Guiding of relativistically intense laser pulses with peak power of 0.85 PW over 15 diffraction lengths was demonstrated by increasing the focusing strength of a capillary discharge waveguide using laser inverse Bremsstrahlung heating. This allowed for the production of electron beams with quasi-monoenergetic peaks up to 7.8 GeV, double the energy that was previously demonstrated. Charge was 5 pC at 7.8 GeV and up to 62 pC in 6 GeV peaks, and typical beam divergence was 0.2 mrad.

Laser plasma accelerators 1 have produced high-quality electron beams with GeV energies from cm-scale devices 2 and are being investigated as hyperspectral fs light sources producing THz to γ-ray radiation 3-5 , and as drivers for future highenergy colliders 6,7 . These applications require a high degree of stability, beam quality and tunability. Here we report on a technique to inject electrons into the accelerating field of a laser-driven plasma wave and coupling of this injector to a lower-density, separately tunable plasma for further acceleration. The technique relies on a single laser pulse powering a plasma structure with a tailored longitudinal density profile, to produce beams that can be tuned in the range of 100-400 MeV with per-cent-level stability, using laser pulses of less than 40 TW. The resulting device is a simple stand-alone accelerator or the front end for a multistage higher-energy accelerator.Producing high-quality electron beams from an accelerator requires electron injection into the accelerating field to be localized in time and space. For laser plasma accelerators (LPAs) that rely on homogeneous plasmas driven with single laser pulses, continuous injection can occur when driving large-amplitude plasma waves (wakefields), resulting in large energy spread. Lower energy spread can be achieved through termination of injection by operating near the injection threshold or by injecting enough charge to suppress the wake amplitude (that is, beam loading). Subsequent termination of the accelerating process at dephasing (that is, when electrons are starting to outrun the accelerating wave) minimizes energy spread. These mechanisms have produced per-cent-level energy-spread beams 2,8-10 , but small changes in parameters can result in large changes in beam quality. As a result, tunability has been limited, necessitating the development of a simple, robust and controlled injection technique combined with an independently controllable accelerating stage.In general, injection of electrons into a plasma wave occurs when the velocity of background electrons approaches the wake phase velocity. Laser-based methods for boosting the electron velocity have been proposed 11,12 and implemented 13,14 to achieve tunable electron beams, but require sophisticated alignment and synchronization of the multiple laser pulses. Injection can also be triggered by introducing electrons into the correct phase of the wake through ionization 15 , but so far the technique has resulted in broad energy-spread beams with high divergence 16,17 . A different approach, that relies on a single laser pulse for powering the LPA, is to momentarily slow down the wake phase velocity to facilitate trapping 18 . The control of the wake phase velocity can be achieved by tailoring the nonlinear plasma wavelength λ p (z) along the longitudinal coordinate z, through control of the electron density n e and the laser parameters. Specifically, λ p (z) = λ p0 (z)F , where the linear plasma wavelength λ p0 (µm) ≈ 3.3 × 10 10 / √ n e (cm −3 ) and F ...

Coherent radiation in the 0.3-3 THz range has been generated from femtosecond electron bunches at a plasma-vacuum boundary via transition radiation. The bunches produced by a laser-plasma accelerator contained 1.5 nC of charge. The THz energy per pulse within a limited 30 mrad collection angle was 3-5 nJ and scaled quadratically with bunch charge, consistent with coherent emission. Modeling indicates that this broadband source produces about 0.3 microJ per pulse within a 100 mrad angle, and that increasing the transverse plasma size and electron beam energy could provide more than 100 microJ/pulse.

Laser-plasma accelerators (LPAs) are capable of accelerating charged particles to very high energies in very compact structures. In theory, therefore, they offer advantages over conventional, large-scale particle accelerators. However, the energy gain in a single-stage LPA can be limited by laser diffraction, dephasing, electron-beam loading and laser-energy depletion. The problem of laser diffraction can be addressed by using laser-pulse guiding and preformed plasma waveguides to maintain the required laser intensity over distances of many Rayleigh lengths; dephasing can be mitigated by longitudinal tailoring of the plasma density; and beam loading can be controlled by proper shaping of the electron beam. To increase the beam energy further, it is necessary to tackle the problem of the depletion of laser energy, by sequencing the accelerator into stages, each powered by a separate laser pulse. Here, we present results from an experiment that demonstrates such staging. Two LPA stages were coupled over a short distance (as is needed to preserve the average acceleration gradient) by a plasma mirror. Stable electron beams from a first LPA were focused to a twenty-micrometre radius--by a discharge capillary-based active plasma lens--into a second LPA, such that the beams interacted with the wakefield excited by a separate laser. Staged acceleration by the wakefield of the second stage is detected via an energy gain of 100 megaelectronvolts for a subset of the electron beam. Changing the arrival time of the electron beam with respect to the second-stage laser pulse allowed us to reconstruct the temporal wakefield structure and to determine the plasma density. Our results indicate that the fundamental limitation to energy gain presented by laser depletion can be overcome by using staged acceleration, suggesting a way of reaching the electron energies required for collider applications.