High-Gradient Plasma-Wakefield Acceleration with Two Subpicosecond Electron Bunches Efthymios Kallos, 1 Tom Katsouleas, 1 Wayne D. Kimura, 2 Karl Kusche, 3 Patric Muggli, 1 Igor Pavlishin, 3 Igor Pogorelsky, 3 Daniil Stolyarov, 3 and Vitaly Yakimenko 3 1 University of Southern California, Los Angeles, California 90089, USA 2 STI Optronics, Bellevue, Washington 98004, USA 3 Brookhaven National Lab, Upton, New York 11973, USA (Received 12 October 2007; published 20 February 2008) A plasma-wakefield experiment is presented where two 60 MeV subpicosecond electron bunches are sent into a plasma produced by a capillary discharge. Both bunches are shorter than the plasma wavelength, and the phase of the second bunch relative to the plasma wave is adjusted by tuning the plasma density. It is shown that the second bunch experiences a 150 MeV=m loaded accelerating gradient in the wakefield driven by the first bunch. This is the first experiment to directly demonstrate high- gradient, controlled acceleration of a short-pulse trailing electron bunch in a high-density plasma. DOI: 10.1103/PhysRevLett.100.074802 PACS numbers: 41.75.Lx, 41.75.Ht, 41.85.Ct, 52.40.Mj Plasma waves can sustain extremely large electric fields that are orders-of-magnitude larger than those in conven- tional radio-frequency accelerators, which are limited by vacuum breakdown to accelerating gradients of up to 150 MV=m [1]. Such large amplitude electron density waves, or wakes, can be excited in plasmas by a laser pulse (laser wakefield acceleration—LWFA), or a relativistic particle beam (plasma-wakefield acceleration—PWFA). Recent LWFA experiments demonstrated quasi-mono- energetic acceleration of self-trapped plasma electrons [2]. Further scaling of LWFA to higher energies, by using higher laser power but larger spot sizes and lower density plasmas, will probably require injecting relativistic elec- tron bunches into a plasma wave, rather than starting with plasma electrons at zero energy. So far, this approach to potentially monoenergetic particle acceleration using plas- mas has not been explored. Experiments using PWFA methods similarly face the challenge of producing low- energy-spread acceleration of an injected relativistic parti- cle bunch. In previous PWFA experiments (see, for ex- ample, Refs. [3 – 5]), a single relativistic electron-bunch both drove the wake and provided the electrons to be accelerated. Using this scheme at SLAC, a record-high energy gain of 42 GeV over 85 cm of plasma was demon- strated [3], albeit with an undesirable 100% spread of the electron energy spectrum. To realize a future collider, such as one incorporating the PWFA afterburner concept [6], a well-defined bunch is required that is suitably phased on the plasma wake of a preceding drive bunch thereby to achieve high efficiency and a small energy spread ( 0:1% which is typical for conventional accelerators). In this Letter, we detail a double-bunch PWFA experiment that allows, for the first time, to achieve a controllable high- gradient acceleration of a witness relativistic electron bunch injected into a plasma wave. Earlier double-bunch PWFA experiments [7] utilized relatively long, picosecond electron bunches in low- density (10 13 cm 3 ) plasmas; wakefields up to 4 MV=m were inferred. Our experiment differs substan- tially from these studies. First, the driver and witness bunches have subpicosecond lengths ( z  100 fs), and are both shorter than the plasma wavelength. Conse- quently, the energy shift of both bunches can be directly observed rather than mathematically extracted, as was required in earlier works. Second, the shorter bunch lengths and the higher plasma densities employed (up to 10 17 cm 3 ) result in generated wakefield amplitudes that are two orders-of-magnitude larger compared to those studies. We show here that, for a fixed bunch-spacing, the plasma density can be chosen such that the drive electron-bunch loses energy that it is transferred through the plasma wave to the second bunch, which, in turn, gains energy with a minimum energy spread. In the experiment, the drive bunch loses about 1.0 MeV over 6 mm propaga- tion in a plasma of 10 16 cm 3 density, and the witness bunch, delayed by 500 fs, gains 0:9 MeV correspond- ing to an average loaded accelerating gradient of 150 MeV=m. The measured energy gain and loss agree well with 2D linear theory calculations. This experiment is the first to generate and directly probe large plasma accel- erating gradients (>100 MeV=m) utilizing a trailing elec- tron bunch. The experiment is performed at Brookhaven National Laboratory’s (BNL’s) Accelerator Test Facility (ATF). A photocathode rf gun followed by a conventional 2.856 GHz (S-band) accelerator produces a 1.5 ps-long (rms), 500- pC, 60 MeV single electron bunch [8] that is compressed and split into two distinct (in time and energy) subpico- second bunches after traveling through a chicane compres- sor and ‘‘dog-leg’’ dipoles downstream from the linac [Fig. 1]. These two bunches are focused transversely to  r  100 m at the entrance of a 10 14 cm 3 –10 17 cm 3 density plasma produced by an ablative capillary discharge [9]. A magnetic spectrometer at the end of the beam line records the energy change imparted to the bunches by the plasma. PRL 100, 074802 (2008) PHYSICAL REVIEW LETTERS week ending 22 FEBRUARY 2008 0031-9007= 08=100(7)=074802(4) 074802-1  2008 The American Physical Society