Laser-driven plasma waves deliver the best beams so far

11 November 2004

Three teams, in the UK, the US and France, have reported breakthroughs in the laser-driven plasma acceleration of electrons. For the first time researchers have been able to create conditions such that the accelerated beam has a low divergence and small spread in energy. This paves the way towards the practical development of compact “table-top” particle accelerators for a variety of applications.


The idea of harnessing the high electric fields generated in laser-driven plasma waves in order to accelerate electrons was first proposed by Toshi Tajima and John Dawson in 1979. The basic principle is to direct an intense laser pulse into a plasma, which sets the plasma electrons oscillating, so creating a relativistic plasma wave in the wake of the pulse. Fields of more than 100GeVm-1-thousands of times greater than achieved with conventional accelerators-can be set up in this way and charged particles can be accelerated as they “surf” the plasma wave. With the advent of high-brightness lasers, this technique of laser wakefield acceleration has in the past decade been demonstrated in several different experiments. In 2003, Karl Krushelnick led a European Union-funded collaboration between the Laboratoire d’Optique Appliquée at the Ecole Polytechnique in Paris, Imperial College London and the Rutherford Appleton Laboratory (RAL) that achieved electron energies of 350MeV over distances of only 1 mm or so in a plasma driven by ultra-short, intense laser pulses. However, the beams produced have had a broad spread in energy (as much as 100%) due to wavebreaking, making them of limited practical use.

In the latest work, the three groups have found different techniques to overcome this problem and localize large numbers of electrons with respect to the plasma wave. The beams generated are much closer to being monoenergetic, with energy spreads down to less than 10%. These findings are described in suCCEessive papers in the same edition of Nature, following earlier reports, for example at the 11th workshop on Advanced Accelerator Concepts in June.


Krushelnick and colleagues from Imperial College, Strathclyde University, the University of California, Los Angeles, and RAL have used the Astra laser in RAL’s Central Laser Facility to focus ultrashort (40fs), intense (0.5J) laser pulses onto a supersonic jet of helium gas. At the same time they have been able to vary the density of the plasma created by varying the pressure of the gas jet. As the plasma density is increased, “wavebreaking” starts to oCCEur, where electrons break free from the plasma wave and can then be accelerated by the remaining wave. At densities below 7×1018cm-3, no energetic electrons emerged from the plasma, but at higher densities the team detected electrons with energies up to 100MeV. Moreover, the energy spectrum measured with a magnetic spectrometer showed narrow spikes.

By carefully controlling the density and increasing the laser power, the team has produced energy spectra with single, narrow peaks, in which the energy spread is as small as 3%, for energies in the range 50-80MeV (Mangles et al. 2004). The researchers explain their observations in terms of “controlled wavebreaking”. Wavebreaking is not always catastrophic and a number of the electrons in the plasma wave can break from the wave, reduce its amplitude but still maintain the wave structure. The differences in the observed spectra correspond to timing of the injected electrons into this relativistic (but decaying in amplitude) plasma wave. For high densities, the wake-field is several plasma wavelengths in duration. SuCCEessive plasma periods can accelerate trapped electron bunches to different energies, producing the multiple spikes in the spectrum. For the single spike feature, it is likely that only the first plasma oscillation is driven to breaking point.

A similar approach has been taken by Malka’s group, building on their previous work in which they demonstrated that they could produce well-collimated energetic beams using laser-driven wakefields, although the beams still had a broad energy spread. In the latest work, the group creates a plasma “bubble” in the wake of a laser pulse that has been compressed by self-focusing. Wave-breaking at the walls of the bubble releases electrons that are accelerated together within the bubble. This technique has allowed the group to claim acceleration of a beam of only 10 mrad divergence, with a total charge of 0.5±0.2 nC at an energy of 170±20 MeV (Faure et al. 2004).

In the US, meanwhile, Wim Leemans and colleagues in the L’OASIS (Laser Optics and Accelerator Systems Integrated Studies) group at the Lawrence Berkeley National Laboratory have taken a different approach. They have created a plasma density channel in hydrogen gas, which is denser towards the edges and serves to guide the driving laser pulse. This overcomes the natural weakening of the pulse caused by diffraction as it propagates thereby extending the distance available for acceleration. Although the L’OASIS team used a laser spot about three times smaller than the one used by the UK and French groups, and hence a diffraction distance roughly ten times shorter, the use of the plasma channels kept the laser beam tightly focused and so highly intense at the entrance of the plasma channel.

In this method, a first “igniter” pulse forms a narrow “wire” of plasma, which a second “heater” pulse enters from the side, expanding it into a broader channel. Lastly, 500 ps later, an intense “driver” pulse is sent through the channel, and this excites the plasma waves that ultimately accelerate electrons from the plasma. By carefully controlling the laser and plasma-channel parameters, the team observed acceleration of electron bunches with a narrow energy spread, for example ±2% at an energy of 86 MeV and 3 mrad (FWHM) divergence, containing around 0.3-0.5 nC of charge (Geddes et al. 2004). The normalized emittance was estimated to be of the order of 1-2π mm-mrad (rms). Once again several factors seem to be involved in keeping the energy spread small, with the most important one being the control of the acceleration distance to match the dephasing distance, i.e. the distance where the electrons start to outrun the wave. Simulations of the process with the particle-in-cell code VORPAL indicate that the laser pulse first self-steepens while propagating in the plasma. As a result larger amplitude waves are excited as the laser pulse propagates deeper into the channel. When the wave amplitude reaches levels sufficient to trap background electrons and the acceleration process is extended to the dephasing distance, momentum bunching oCCEurs and this results in a narrow energy spread for the beam. Combined with sufficient beam loading to suppress trapping in trailing accelerating buckets, this leads to the quality of the electron beams observed, with low divergence and energy spread.

These results represent a great achievement, but all three groups point to the need for further work, on efficiency and shot-to-shot stability for example. Also it is still far from clear as to how an accelerator based on this technique could be “staged” to reach the teravolt energies now generally required for research at the high-energy frontier. However, the results provide great hope for progress towards compact, high-brightness machines operating in the giga-electronvolt region. These would have many applications, for example in materials science, ultrafast chemistry and medicine.

Further reading

C G R Geddes et al. 2004 Nature 431 538.
J Faure et al. 2004 Nature 431 541.
S P D Mangles et al. Nature 431 535.

bright-rec iop pub iop-science physcis connect