1983 IEEE. Personal use of this material is permitted.

However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

IEEE Transactions on Nuclear Science, Vol. NS-30, No. 4, August 1983

SIMULATION OF PREACCELERATED PARTICLES IN THE PLASMA BEATWAVE ACCELERATOR D. J. Sullivan Mission Research Corporation Albuquerque, New Mexico 87106 Abstract One-dimensional relativistic and electromagnetic simulations show the Plasma Beatwave Accelerator to be a viable approach to obtaining large accelerating The plasma wave excited. by Raman Forward fields. Scattering has a longitudinal electric field on the order of 100 MeV/cm to 10 GeV/cm based on plasma densities of 1016 cm-3 to IOzo cm-3. The use of optical mixing to grow a plasma wave of the proper frequency above the noise level greatly reduces the laser intensity needed to drive the Raman instability. TeV energies for plasma electrons or preaccelerated charged particles in short distances appear feasible. Results where an injected electron beam is accelerated coherently from 1 MeV to 20 MeV in one millimeter are preTwo- and three-dimensional effects must still sented. be resolved. Introduction laser beams are capable of producing electromagnetic high fieids Volts/cm. The Plasma Beatwite Acceleri: 10 -10 tar'** is one 'of a number of particle accelerator concepts utilizing the high fields inherent in intense laser beams. In this case, EM waves interact nonlinearly with an underdense plasma through optical mixing or Raman Forward Scattering (RFS) to excite large amplitude$plasma waves. The plasma self-fields provide the acceleration. Thus, the concept is a collective effect accelerator. The longitudinal electrostatic field of the plasma depends on the plasma density and is typically on the order of mcwp/e, where m is the electron rest mass, e is the electron charge, c is the speed of light and wp is the electron plasma freFor a lo* cm-3 plasma this expression is quency. equal to 1 GeV/cm. It explains the very high accelerating fields obtainable with this mechanism. Because previous papers,3-6 cover the theory in detail, we will immediately discuss the simulations. Our results address optical mixing, Raman Forward wave synchronism, Scattering, streaming instabilities and frequency matching. Only the wave-wave processes will be reviewed here. A simulation which coinjects an electron beam with the laser pulse will be described in detail. Simulations Simulations were carried out using a two-dimenfully relativistic and electromagnetic parsional, The code can solve self-consisticle-in-cell code. tently for the time dependent trajectories of tens of thousand of plasma particles over thousands of plasma All variables are expressed in dimensionless periods. length is in units of c/up; time Therefore, terms. is measured in units of up-l; and particle velocity is given by vi = sir (i = 1, 2, 3), where wp iS the initial electron plasma frequency. In simulations on the Beatwave Accelerator two are launched electromagnetic waves plane-polarized Periodic boundary condiinto a Cartesian geometry. tions in the transverse (y) direction make configuraIn general, tion space effectively one-dimensional. the simulation has 1250 cells in the longitudinal (2) The cells direction modelling a length of 100 C/W,. Intense trgnsvfl;se

3159

in the y direction appeared uniform because of periodEach of these macrocells initially contained icity. The risetime of the laser pulses was 24 particles. The ions were taken to be an typically 25 up-l. infinitely massive neutralizing background except in Different simulations were made in which the one run. temvalues of otJup I wlbp, Eo and electron were varied where wo and W, are the perature, T,, * two 1 aser frequencies. a vacuum region of 10k C/W, long was Usually, left between the left hand boundary and the plasma in order to accurately determine the dynamics of laser When the laser pulse injection into the plasma. the nonlinear ponderbmotive encounters the plasma, force resulting from the intensity gradient causes the This continues until the force electrons to snowplow. arising from charge separation is greater than the and the electrons attempt to ponderomotive force, restore the charge imbalance by moving in the negative This motion initiates a train of large z direction. amplitude plasma waves. The one drawback evident in the single wavepacket This scheme's5 was the short pulse (r < 2~ up-'). implied an instantaneous risetime to very high intenIt is obviously unattainable. The beatwave sities. on the other hand, is a practical means of approach, obtaining the required intensity gradient necessary for the nonlinear ponderomotive force to establish The beatwave results large amplitude plasma waves. coherent EM sources w. and w1 from two parallel This can be accomplished by where w. - w1 = w . utilizing two separa ! e colinear lasers or exciting two appropriate bands of the same laser.3 Nonlinear Wave-Wave Processes

An alternative method of depicting the beatwave acceleration is in terms of two nonlinear wave-wave At low intensities where E. (0,l) < interactions. mco (0,1)/e optical mixing can result in large amplitude plasma waves.',* This occurs provided frequency of the elec "e ro- = Then, the beat wo * q. magnetic waves is in resonance with the plasma, and electron density fluctuations can grow. The result is linear growth of the plasma wave amplitude in time until relativistic plasma effects cause a frequency mismatch and wave saturation. The second interaction is the Forward Raman Scattering Instability.1*2,g As the laser waves transit the plasma, they undergo successive scatterings where IQ = ho - nwp. Alternatively, this may be This is a parametric viewed as k, = k, - nkp. three-wave process where a large electromagnetic wave interacts with a plasma to generate another forward moving electromagnetic wave and a plasma wave. Conservation of energy and momentum require the frequency and wavenumber matching conditions noted above. As is usual for parametric processes, the pump wave must exceed a certain intensity threshold for the instability to take place." The two wave-wave interactions, however, are not to low intensities mutually exclusive - one limited Rather, the plasma wave and the other to high. excited by optical mixing helps drive the parametric To examine this a simulation with vosc instability.

UO18-9499/83/0800-3159$Ol.OD~1983

IEEE

3160 (0,1)/c s eEo(O,l)/mcw(O,l) = .04 and .05 was conThe plasma was cold at WIT = 0. The linear ducted. growth of the'plasma wave is depicfed in Figure 1. At fluctuations are this time (up? = 150) the density 10% of the initial density. The analytical equation' indicates saturation should occur at a fluctuation 'time histories of the laser Indeed, level of 5%. fields indicate a cascade to lower frequencies in of an RFS instability. By steps of wp indicative field amplitude has saturated @PT = 300 the plasma the density variations are as high ds at .2 mcwp/e, and the RFS instability is .75 of the initial value, as expected, these field However, fully developed. levels are not sufficient to trap plasma electrons at rest. :::10 -1 I 1.00, I

-10.0 0.0

, 26.0

, 60.0

I 14.6

I 99.9

B d

103 101 10-l -3' 10 .n -6 l 0.0

s 3 d 3 a

z 0.60 3 2 0.0 -N -0.60, L-, w

-1.00 1 0.0 Fig. 1. ,Time longitudinal meters are mcwp/e. / 31 + I ;:I, plasma of the Simulation paraand E,(0,1)=0.2 L 112 I 160

9.8

19.6 w/wp

29.6

39.3

Fig.

2. Time history and the power spectrum inside the plasma of the longitudinal electric Simulation parameters are wp=10.6 wp, field. ~1=9.6 up and Eo(0,1)=5.0 mcwp/e.

history inside the electric field. 'ao=5.0 UP, wi=4.00p

T . 0.60 e s 0.00 7-I ~-0.60 -1.00' 0.0 ,* B 't f *s OI z a 10 n I 26.0 i 60.0

The importance of this result must be stressed. It indicates that large amplitude plasma waves can be grown from relatively low intensity laser sources, due Consider the to RFS excitation by optical mixing. cold plasma simulation just discussed. If wo is 2.36~10'~ set-l (X7;:;1ym),t$en A1 is 8 urn and the The equivalent laser cm- . plasma density is intensity is only 1.3~10'~ W/cm*, yet the plasma wave accelerating field is 166 &V/cm. This lower intensity makes the Plasma Beatwave Accelerator an extremely attractive candidate for The application to high energy physics accelerators. lower intensity implies the focal spot size may be larger, thus lessening diffraction of the laser pulse in the plasma. The fact that plasma electrons are not trapped is advantageous for injection of preaccelerAs has been shown in other ated bunched particles. particles lead to wave numbers of trapped work , large damping and loss of synchronism. At higher intensities the RFS instability grows Figure 2 depicts exponentially to a saturation level. the plasma wave amplitude and spectrun for the case of Note the presence of spectral vosc (0,1)/c - 0.5. componentsatw=wo+W1 aswellasti=wo-w = Graphic evidence of this wp and higher harmonics. instability is also given in Figure 3, which is the time history and power spectra of the laser waves inside the plasma.This downward cascade of energy into signature of Raman instamultiples of up is a clear The upward cascade is the result of multiple bility. four-wave interactions in which each wave is coupled to all other waves in a multiwave parametric process.ll The sidebands are often referred to as Stokes (downshifted) and Anti-Stokes (upshifted) lines. is particularly interestlny, This simulation because the parameters model a CO, laser emitting in the 9.6 pm and 10.6 urn bands at a combined intensity The laser pulse is incident on a of 1.3~10'~ W/cm*. Excitation of these CO, lines hot 1Ol7 cmq3 plasma. at this intensity and production of such a plasma is The higher saturation not beyond current technology. level of the electrostatic field is only a factor of

'

"' 'II' 1 74.9

" 99.9

IO6 IO3 101 0.0

9.8

19.6 w/w

29.6

39.3

Fig.

P 3. Time history and the power spectrum inside the plasma of the two Plane polarized laser wave electric fields. Parameters are wo=10.6 wp, ol=9.6wp and Eo(0,1)=5.0 mcwp/e.

three larger in this simulation than in the run where optical mixing was important. This implies a field of 188 #V/cm based on the lower plasma density. The combined laser intensities here, however, are equivalently 100 times larger. Nevertheless, there are some advantages to this higher intensity. The high value of v osc/c and its associated relativistic effects lead to particle trapping and acceleration of a small number of electrons. This also occurs if Te = 0 for the same parameters. The trapped particles originate at wavelength intervals where Byy is a maximum and are not associated with any overall temperature. These results are important if the plasma is to be the source of accelerated particles. In addition, the particle bunching at higher intensity is more pronounced and coherent. Ps a reference, assume the focal spot size for the laser pulse described above is 10m4 cm*. Then, based on simulation results, a particle pulse occurs every .18 psec, contains 5x10" electrons, carries a peak current of 96 kA at a current density of 960 Rgamps/cm*. The main reason for belaboring this point is to indicate the potential the plasma has for confining and accelerating high density bunches of preaccelerated ions.

3161

_ T 3 ii

1.00

0.00

7 \ 9 ;i 2 0.0 25.0 Plasma 50.0

15.0

0.0 -16.0 0.0 25.0 Plasma 60.0

74 p3 -1.00

Beam

Beam

%(C/WJ 1
Fig. 4. Plots of a) plasma wave electric field, electron transverse, e ) beam longitudinal, tion parameters are ~~-10.6 tip, wl=9.6 op,

AC/W p)
c) plasma electron b) laser electric field, and f) beam transverse momentum phase spaces E,(O,l) = 5.0 mcwp/e and Te = 0 kev. 4. longitudinal, at WIT = 74.0. d) plasma Simula-

A 1 t&V electron beam with a density of 1Ol4 cm-3 was coihjected with a 1016 W/cm CO, laser emitting at Thus, the beam to plasma density 10.6 pm and 9.6 urn. As before, a The plasma was cold. ratio was -001. of plasma electrons are trapped and small number accelerated resulting in a monotonically decreasing Because the electron beam was not distribution. its accelerated emittance is bunched at injection, Notably, the beam particles bunch at a phase large. in the wave which is different from the plasma elecYet no streaming instability results. tron bunches. Plots of various quantities of interest are given in The total longitudinal distance is 0.8 mm Figure 4. A more conclusive test for the given plasma density. using prebunched, monoenergetic ions at a laser intensity which does not allow plasma trapping is needed. If this indicates a problem, we will lower the ratio of beam to plasma densities to determine at what point These studies will also the growth is negligible. Both issues will help address emittance growth. establish the expected beam luminosity as it exits the accelerator. Acknowledgement This work was supported by the Air Force kapons The author is pleased to acknowledge many Laboratory. useful discussions with Drs. B. Godfrey, C. Joshi, A. Sessler, T. Tajima and W. Willis.

T. Tajima and J. M. Dawson, in Laser Acceleration 11 (A of Particles, edited by P. J Ch Conference Proceedings, No. 91,' Newa?iFk, '1982fy p. 69. D. J. Sullivan and B. B. Godfrey, Nut. Sci. NS-28. 3395 (1981). and B. J. Sullivan Acceleration of Particles, nannell (AIP Conference New York: 1982), p. 43. IEEE in by Trans. Laser P. J. No. 91, Watson, Lett.

5. 6.

D.

B.

Godfrey, edited Proceedings, K.

7. 8. 9.

A. N. Kaufman, and B. I. Cohen, Phys. Rev. Lett. 29-, 581 (1972). M. N. Rosenbluth -29, 701 (1972). and C. S. Liu,

M. Rev.

Phys.

C. Joshi, T. Tajima, and N. A. Ebrahim,

J. M. Oawson, H. A. Baldis, Phys. Rev. Lett. 3, 1285

(1981). 10. 11.

K. Estabrook, W. L. Phys. Rev. Lett. 45-, D. Eimerl, Phys. Rev. Kruer, and 8. 1399 (1980). F. A. Lasinski, Pasiner,

R. S. Hat-grove, and Lett. 46, 651 (1981).

J.

References 1. T. Tajima and J. 267 (1979). M. Dawson, Phys. Rev. Lett. 5, NuC.

2. 3.

T. Tajima and J. M. Dawson, Sci. NS-28, 3416 (1981). in Laser C. Joshi, edited by P. J Proceedings, No. 9i,

IEEE

Trans.

Acceleration of Particles, (AI Conference 11 Nei"!zrk: 1982)', p. 28.

Ch