Analytical and Numerical Studies of Dark Current in Radiofrequency Structures for Short-Pulse High-Gradient Acceleration Gaurab Rijal, Michael Shapiro, Xueying Lu Abstract High-gradient acceleration is a key research area that could enable compact linear accelerators for future colliders, light sources, and other applications. In the pursuit of high-gradient operation, RF breakdown limits the attainable accelerating gradient in normal-conducting RF structures. Recent experiments at the Argonne Wakefield Accelerator suggest a promising approach: using short RF pulses with durations of a few nanoseconds. Experimental studies show that these đ’Ș ⁥ ( 1 ​ ns ) \mathcal{O}\!\bigl(1\,\text{ns}\bigr) RF pulses can mitigate breakdown limitations, resulting in higher gradients. For example, an electric field of nearly 400 MV/m was achieved in an X X -band photoemission gun driven by 6-ns-long RF pulses, with rapid RF conditioning and low dark current observed. Despite these promising results, the short-pulse regime remains an under-explored parameter space, and RF breakdown physics under nanosecond-long pulses requires further investigation. In this paper, we present analytical and numerical simulations of dark current dynamics in accelerating cavities operating in the short-pulse regime. We study breakdown-associated processes spanning different time scales, including field emission, multipacting, and plasma formation, using simulations of the X X -band photogun cavities. The results reveal the advantages of using short RF pulses to reduce dark current and mitigate RF breakdown, offering a path toward a new class of compact accelerators with enhanced performance and reduced susceptibility to breakdown. I Background and Introduction Achieving a high accelerating gradient is crucial for compact linear accelerators, but the presence of radiofrequency (RF) breakdown poses a significant challenge. Recently, acceleration with RF pulses on the order of a few nanoseconds has emerged as a promising approach for overcoming long-standing limitations in the gradient achievable by mitigating the impact of RF breakdown Döbert et al. 2004 ; Döbert et al. 2007 ; Yu et al. 2001 ; Wuensch et al. 2003 ; Tan et al. 2022 ; Shao et al. 2022 ; Merenich et al. 2024 ; Freemire et al. 2023 . These short RF pulses have been generated by various approaches including RF pulse compressors Ivanov et al. 2013 ; Samsonov et al. 2004 ; Gu et al. 2025 , and power extractors Syratchev et al. 2008 ; Cappelletti et al. 2011 ; Picard et al. 2022 ; Peng et al. 2019 ; Shao et al. 2020 based on the structure wakefield acceleration (SWFA) concept Lu et al. 2022 . SWFA is one promising advanced accelerator concept, where a high-charge drive beam excites wakefield in metallic or dielectric structures in vacuum, and then the excited wakefield can be used to accelerate a main beam at high gradients. There are two schemes of SWFA: collinear wakefield acceleration, where the drive and main beams travel in the same structure, whereas in two-beam acceleration (TBA) the beamlines are decoupled. In the TBA scheme, a power extractor is designed to extract the wakefield generated by the drive beam, often a train of high-charge electron bunches. Each electron bunch in the train excites a short RF pulse as it traverses the power extractor, and when the wakefield pulses generated from multiple electron bunches are coherently added, a combined RF pulse with high peak power can be produced. Using this approach, RF pulses with a duration of a few nanoseconds and a peak power of over 500 MW have been generated at X -band Picard et al. 2022 ; Peng et al. 2019 . These short RF pulses can then be transferred and coupled into accelerating structures, where gradients exceeding the limit in conventional accelerators (100 MV/m for X -band for example) have been demonstrated in a series of experiments Jing et al. 2022 . One of these experiments at the Argonne Wakefield Accelerator (AWA) facility successfully demonstrated an electric field approaching 400 MV/m on the photocathode surface in an X -band photogun with (1+1/2)-cell resonant cavities Tan et al. 2022 . These cavities are powered by approximately 9 ns long RF pulses extracted from an eight-electron-bunch train with a total charge of about 400 nC using an X -band power extractor Peng et al. 2019 . Figure 1 (a) shows the photogun cavities, comprising a full cell and a half cell, and an input waveguide port. A coaxial coupler was designed to maintain good field symmetry at the photocathode. A directional coupler (not shown) was attached to the waveguide port to measure the input and reflected RF signals. The cavities operate in the TM 010,π mode with a center frequency of 11.7 GHz, with the electric field distribution shown in Fig. 1 (b). The input RF signal exhibits a time profile of approximately 3 ns rise, 3 ns flattop and 3 ns fall, as shown in Fig. 1 (c). This time profile is determined by the design of the power extractor and the drive bunch train length. The electric field at the center of the photocathode for an input pulse with a peak power of 200 MW is also shown in Fig. 1 (c). Figure 1: Model and RF design of the X -band photogun cavities driven by short RF pulses. (a) Cross-sectional view of the copper cavities along the central plane. (b) Normalized electric field distribution at 11.7 GHz. (c) Electric field on the photocathode surface for an input RF pulse with 200 MW peak power; the inset shows the time profile of the input RF signal, with a 3 ns rise, 3 ns flattop and 3 ns fall. The encouraging high gradient demonstrated experimentally points to the need for a deeper physics understanding of RF breakdown processes in the under-explored parameter space of \mathcal{O}\!\bigl(1\,\text{ns}\bigr) pulses. An empirical scaling law Grudiev et al. 2009 , established from breakdown studies in klystron-powered accelerating structures with \mathcal{O}\!\bigl(100\,\text{ns}\bigr) RF pulse durations, relates the breakdown rate (BDR) to the accelerating gradient E_{a} and the pulse length t_{p} as E_{a}^{30}\times t_{p}^{5}/\text{BDR}=\text{constant} , suggesting that higher gradients can be achieved with shorter pulses. However, recent experiments using \mathcal{O}\!\bigl(1\,\text{ns}\bigr) long pulses revealed a different scaling behavior Merenich et al. 2024 ; Shao et al. 2022 and interesting dark current behavior Tan et al. 2022 . During the RF conditioning of the short-pulse X -band photogun, very low dark current was observed. Despite monitoring with both a Faraday cup and an integrating current transformer (ICT), no clear signal was detected, suggesting an upper limit of approximately 1 pC in dark charge. Dark-current-induced beam loading was also observed in this experiment, as shown in Fig. 2 (a). The ratio R of the measured and simulated peak reflected RF signal was close to unity in the absence of RF breakdown or dark current [Figs. 2 (b) and 2 (d)], but dropped below unity during a stage where beam loading occurred [Fig. 2 (c)]. These observations call for further investigation of the dark current dynamics in the short-pulse regime. Dark current refers to the unwanted flow of electrons in the absence of an injected beam Chao et al. 2013 , which can coexist with the primary beam and cause instabilities, power losses, and structural damage due to electron bombardment Wilson 2001 . The growth of dark current may also contribute to RF breakdown, which typically manifests as vacuum arcs that can disrupt RF power flow and may even cause irreversible damage to accelerating structures. RF breakdown is strongly influenced by the localized electromagnetic field on the inner surfaces of the structure, as well as by interactions between these fields and dark electrons Simakov et al. 2018 . A range of physical processes across various timescales are associated with dark current dynamics. Figure 2: RF conditioning of the X -band photogun cavities. (a) Conditioning history of the peak electric field E_{0} at the photocathode surface (blue), and the ratio of the measured to simulated peak reflected RF signal R (orange). (b-d) Reflected RF signals at the corresponding stages denoted by the vertical lines in Panel (a). This figure is reprinted with permission from Tan et al., Phys. Rev. Accel. Beams 25, 083402 (2022) under the terms of the Creative Commons Attribution 4.0 International license. A common source of dark electrons is field emission from surface regions with high local fields Jensen et al. 2008 ; Jimenez et al. 1994 . The local field E_{\text{local}} is often significantly higher than the applied macroscopic field E_{\text{applied}} due to surface irregularities, and can be expressed as E_{\text{local}}=\beta E_{\text{applied}} , where \beta is the field enhancement factor. Material properties such as work function, roughness, and composition can strongly influence field emission Fowler and Nordheim 1928 ; Alpert et al. 1964 ; Norem et al. 2005 ; Ahearn 1936 . Field emission commonly occurs in high-field regions, where pulsed heating – localized surface temperature rise via Joule heating – can vaporize emitting sites, triggering explosive electron emission and surface modification, which may in turn further enhance field emission Wang and Loew 1997 . Electron multipacting is another important mechanism contributing to dark current growth Vaughan 1988 ; Bienvenu et al. 1992 ; Larciprete et al. 2013 ; Valizadeh et al. 2014 ; Han et al. 2008 ; Mori et al. 2021 ; Power et al. 2004 ; Yater 2023 ; Xu et al. 2019 . In this process, emitted electrons from the surface interact with the RF field and may return to the surface again, releasing secondary electrons. Under certain resonant conditions, this process repeats, leading to exponential electron multiplication. Plasma can form as the electron density increases Neuber et al. 1999 ; Valfells et al. 2000 ; Sun et al. 2018 . This plasma can short-circuit RF structures, trigger further ionization, and escalate into arcing. We report analytical and numerical studies on the dynamics of dark current in the short-pulse acceleration regime, using the X -band cavities shown in Fig. 1 as a representative case. Section II presents an analytical theory of electron multipactor, where we identify the multipacting resonance conditions from electron trajectory calculations and evaluate the secondary electron yield (SEY). Section III describes simulations of field emission and multipacting processes using the cst Dassault SystĂšmes Particle Studio, analyzing the impact of pulse shape, pulse length and structure gradient on dark current growth. Section IV discusses the formation of multipacting electron clouds and their influence on RF pulse distortion. Finally, Sec. V summarizes the key findings and outlines future directions for high-gradient short-pulse RF acceleration research. II Theory of Electron Multipacting In this section, we present an analytical treatment of electron multipacting by introducing a model which incorporates spatial variations in the electric and magnetic fields of a modified pillbox mode to capture electron motion near the cavity sidewalls. With this analytical model, we can precisely identify multipacting resonance modes in cylindrical accelerating cavities. Previous analytical models have explored electron multipacting behavior in various RF configurations Kishek and Lau 1998 ; Sazontov et al. 2011 ; Iqbal et al. 2023 ; Shemelin 2013 ; Lyneis et al. 1977 ; Gonz’alez-Iglesias et al. 2024 . For example, Ref. Kishek and Lau 1998 presents a theory of single-sided multipacting on a dielectric surface with an RF electric field and a DC electric field perpendicular to the dielectric surface. The DC electric field is necessary to extract electrons from the dielectric. Ref. Sazontov et al. 2011 investigates the effect of the RF magnetic field on multipacting on dielectric surfaces. These results are important for evaluating multipacting behavior in dielectric windows. Multipacting on metallic surfaces of accelerator structures has also been studied. Ref. Iqbal et al. 2023 investigates two-sided multipacting between two metal electrodes. Ref. Shemelin 2013 analyzes the single-sided multipacting on cavity sidewalls under the influence of normal electric fields. The study identifies electron trajectories that return to the emission surface and contribute to sustaining the discharge. In this work, we study electron multipacting in crossed RF fields near the sidewall of a pillbox cavity, focusing on closed electron trajectories that originate from and return to the sidewall. These trajectories are of particular interest, as they are resonant with the RF field and are expected to play a dominant role in the multipactor phenomenon. Our analysis includes an axial electric field E_{z} , a radial electric field E_{r} , and an azimuthal magnetic field B_{\theta} . The radial component E_{r} arises near the cavity sidewall due to the presence of the cavity iris. We modify the field distribution in the pillbox cavity to approximate conditions near the sidewall region of the X -band photogun cavities. We assume an axisymmetric field distribution, so electron motion is confined to the r – z plane. This assumption holds as long as any asymmetries or perturbations that could induce out-of-plane motion remain negligible. These approximations yield predictions in good agreement with simulations of the actual cavity, as presented in Sec. III . We begin by calculating secondary electron trajectories and then evaluate the SEY in both single-impact and multipacting resonance conditions. Figure 3: Schematic representation of the multipactor analysis setup. (a) Simulation geometry used for electron trajectory evaluation. (b) Crossed RF field configuration showing an electron trajectory under combined E_{r} , E_{z} , and B_{\theta} fields. The parameters v_{0} (initial velocity) and \alpha (emission angle) are defined at emission, while r and z denote the radial and longitudinal coordinates, respectively; a and L are the cavity radius and length, respectively. II.1 Secondary electron trajectory calculations Under the assumptions outlined above, we investigate electron multipacting near the sidewall of a pillbox cavity. The cavity has a radius of a=9.49 mm and a length of L=6.50 mm, corresponding to the full-cell dimensions of the X -band photocathode cavities. The configuration used for the electron trajectory calculations is illustrated in Fig. 3 . The origin, r=z=0 , is defined at the center of the pillbox cavity. An electron is emitted from the sidewall at t=0 with an emission angle \alpha_{0} , initial velocity v_{0} , and initial RF phase \phi_{0} . The initial axial position is z(t=0)=z_{0} , and the radial distance from the sidewall is \Delta r(t=0)=r(t=0)-a=0 . We assume a steady-state field, with the axial electric field amplitude at the cavity center denoted by E_{z0} . The equations of motion for the electron are: \displaystyle m\frac{dv_{z}}{dt} \displaystyle=-eE_{z}-ev_{r}B_{\theta}, (1) \displaystyle m\frac{dv_{r}}{dt} \displaystyle=-eE_{r}+ev_{z}B_{\theta}, (2) where v_{r}=dr/dt is the radial velocity, v_{z}=dz/dt is the axial velocity, and e and m are the charge and mass of the electron, respectively. Figure 4: Comparison of RF field amplitudes from cst simulations and analytical approximations for the full-cell cavity with radius a=9.49 mm and length L=6.50 mm. The field amplitudes— B_{\theta} in (a), E_{r} in (b), and E_{z} in (c)—are all scaled to an axial field amplitude of E_{z0}=100 MV/m at the cavity center. The shaded region in (c) indicates the radial range near the sidewall where the analytical field approximation is applied. Solid red curves represent cst simulation data, while black dashed lines indicate the analytical field amplitudes used in the model. As secondary electrons remain localized near the cavity sidewall, we approximate the local RF fields as linear in \Delta r=r-a , the radial displacement from the sidewall. The radial electric field E_{r} is given by: E_{r}(\Delta r,z)=E_{r_{0}}\left(1+S\cdot\Delta r\right)\cos\left(\frac{\pi z}{L}\right)\cos(\omega t+\phi_{0}), (3) where E_{r0} is the peak amplitude of E_{r} , \omega is the angular frequency of the RF field, and S=dE_{r}/dr , fitted from the simulated field distribution. The axial electric field E_{z} is: E_{z}(\Delta r,z)=E_{z0}P\Delta r\cos(\omega t+\phi_{0}), (4) where E_{z0} again denotes the E_{z} field amplitude at the cavity center, and P=(\mathrm{d}E_{z}/\mathrm{d}r)/E_{z0} . The azimuthal magnetic field B_{\theta} is modeled as: B_{\theta}(\Delta r,z)=\left[(B_{0}+S_{1}\cdot\Delta r)+(B_{1}+S_{2}\cdot\Delta r)\right.\\ \left.\times\sin\left(\frac{\pi z}{L}+\zeta\right)\right]\sin(\omega t+\phi_{0}), (5) where the coefficients B_{0} , B_{1} , and \zeta are obtained by fitting to the simulated B_{\theta} distribution. S_{1}=dB_{0}/dr and S_{2}=dB_{1}/dr represent the radial gradients of the respective components near the sidewall. A comparison between the cst -simulated field in the full cell of the X -band cavities and the analytical approximations in Eqs. 3 – 5 is shown in Fig. 4 . While we employ the above field distributions to calculate secondary electron trajectories for the remainder of this section, we have also developed a simplified theory neglecting the radial variation of E_{z} and B_{\theta} . This approximation yields closed-form expressions for multipacting resonance conditions, useful for rapid design estimates. The simplified theory, presented in Appendix A , agrees well with the full model in the low-field regime. For accurate trajectory tracking and SEY evaluation, however, we retain full radial dependence in all RF field components. II.2 Single-cycle SEY calculations The SEY, \delta , quantifies the number of secondary electrons emitted per incident electron. Here, we evaluate \delta after a single emission–return cycle, defined as the electron’s first re-impact with the cavity sidewall following its initial emission. The SEY depends on both the impact energy W_{i} and the angle of incidence \theta_{i} . We use Vaughan’s empirical model Vaughan 1993 to evaluate \delta : \delta=\delta_{m}\left[w\exp(1-w)\right]^{k},\quad\text{where}\quad w=\frac{W_{i}}{W_{m}}. (6) Here, \delta_{m} is the maximum yield, and W_{m} is the corresponding impact energy at which this maximum occurs. Both parameters increase with the angle of incidence \theta_{i} according to: \delta_{m}=\delta_{\text{max}}\left(1+\frac{\theta_{i}^{2}}{2\pi}\right),\quad W_{m}=W_{\text{max}}\left(1+\frac{\theta_{i}^{2}}{2\pi}\right), (7) where \delta_{\text{max}}=2.1 and W_{\text{max}}=150~\text{eV} are the values for normal incidence. The exponent k in Eq. 6 depends on the normalized energy w and is defined piecewise: k=\begin{cases}0.56&\text{if }w<1,\\ 0.25&\text{if }w>1.\end{cases} (8) Figure 5: Single-cycle SEY ( \delta ) map for the full cell of the X-band photogun cavities for different combinations of initial axial emission position z_{0} and RF field amplitudes, characterized by the axial electric field at the cavity center E_{z0} . The color scale indicates the magnitude of \delta . To compute the SEY, we begin by extracting the impact energy W_{i} and incidence angle \theta_{i} from the electron trajectory. The axial impact location is denoted by z_{i} , and the RF phase at impact is given by \phi_{i}=\omega\tau+\phi_{0} , where \tau is the time of flight. We assume an initial kinetic energy of 2 eV for secondary electrons, consistent with the emission model in Ref. Vaughan 1988 , which corresponds to the minimum energy needed to overcome the initial retarding field. We then perform a parameter sweep over the electron emission angle \alpha_{0} and initial RF phase \phi_{0} , across a range of axial emission positions z_{0} and field amplitudes characterized by the axial electric field amplitude at the cavity center, E_{z0} . All field components— E_{z} , E_{r} , and B_{\theta} —are scaled proportionally with E_{z0} , which serves as the reference parameter for the sweep. SEY values exceeding unity are retained and averaged to generate a representative single-cycle SEY contour map. The resulting map is shown in Fig. 5 , highlighting regions in E_{z0} – z_{0} space that are susceptible to enhanced secondary emission and the potential onset of electron multipacting. Figure 6: Sample multipacting resonance modes ( N=1\text{--}7 ) calculated for the RF field distribution in the full cell of the X -band photocathode cavities. The corresponding axial field amplitudes E_{z0} at the cavity center for each order are 106, 61, 39, 31, 26, 22, and 19 MV/m, respectively. II.3 SEY under multipacting resonance We now turn to the evaluation of SEY under multipacting resonance conditions. Multipacting resonance occurs when secondary electrons emitted from a surface are accelerated by an RF field such that their time of flight \tau is synchronized with the RF period T Shemelin 2013 ; Xu et al. 2019 . The resonance order N is given by N=\tau/T . When the SEY associated with a given resonance exceeds unity, the number of secondary electrons can increase with each RF cycle, potentially leading to an electron avalanche during a long RF pulse. To identify resonance conditions, we begin with a specified initial RF phase \phi_{0} and axial emission position z_{0} , and calculate the electron trajectory over multiple iterations. A convergence criterion is applied, requiring both the RF phase \phi at which the electron leaves the surface and the axial position z to stabilize within a tolerance of 1\times 10^{-6} over ten consecutive iterations. Figure 7: Multipacting susceptibility plots at resonance showing the SEY \delta at various E_{z0} with respect to (a) resonant axial emission position z_{\text{res}} , (b) resonant emission angle \alpha_{\text{res}} , and (c) resonant emission RF phase \phi_{\text{res}} . Figure 8: Simulated electron trajectories emitted from (a) the iris and (b) the photocathode surface, under a peak axial electric field amplitude of 325 MV/m at the center of the photocathode surface. Color-coded lines represent trajectories of electrons launched from different initial positions. Figure 6 shows trajectories of multipacting resonance modes in the \Delta r\text{--}z plane for various resonance orders N . SEY values are calculated for each mode across a range of initial conditions. The corresponding susceptibility map in Fig. 7 plots SEY \delta as a function of E_{z0} and three key parameters: (a) resonant axial position z_{\text{res}} , (b) emission angle \alpha_{\text{res}} , and (c) emission phase \phi_{\text{res}} . These plots reveal distinct bands of sustained multipacting. As E_{z0} increases, dominant resonance orders shift from higher ( N=7 ) to lower ( N=1 ), a typical trend in RF multipactor behavior. Identifying these bands provides insight into multipacting onset and informs suppression strategies. This analysis assumes steady-state RF fields. In short-pulse operation, transient fields limit the time available for multipacting to develop, thereby suppressing secondary electron growth. The next section presents time-domain simulations of dark current dynamics in the short-pulse regime. III Dark Current Simulations To investigate dark current behavior under short-pulse RF excitation, we perform numerical simulations using the cst Particle-In-Cell (PIC) and particle tracking solvers. This complements the steady-state analytical model by capturing transient effects in realistic cavity geometries. We begin by modeling field emission from high-gradient regions, such as the cavity iris and photocathode. These field-emitted electrons can act as seed particles for multipacting-driven growth and subsequent dark current buildup, as discussed later. III.1 Field Emission Simulations Field emission is a critical process in RF cavities, where strong local electric fields can extract electrons from metal surfaces. These electrons may disrupt the primary beam Bane et al. 2005 ; Zheng et al. 2015 , trigger RF breakdown Wang and Loew 1997 , or initiate dark current through multipacting. Extensive efforts have been made to model and characterize field emission across various RF structures Bienvenu et al. 1992 ; Wu et al. 2017 ; Norem et al. 2005 ; Huang et al. 2015 . Field emission simulations were performed using the cst PIC solver to investigate the low dark current observed during RF conditioning of the X -band photogun cavities. As discussed in Sec. I , no dark current signal was detected above the noise threshold Tan et al. 2022 . This can be attributed to two factors: (1) the short RF pulse limits the total emitted charge, and (2) emitted electrons must traverse a long, narrow beam pipe (the inner conductor of the coaxial coupler) to reach diagnostics Rijal et al. 2024 . This is illustrated in Fig. 8 : in (a), most electrons emitted from the iris strike nearby surfaces; in (b), only electrons emitted near the center of the photocathode and within a narrow angular range are able to exit the beam pipe. Simulations estimate that only about 10^{-5} of the total electrons emitted from the iris reach the end of the beam pipe. The electrons that strike cavity surfaces may initiate multipacting when resonance conditions are met. Figure 9: Multipacting simulation setup at the middle cutting plane of the X -band cavities, divided into the full cell, and Region I and Region II in the half cell. The reference points P_{1} , P_{2} and P_{3} denote the centers of each cavity, respectively. The relevant dimensions are: R 1 = 8.09 mm, R 2 = R 3 = 9.49 mm, Z 1 = 2.40 mm, Z 2 = 2.60 mm, and Z 3 = 6.50 mm. Figure 10: Multipacting resonance trajectories simulated in the full cell (defined in Fig. 9 ). The corresponding axial field amplitudes E_{z0} for resonance orders N=1\text{--}7 , evaluated at point P_{3} (the center of the full cell), are 106, 61, 39, 31, 26, 22, and 19 MV/m, respectively. These trajectories show good agreement with the analytical results in Fig. 6 , as the RF field distribution in the full cell is well approximated by the analytical model. Figure 11: Multipacting resonance trajectories in the half-cell for (a) Region I and (b) Region II. In Region I, the axial fields E_{z0} evaluated at point P_{1} for resonance orders N=1\text{--}4 are 112, 61, 39, and 28 MV/m, respectively. In Region II, the axial fields E_{z0} evaluated at point P_{2} for resonance orders N=1\text{--}5 are 95, 46, 30, 25, and 21 MV/m, respectively. III.2 Multipacting simulations in the short-pulse regime Multipacting behavior was simulated in cst using the realistic geometry of the X -band photocathode cavities driven by short RF pulses. Two types of simulations were performed, following the previous analytical studies: particle tracking simulations to identify multipacting resonance conditions by analyzing individual electron trajectories, and PIC simulations to evaluate dark current growth. In particle tracking simulations, electron trajectories were analyzed for emission from the sidewalls of the three regions as defined in Fig. 9 , the full cell, and Region I and Region II in the half cell. Sample trajectories satisfying multipacting resonance conditions are presented in Figs. 10 and 11 . The resonant trajectories and the associated electric field amplitudes in the full cell (Fig. 10 ) exhibit excellent agreement with the analytical results shown in Fig. 6 due to the similarity in the RF field pattern, providing further validation of the analytical model. Figure 12: Temporal evolution of the electric field amplitudes and the number of multipacting electrons for (a) Region I, (b) Region II, and (c) the full cell. In each subplot, the on-axis electric field at point P_{1} , P_{2} , or P_{3} (red solid line), the electric field across the photocathode surface (black dashed line), and the number of multipacting electrons (blue solid line) are shown. For Region I, the dark current peaks corresponding to N=1\text{--}4 appear at 113, 66, 43, and 30 MV/m; for Region II, the peaks for N=1,2,4,5 are at 100, 47, 28, and 20 MV/m; and for the full cell, the peaks for N=1,2,3,5 are at 113, 51, 35, and 27 MV/m. Higher-order resonance modes ( N\geq 6 ) also contribute to later peaks in the full cell. Figure 13: Number of multipacting electrons in Region I for RF pulses with a time profile of 3 ns rise, 3 ns flattop, and 3 ns fall at various peak power levels: 200 MW (red), 100 MW (blue), 50 MW (cyan), and 25 MW (black). Dashed curves represent the axial electric field amplitude at the center of Region I ( P_{1} ) while the corresponding solid curves in matching colors represent the number of multipacting electrons N_{\text{MP}} . In PIC simulations, we use again the short input RF pulse with a 3 ns rise, 3 ns flat top, and 3 ns fall and a peak power of 200 MW. Seed electrons were introduced following the method in Ref. Xu et al. 2019 . Secondary emission was modeled using the Vaughan formalism Vaughan 1993 , with a maximum yield \delta_{max}=2.1 at an impact energy W_{i}=150 eV. Figure 12 shows the temporal evolution of the electric field and the number of multipacting electrons in the three regions. Peaks in the electron population correspond to distinct multipacting resonance orders, and the gradient at resonance are in strong agreement with the multipactor resonance trajectory analysis, confirming that electron growth is driven by multipacting resonances. Figure 14: Temporal evolution of the electric field (red) and number of multipacting electrons (blue) using a longer RF pulse with a 30 ns rise. To further confirm that dark current growth arises from multipacting resonances, we varied the peak power of the input RF pulses while maintaining a fixed temporal profile. The resulting time evolution of the number of multipacting electrons is shown in Fig. 13 . For a given cavity geometry, the field gradients at which multipacting resonance occurs are fixed. Adjusting the input power changes the time at which these gradients are reached, thereby shifting the onset of the resonance modes. The dark current peaks consistently occur at the same axial field amplitudes across different power levels, matching the resonance orders identified earlier. Each mode persists for a duration that depends on the field ramp rate. At lower gradients, the field ramp rate is reduced, so the peak dark current is higher. These analyses show that short RF pulses can reduce dark current. Fig. 14 shows the simulation results with longer RF pulses as a comparison, where the gradual rise in field amplitude allows resonance conditions to persist longer, resulting in rapid growth of the multipacting electron population. In contrast, short RF pulses sweep through each resonance condition more quickly, limiting the duration over which the field remains within the amplitude window required for sustained multipacting. As a result, the buildup of secondary electrons is effectively suppressed. This comparison demonstrates that the short-pulse regime reduces dark current by limiting the time available for resonance-driven electron multiplication. IV Impact of dark electrons As dark electrons accumulate through field emission and multipacting, electron clouds can form. To observe this process, we performed cst PIC simulations with the cavity operated at the multipacting resonance conditions identified above. Seed electrons were briefly introduced near the cavity sidewall to initiate secondary emission. When the field satisfied resonance conditions, the electron population grew exponentially, forming a localized cloud near the sidewall. Figure 15: Normalized reflected RF power as a function of time for different plasma electron densities n_{e} , modeled as a 0.2 mm thick layer on the cavity sidewall. All cases use the same short RF input pulse. Higher plasma densities lead to further reductions in the reflected power. A critical consequence of multipacting-induced electron clouds is dark-current-induced beam loading. To model this effect, we approximate the electron cloud as a cold plasma layer with a thickness of 0.2 mm, located along the cavity sidewall. The plasma response is described using Drude dispersion, appropriate for an electron-only, collisionless medium. The effective permittivity of the layer is determined by an assumed plasma electron density n_{e} . Figure 15 shows the normalized reflected RF power for different plasma densities. As n_{e} increases, the reflected power decreases due to greater energy absorption by the plasma Wuensch 2002 . This behavior is consistent with the experimentally observed drop in reflected power during conditioning of the X-band photocathode cavities (see Fig. 2 ). V Conclusion and discussion Recent advances in short-pulse acceleration have opened new opportunities for achieving ultra-high gradients, but they also demand a reexamination of the underlying RF breakdown mechanisms on previously underexplored timescales. Traditional breakdown studies, grounded in long-pulse operation and steady-state field assumptions, fall short in capturing the transient physics unique to nanosecond-long RF pulses. In this work, we use analytical and numerical methods to show that RF pulses on the order of a few nanoseconds can effectively suppress dark current growth by disrupting the timing required for multipacting resonance. The analytical model identifies clear resonance bands, while time-domain simulations reveal how rapid field ramping shortens the window for sustained development of multipacting resonances. These results provide strong evidence that short-pulse operation alters the onset dynamics of dark current and RF breakdown. Our simulation results are also consistent with experimental observations of low dark current and beam loading in the X -band photogun cavities at AWA. To advance the understanding of RF breakdown in the short-pulse regime, future work should focus on modeling the dynamics of plasma formation and its interaction with time-varying RF fields, ideally using first-principles simulation tools. Conventional breakdown models often treat plasma evolution as a multi-stage process, with nanosecond-scale electron dynamics coupled to slower changes in surface morphology. However, in the short-pulse regime, where the electromagnetic fields vary rapidly in time, these transient interactions may dominate the breakdown initiation process. Investigating these fast-timescale effects will be essential for refining our understanding of the coupled physics in RF breakdown, and short-pulse experiments provide a promising platform for isolating and studying these mechanisms. Acknowledgements. The authors thank Dr. Haoran Xu and Dr. Omkar Ramachandran for valuable discussions. This research was supported by the U.S. Department of Energy, Office of Science, Office of High Energy Physics under Award DE-SC0021928 and DE-SC0022010. Appendix A Simplified theory of multipactor in crossed RF fields with closed-form solutions We present a simplified theoretical framework to describe the multipacting process near the cavity sidewall. This analysis focuses on the essential dynamics governing electron motion under RF electric and magnetic fields. The key distinction between this simplified theory and the analytical model in Sec. II is the neglect of the radial variation of E_{r} and B_{\theta} , which allows for closed-form solutions of the multipacting resonance conditions. The schematic setup for this analysis is the same as that shown in Fig. 3 . We assume that E_{r} and B_{\theta} are independent of r near the sidewall, so E_{r} is expressed as: E_{r}(z)=E_{r0}\cos\left(\frac{\pi z}{L}\right)\cos(\omega t+\phi_{0}), (9) where E_{r0} is the peak amplitude of E_{r} , \omega is the angular frequency of the RF field, and \phi_{0} is the RF phase. The azimuthal magnetic field B_{\theta} is approximated as: B_{\theta}(z)=\left[B_{0}+B_{1}\sin\left(\frac{\pi z}{L}+\zeta\right)\right]\sin(\omega t+\phi_{0}), (10) where the coefficients E_{r0} , B_{0} , B_{1} , and \zeta are determined by fitting to cst simulation results of the field distribution near the sidewall. The axial electric field E_{z}(r) retains its radial dependence and is written as: E_{z}(r)=E_{z0}P\Delta r\cos(\omega t+\phi_{0}) (11) Here, E_{z0} denotes the axial field amplitude at the cavity center, and \Delta r=r-a is the distance from the sidewall. The local radial derivative of the axial field is defined as P=\frac{dE_{z}/dr}{E_{z0}} , which, under Maxwell’s equations, can be approximated as P=\frac{i\omega B_{\theta}}{E_{z0}} . Under this approximation, the radial electric field component E_{r} can be neglected in the equation for the axial velocity v_{z} (see also Ref. Shemelin 2013 ). Using the field distribution given in Eqs. ( 9 )–( 11 ), we substitute them into the equations of motion: \frac{dv_{z}}{dt}=-\frac{e}{m}\frac{d}{dt}\left[B_{\theta}\,\Delta r\sin(\omega t+\phi_{0})\right]. (12) Integrating Eq. ( 12 ) using the initial condition v_{z}(t=0)=v_{0}\sin(\alpha_{0}) , we obtain v_{z}(t) , which is then substituted into the equation for v_{r} : \frac{dv_{r}}{dt}=-\frac{e}{m}E_{r}\cos(\omega t+\phi_{0})+\frac{e}{m}v_{0}\sin\alpha_{0}\cdot B_{\theta}\sin(\omega t+\phi_{0})\\ -\left(\frac{e}{m}\right)^{2}B_{\theta}^{2}\,\Delta r\frac{1-\cos(2\omega t+2\phi_{0})}{2}. (13) The oscillatory terms with frequency 2\omega are neglected, as their contribution is negligible in the low-field regime. Their influence becomes significant at higher fields, as demonstrated by the phase-space trajectories in Fig. 16 . In this example, the approximation remains valid for multipacting orders N\geq 2 . Figure 16: Phase-space trajectories for N=2 , E_{z0}=60\ \mathrm{MV/m} (left) and N=1 , E_{z0}=113\ \mathrm{MV/m} (right) cases with and without the 2\omega term in Eq. ( 13 ). The approximation shows good agreement for lower fields when N\geq 2 , but diverges for N=1 at high fields. By omitting the 2\omega terms, Eq.( 13 ) simplifies to the following second-order differential equation: \frac{d^{2}\Delta r}{dt^{2}}+\frac{1}{2}\left(\frac{e}{m}\right)^{2}B_{\theta}^{2}\Delta r=-\frac{e}{m}E_{r}\cos(\omega t+\phi_{0})\\ +\frac{e}{m}v_{0}\sin\alpha_{0}\cdot B_{\theta}\sin(\omega t+\phi_{0}). (14) Following the approach in Ref. Miller 1958 , this equation takes the form of a driven harmonic oscillator: \ddot{\Delta r}+\Omega^{2}\Delta r=f(\omega t), (15) where f(\omega t) represents the RF driving term, and the characteristic frequency \Omega is defined as: \Omega=\frac{1}{\sqrt{2}m}eB_{\theta}. (16) This solution holds under the condition \Omega\ll\omega , but begins to deviate from full numerical results when \Omega\gtrsim 0.25\,\omega . For the system studied here, the condition of \Omega=\omega occurs at E_{z0}=260.6~\mathrm{MV/m} , indicating the upper limit of validity for the approximation. The multipacting resonance condition is then given by: \Omega NT=\pi, (17) which can be rearranged as: \quad\frac{eB_{\theta}}{m}=\frac{\omega}{N\sqrt{2}}, (18) where T=(2\pi)/\omega is the RF period and N is again the multipacting order. While this simplified approach neglects the radial variation of the transverse fields E_{r} and B_{\theta} , and thus cannot predict the secondary electron yield (SEY) with the same accuracy as the full analytical model, it offers a closed-form expression that effectively approximates multipacting resonance conditions, especially in the lower-field regime. Multipacting Full Analytical Simplified Analytical Resonance E_{z0} min E_{z0} max E_{z0} min E_{z0} max Order ( N ) (MV/m) (MV/m) (MV/m) (MV/m) 1 91.1 120.0 116.0 149.0 2 54.0 79.0 58.0 74.0 3 38.0 49.5 39.0 49.0 4 29.4 36.1 29.0 37.0 5 24.0 28.2 24.0 29.0 6 20.3 23.0 20.0 24.0 7 18.0 19.8 17.0 21.0 Table 1: Comparison of E_{z0} ranges for various multipacting resonance orders N , calculated using the simplified and full analytical models. The simplified model captures the scaling trend well, particularly for higher-order resonances ( N\geq 2 ). Table 1 summarizes the predicted ranges of E_{z0} corresponding to various multipacting orders, as computed using both the simplified and full analytical models. These ranges are obtained by sweeping over the resonant parameters z_{\text{res}} , \alpha_{\text{res}} , and \phi_{\text{res}} . Notably, good agreement is observed for higher-order multipacting modes ( N\geq 2 ), indicating that the simplified model effectively captures the key scaling behavior.