of k_{\parallel} for k_{\perp}=0 and \Delta=0.1\penalty\ \text{eV} for (a) Gaussian and (b) deformed pulses. The field amplitude is E_{0}=40\penalty\ \text{kV}/\text{cm} . Pulse durations are \tau=20\penalty\ \text{fs} (solid), \tau=30\penalty\ \text{fs} (dotted), and \tau=50\penalty\ \text{fs} (dashed), corresponding to \gamma_{\text{D}}=1.25 , 0.833 , and 0.5 , respectively. For clarity, curves are rescaled by the factors indicated in the panels. To gauge the expected signal, it is useful to translate the post-pulse carrier distribution into experimentally accessible observables. While our microscopic output is the post-pulse conduction-band occupation in momentum space, the most direct observables are pump-induced changes in THz-to-mid-IR conductivity/absorption/reflectivity, which probe momentum-weighted integrals of the distribution. An order-of-magnitude estimate for the total number of excited carriers N in the illuminated region is obtained by integrating the distribution over the characteristic momentum scale k_{\Delta}\equiv\Delta/(\hbar v_{\text{F}})\sim 10^{8}\penalty\ \text{m}^{-1} (for \Delta\simeq 0.1\penalty\ \text{eV} ). Including spin and valley degeneracy ( g=4 ), one finds N\sim gf_{\text{max}}(k_{\Delta}L)^{2}/(4\pi) , where f_{\text{max}} is the peak occupation and L is the characteristic interaction length. For a far-field spot of linear size L\sim 10 – 100\penalty\ $\text{\,}\mathrm{\SIUnitSymbolMicro m}$ , this corresponds to a non-negligible number of photoexcited carriers per pulse even for f_{\text{max}}\sim 10^{-8} , implying a measurable pump-induced modification of the low-energy response (conductivity/absorption) in standard ultrafast infrared/THz probes. To quantify the strength of spectral fringes we introduce a simple extrema-based “wiggliness” measure. For each longitudinal momentum distribution, we identify the sequence of successive local maxima and minima (including the endpoints of the plotted interval) and sum the absolute changes of the spectrum between neighboring entries. Normalizing this total variation by the peak value yields a dimensionless fringe-visibility measure \nu : a smooth, single-peaked spectrum gives \nu\simeq 1 , while resolved oscillations increase \nu>1 . Using \nu as a compact diagnostic, Fig. 4 summarizes where the interference sets in for the deformed single-lobe pulse. Panel (a) shows \nu in the (\tau,\Delta) plane at fixed E_{0}=40\penalty\ \text{kV}/\text{cm} , and panel (b) shows \nu in the (\tau,E_{0}) plane at fixed \Delta=0.1\penalty\ \text{eV} . In both cases the enhanced-fringe region tracks the nonadiabatic crossover \gamma_{\text{D}}\simeq 1 , consistent with the TP dominance transition picture. Experimentally, the fringe-onset maps indicate where the momentum distribution is most sensitive to pulse shape and therefore where the largest pump-induced changes in low-energy optical response are expected. In far-field pump–probe measurements this sensitivity can be accessed through transient absorption/reflectivity in the THz-to-mid-IR range, which probes momentum-shell averages set by the interband resonance condition 2\varepsilon_{\mathbf{k}}\simeq\hbar\Omega and is therefore naturally sensitive to structured post-pulse populations dawlaty_apl_2008 ; horng_prb_2011 ; jnawali_nano_lett_2013 . In particular, after angular averaging the post-pulse distribution can be recast as an energy-resolved spectrum dN/d\varepsilon ; in the parameter range of Fig. 4 this spectrum exhibits oscillations with characteristic spacing \delta\varepsilon_{\text{fr}}\sim 20\penalty\ \text{meV} (see Supplemental Material SM ). Since interband probes address 2\varepsilon through \hbar\Omega , detecting this modulation requires a spectral resolution on the order of \Delta(\hbar\Omega)\lesssim 2\delta\varepsilon_{\text{fr}}\sim 40\penalty\ \text{meV} , which is feasible with narrowband ultrafast probes. In device geometries, the same nonequilibrium carrier population can also generate transient currents and radiation signals whose connection to field-driven Landau–Zener/Schwinger carrier creation has been discussed in the graphene literature dora_prb_2010 ; lewkowicz_prb_2011 ; gavrilov_prd_2012 ; gavrilov_universe_2020 ; oladyshkin_prb_2017 . Finally, note that finite spectral/momentum resolution and dephasing primarily reduce fringe contrast; the energy-domain analysis in the Supplemental Material shows the modulation survives realistic broadening scales. Figure 4: Dimensionless fringe-visibility measure \nu for the deformed single-lobe pulse e(z)=\mathrm{exp}(-z^{2}-z^{4}) in the gapped-graphene Dirac model. Panel (a) shows \nu in the (\tau,\Delta) plane at fixed field amplitude E_{0}=40\penalty\ \text{kV}/\text{cm} . Panel (b) displays \nu in the (\tau,E_{0}) plane at fixed gap parameter \Delta=0.1\penalty\ \text{eV} . The red curves mark the nonadiabatic crossover \gamma_{\text{D}}=1.0 , along which the onset of pronounced oscillatory fringes is observed. A possible concern is that an idealized strictly single-lobe field has a nonvanishing time integral, whereas a freely propagating far-field pulse cannot carry a static (zero-frequency) component. In practice, waveform-asymmetric single-/half-cycle transients are accompanied by a weak, long tail of opposite sign that restores zero net field area, and similar effective single-lobe driving can also be implemented through near-field coupling from antennas or circuit elements schmuttenmaer_2004 ; isgandarov_2021 . In the Supplemental Material SM , we explicitly verify that the fringe formation persists for zero-area pulses obtained by adding such a compensating tail. Discussion and outlook. Our results show that nonperturbative spectra can depend qualitatively on the global analytic structure of otherwise smooth, carrier-free pulses. In QED this yields “interference without subcycles” driven by turning-point dominance transitions rather than real-time carrier oscillations, and we demonstrate the same mechanism in a realistic solid-state analog. A natural question is whether this sensitivity can be controlled experimentally. In practice, experiments shape only the real-time waveform; the turning-point picture provides an organizing principle that predicts when small, smooth deformations of a single-lobe pulse trigger a dominance switch among saddles and hence qualitative spectral changes. Importantly, the effect is robust across broad families of smooth single-lobe profiles, so it does not rely on fine tuning. Conversely, this sensitivity also suggests a nonperturbative pulse diagnostic: fringe onset and evolution can fingerprint the effective field experienced by the carriers, complementing conventional waveform characterization. Epitaxial graphene naturally corresponds to far-field pump–probe measurements on an extended sheet, where the relevant interaction length L is set by the smaller of the illuminated spot size and the device dimension along the field direction. A kinematic estimate for ballistic escape is t_{\text{esc}}\sim L/v_{\text{F}} dora_prb_2010 . For typical far-field spots L\sim 10 – 100\penalty\ $\text{\,}\mathrm{\SIUnitSymbolMicro m}$ one finds t_{\text{esc}}\sim 10 – 100\penalty\ \text{ps} , so for the few-cycle mid-IR/near-IR driving emphasized here ( \tau\sim 10 – 100\penalty\ \text{fs} ) the creation stage satisfies \tau\ll t_{\text{esc}} and transport out of the driven region is negligible. Relaxation and dephasing primarily add broadening that reduces fringe contrast, while leaving the dominance-transition onset scale unchanged. Acknowledgements. We are grateful to Prof. I. S. Terekhov for valuable discussions. The study was funded by the Foundation for the Advancement of Theoretical Physics and Mathematics BASIS (Project No. 25-1-3-48-1). The work of M. A. Dorodnyi was performed at the Saint Petersburg Leonhard Euler International Mathematical Institute and supported by the Ministry of Science and Higher Education of the Russian Federation (agreement No. 075–15–2025–343). Spectral fringes without subcycles in Schwinger pair production and Dirac materials I. A. Aleksandrov, M. A. Dorodnyi, and E. D. Akimkina Supplemental Material This Supplemental Material is organized as follows. Appendix A summarizes the semiclassical turning-point framework (including the Stokes/prefactor conventions) used to interpret the spectra in scalar and spinor QED. Appendix B contains several additional examples of the particle momentum distributions in the case of deformed pulses e_{a}(z)=\mathrm{exp}(-z^{2}-az^{4}) . Appendix C derives the quantum kinetic equations for the gapped Dirac model and records the coefficient functions used in our numerics. A. WKB APPROXIMATION IN SCALAR AND SPINOR QED Here we briefly recap the main WKB expressions which allow one to approximately evaluate the momentum spectra n_{\mathbf{p}} . In the case of a linearly polarized field E(t)=-\dot{A}(t) , we define the adiabatic energy Q_{\mathbf{p}}(t)=\sqrt{\pi_{\perp}^{2}+[p_{\parallel}-eA(t)]^{2}}, (S1) where \pi_{\perp}=\sqrt{m^{2}+p_{\perp}^{2}} . Assuming that the vector potential is a regular function of a complex variable t , we identify the upper-half-plane turning points (TPs) for given \mathbf{p} : Q_{\mathbf{p}}(t_{i})=0. (S2) The complex conjugates t_{i}^{*} are also TPs. We then introduce \rho_{i}(t)=\int\limits_{t_{i}}^{t}Q_{\mathbf{p}}(t^{\prime})dt^{\prime}. (S3) It is clear that \rho_{i}(t_{i})=0 . The Stokes lines are defined as the following sets: \mathcal{S}_{i}=\big\{t\in\mathbb{C}:\mathrm{Re}\,\rho_{i}(t)=0\big\}. (S4) It turns out that t_{i} and t_{i}^{*} are connected by the corresponding Stokes line \mathcal{S}_{i} taya_jhep_2021 . The intersection of \mathcal{S}_{i} and the real axis is denoted by s_{i} . According to Refs. dumlu_dunne_prd_2011_1 ; taya_jhep_2021 , in scalar QED the momentum distributions can be obtained by the following approximate recipe: n_{\mathbf{p}}\approx\sum_{i}\mathrm{e}^{-2\sigma_{\mathbf{p}}^{(i)}}+2\sum_{i0), (S13) we provide here additional numerical examples of the momentum distributions in scalar QED. We compare two pulses with a=0.5 and a=2.0 , respectively. The critical values of the Keldysh parameter ( 5 ) amount to \gamma_{*}(0.5)\approx 2.08 and \gamma_{*}(2.0)\approx 1.01 . The example spectra are displayed in Fig. S1 . As anticipated from the \gamma_{*}(a) indicator, the onset of pronounced fringes follows the location of the turning-point dominance transition. For a=0.5 (panel a), the spectrum at \gamma\simeq 2 already lies close to \gamma_{*}(0.5)\approx 2.08 and exhibits clear oscillatory modulation, while moving deeper into the adiabatic regime ( \gamma=1,0.5 ) suppresses the fringes toward a smooth single-peaked shape. For the more strongly deformed profile a=2.0 (panel b), the critical value is much smaller, \gamma_{*}(2.0)\approx 1.01 , and correspondingly the qualitative change occurs already around \gamma\simeq 1 : spectra above and below this threshold display markedly different fringe visibility. These additional examples therefore corroborate the \gamma_{*}(a) -based diagnostic and demonstrate that the semiclassical mechanism is robust across the deformation family ( S13 ), rather than being specific to the single case a=1 . Figure S1: Momentum distributions of the produced particles as a function of longitudinal momentum p_{\parallel} for p_{\perp}=0 in scalar QED for a deformed pulse ( S13 ) with (a) a=0.5 and (b) a=2.0 . The numerical results are obtained for a peak field E_{0}=0.1E_{\text{c}} , where E_{\text{c}}=m^{2}/|e| is the critical field strength. Pulse durations are \tau=5m^{-1} (solid), \tau=10m^{-1} (dotted), and \tau=20m^{-1} (dashed), leading to \gamma=2.0 , 1.0 , 0.5 , respectively. For clarity, the plotted number densities are rescaled by the multiplicative factors indicated in the panels. C. DERIVATION OF THE QKE SYSTEM IN GAPPED GRAPHENE Here we present a derivation of the QKE system in a massive ( 2+1 ) Dirac model. The one-particle Hamiltonian in the presence of a time-dependent electric field is \mathcal{H}(t)=v_{\text{F}}\,\boldsymbol{\sigma}\cdot[-i\hbar\boldsymbol{\nabla}-e\mathbf{A}(t)]+\Delta\sigma_{z},\qquad\mathbf{E}(t)=-\dot{\mathbf{A}}(t), (S14) where e<0 is the electron charge, v_{\text{F}} is the Fermi velocity, and \Delta the Dirac mass term (half the band gap), \Delta=E_{\text{g}}/2 . We assume that the electric field vanishes unless t_{\text{in}}0). (S56) Figure S3 compares longitudinal momentum spectra in gapped graphene for several values of \eta . Panel (a) confirms that in the smooth regime at relatively large \tau (small \gamma_{\text{D}} ) the spectra remain single-peaked even after enforcing zero net area of the external field. Panel (b) shows that the pronounced fringe pattern for a shorter main lobe persists for all tested zero-area pulses and becomes essentially \eta -independent for \eta\lesssim 0.02 (i.e. T\gtrsim 50\tau ), where the spectra nearly coincide pointwise with the \eta=0 results of the main text. The range \eta\sim 0.02 – 0.1 used here is representative of experimentally generated quasi-half-cycle waveforms, in which a dominant lobe is followed by a longer, weaker compensating tail schmuttenmaer_2004 ; isgandarov_2021 . Figure S3: Longitudinal momentum distributions of the carriers in gapped graphene for k_{\perp}=0 , \Delta=0.1\penalty\ \text{eV} and E_{0}=40\penalty\ \text{kV}/\text{cm} , driven by explicitly zero-area pulses ( S55 ) built from the deformed envelope e(z)=\mathrm{exp}(-z^{2}-z^{4}) . Curves correspond to \eta=0 (solid), \eta=0.02 (dashed), and \eta=0.1 (dotted). The main-lobe duration is (left) \tau=50\penalty\ \text{fs} and (right) \tau=30\penalty\ \text{fs} . For visibility, the data in panel (a) are multiplied by a factor of 2 .