A table-top laser-based source of femtosecond, collimated

ultra-relativistic positron beams

G. Sarri,1 W. Schumaker,2 A. Di Piazza,3 M. Vargas,2 B. Dromey,1 M. E.
Dieckmann,1 V. Chvykov,2 A. Maksimchuk,2 V. Yanovsky,2 Z. H. He,2 B. X. Hou,2
J. A. Nees,2 A. G. R. Thomas,2 C. H. Keitel,3 M. Zepf,1, 4 and K. Krushelnick2
School of Mathematics and Physics, The Queen’s University of Belfast, BT7 1NN, Belfast, UK
Center for Ultrafast Optical Science, University of Michigan, Ann Arbor, Michigan 48109-2099, USA
Max-Planck-Institut für Kernphysik, Saupfercheckweg 1, 69117 Heidelberg, Germany
Helmholtz Institute Jena, Fröbelstieg 3, 07743 Jena, Germany
(Dated: April 22, 2013)
arXiv:1304.5379v1 [physics.acc-ph] 19 Apr 2013

The generation of ultra-relativistic positron beams with short duration (τe+ ≤ 30 fs), small
divergence (θe+ ' 3 mrad), and high density (ne+ ' 1014 − 1015 cm−3 ) from a fully optical setup
is reported. The detected positron beam propagates with a high-density electron beam and γ-rays
of similar spectral shape and peak energy, thus closely resembling the structure of an astrophysical
leptonic jet. It is envisaged that this experimental evidence, besides the intrinsic relevance to laser-
driven particle acceleration, may open the pathway for the small-scale study of astrophysical leptonic
jets in the laboratory.

Creating and characterizing high-density beams of rel- [11–14]. Despite the intrinsic interest of these results,
ativistic positrons in the laboratory is of paramount im- the low energy and broad divergence reported (Ee+ ≤ 20
portance in experimental physics, due to their direct ap- MeV and θe+ ≥ 350 mrad , respectively) still represent
plication to a wide range of physical subjects, including clear limitations for future use in hybrid machines.
nuclear physics, particle physics, and laboratory astro-
The possibility of generating high density and high en-
physics. Arguably, the most practical way to generate
ergy electron-positron beams is of central importance also
them is to exploit the electromagnetic cascade initiated
for astrophysics, due to their similarity to jets of long
by the propagation of an ultra-relativistic electron beam
gamma-ray bursts (GRBs), which are ejected as a conse-
through a high-Z solid. This process is exploited to gen-
quence of the collapse of super-massive stars [15]. These
erate low-energy positrons in injector systems for con-
structures, despite extensive numerical studies [16], still
ventional accelerators such as the Electron-Positron Col-
present enigmatic features which are virtually impossi-
lider (LEP) [1]. In this case, an ultra-relativistic elec-
ble to address by simply relying on direct observations.
tron beam (Ee− ≈ 200 MeV) was pre-accelerated by a
A possible solution might be represented by reproduc-
LINAC and then directed onto a tungsten target. The
ing small scale electron-positron jets (required bulk flow
resulting positron population, after due accumulation in
Lorentz factor of the order of 100 - 1000) in the labo-
a storage ring, was further accelerated by a conventional,
ratory. Although GRB jets may well have a weak large
large-scale (R ≈ 27 km), synchrotron accelerator up to a
scale magnetic field [17], the external shock is exclusively
peak energy of 209 GeV. The large cost and size of these
mediated by self-generated micro-scale magnetic fields.
machines have motivated the study of alternative particle
A purely electronic jet would present toroidal magnetic
accelerator schemes. A particularly compact and promis-
fields whose strength and structure would be compara-
ing system is represented by plasma devices which can
ble to the microscale fields that develop in response to
support much higher accelerating fields (of the order of
the filamentation instability [18] and modify the shock
100s of GV/m, compared to MV/m in solid-state acceler-
physics. The presence of the highly mobile positrons
ators) and thus significantly shorten the overall size of the
would reduce the overall magnetisation of the jet (along-
accelerator. Laser-driven generation of electron beams
side with the amplitude of the electrostatic fields driven
with energies per particle reaching [2–5], and exceeding
by a charge separation [19]), simplifying the interpreta-
[6], 1 GeV have been experimentally demonstrated and
tion of the experimental data and their comparison with
the production of electron beams with energies approach-
the astrophysical scenario.
ing 100 GeV is envisaged for the next generation of high-
power lasers (1 - 10 PW) [7]. Hybrid schemes have also Here we experimentally demonstrate the possibility
been proposed and successfully tested in first proof-of- of producing collimated and high-density ultrarelativis-
principle experiments [8, 9]. On the other hand, laser- tic positron beams in a fully laser-driven configura-
driven low energy positrons (Ee+ ≈ 1−5 MeV) have been tion. Their low divergence and short duration (com-
first experimentally obtained by C. Gahn and coworkers parable to those of the incoming laser-driven electron
[10] and recently generated during the interaction of a beam) suggest the possibility of applying this generation
picosecond, kiloJoule class laser with thick gold targets scheme to current laser facilities towards the construc-
tion of a fully-optical generator of high-quality, ultra-

&'A$# $"# 9: 8+C)#D'*)).)# # = 0?#$1#@. Typical signal of the electron beam as recorded on the LANEX screen. we can then a plasma wavelength (λpe = 2πc/ωpe ≈ 10 µm) imply. by a magnetic spectrometer consisting of a rial. Furthermore. 2 ?#$1#@. pend on Zα. 1. 1. prior to any shot with a high-Z n is the number of atoms per unit volume in the mate- solid target. with A and mp being the atomic num- electrons).5% of is mediated by a virtual photon in the electron field [26].The laser-accelerated electron beam Ta. a. express the scaling as Np ∝ (Z 2 /A)j .E# 3& . b. doped with 2.&8'# !"#$%& 5"6#2# .4 to pared to γe+ ≤ 10 in [10–12]) of these beams are fi. versity of Michigan [20]. 6.8 µm. the positrons inside J and duration τL = 30 fs. Pb) and thicknesses (from 1. Plastic and Teflon shielding was inserted to reduce the noise due to low energy divergent particles and x-rays.a) was induced by both the low-energy electrons and gamma- carried out using the HERCULES laser system at the rays generated. Higher-order multi-step cascade processes sity of IL ≈ 6 × 1018 W/cm2 ).E# # 4& !"#$%&# '+. Electron bunches obtained in similar condi. separate the electrons from the positrons which were then This.< '()&*+. which delivered a laser beam with a central wavelength λL = 0. de- pressure were chosen in order to stay slightly above the pending on the ratio between the thickness of the target threshold for ionisation injection [21]. constant). and j = 1 for 0. by using the calibration curves reported in [25]. finally open up a realistic possibility 2% [24]). the inferred recorded onto an Image Plate (IP). without a solid target and d. Sn.&'A$# 8+C)#D'*)). Laser power and gas-jet may also significantly contribute to pair production.a). in which pair production a 3 mm wide supersonic He gas-jet. Once fully ionised. relativistic positron beams. with a backing pressure of 5. process [28]. Sn. extracted spectrum.5 mrad (see Figs. 15 cm long pair of magnets and a LANEX screen. 1. and the neutron mass. N2 . the positron of approximately 2.-!#*&.-& 0$1#B+. ber and the proton mass. c. Due to the small dif- electron/positron balance. Plas- tic shielding was inserted to reduce the noise on the IP The experiment (shown schematically in Fig. the trident process and j = 2 for the two-step cascade The arrangement of the spectrometer did not allow us to process (we neglect here Coulomb corrections. Its yield Ne+ is expected to scale as: Ne+ ∝ (Z 2 nd)j . ference in positron and electron stopping power (below tense gamma-rays./012& #3& !"# 05#$1# >#$1# %&'()*# 0#$1#2)3+4# 0?#$1#@. If we maintain tions have been shown to have a length comparable to the areal mass density (σ = ρd) constant. and co-propagation with in. the mass density of the solid tar- ceeding 80 MeV to be of the order of 50 pC (3 × 108 get is ρ ≈ Amp n.e). The region labelled with γ noise is predominantly exposed by the γ-rays escaping the solid target. onto the edge of troproduction (trident process).8 T.&'A$# # '("#$%&# 7#. materials (Cu. This laser beam was fo. The same magnetic spectrometer was used to nally comparable to those of astrophysical leptonic jets. the signal recorded was absolutely calibrated of studying the dynamics of such jets in the laboratory. the high-Z target can be generated via either direct elec- cussed. Neglecting the difference between the proton indicated the charge carried by electrons with energy ex. where spectrum was measured./)*# FG# 0#$1#2)3+4# Figure 1.8 In these experimental conditions. respectively.5 bar. first emits a real photon (bremsstrahlung) [27] which then The focal spot size was measured to have a radius of 23 produces an electron/positron pair via the Bethe-Heitler µm which contained 50% of the laser energy (peak inten. This interaction and the radiation length of the material [29]. using an f /20 off-axis parabola.d and 1. We have thus per- ing a typical temporal duration comparable to that of formed a series of shots for different materials (Cu. the laser pulse [23]. d is the thickness of the solid target. energy EL = 0. during the laser-gas and Center for Ultrafast Optical Science (CUOS) in the Uni. in conjunction with the low divergence. Typical positron signal as recorded by the Image Plate. which de- resolve electron energies below 80 MeV. Top-view of the experimental setup. with α ≈ 1/137 being the fine-structure obtained using the calibration curves reported in [22].4 mm). com.the measured interacted with mm-size high-Z solid targets of different high positron Lorentz factors (γe+ ' 200 − 400. Typical spectra. or via a two-step “cascade” process where the electron this corresponds to an electron density of 9 × 1018 cm−3 . Pb) adjusting the target thickness so that the areal . Ta. at wide angles. electron-solid target interactions (see Fig. Keeping the delivered a reproducible electron beam with a divergence parameters of the electron beam constant.

6 mm.30 ± 0. inferred from the signal recorded on the 4.6 63 simulations (NT in Table I). The positron yield over the entire Cu 5.9 350 Ne+ ∝ d2 for the two-step process). which implies that the contribution of higher-order cas- tron beam through a solid target [30]. of-magnitude estimate. respectively. there is total screening  if the parame- ter S ≡ αZ 1/3 ε(ε − ~ω)/ ~ωmc2 is much larger than unity. As expected. λC = ~/mc = 3.3 mm of Cu. numerical simulations performed with the nuclear magnetic cascade. as extracted from matching numerical simulations. In order to qualitatively explain the reported experimental results. as extracted from matching FLUKA Sn 6. the positron beam has a function of Z 2 /A.5 ± 0. as a side of the solid target. where the Thomas-Fermi model of the atom has been assumed [29].1 122 thus taken varying the thickness of the solid target d for Ta and Pb (second four rows in Table I and Fig. is fulfilled. which accounts for except one where d = 4.1 ± 0. In !"#$%& this regime. which is sufficiently large for the present estimate.3 ± 0.7 ± 0.7 ± 0.2 0. 2). with f (x) = P ∞ 2 2 2 `=1 x /`(` + x ). The contribution of two-step process NT refers instead to the total yield of positrons with Ee+ > is expected to exceed the one of the trident process if 1MeV. . physics Monte-Carlo code FLUKA.3 0. In all of the predominance of a two-step process for the electro- cases. the positron yield is seen to scale with d2 Table I.3 3. Within this energy range.3 1. in all the considered cases.2 mm for Ta.8 78 upon the target thickness (Ne+ ∝ d for the trident and Ta 4.8 3.3 ± 0. positron number of (2. Keeping j as a free imum density is located at the close vicinity of the rear parameter. we obtain these data to be best fitted.a). cade processes can generally be neglected for an order- duce the experimental data well (dashed lines in Fig.1 0. with γ ≈ 300 being the relativistic factor of quantitatively confirmed by integrating the experimental the incoming electron beam (a larger number of cascade spectra in the range 90 < Ee+ (MeV)< 120 (see Table steps implies in general a broader angular distribution of I and Fig. (b) 6.4 2. A further indication as to what process domi. Lrad (Sn) = 12 mm.0 ± 0. the radiation refer to the energy window 90 < Ee+ (MeV)< 120 as obtained length Lrad with the range of target thicknesses d used from the experiment and FLUKA simulation.3 31 positron spectrum. and L0 = log(183Z −1/3 ) − f (Zα). see Supplementary Material).7 ± 0. electromagnetic cascades during the passage of an elec.2 3. and by including Coulomb corrections.3 2.7 ± 0.28) × 105 is obtained for the Due to the divergence of the positron beam. Pb 4.1 0.3 240 nates is given by the dependence of the positron yield Ta 1. d(mm) θe+ (mrad) Nexp ×105 Nsim ×105 NT × 105 one (see Fig.1 190 trend.1 ± 0. 3. the positron yield increases observed angular divergence of the positron beams of the for materials with higher atomic number.3 ± 0.3 ± 0.3 2. see first four rows in Table I).b). where n is the number of atoms per unit volume. Thus. In order to support this statement the. All the measured Lrad (Cu) = 15 mm.1 mm of Pb. for each material. For an order-of magnitude estimate of Lrad . This is also corroborated by the As theoretically predicted. The first four rows illustrate the results from targets in both cases. As seen by comparing IP (which was absent if no solid target was inserted in these values with those in Table I. the radiation length is approximately given by [29] Figure 2.3 2. are able to repro. Experimental (solid lines) and simulated (dashed 1 Lrad = . with same areal density. nesses are always such that the inequality d/Lrad & 10−2 showed a monotonically decreasing profile with approxi.3 ± 0. Lrad (Ta) = positron spectra. (1) lines) positron spectra for (a) 5.4 mm of 4α(Zα)2 nλ2C L0 Sn.2 3. Eq. 2).4 2.9 × 10−11 cm is the Compton wave- length. 3 d/Lrad & 10−2 [26].2 0.1 mm and Lrad (Pb) = 5.1 ± 0. it is S & 4 in the worst case of !"#$%& Cu.3 1.4 2.8 ± 0. in the experiment.3 0. Here. This trend is order of 1/γ.3 2. 3). Moreover.2 2.8 2. For an electron with energy ε emitting a photon !"#$%& !"#$%& with energy ~ω. the material thick- the electron beam path.6 ± 0.7 values for the materials employed in the experiment: g/cm2 . a maximum the final particles). (c) 2. we can assume here to be in the total-screening regime. is seen to follow a similar Ta 2.2 0. A series of shots was Pb 2.8 ± 0. (1) provides the following mass density was kept constant for each material (σ ≈ 4.3 3. its max- material with the highest Z (Pb).7 60 Pb 2. The positron yield Nexp and Nsim oretically we compare.2 0. we can roughly estimate ε ∼ ~ω ∼ 100 MeV. if j = 2.1 confirming the domi- nance of the cascade process with respect to the trident Mat. in agreement with the experimental indication mately 103 positrons/MeV (solid lines in Fig.8 mm of Ta and (d) 4. it is d < Lrad . 3.

in the energy range ulations show that such a positron beam co-propagates 90 < Ee+ (MeV)< 120. all charge of 1 nC. The high leptonic density would in fact be noted that the positron beam would be inherently allow for the study of the propagation of these jets in synchronised with the laser. FLUKA simulations of the interaction of a pencil- bursts associated with these structures (indication of the like monoenergetic electron beam (Ee− = 100 GeV. The dashed line represents the best quadratic fit. the generation mechanism proposed.#. velopments in laser technology. low-divergence achievable with our proposed generation However. (a) Measured positron yield. their composition ponentially decreasing energy spectrum (maximum en- would be similar to the jets reported here. in the conser- toroidal magnetic fields would be virtually zero.% means that the total amount of positrons with energy Ee+ > 1 MeV will be of the order of 3. indicating an overall positron density of about 2 × 1014 cm−3 . Lines give the ne ≈ 2 × 1015 cm−3 . as recently demonstrated 11 2 15 −3 −10 A/m (assuming ne ≈ 2 × 10 cm ) inducing an by Blumenfeld and collaborators [8]. by varying mechanism would prove fundamental for efficient injec- the electron beam characteristics and target thickness. Despite the different gen. for Ta (blue full circles) and Pb (red with an electron beam with an average density of about empty circles) for different target thicknesses. The normalised is possible to significantly modify this percentage.2 mm of Ta. constant areal density as a function of Z 2 /A. For 4. applied to the near-term de. The positron contribution on the best quadratic fits.1%BW). with approximately 106 positrons to their co-propagation with a high-density gamma-ray with energy between 70 and 80 GeV). for different materials but a null component of positive ions. allowing for the possibility much more rarefied plasmas. Di- the propagation of these jets in the interstellar medium. this !"#$%&'%(%'& . the interaction of a GeV-like laser-accelerated each monoenergetic component. by as- suming that the positron beam will retain the temporal !"#$% duration of the initial electron beam (τe− ≤ 30 fs. as εn ≈ γζθ. a divergence of beam of similar size and duration (FLUKA simulations the order of θ ≈ 10 µrad and an overall charge compa- indicate a gamma-ray brilliance of the order of 1019 −1020 rable to that of the incoming electron beam. It must also instability [11]. for instance. allowing vative case of a 100 µm source size. Particle- ph/s/mm2 /mrad2 /0. being ζ the electron beam with a thicker tantalum target (d ≈ 2 beam source size. in the en- leptonic beam will therefore be of the order of 10% with ergy range 90 < Ee+ (MeV)< 120. matter asymmetries in a highly non-linear regime. in a scenario comparable to of both electron-laser and positron-laser interactions. 4 transverse diameter of the order of 150 µm and. it tion of the positrons into such devices. [23]).-. rect comparison between these two experimental scenar- The proposed mechanism for the generation of ultra. thus reducing the normalised sity of the order of 1016 − 1017 cm−3 . This is comparable one to unveil the microphysics induced by the presence to the positron emittance measured after the injection of small-scale magnetic fields generated by filamentation stage of LEP (εLEP ≈ 60π mm mrad [33]). see Table I) the den- sity of positrons with an energy between 90 and 120 MeV . In the case of maximum yield (4. might also be relevant The authors acknowledge the funding schemes NSF . In the experimental results driven plasma wakefield accelerators can also be subse- reported here. ios might allow for the testing of possible matter/anti- relativistic positron beams.5 × 107 . an element in favor to the latter to the construction of all-optical electron-positron collid- is the power-law continuum spectra of the gamma-ray ers. (b) Measured positron yield. FLUKA sim- Figure 3. FLUKA simulations indicate that. over- jets being optically thin) [31]. Due to the ultra-relativistic nature of cm) is expected to generate a high-density. &#'(#)*+% a longitudinal length of the order of cτe− ≤ 9 µm.2 mm Ta. the emittance to be εn ≈ ζ ≈ 30π mm mrad.% is of the order of 2. In this case.3 × 1012 cm−3 . We compare now our experimental results with the electron-positron astrophysical jets. The extremely azimuthal magnetic field of the order of |Bφ | ≈ 30 T. less of the positron energy. also thanks ergy of Ee+ = 80 GeV. see [7]) with a 2 cm thick Ta target eration mechanism (pair production from gamma-gamma indicate the production of a positron beam with an ex- instead of gamma-nucleus collisions). FLUKA simulations indicate that this energy window contains approximately 1% of the total positron yield. γθ ≈ 1 rad regard- tral electron-positron beam with an overall leptonic den. Even though a de- bate is still open as to whether these jets are predom- inantly constituted by an electron-proton plasma or by electron-positron pairs. the excess of electrons in the beam implies quently employed to further increase the positron en- a net current density of the order of Je ≈ −(n− + e −ne )ec ≈ ergy in a metre-scale device. The overall positron beam intensity can thus be estimated to be of the order of 1019 erg s−1 cm−2 . purely neu. For emittance of such a positron beam can be expressed.

C. Leemans et al. Katkov. Glinec et al. Wang et al. (2007). 245 (2009). J. [18] M. 5 CAREER (Grant 1054164) and NSF/DNDO (Award No. 596... Nat. Gahn et al. C. 2662 (2000). Phys. C. J. 014802 (2004). K. 390. 12. [27] H. V. [32] G. [19] M. Appl. Rev. J. 526. 77. . Phys. Rev. H. Photon. pp. [16] L. Tanaka et al. Clayton et al. Beresteskii. Instrum. 2109 (2008). (2007). 575 (1999). Plasma Phys. 93. CERN-PS-90-56-LP (1990). Rinolfi. E. Stenflo.. Koch and J.. 105. B. A. 215001 [2] S. Rep. [24] F. Rev. Mon. Potier and L. Teoret Fiz. 3. Bossart et al. Phys. H. 7. Phys. Battistoni et al. 93. [9] P. Astrophys. P. Phys.. ST Accel. Mod. Gallant and A. [10] C. 045006 (2010). don Press. Rev. Hafz et al. Phys. 38 (1954). L121 (2003). Hoshino. 1177 (2012). I. P. 987 (2000).. 061301 (1959). 1954). [11] H. Fusion 51. Kneip et al. Rev. 314.. [21] C. M.. Nat. [14] A. grateful to A. Sarri et al. Astrophys. Proceedings 896. 6th European Particle Acceler- Phys. 88:2. [26] V. [23] S. M. Phys. 103. P. 88 (2008). Mod. N. 77. [6] X. Lu et al. A. Controll. Nat. 2. Rev. Rev. [29] V. Baier and V. Phys. [30] G. Mangles et al. Voitkiv for stimu. Carlson. Y. Shukla and L. [15] S. Keitel. 571 (2008).. Lundh et al. F021166). W. Quantum Electrodynamics (Butterworth-Heinemann. T. [31] M. Not. 426. 219 (2011). 31 [12] H. 405. R. Hatsagortsyan and C. Silva et al. shop.. lating discussions. Pis’ma v Zh. 2008). Soc.859-861 (1998). 105001 (2009). Arons. E. 015003 (2010).. Lett. Rev. Chen et al. Phys. ator Conference. Rev. Z. AIP Conf. 697 (1999). 065015 (2009). Müller. Rev. Blumenfeld et al. 105003 (2010). Plasmas 9. Gahn et al. Oxford. Muggli et al. New J. O. Èksper. Advanced Accelerator Concepts Work. [25] K. Keitel. Rohrlich and B. Woosley... [5] O. A. GS wishes to acknowledge the support from [17] K. Astrophys. Wiersema et al. Beams 10. E. 2. Phys. Lett. Lett.. Phys. K. 2 the Leverhulme Trust (Grant: ECF-2011-383). 454 (1992). 920 [7] W. Oxford. Milstein and to A. Lett. Astron. Lett. Medvedev and A. Sci. Pitaevskii... Rev. [3] W. 013507 (2005). B. (Claren- [8] I. [4] N. Stockholm. Di Piazza. Nature 445. Nat. Phys. B. 76. Chen et al. Motz. [1] R. Photon. Lifshitz and L.. Heitler The Quantum Theory of Radiation.. J. Phys.. 102. D. Yanovsky et al. 103301 (2006). Müller and C. 105.. 035002 (2009).. Lett. 696 (2006). P. Sci. 741 (2007). E. Instr. P. (2006). [22] Y. ADP is (2012). Sweden. [20] V. Astrophys. Dieckmann. Phys. Piran. Austin 2012.. M. [28] W. 31. Phys. Rev. Langdon. J. 84. 273 (1993). [33] J. Optics Express 16. [13] C. Loeb. 96. Lett.