Spectral modification of shock accelerated ions using hydrodynamically shaped gas target
O. Tresca, N. P. Dover, N. Cook, C. Maharjan, M. N. Polyanskiy, Z. Najmudin, P. Shkolnikov, I. Pogorelsky
SSpectral modification of shock accelerated ions using hydrodynamically shaped gastarget
O. Tresca , N. P. Dover , N. Cook , C. Maharjan ,M. N. Polyanskiy , Z. Najmudin , P. Shkolnikov , and I. Pogorelsky Accelerator Test Facility, Brookhaven National Laboratory, Upton, New York 11973, USA John Adams Institute for Accelerator Science, Blackett Laboratory,Imperial College London, SW7 2BZ, United Kingdom and Stony Brook University, Stony Brook, New York 11794, USA (Dated: September 20, 2018)We report on reproducible shock acceleration from irradiation of a λ = 10 µ m CO laser onoptically shaped H and He gas targets. A low energy laser prepulse ( I (cid:46) Wcm − ) wasused to drive a blast wave inside the gas target, creating a steepened, variable density gradient.This was followed, after 25 ns, by a high intensity laser pulse ( I > Wcm − ) that producesan electrostatic collisionless shock. Upstream ions were accelerated for a narrow range of prepulseenergies ( >
110 mJ & < (cid:38) µ m), broadband beams of He + andH + were routinely produced, whilst for shorter gradients ( (cid:46) µ m), quasimonoenergetic accelerationof proton was observed. These measurements indicate that the properties of the accelerating shockand the resultant ion energy distribution, in particular the production of narrow energy spreadbeams, is highly dependent on the plasma density profile. These findings are corroborated by 2DPIC simulations. The light and thermal pressure associated with intenselasers causes compression and heating when incident onan overdense plasma. This can generate a piston thatlaunches electrostatic collisionless shocks into the plasma[1]. Upstream ions can be reflected off the shock creatingan ion population accelerated to twice the velocity of thedriving shock [2–5]. This mechanism has been demon-strated in recent experiments using intense CO lasersinteracting with gas jets [6, 7] where multi-MeV protonbeams with energy spread smaller than 4% have beenreported.Gas jets have also been proposed as a source of highpurity, high-Z ion beams. This is in contrast with themultiple species beams generated from solid targets viasheath acceleration [8, 9]. For solid targets, protons fromsurface contaminants are preferentially accelerated due totheir higher charge-to-mass ratio, making the productionof impurity free high-Z beams challenging. Generatinghigh purity helium ion beams from gas jets would be ofinterest for nuclear and medical physics applications [10].Laser driven longitudinal [11] and transverse [3, 12, 13]acceleration of helium ions from laser irradiation of gasjets has been observed experimentally. However, theseprevious studies were conducted in underdense plasmasrequiring laser intensities > Wcm − . Collisionlessshock acceleration, on the other hand, allows the produc-tion of directed beams at lower intensities with sizeablenumber [6], and with favourable intensity scaling [5].It is thought that the generation and subsequent prop-erties of shock accelerated ion beams are highly depen-dent on the initial plasma density distribution [5, 14].Gas jet targets used for laser-plasma interactions typi-cally have initial density scale lengths l > µ m, whichis unsuitable for efficient shock generation. In previous experiments using intense CO lasers, this profile wasmodified by the pulse train inherent to the laser system[6, 7], which is difficult to control and reproduce. Re-cent breakthroughs in CO laser technology have madesingle pulses of intense radiation [15] possible. This al-lows reproducible interaction of the laser with the target.However it requires a different method to modify the den-sity profiles. One method investigated previously is theuse of optically generated hydrodynamic blast waves vialaser solid interaction [16, 17].In this paper, we demonstrate the use of a controlledlow energy prepulse focused in dense gas to generate blastwaves that hydrodynamically shape the gas targets tomake them suitable for ion acceleration. Interaction ofthe high intensity laser pulse with the modified plasmadensity profile leads to reproducible shock acceleration.In particular, we present the first measurements of shockaccelerated, > laser beam, λ = 10 . µ m, withenergy up to 11 J per pulses of 5 ps full-width-half-maximum (FWHM), provided a peak power of 2.2 TW.An f / w =65 µ m, resulting in peak intensity I = 2 . × Wcm − ( a = 1 . ∼ µ m above the the nozzle. At this po-sition the initial gas density profile, along the laser axis, a r X i v : . [ phy s i c s . p l a s m - ph ] M a r was near triangular with a linear gradient of ∼ n e ∼ n c [18].The ion beam was characterized using a laser-axisThomson parabola spectrometer, opening angle of 2 × − sr, and detected with BC-408 polyvinyl-toluenescintillator coupled to an EMCCD camera. The ion-to-detected-photon yield was calibrated using the Tandemproton accelerator at Stony Brook University [19]. Thecalibration was scaled to He + by introducing a scalingfactor of 0 . ± . E pp ,up to 1 J in 5 ps. The resulting plasma density distri-bution was characterized using optical probing. Two,time-adjustable, ∼
10 ps long, λ = 532 nm synchronizedpulses were used for shadowgraphy and interferometry.They were set to allow characterization just before andafter the drive pulse on a single shot.Fig. 1a shows the experimentally observed plasma den-sity map 100 ps before the main laser plasma interaction(LPI) using He, similar images are found using H . Forthis shot, a prepulse E pp ≈
220 mJ was incident on he-lium gas of peak neutral density n He = 0 . × cm − ,corresponding to a fully ionized plasma density n e =1 . n c . The plasma density was obtained from interfer-ometry assuming cylindrical symmetry around the laseraxis, thus allowing Abel inversion. The optical path dif-ference due to the unionized gas is small compared tothat of the plasma and is therefore neglected. Plasmaformation was observed over a large volume with a peakdensity significantly lower than the fully ionized density,and shows no structure. However, probing the plasma300 ps after the LPI revealed the characteristic high den-sity shell of a blast wave (fig. 1b). The shell is observedto expand at velocity < ms − . Hence the dramaticchange in the plasma appearance cannot be due to hy-drodynamic motion originating from the main pulse in-teraction.The effect of the prepulse was modelled using the flash hydrodynamics solver in 2D cylindrical geometry.Helium was initialized following the experimental gas jetprofile with n He = 0 . × cm − . The prepulse energywas modelled by depositing thermal energy evenly intoa cylinder of gas at the laser focus (length 600 µ m andradius 80 µ m) centered near the initial critical densitysurface; n He = n c . Within 1 ns, a collisional blast waveis formed and expands out into the cold gas [21, 23]. Theblast wave forms a cavity with low density but high elec-tron temperatures at the center, surrounded by a coldbut high density shock moving outwards. Fig. 1c showsthe electron density at 25 ns after the prepulse, calculatedfrom the Saha equation, for an initial 3 mJ of energy ab-sorbed in the plasma. Only the center of the cavity ishot enough to be ionized and detectable by interferom-etry, in agreement with the experimental measurements c) d)a) b) z (mm)0 1 2 r ( mm ) -.8-.40.4.8 n He (n c )012z (mm)0 1 2 r ( mm ) -.8-.40.4.8 n e (n c ) 0.05.1 300 ps after pulsez (mm)0 1 2 r ( mm ) -.8-.40.4.8 012n e (n c )Before intense pulsez (mm)0 1 2 r ( mm ) -.8-.40.4.8 n e (n c )0.05.1Experiment ExperimentHydro simulation Hydro simulationw FIG. 1. Electron plasma density from He targets, n e frominterferometry: (a) immediately before, and (b) 300 ps afterthe LPI, for E pp ≈
220 mJ. (c) n e and (d) total helium atomicdensity (irrespective of ionization), n He from hydro simula-tions after 25 ns of expansion for 3 mJ absorbed energy. prior to the LPI in fig. 1a. Inspection of the neutral he-lium density (fig. 1d) reveals the blast wave, which has acavity wall of peak density given by that of a strong shock( γ + 1) / ( γ − n i , where n i is the ion density and γ , theratio of specific heats (5/3 for He, 1.4 for H )[21]. There-fore, using hydrogen with comparable laser parametersgives a higher density enhancement as well as a reducedscale length [22]. Even though this neutral profile exists a)b) y ( mm ) y ( mm ) FIG. 2. Interferograms 300 ps after the LPI with He; (a)without prepulse, (b) with prepulse ideal for ion acceleration, E pp ≈
150 mJ, (c) and with prepulse too large for ion accel-eration, E pp ≈ .
27 J. (d-f) corresponding n e /n cr . Main pulse a E pp ( J ) I on s / ( M e V s r) H , E pp =4 mJH , E pp =23 mJHe, E pp =150 mJa) b) FIG. 3. (a) Average flux of accelerated helium ions (/sr) > . E pp and main pulse a . Thewhite region indicates laser parameters tested experimentally.(b) Proton spectra for a H target with E pp ≈ + spectrum for a He target with E pp ≈
150 mJ (solidlines). The dotted lines are the detection thresholds, and thedashed red line is the deconvolved spectrum for E pp ≈ before the LPI, it is only due to ionization by the fastelectrons generated during the main interaction that itcan be directly observed. Once ionized, the blast waveprovides the ideal small scale-length plasma in which togenerate collisionless shock waves.Interferograms for three different prepulse levels withhelium gas are shown in fig. 2a-c, with corresponding elec-tron densities in fig. 2d-f. The images are all taken 300 psafter the LPI to reveal the full blast wave structure. Sim-ilar images are obtained with H . To account for the ver-tical density gradient in the initial gas density, azimuthalsymmetry is assumed in each half-cylinder above and be-low the laser axis. These regions were processed individ-ually by Abel inversion, with continuity assumed at theirinterface.With no prepulse (fig. 2 a & d), a small cavity of di-ameter ≈ µ m forms around the laser focal position.The intense laser has self-focussed and channelled partway into the long density scale length ( l ≈ E pp ≈
150 mJ (fig. 2 b& e) shows a significant difference in the plasma distri-bution. A blast wave has been generated, so that theintense pulse interacts with a profile with steeper densitygradient, ≈ µ m, and higher peak density. Energeticions were consistently generated in this regime. For fur-ther increase in E pp (fig. 2 c & f), the prepulse inducedblast wave propagates deeper into the jet and ultimatelythrough it, leaving only an underdense remnant alongthe laser axis. For this higher prepulse, no ion beam wasobserved. Probing in this case, showed little change tothe shape of the plasma before and after the LPI, exceptfor a doubling of the density in the walls, consistent withionization from He + to He .Control of the helium density profile generated by theblast wave is essential for reproducible ion acceleration. Fig. 3a shows the flux of detected He + ions (color scale)as a function of main pulse normalized vector poten-tial, a , and prepulse energy, E pp . Accelerated ions wereonly observed over a narrow range of prepulse energies,110 mJ (cid:46) E pp (cid:46)
220 mJ, showing the importance of op-timizing the density profile as in fig. 2 b & d. Within thisrange of E pp and a o ≈ .
4, the total flux of acceleratedions remains stable shot-to-shot at ≈ ions/sr.Fig. 3b shows a He + spectra obtained with E pp =150 mJ and a main pulse a = 1 .
4, resulting in a broadenergy distribution up to 1.5 MeV. This spectrum wastypical for all the observed beams. Only He + ions areobserved although the main pulse laser intensity exceedsthe threshold value to double ionize helium via field ion-ization ( I (cid:38) × Wcm − ). The observed single chargestate is the result of charge transfer as the ions traversethe plasma and un-ionized gas [12, 24], and recombina-tion as the accelerated ion beam co-moves with a lowtemperature electron cloud to the detector.For fixed n i and absorbed energy E abs , a blast wavein hydrogen will give a steeper density ramp than forhelium due to the lower γ [22]. Spectra from shotswith H for different prepulse levels E pp = 4 mJ and23 mJ are also shown in fig. 3b. Note that a signifi-cantly smaller prepulse was required for ion generationwith hydrogen; observation of the relative blast wavesize implies a significantly higher energy absorption. For E pp = 4 mJ, a quasi-monoenergetic beam is observedwith peak energy ≈ . E/E ≈ ≈ . E pp = 23 mJ. As theradius for a spherical blast-wave r bw ∝ ( E abs /n i ) / [21],lower E pp results in less expansion, providing a steepertarget density gradient. Not only does this techniqueallow the production of reproducible shock acceleratedions, but for sufficiently steep gradients spectral shapingis also achievable.The LPI was modelled using the 2D PIC code epoch .The plasma was initialized with 30 particles per cell ofcold ions (He or H + ) and electrons, in cells of size λ/ a = 1 . w = 65 µ m at focus, pulselength τ = 5 ps (FWHM), and was linearly polarizedin the transverse plane. The initial density distributionwas varied to investigate the effect of the optical densitytailoring.To model the LPI with no prepulse, the initial densityprofile was set to rise linearly from 0 to 2 n c over 800 µ mwith He . The laser self-focuses and bores a channelin the plasma, shown in fig. 4a. All the laser energy isexpended in the underdense region and no forward shockis formed, although lower velocity transverse electrostaticshocks are observed [3].The effect of the prepulse was simulated using a plasmadensity distribution extracted from the interferometry for E pp ≈
150 mJ (green box in fig. 2b). The profile wastaken along a line radially off-set from the laser axis inorder to reduce errors introduced by the Abel inversion.In the blast-wave ramp the density is well approximatedby n = n max e − ( z − z ) /l exp , where n max ≈ n c is the peakdensity, z is the position of the peak and l exp ≈ µ m isthe scale length. In this simulation, the laser pulse showsa modest increase in intensity due to self-focussing in theramp, but the short ramp length means the laser is notadversely absorbed. The pulse penetrates up to the criti-cal surface where a combination of radiation and thermalpressure launches a collisionless shock (fig. 4b). Ions arereflected at the shock front to twice the shock speed v s .The result agrees well with optical shadowgraphy 300 psafter the LPI (inset of fig. 4b), which shows remnantsof the accelerating collisionless shock emerging from thecollisional prepulse induced blast wave.The z − p z phase space (fig. 4c) demonstrates that theions originate from the position of the shock, but alsothat the ions are reflected with a large energy spread.Fig. 5a shows the position of the shock front and criticalsurface on the laser-axis as a function of time for this sim-ulation. The initial shock velocity v sh ≈ × ms − ex-ceeds the measured hole-boring velocity v hb = 3 . × ≈ (cid:112) I L /n i m i c [26, 27]. Ions reflected off the shock at itspeak velocity would gain ∼ . ∼ . ∼ b)c) d)a) 021 630t = 12 ps t = 12 psz-p z ROI r ( μ m ) -100-50050100 z ( μ m)0 200 r ( μ m ) -80-4004080 z ( μ m)0 100 200z ( μ m)80 120 160 μ m t = 400 ps z ( μ m)50 100 150 200 p z / m H ( m s - ) x10 , l exp = 20 μ m126 p z / m H e ( m s - ) t = 12 ps0 He, experimentally measured profile, l exp ≈ μ m He, experimentallymeasured profileHe, no prepulse400 FIG. 4. Results of PIC simulations. Ion density map 12 psafter LPI for: (a) no prepulse, (b) experimental initial profilefor E pp = 150 mJ obtained from fig. 2e. (inset is optical shad-owgraphy showing shock remnants) and (c) z - p z phase space.(d) The z - p z phase space for for hydrogen plasma with expo-nential ramp ( l exp = 20 µ m) I onnu m be r [ a r b ] Time (ps) P o s i t i on ( μ m ) Energy (MeV/u)4 8 12801001207090110 5.1 μ m ps -1 μ m ps -1 H shockHe shockHe n e = 1 n c a) b) 0 0.5 1 1.510 H, l = 20 μ mDetect lim (exp)He, no prepulseHe, prepulse10 μ m ps -1 FIG. 5. (a) Critical surface (red) and shock (blue) positionwith time for simulation shown in fig. 4b&c, and shock po-sition for hydrogen simulation in fig. 4d. (b) Comparison ofHe spectra for 800 µ m linear ramp (no prepulse) (green),prepulse fig. 4a&b (blue), and for hydrogen with exponentialprofile, l exp = 20 µ m (black). reflecting ions, resulting in the large momentum spread infig. 4b and the corresponding broadband axial ion spec-trum in fig. 5b. The shock deceleration is mainly dueto spherical expansion of the front. However, the ex-periment showed spectrally peaked proton beams wereproduced from sharper density gradients using hydrogen.Density profiles with an exponential ramp, and variablelength scale, were used to reproduce the sharper profilesachievable with hydrogen. At the edge of the blast-wave,the density dropped to n max /
6, the density drop for astrong shock, imitating the experimental profile.A scale length l exp = 50 µ m produces a broadbandbeam, as described for helium. However, shortening thescale length to l exp = 20 µ m results in the shock breakingthrough the density discontinuity at the blast-wave front.The sudden change in thermal pressure ahead of the fronttriggers the shock immediate dissipation, just at the endof the LPI (fig. 5a). The axial z − p z phase space, fig. 4d,demonstrates the formation of a single bunch with veloc-ity near 2 v s ≈ × m/s. Particle reflection stops andthe peaked spectrum is maintained (fig. 5d). This gen-eration of peaked spectra differs from thermally drivenshocks in isothermal plasmas [5, 7], where reflection at auniform speed persists over a longer time scale. Furthersimulation with l exp < µ m with helium also demon-strated spectrally peaked beams. Producing such gra-dients in helium would be experimentally achievable byshortening the time between the prepulse and the drivepulse, which was not possible with our experimental ge-ometry.In this work, we have observed shock acceleration ofprotons and helium ions. The initial plasma density pro-file is found to be critical to achieve not only shock accel-eration but also spectral control of the ion beam. Usinghydrogen, we found that long density gradients lead tothe production of broadband shock accelerated beams,while steeper gradients allow for the generation of quasimonoenergetic beams. Helium ions have also been ac-celerated using this scheme, demonstrating that it couldbe extended to other high Z gaseous species, providing aroute towards a wide variety of easily replenishable high-repetition rate ion sources for future nuclear physics ap-plications.Work supported by the US DOE Grant DE-FG02-07ER41488, and UK EPSRC grant EP/K022415/1. flash was developed by the DOE NNSA ASC and NSF-supported FCCS at the U. Chicago. epoch develop-ment was supported by EPSRC grants EP/G054940/1,EP/G055165/1 and EP/G056803/1. Computing re-sources provided by Imperial College HPC services andNERSC (mp1401), supported by DOE Contract DE-AC02-05CH11231 and BNL/LDRD No. 12-032. [1] J. Denavit, Phys. Rev. Lett., , 3052 (1992).[2] L. O. Silva et al., Phys. Rev. Lett., , 015002 (2004).[3] M. Wei, et al., Phys. Rev. Lett., , 155003 (2004).[4] L. M. Chen et al., Phys. Plasmas, , 040703 (2007).[5] F. Fiuza et al., Phys. Rev. Lett., , 215001 (2012).[6] C.A.J. Palmer et al., Phys. Rev. Lett., , 014801(2011).[7] D. Haberberger et al., Nat. Phys., , 95-99 (2012).[8] E. L. Clark et al., Phys. Rev. Lett., , 1654 (2000). [9] R. Snavely et al., Phys. Rev. Lett., , 2945 (2000).[10] H. Daido et al. Rep. Prog. Phys. , 056401 (2012).[11] L. Willingale et al., Phys. Rev. Lett., , 245002 (2006).[12] F. Sylla et al., Phys. Rev. Lett., , 115003 (2012).[13] A. Lifschitz et al., New J. Phys., , 033031 (2014).[14] N. Dover et al., arXiv:1205.4558.[15] M. N. Polyanskiy et al., Opt. Express, , 7717 (2011).[16] C. T. Hsieh et al., Phys. Rev. Lett. , 095001 (2006).[17] D. Kaganovich et al., Phys. Plasmas, , 120701 (2011).[18] Z. Najmudin et al., Phys. Plasmas, , 056705 (2011).[19] N. Cook, O. Tresca, and R. Lefferts, JINST, Luminescent Materials (Springer, Berlin, 1994).[21] Y. B. Zel’dovich and Y. P. Raizer,
Physics of ShockWaves and High-Temperature Hydrodynamic Phenomena (Academic Press, New York, 1967).[22] L. I. Sedov,
Similarity and Dimensional Methods in Me-chanics (Academic Press, New York, 1959)[23] Yu. P. Raizer, Sov. Phys. Usp.,
650 (1966).[24] A. Itoh, M. Asari and F. Fukuzawa, J. Phys. Soc. Jpn., , 943 (1980); A. Itoh, K. Ohnishi and F. Fukuzawa, J.Phys. Soc. Jpn., , 1513 (1980).[25] G. Sorasio et al. Phys. Rev. Lett, , 045005 (2006).[26] S. C. Wilks et al., Phys. Rev. Lett., , 1383 (1992).[27] A. P. L. Robinson et al. Plasma Phys. Control. Fusion,51