betaFeedback

WorksPhysics and Astronomy

Resonant Excitation of Surface Plasmon for Wakefield Acceleration by Beating GW Lasers on Smooth Cylindrical Surface

Abstract

We present a theoretical and numerical study of resonant surface-plasmon (SP) excitation driven by the beating of two co-propagating laser pulses on a smooth cylindrical plasma-vacuum interface. Analytical expressions for the SP dispersion relation, field amplitude, geometric coupling factor, and resonance conditions are derived and validated by fully three-dimensional particle-in-cell simulations. We reveal that curvature-induced geometric effects can substantially modify the SP dispersion and enable resonant matching by laser beat waves. This is inaccessible in planar geometries or with a single laser. Under matched resonance conditions, a high-amplitude SP-based wakefield can be generated by a few gigawatt lasers, placing this mechanism within reach of state-of-the-art fibre lasers. It therefore opens a route toward portable laser-driven plasma wakefield accelerators.

Surface plasmons (SPs) are collective electron oscillations localised at the interface between a conductor and an insulator [1]. In conventional configurations, SP excitation requires momentum-matching techniques to bridge the mismatch between photon and plasmon wavevectors [3, 4] [2], which becomes increasingly restrictive for high-intensity drivers and limits operational flexibility.

Recent progress in high-power laser-driven grate-free surface plasmons (SPs) on a smooth surface has opened a new route toward ultra-high gradient, ultra-compact plasma-based particle acceleration and radiation generation [5, 6, 7, 8, 9]. In this approach, SPs are excited when the translational symmetry is broken during a laser pulse scattering with a smooth surface of a finite edge [10]. The efficiency depends primarily on the mode overlap between the incident laser field and the SP eigenfield. As surface geometry strongly shapes the SP eigenmodes, it enables the precise mode selection and controlled manipulation. In particular, cylindrical surfaces, such as the inner wall of a microtube, have been shown to offer strong potential for SP-based wakefield acceleration and electron-beam modulation [8]. In the linear regime driven by a weak laser pulse, electrons are confined on the surface, and the SP field can leak into the vacuum channel to form the leaky wakefield, capable of accelerating both positively and negatively charged particles [9]. In the highly nonlinear regime, SPs can in principle sustain bubble wakefields approaching hundreds of TV/m inside nanostructured vertically aligned carbon nanotubes (VCNTs) [11]. VCNT targets not only offer favourable thermal and electrical properties for efficient SP excitation, but also provide substantial flexibility in achievable plasma density, spanning 102010^{20}–1024 cm−310^{24}\,\mathrm{cm}^{-3} [12].

However, these concepts typically rely on high-power lasers, ranging from hundreds of terawatts (TW) to petawatts (PW), which remain costly and bulky for many laboratories, and are poorly suited to applications where flexibility and portability are important. Moreover, target survivability under strong electromagnetic (EM) fields of high-power lasers often constrains operation in an effectively single-shot mode, severely limiting the repetition rate. Low-power lasers can be more compact, economical and comparatively target-friendly, such as fibre lasers, which can deliver up to 5 gigawatts (GW) power on target [13]. This power is generally insufficient to directly drive a wakefield of amplitude high enough for electron self-injection and acceleration inside the gaseous bulk plasma [14]. Beating two laser pulses can resonantly enhance the wakefield, which has been explored previously in underdense uniform plasmas [15] and on a magnetised planar surface for THz generation [16]. However, a viable approach is lacking to use ultralow-power lasers to drive strong SP for relativistic particle acceleration in ultra-compact form.

In this Letter, we demonstrate that SP-based wakefields with amplitudes up to a few TV/m can be resonantly generated by the beating of two GW-level laser pulses on a smooth cylindrical microtube surface, enabling electron self-trapping and acceleration. We develop a self-consistent theoretical framework that yields analytical expressions for the SP dispersion relation, field amplitude, geometric coupling factor, and resonance conditions, which are quantitatively validated by fully 3D particle-in-cell (PIC) simulations. We show that curvature-induced geometric effects substantially modify the SP dispersion, thereby enabling resonant phase matching with laser beat waves, which is inaccessible in planar geometries. Under matched resonance conditions, low-power lasers can efficiently drive high-amplitude SP-based wakefields. Our results further indicate that laser powers of only a few GW are sufficient to generate SP fields at the tens of GV/m level, identifying fibre lasers as practical drivers. This work thus opens a pathway toward microscale electron accelerators for portable applications, such as endoscopic sources [17].

Beat-wave-driven SP excitation on a smooth cylindrical surface can be demonstrated by a fully 3D PIC simulation done by WarpX [18] as shown in Figure 1. Two laser pulses of wavelengths λ1=0.8\lambda_1 = 0.8 µm and λ2=0.6\lambda_2 = 0.6 µm copropagate inside a microtube of a vacuum channel radius a=1.5a = 1.5 µm. The normalised laser strength a0,1=a0,2=0.8a_{0,1} = a_{0,2} = 0.8 and the root-mean-square (RMS) pulse durations are the same, τL=50\tau_L = 50 fs. These parameters give the laser peak power of 200 GW and 320 GW, respectively. The carbon atom density is 1×10201 \times10^{20} cm−3^{-3}, which can be structured by VCNT forest with a tube density of 1×10111 \times10^{11} tube/cm2^2 giving the free electron density about 4×10204 \times10^{20} cm−3^{-3}, as such laser pulses can fully ionise the σ\sigma-bond of the carbon atom. The ionisation energies are modified with the CNT properties in simulations. The RMS laser spot size is w0=3.0w_0 = 3.0 µm, which gives 60% of on-wall intensity. The simulation window is 8 µm × 8 µm × 40 µm with 192 × 192 × 768 cells in xx, yy and zz directions, respectively. There are 2×2×22 \times2 \times2 microparticles in each cell. Two vacuum sections of 20 µm are placed longitudinally at the head and tail of the target for laser initialisation and electron beam diagnostics, respectively. See more details in the Supplementary Materials.

Image not extracted

Figure 1. PIC results after two laser pulses co-propagating 40 µm inside a solid microtube: (a) electron density (grey) and beating laser field EyE_y (blue-red) in xzxz plane. (b) Acceleration field EzE_z in xzxz plane. (c) Energy spectrum of the electron beam accelerated. (d) radial phase space (r,ur)(r,u_r) of the accelerated electron beam. The vacuum and plasma regimes are on the left and right, respectively. (e) and (f) show the transverse (indicated by green dashed line in (b)) and longitudinal line plots of on-axis (x=0x = 0 µm) EzE_z. Black solid lines are from PIC, and orange dashed lines are from analytical calculation with ϱ=1.05\varrho= 1.05. (g) Energy spectrum of the electron beam accelerated inside the vacuum channel.

The laser pulses can stably beat inside the channel and drive the SP as shown in Figure 1 (a) and (b), where the structure of the wakefield is capable of accelerating both positively and negatively charged particles. The longitudinal wakefield EzE_z of amplitude 0.3 TV/m can be resonantly generated inside the vacuum channel, which accelerates totally 0.2 nC electrons to the energy of up to 10 MeV over a 40 µm interaction length, as shown in Figure 1(c). This indicates an averaged peak acceleration gradient of Gpeak≃0.25G_{\mathrm{peak}} \simeq0.25 TeV/m. While these electrons are confined and accelerated on the surface, the majority of them can propagate longitudinally from the channel due to the strong EzE_z field. There is also a small fraction of electrons, e.g. about 20 fC from this simulation, which can escape separately from the surface to be trapped during the laser scattering with the vertical edge of the microtube [9] and accelerated by the wakefield inside the vacuum channel to produce a narrow energy spectrum as shown in Figure 1(g). The energy spectrum is peaked at 0.8 MeV, which indicates the mean acceleration gradient of Gmean≃20G_{\mathrm{mean}} \simeq20 GeV/m. This is a general picture of the beating wave drive SP and its wakefield acceleration.

SP-based wakefield excitation can be understood as a stimulated Raman scattering (SRS) process [19]. At the same time, the beat wave-driven SP is not identical to conventional bulk SRS. This interaction can be self-consistently described using Maxwell’s equations coupled to the cold-fluid model for the plasma electrons, formulated in cylindrical coordinates (r,φ,z)(r,\varphi,z). Inside the vacuum channel, 0≤r<a0 \le r < a, the permittivity is ϵv=1\epsilon_v = 1. In the plasma wall, r≥ar \ge a, ϵp(ω)=1−ωp2/[γ0ω(ω+iν)]\epsilon_p(\omega) = 1 - \omega_p^2/[\gamma_0\omega(\omega+ i\nu)] where ωp2=4πn0e2/me\omega_p^2 = 4\pi n_0 e^2/m_e. n0n_0 is the equilibrium density of cold electron plasma. γ0\gamma_0 is the effective relativistic factor of the electron fluid from the quiver or drift. ν\nu is an effective collision or damping rate. Here, we seek axisymmetric modes with a predominantly longitudinal accelerating field. The possible mode is the fundamental TM mode, TM0\mathrm{TM}_0, which can be driven by ponderomotive [9] or space charge force [11]. The electromagnetic (EM) fields can be written as E=(Er,0,Ez)\mathbf{E} = (E_r,0,E_z) and B=(0,Bφ,0)\mathbf{B} = (0,B_\varphi,0).

For the TM0\mathrm{TM}_0 mode with ei(kz−ωt)e^{i(kz-\omega t)} dependence, the eigen wave equation for EzE_z is given as [8]

(∂2∂r2+1r∂∂r)Ez+(ϵjω2c2−k2)Ez=0,(1)\left(\frac{\partial^2}{\partial r^2}+\frac{1}{r}\frac{\partial}{\partial r}\right)E_z+\left(\epsilon_j\frac{\omega^2}{c^2}-k^2\right)E_z=0, \tag*{(1)}

where jj denotes vv or pp, representing the equation in vacuum or plasma. kk and ω\omega represent the propagation constant and oscillation frequency of the mode, respectively. Therefore, we can define the radial evanescent constants as κv=k2−ω2/c2\kappa_v=\sqrt{k^2-\omega^2/c^2} and κp=k2−ϵp(ω)ω2/c2\kappa_p=\sqrt{k^2-\epsilon_p(\omega)\omega^2/c^2}. The solution is given inside the vacuum as Ez(v)(r,z,t)=AI0(κvr)ei(kz−ωt)+c.c.E_z^{(v)}(r,z,t)=AI_0(\kappa_v r)e^{i(kz-\omega t)}+\mathrm{c.c.} and inside the plasma as Ez(p)(r,z,t)=CK0(κpr)ei(kz−ωt)+c.c.E_z^{(p)}(r,z,t)=CK_0(\kappa_p r)e^{i(kz-\omega t)}+\mathrm{c.c.}, where I0I_0 and K0K_0 are modified Bessel functions of first and second kinds, respectively. AA is a constant and C=AI0(κva)/K0(κpa)C=AI_0(\kappa_v a)/K_0(\kappa_p a). The radial electric field Er(j)E_r^{(j)} and azimuthal magnetic field Bϕ(j)B_\phi^{(j)} are given as Er(j)=(ik/κj2)dEz(j)/drE_r^{(j)}=(ik/\kappa_j^2)dE_z^{(j)}/dr and Bϕ(j)=(iωϵjk/κj2)dEz(j)/drB_\phi^{(j)}=(i\omega\epsilon_j k/\kappa_j^2)dE_z^{(j)}/dr.

The dispersion function is defined by considering the boundary conditions at the interface r=ar=a where tangential EzE_z and BϕB_\phi are continuous as Ez(v)(a)=Ez(p)(a)E_z^{(v)}(a)=E_z^{(p)}(a) and Bϕ(v)(a)=Bϕ(p)(a)B_\phi^{(v)}(a)=B_\phi^{(p)}(a), given by

D(ω,k)=ϵvκvI1(κva)I0(κva)+ϵpκpK1(κpa)K0(κpa)(2)\mathcal{D}(\omega,k)=\frac{\epsilon_v}{\kappa_v}\frac{I_1(\kappa_v a)}{I_0(\kappa_v a)}+\frac{\epsilon_p}{\kappa_p}\frac{K_1(\kappa_p a)}{K_0(\kappa_p a)} \tag*{(2)}

where the eigenmode satisfies D(ω,k)=0\mathcal{D}(\omega,k)=0. The phase velocity is vph=ω/kv_{\mathrm{ph}}=\omega/k and group velocity vg=−(∂D/∂k)/(∂D/∂ω)v_g=-(\partial\mathcal{D}/\partial k)/(\partial\mathcal{D}/\partial\omega). In the large-radius limit ka≫1ka\gg1 or more precisely κj−1≪a\kappa_j^{-1}\ll a, the dispersion in Eq. (2) recovers to that on a planar surface with a curvature correction in order of O(1/ka)O(1/ka). The phase velocity is shown in Figure 2, where this curvature correction leads to the divergence for different plasma densities. In contrast, all vph/cv_{\mathrm{ph}}/c collapse onto the same curve on planar surfaces. This is the fundamental reason for the advantage of curved surfaces, as discussed later. Generally, in the vicinity of the SP resonance, ω/ωp→1/2\omega/\omega_p\to1/\sqrt{2}, the wavenumber diverges k→∞k\to\infty, while both vphv_{\mathrm{ph}} and vgv_g vanish as vph→0v_{\mathrm{ph}}\to0 and vg→0v_g\to0, corresponding to a standing wave strongly localized at the plasma–vacuum interface. In contrast, for ω/ωp≪1\omega/\omega_p\ll1, the dispersion approaches the light line, with vph≈vg≈cv_{\mathrm{ph}}\approx v_g\approx c, which indicates a weakly bound SP. Notably, on a cylindrical interface, the phase velocity remains significantly higher near the resonant line than in the planar case due to curvature effects. This enhancement is particularly pronounced for lower plasma densities and therefore allows the resonant SP excitation at a near-critical-density plasma for relativistic particle acceleration.

Image not extracted

Figure 2. Phase velocity vphv_{\mathrm{ph}} of SP on planar and cylindrical surfaces for different plasma density, calculated with a=1.0 μma=1.0\ \mu\mathrm{m} by Eq. (2).

A physical SP field depends on the specific drive for the eigen wave equation in Eq. (1). Let’s consider two co-propagating linearly polarised laser pulses of sufficiently long duration as aL(r,z,t)=e⊥{a0,1f1(r)ei(k1z−ω1t)+a0,2f2(r,t)ei(k2z−ω2t)}\mathbf{a}_L(r,z,t)=\mathbf{e}_\perp\{a_{0,1}f_1(r)e^{i(k_1z-\omega_1t)}+a_{0,2}f_2(r,t)e^{i(k_2z-\omega_2t)}\}, where fl(r)f_l(r) is the laser transverse profile and a0,la_{0,l} is the amplitude of normalised vector potential with ll denoting either 1 or 2 for the first and second laser pulses, respectively. Here, we consider two laser pulses of the same profiles as fl(r)=f(r)f_l(r)=f(r), which does not introduce any new physics. The frequencies and wavenumbers satisfy ωl≃ckl\omega_l\simeq ck_l in vacuum.

They create a slowly varying beating intensity of the time-averaged squared amplitude ⟨aL2(r,z,t)⟩=a0,12f2(r)+a0,22f2(r)+2∣a0,1∣∣a0,2∣f2(r)cos⁡(Δkz−Δωt)\langle a_L^2(r,z,t)\rangle=a_{0,1}^2f^2(r)+a_{0,2}^2f^2(r)+2|a_{0,1}||a_{0,2}|f^2(r)\cos(\Delta kz-\Delta\omega t) with Δk=k1−k2\Delta k=k_1-k_2 and Δω=ω1−ω2\Delta\omega=\omega_1-\omega_2. The cycle-averaged ponderomotive potential for electrons near the surface is given by Up(z,t)≃mec2⟨aL2(r,z,t)⟩/4U_p(z,t)\simeq m_ec^2\langle a_L^2(r,z,t)\rangle/4. The beat ponderomotive force on electrons is Fpbeat=−∇Up=Fp,rer+Fp,zez\mathbf{F}_p^{\mathrm{beat}}=-\nabla U_p=F_{p,r}\mathbf{e}_r+F_{p,z}\mathbf{e}_z, where the longitudinal component is given by Fp,z(z,t)=[mec2a0,1a0,2Δksin⁡(Δkz−Δωt)]/2F_{p,z}(z,t)=[m_ec^2a_{0,1}a_{0,2}\Delta k\sin(\Delta kz-\Delta\omega t)]/2. For a solid surface with an optical laser beam where ωpω0/c≫1\omega_p\omega_0/c\gg1, the longitudinal beat force has much better overlap with the SP eigenmode than the transverse one. Therefore, Fp,zF_{p,z} dominantly modulates the surface electron density and current at the beat frequency and wavenumber (Δω,Δk)(\Delta\omega,\Delta k). The radial force only creates the necessary charge separation across the vacuum-plasma interface to excite the surface mode. The feature is highly beneficial for maintaining the stability of the interface. See more discussions in the Supplementary Materials.

The physical field can be written as ESP(r,z,t)=Re{E(j)(r)A(z,t)ei(kspz−ωspt)}\mathbf{E}^{\mathrm{SP}}(r,z,t)=\mathop{\mathrm{Re}}\{\boldsymbol{\mathcal{E}}^{(j)}(r)A(z,t)e^{i(k_{\mathrm{sp}}z-\omega_{\mathrm{sp}}t)}\}, where A(z,t)A(z,t) is complex amplitude. ωsp\omega_{\mathrm{sp}} and kspk_{\mathrm{sp}} are the eigen frequency and wavenumber and satisfy the dispersion relation in Eq. (2). E\boldsymbol{\mathcal{E}} is the normalized eigenmode profile along z^\hat{\mathbf{z}}-direction as E=E/NSP\boldsymbol{\mathcal{E}}=\mathbf{E}/\sqrt{\mathcal{N}_{\mathrm{SP}}} where NSP\mathcal{N}_{\mathrm{SP}} is the mode energy as NSP=116π∫(∣B∣2+∣E∣2)d2r⊥\mathcal{N}_{\mathrm{SP}}=\frac{1}{16\pi}\int(|\mathbf{B}|^2+|\mathbf{E}|^2)d^2r_\perp with ∫d2r⊥\int d^2r_\perp denoting the integration over the cross-section transverse to zz-direction. Projecting the Maxwell equations onto the eigenmode yields a driven, damped oscillator for A(z,t)A(z,t) as

(∂2∂t2+2ν∂∂t+ωsp2−vg2∂2∂z2)A(z,t)=SSPu0,1a0,2Δkei(Δkz−Δωt)(3)\left(\frac{\partial^2}{\partial t^2}+2\nu\frac{\partial}{\partial t}+\omega_{\mathrm{sp}}^2-v_g^2\frac{\partial^2}{\partial z^2}\right)A(z,t)=S_{\mathrm{SP}}u_{0,1}a_{0,2}\Delta k e^{i(\Delta kz-\Delta\omega t)} \tag*{(3)}

where SSPS_{\mathrm{SP}} is a coupling coefficient that depends on the overlap between the laser ponderomotive source and the SP eigenmode. It is calculated as

SSP=∫d2r⊥ E∗(r)⋅Fpbeat(r)≈−iπen0c22γ0ωsp2∫0∞r drEz∗(r)f2(r).(4)S_{\mathrm{SP}}=\int d^{2}r_{\perp}\,\mathbf{E}^{*}(r)\cdot\mathbf{F}_{p}^{\mathrm{beat}}(r) \approx-\frac{i\pi e n_{0}c^{2}}{2\gamma_{0}\omega_{\mathrm{sp}}^{2}}\int_{0}^{\infty}r\,dr E_{z}^{*}(r)f^{2}(r). \tag*{(4)}

where the radial contribution is neglected. See more details in the Supplementary Materials.

For a monochromatic near-resonant drive with slowly varying A(z,t)A(z,t), the steady-state solution is approximately

A(z,t)≈−SSPa0,1a0,2Δkωsp2−vg2Δk2−Δω2−2iνΔωei(Δkz−Δωt).(5)A(z,t)\approx-\frac{S_{\mathrm{SP}}a_{0,1}a_{0,2}\Delta k}{\omega_{\mathrm{sp}}^{2}-v_{g}^{2}\Delta k^{2}-\Delta\omega^{2}-2i\nu\Delta\omega}e^{i(\Delta k z-\Delta\omega t)}. \tag*{(5)}

The peak amplitude is then scaled as

Ez,max⁡wake≈GSP a0,1a0,2 ESP,0,(6)E_{z,\max}^{\mathrm{wake}}\approx G_{\mathrm{SP}}\,a_{0,1}a_{0,2}\,E_{\mathrm{SP},0}, \tag*{(6)}

where the natural SP field scale is defined by ESP,0=mecωsp/eE_{\mathrm{SP},0}=m_{e}c\omega_{\mathrm{sp}}/e. The dimensionless geometric factor GSPG_{\mathrm{SP}} measures how efficiently the beat ponderomotive force deposits energy into the normalised SP mode, and is defined by

GSP≡∣Ez(a)∣ESP,0∣SSP∣Δkνeff,(7)G_{\mathrm{SP}}\equiv\frac{\lvert E_{z}(a)\rvert}{E_{\mathrm{SP},0}}\frac{\lvert S_{\mathrm{SP}}\rvert\Delta k}{\nu_{\mathrm{eff}}}, \tag*{(7)}

where the effective damping is defined by νeff=(ωsp2−vg2Δk2−Δω2)2+(2νΔω)2\nu_{\mathrm{eff}}=\sqrt{(\omega_{\mathrm{sp}}^{2}-v_{g}^{2}\Delta k^{2}-\Delta\omega^{2})^{2}+(2\nu\Delta\omega)^{2}}. It is easy to see that GSPG_{\mathrm{SP}} carries the information of geometry, damping and dispersion and depends on the material properties. GSPG_{\mathrm{SP}} is large if 1) ∣Ez(a)∣\lvert E_{z}(a)\rvert is large (e.g. the mode is strongly surface localised); 2) ∣SSP∣\lvert S_{\mathrm{SP}}\rvert is large (e.g. the laser beam is well matched to the mode radial structure); 3) detuning is small (e.g. the beam matches the SP dispersion); 4) ν\nu is small (e.g. modest damping). The analytical solution of the SP field agrees very well with the PIC simulation, as shown in Figure 1 (e) and (f).

The resonance condition in Eq. (5) is ωsp2−vg2Δk2−Δω2≈0\omega_{\mathrm{sp}}^{2}-v_{g}^{2}\Delta k^{2}-\Delta\omega^{2}\approx0 for weak damping. The exact resonance occurs when Δω=ωsp\Delta\omega=\omega_{\mathrm{sp}} and Δk=ksp\Delta k=k_{\mathrm{sp}}. The cylindrical resonant condition is obtained by satisfying the dispersion relation in Eq. (2) as D(Δω,Δk;ne)=0D(\Delta\omega,\Delta k;n_{e})=0. This means to choose a matched plasma density such that the cylindrical SP intersects the light line exactly at the beat frequency. If the radial decay of the SP field length in the core is much larger than the tube radius, as shown in Figure 4, we have lim⁡κva→0I1(κva)/(κvI0(κva))=a/2\lim_{\kappa_{v}a\to0}I_{1}(\kappa_{v}a)/(\kappa_{v}I_{0}(\kappa_{v}a))=a/2. As a result, the cylindrical resonant condition can be reduced to

Image not extracted

Figure 4. Radial distribution of longitudinal component of electric field EzwakeE_{z}^{\mathrm{wake}} for different values of detuning rate ϱ\varrho defined by ϱ=(Δω−αΔk)/(ωsp−αksp)−1\varrho=(\Delta\omega-\alpha\Delta k)/(\omega_{\mathrm{sp}}-\alpha k_{\mathrm{sp}})-1. Here, ν=0\nu=0. The light blue regions indicate the plasma wall.

12+ϵp(Δω)ζK1(ζ)K0(ζ)=0,(8)\frac{1}{2}+\frac{\epsilon_{p}(\Delta\omega)}{\zeta}\frac{K_{1}(\zeta)}{K_{0}(\zeta)}=0, \tag*{(8)}

where ζ=ωpa/γ0c\zeta=\omega_{p}a/\gamma_{0}c. The resonant condition is numerically calculated and shown in Figure 3. It is seen that for a

Image not extracted

Figure 3. Resonant condition between matched plasma density nen_{e} and second laser wavelength λ2\lambda_{2}, where λ1=0.8 μm\lambda_{1}=0.8\ \mu\mathrm{m}, a=1.5 μma=1.5\ \mu\mathrm{m} and γ0=1.0\gamma_{0}=1.0, calculated by Eq. (8).

finite density, the SP mode lives around ω∼O(ωp)\omega\sim O(\omega_{p}), and does not extend down to arbitrarily small ω\omega on the light line where Δω→0\Delta\omega\to0. This implies that a single on-axis laser cannot resonantly excite a finite-frequency SP without additional asymmetry. The most efficient excitation occurs when the cylindrical SP mode admits vphv_{\mathrm{ph}} close to cc, while the field enhancement benefits from approaching the bound regime near the resonant line. The curvature allows the SP to be significantly faster than in the planar case at a similar density. Eq. (8) implies a vphv_{\mathrm{ph}} close to cc while staying in a near-critical regime as shown in Figure 2. That’s the key advantage of the cylindrical surface.

Eq. (8) is a joint condition on the lasers and the plasma dispersion and geometry, with a trade-off controlled by detuning and damping. The field amplitude is significantly limited by detuning on a cold plasma surface, as shown in Figure 4. By linearising the dispersion around (ωsp,ksp)(\omega_{\mathrm{sp}},k_{\mathrm{sp}}) near the resonance, we have ω2≈ωsp2+2ωspδω\omega^{2}\approx\omega_{\mathrm{sp}}^{2}+2\omega_{\mathrm{sp}}\delta\omega and k2≈ksp2+2kspδkk^2 \approx k_{\rm sp}^2+2k_{\rm sp}\delta k with δω=Δω−ωsp\delta\omega=\Delta\omega-\omega_{\rm sp} and δk=Δk−ksp\delta k=\Delta k-k_{\rm sp}. The effective damping is reduced to

νeffres≃2ωspν(δω+αδk)2+ν2withα=vg2kspωsp2.\nu_{\rm eff}^{\rm res}\simeq\frac{2\omega_{\rm sp}\nu}{(\delta\omega+\alpha\delta k)^2+\nu^2}\quad\text{with}\quad\alpha=\frac{v_g^2k_{\rm sp}}{\omega_{\rm sp}^2}.

This will be a major experimental concern due to the inevitable laser-plasma mismatch in a real experiment.

In PIC simulations, GSPPICG_{\rm SP}^{\rm PIC} can be simply calculated by GSPPIC=Ez,max⁡wake/(a0,1a0,2ESP,0)G_{\rm SP}^{\rm PIC}=E^{\rm wake}_{z,\max}/(a_{0,1}a_{0,2}E_{\rm SP,0}) and presents the resonant enhancement, while the acceleration field scales with a0,1a0,2a_{0,1}a_{0,2}, agreeing well with the theory, as shown in Figure 5(a). Resonant SP excitation enables low-power lasers to act as effective drivers, as shown in Figure 5(b), where a0,1=a0,2=0.01a_{0,1}=a_{0,2}=0.01 corresponds to a 0.15 GW peak power for λ1=0.8\lambda_1=0.8 μm and a 0.25 GW for λ2=0.6\lambda_2=0.6 μm with an acceleration field of ∼1\sim1 GV/m. It can be seen that SP growth is much faster in the low-power regime. This is due to the small detuning rate caused by the weak surface density perturbation.

Image not extracted

Figure 5. (a) GSPG_{\rm SP} as a function of plasma density nen_{\rm e} from theory (normalised, blue solid) and PICs (red-dot). (b) Amplitude of on-axis acceleration field Ez,max⁡wakeE^{\rm wake}_{z,\max} as a function of a02a_0^2 from theory (blue solid) and PICs (red-dot). Here, a0,1=a0,2=a0a_{0,1}=a_{0,2}=a_0. The other parameters are λ1=0.8\lambda_1=0.8 μm, λ2=0.6\lambda_2=0.6 μm, a=1.0a=1.0 μm and w0=2.0w_0=2.0 μm. In (a), a0=0.6a_0=0.6, corresponding to laser peak powers P1,peak=48P_{1,\mathrm{peak}}=48 GW and P2,peak=80P_{2,\mathrm{peak}}=80 GW, respectively. In (b), ne=6×1020n_{\rm e}=6\times10^{20} cm−3^{-3}, close to the resonant density. The PIC results are obtained after the laser pulses co-propagate 40 μm inside a microtube.

SP excitation occurs as soon as the outermost electrons of the surface material can be ionised by the laser field. For example, on a metallic crystalline surface, such as VCNT, laser pulses with a peak power of order 10 MW are sufficient to ionise the first π\pi-bond electrons, thereby initiating SP resonant growth. However, at such low intensities, the resulting SP field remains too weak to trap electrons. Current PIC simulations in Figure 5(b) indicate that efficient electron trapping occurs for a0>0.2a_0>0.2, corresponding to a laser power threshold of approximately 4 GW.

Acknowledgments

Importantly, this power level lies well within the capabilities of currently available high-power fibre laser systems [13], demonstrating the experimental feasibility of resonant SP-driven wakefield acceleration.

In summary, the theory of resonant SP excitation driven by beating lasers presented in this letter can be readily generalised to other drivers, including the Lorentz or ponderomotive force of a laser field and the space-charge force of a charged-particle beam. In each case, Eq. (3) is modified by the appropriate source term, and the corresponding eigenmodes become dominant depending on the coupling efficiency. Geometric effects can be treated similarly, and the efficiency can be improved by using structured drivers, such as Laguerre-Gaussian (LG) lasers. While this scheme works with lasers of a wide range of power from GW to PW, a particularly attractive prospect is to accelerate a well-collimated electron beam up to a few MeV inside a solid microtube using portable GW lasers over a 10s of μm scale, while retaining substantial flexibility in energy spectrum, brightness, and beam size. Technically, this could be realised by attaching an ultrathin film to the output of standard laser optics. Such films can be fabricated with existing techniques using VCNT with well-structured channels. Therefore, this beat-wave-driven scheme merits consideration as a platform technology with potential impact across materials science, medicine, computing, communications, and energy.

This work has been supported by the Science and Technology Facilities Council (STFC) through the Cockcroft Institute core grant UKRI1887. Guoxing Xia acknowledges the support from the Cockcroft Institute Core Grant No. ST/V001612/1. Javier Resta-López acknowledges support by the Generalitat Valenciana under grant agreement CID2022/2019/058. Bin Liu acknowledges the support of Guangdong Basic and Applied Basic Research Foundation (Grant No. 2025A1515011532) and Guangdong Major Talent Project (Grant No. 2023CX10Y036). This work made use of the Barkla High Performance Computing facilities at the University of Liverpool.

References

References

  1. [3]ica 31, 213 (1941).
  2. [4]A. Otto, Excitation of nonradiative surface plasma waves in silver by the method of frustrated total reflection, Zeitschrift für Physik A Hadrons and nuclei 216, 398 (1968).DOI
  3. [5]X. F. Shen, A. Pukhov, and B. Qiao, Monoenergetic high-energy ion source via femtosecond laser interacting with a microtape, Physical Review X 11, 041002 (2021).DOI
  4. [6]J. Sarma, A. McIlvenny, N. Das, M. Borghesi, and A. Macchi, Surface plasmon-driven electron and proton acceleration without grating coupling, New Journal of Physics 24, 073023 (2022).DOI
  5. [7]A. McCay, A. McIlvenny, A. Macchi, L. Romagnani, P. Martin, O. Cayanan, D. P. Mollov, L. Lancia, T. Dzelzainis, H. Ahmed, J. Sarma, S. Kar, D. Margaronne, and M. Borghesi, Surface wave electron acceleration from flat foils at parallel laser incidence, Physical Review Letters 135, 145001 (2025).DOI
  6. [8]B. Lei, H. Zhang, D. Seipt, A. Bonatto, B. Qiao, J. Resta-López, G. Xia, and C. Welsch, Coherent synchrotron radiation by excitation of surface plasmon polariton on near-critical solid microtube surface, Phys. Rev. Lett. 135, 205001 (2025).
  7. [9]B. Lei, H. Zhang, C. Bontoiu, A. Bonatto, P. Martín-Luna, B. Liu, J. Resta López, G. Xia, and C. Welsch, Leaky surface plasmon-based wakefield acceleration in nanostructured carbon nanotubes, Plasma Physics and Controlled Fusion 67, 065036 (2025).DOI
  8. [10]G. I. Stegeman, R. F. Wallis, and A. A. Maradudin, Excitation of surface polaritons by end-fire coupling, Optics Letters 8, 386 (1983).pubmed.ncbi.nlm.nih.gov/19718123
  9. [11]B. Lei, H. Zhang, C. Bontoiu, A. Bonatto, J. Resta-López, G. Xia, B. Qiao, and C. Welsch, 100s tev/m-level particle accelerators driven by high-density electron beams in micro structured carbon nanotube forest channel, New Journal of Physics 27, 084301 (2025).arxiv.org/abs/2502.08498
  10. [12]M. Yu, H. H. Funke, J. L. Falconer, and R. D. Noble, High density, vertically-aligned carbon nanotube membranes, Nano Letters 9, 225 (2009).DOI
  11. [13]A. Lekosiotis, F. Belli, C. Brahms, M. Sabbah, H. Sakr, I. A. Davidson, F. Poletti, and J. C. Travers, On-target delivery of intense ultrafast laser pulses through hollow-core anti-resonant fibers, Optics Express 31, 30227 (2023).arxiv.org/abs/2305.16911
  12. [14]E. Esarey, C. B. Schroeder, and W. P. Leemans, Physics of laser-driven plasma-based electron accelerators, Reviews of Modern Physics 81, 1229 (2009).DOI
  13. [15]Y. Kitagawa, T. Matsumoto, T. Minamihata, K. Sawai, K. Matsuo, K. Mima, K. Nishihara, H. Azechi, K. A. Tanaka, H. Takabe, and S. Nakai, Beat-wave excitation of plasma wave and observation of accelerated electrons, Physical Review Letters 68, 48 (1992).pubmed.ncbi.nlm.nih.gov/10045109
  14. [16]A. Chamoli, D. N. Gupta, and V. Kumar, Surface plasma wave excitation using laser beat-wave and terahertz generation, Radiation Effects and Defects in Solids 179, 962 (2024).
  15. [17]D. Roa, J. Kuo, H. Moyses, P. Taborek, T. Tajima, G. Mourou, and F. Tamanoi, Fiber-optic based laser wakefield accelerated electron beams and potential applications in radiotherapy cancer treatments, Photonics 9, 403 (2022).DOI
  16. [18]J. L. Vay, A. Almgren, J. Bell, L. Ge, D. P. Grote, M. Hogan, O. Kononenko, R. Lehe, A. Myers, C. Ng, J. Park, R. Ryne, O. Shapoval, M. Thévenet, and W. Zhang, Warp-x: A new exascale computing platform for beam–plasma simulations, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 909, 476 (2018).DOI
  17. [19]T. M. Antonsen and P. Mora, Self-focusing and raman scattering of laser pulses in tenuous plasmas, Physical Review Letters 69, 2204 (1992).

Paper details

vlnc:W_454m76ju
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents