c. e. clayton et al- demonstration of plasma beat wave acceleration of electrons from 2 mev to 20...

8/3/2019 C. E. Clayton et al- Demonstration of Plasma Beat Wave Acceleration of Electrons from 2 MeV to 20 MeV

1/5

Demonstration of Plasma Beat Wave Acceleration ofElectrons from 2 MeV to 20 MeVC. E. Clayton, K. A. Marsh, M. Everett, A. Lal, and C. Joshi

Department of Electrical EngineeringUniversity of California at Los Angeles56- 125B Engineering IVLos Angeles, CA 90024We describe the results from recent experiments1 on theplasma beat wave accelerator (PBW A) scheme at UCLA. Arelativistic electron plasma wave (which is the acceleratingstructure) is resonantly excited in a plasma by the beating oftwo co-propagating electromagnetic waves (obtained from aCO2 laser operating simultaneously on two wavelengths). A

2 MeV, 200 mA (peak-current) electron beam, roughly 1 nsec(FWHM) in duration is used as a source of test particles formeasuring the longitudinal fields o f the plasma wave whichitself is moving with a relativistic Lorentz factor of about 34.Accelerated electrons are energy-selected with an imagingsector magnet and detected simultaneously with a cloudchamberand surfacebarrier detectors. Initial experimentsshowthat electrons are acceleratedup to 20 MeV over roughly 1 cm(the uniform length of plasma) indicating an gradient ofaccelerationof more than 1.8 GeV/m.

(eV/cm) for n, expressed in electrons/cm3. This gives 1


2/5

plasma/plasmawave are given in Table 1. An important pointto note is that the Rayleigh range is about 1 cm and this willbe one of the limiting factors in the experiment. The electricfields o f the wave are probed directly by injecting electronsfrom an electron linac.7 The linac parameters are alsosummarized n Table 1. The electrons are focused through aVacuumchamber PlasmaA/ source

Two-frequencylaser beam

\Electron Focusing lenslinac mirror Electroh /diagnosticsFigure 2: Cartoon overview of the main experimentalapparatus.small hole drilled in the large CO2 laser focusing m irror,allowing the electrons and laser to propagate collinearly as isnecessary or the experiment. The electrons arc focused to aspot of about 260 p.m diam which is smaller than the laserspot allowing good coupling of the electrons to the plasmawave. A double-focusing sector magnet selects he energy ofthe accelerated electrons which are detected by the electrondiagnostics. These will be discussed n more detail in a latersection.

0.15L I I , 1 I A

-300 -200 -100 0 100 200 300Time (psec)Figure 3: Bottom graph shows a calculation of the plasmawave amplitude vs. time for our laser parameters and the laserrising at t=Opsec. The model cannotpredict what happens laterin time due to ion instabilities. The top image is an actual streakof Chercnkov light from the linac electrons with the same timescale as the graph.

MODELING OF THE EXPERIMENTWith the laser parameters isted in Table 1, one can modelthe expcriment8 to find out what sort of dynamics to expect inthe experiment. For example, the plasma formation can bemodeled using the proven tunneling ionization theory9 and the

Table 1: Experimental parameters for the laser, plasma,na wave, and electron injector.LaserSourceWavelengthsEnergy per line (typ.)Spot radius w.Rayleigh range 2zoElectron quiver vel.Pulse risetimePulse FWHM PlasmaSourceDensityGasPlasmaperiod vpvlPlasmawave wavelengthLorentz factor yphAccel. gradient for 11% wave

CO2 laser10.591, 10.289tun6OJ, 15J150 pm1.3 cm0.17,0.07150 psec300 psecTunnel ionization8.6 x 1015 cmv3Hydrogen1.2 psec360 t.trn341 GeV/m

ElectronsSource RF LINACEnergy 2 MeVPeakcurrent 200 mAEmittance 6n mm-mradFocusedspot radius 130 f.tmRF frequency 9.3 GHzMicropulse separation 107 psecElectrons per micropulse 1.7 x 107Micropulse length 10 psec

time-dependentdensity one thcrcby calculates can be insertedinto a computation of the plasma wave growth. The result ofsuch a modeling is shown in Fig. 3. The plasma reaches ullionization at best focus after about 25 psec into the 150 psecrise of the laser pulse. At this point, the plasma wave beginsto grow until relativistic saturation occurs after about 70 pscc.This particular calculation was for wpe = Am If one sets 6.1~~slightly above &I, the relativistic saturation is delayed and apeak wave amplitude of 40% can be expected.lOThe timing of the CO;! laser and the 1 nsec FWHM linacmacropulse is set to within flO0 psec. However, due to thetechnique by which the CO;! pulse is generated, it is notpossible to phase he exact micropulse timing to the timing ofthe peak fields in the plasma wave. A typical streak of theelectrons (from Chercnkov light) is shown in the image abovethe graph in Fig. 3. One can see that there will be roughly a50% probability that a micropulse will interact with a field ofat least half the maximum fields of the plasma wave. Theprobability of interacting with higher and higher fieldscontinues to drop until there is an approximately 10% chanceof a micropulse interacting with the peak plasma wave fields.

DETAILED EXPERIMENTAL SETUPA more detailed view of the experimental setup s shown inFig. 4. The CO2 laser and the electrons are both focused nto

2552PAC 1993


3/5

the plasma which is located at the center of a cross piece in thevacuum chamber. The back- and forward-scatteredCO2 lightas well as the Thomson scatteredprobe beam are indicated byarrows. These hree scattering diagnostics comprise he opticaldiagnosticsof the wave amplitude mentioned earlier.

budCtamh

Figure 4: Detailed view o f the experimental setup.Injected electrons which do not interact with the plasmawave follow a small-radius trajectory and are dumped in aplastic beam dump as shown in Fig. 4. Electrons which areaccelerated ollow a larger radius trajectory and are detected none of the electron diagnostics: a silicon surface barrierdetector SBD) or a diffusion cloud chamber.The SBD, along with a preamplifier, produces a signal ofabout 20 mV per electron and can thus easily detect singleelectrons. It also has the advantageof providing a quantitative

measurement f the number of electrons over a large dynamicrange (by changing the preamplifier gain). But it has onemajor disadvantageand that is its sensitivity to x-rays. In fact,one cannot discern the difference betweenan electron signal andan x-ray signal except through the statistics of many data shotsand null tests. For the experimental setup shown in Fig. 4,the contribution from x-ray noise varies from about 30electrons worth of signal when the electron spectrometer s setto 3.5 MeV down to < 1 electron worth of signal when thespectrometer s set to 20 MeV.The other electron detector is a simple, home-madediffusion cloud chamber11 which uses a methanol bath at dryice temperature n a chamber of 1 ATM of dry air. A piece offelt wicks the methanol up to the top of the chamber and themethanol vapor falls back into the bath, going through aregion of supersaturation. The electrons ionize the air and themethanol condenseson the ions along the electrons path. Thechamber s shielded n lcad and surroundedby a coil which canbe energized to provide a = 260 G vertical magnetic fieldthroughout the active region of the cloud chamber. The tracksarc recorded by frame-grabbing CCD camera. The bigadvantage of the cloud chamber aver the SBD is that it isessentially immune to x-ray noise. Although x-rays do

produce some onization of the gas and thus visible tracks, thetracks are typically short and kinked as they are due to lowenergy (~50 keV) electrons which are kicked out by the x-rayabsorption in the dry air. The actual signal due to acceleratedelectrons should appearquite distinct from these racks. Theyshould be straight tracks, traversing the entire field of view ofthe camera and they should appear to originate form theaperture n the chamber.EXPERIMENTAL RESULTS

Null testsIdeally, the presence of signals on the two electrondetectors should only occur if the three main components ofthe experiment are present. These are; (1) the plasma, (2) theinjected electrons, and (3) the beat wave. If any of these areturned off, we would expect that there would be no signal onthe electron detectors unless some other physics is occurringbesides beat wave acceleration. The result of these null testsare summarized n Table 2. As indicated in the third column,we seeno evidence of acceleration by the laser beam tself, o f

acceleration of non-injected electrons (from for example,instability-heated electrons), or of spontaneousgeneration ofan accelerating structure through some mechanism other thanbeat wave (as n, for example, Raman orward scattering).Cloud chamber data

When the plasma, external electron source, and two-frequency laser are fired up simultaneously in about 140 mT o fhydrogen gas, racks are observed n the cloud chamberas seenin Fig. 5. For the image in Fig. 5(a), the electronspectrometer was set to direct 5.2 MeV electrons into theaperture of the cloud chamber and the magnetic fieldsurrounding the cloud chamber was off. The tracks have thecharacteristics of high-energy electrons discussedearlier; thatis, straight lines, long range, and directionality. The scatter nthe angles is probably dominated by scatter in the 25 pmMylar vacuum-exit window and the = 6 cm of 1 ATM airbetween the vacuum window and the beginning of the field ofview. Subsequent calibrations indicate that Fig. 5(a) isTable 2: Summary of the key null tests performed.

ConditionLinac & 2-frequencylaser but no plasma(chamberevacuated).2-frequency laser &plasma but no linac(linac beamblocked).

Single-line laser &linac & plasma (nosecond requency).

ResultOnly usual*x-ray noise.No signalat all$.

Only usual*x-ray noise.

No trapping ofbackgroundplasma electronsor SRS-generatedtail.No substantiallevel of stimulatedRaman forwardscattering.* Usual x-ray noise - I electronsworth of signal.$ No s ignal => less than 1 electrons worth of signal.

ConclusionNo acceleration bylaser only.

2553PAC 1993


4/5

composed of about lo3 tracks.12 For Fig. 5(b), the externalmagnetic field was energized causing the electron trajectoriesand therefore the tracks to bend. For this shot, the electronspectrometer was again set to 5.2 MeV as in Fig. 5(a). Animage with a small number of tracks was chosen in order tosee ndividual tracks. Three theoretical trajectories calculatedfor 5.2 MeV electrons are overlaid onto this image. The goodagreement ndicates that the electrons were within 10% of 5MeV; i.e., 2.5 times higher energy than the injection energyof 2 MeV.

(a) apertureAcceleratedelectrop track Low energynoise

Calculat& trajectoryd- 7.25 cm -b( wFigure 5: Images of tracks in cloud chamber w ith the electronspectrometer set to send 5.2 MeV electrons into the aperture ofthe cloud chamber. (a) No external magnetic field. Relativelocation of the aperture is shown. (b) With 260 G external field.Calculated trajectories are overlaid as the thin white curves.

SwjCace arrier ahector dataWhile the cloud chamber dramatically confirms theexistence of acceleratedelectrons, it is not a very quantitative

diagnostic. The SBD, however, is quantitative since the signallevel is directly proportional to the number of electronsstriking the silicon detector. Figure 6 shows the SBD signalvs. fill pressureof H2 gas n the vacuum chamber or a varietyof electron spectrometersettings. Since we expect that the gasis fully ionized and since the beatwave is a resonantphenomena,we can predict the range of pressures ver which asubstantial level of plasma wave will exist. This is shown bythe hatched bar along the pressure axis near 135 mT.Experimentally, we find that we must overfill the chamber by

about 8% or so. This is probably due to hydrodynamicexpansion of the plasma column during the =I00 psec timescale of the interaction. The individual points on Fig. 6 aresingle laser shots. All the shots which show electrons signalsabove the x-ray noise (which is around 20&300 mV) alsoshow evidence for a large-amplitude plasma wave on the threeoptical diagnostics.* Shots with small electron signals

100 120 140 160 180 200Pressure (mTorr)Figure 6: SRD signal vs. vacuum-chamber fill pressure forvarious settings of the electron spectrometer. The open pointsare from Detector A and the solid point is from a calibration ofthe intensity of the tracks in Fig. 5(a). The hatched bar is therange of the expected resonance.

near 140 mT either did not have a plasma wave (according tothe optical diagnostics) due to lack of IWO frequencies in thelaser or they did have a plasma wave but the electrons wereunsynchronized due to the micropulse structure of the linacbeam. The data in Fig. 6 were taken with the experimentalarrangement of Fig. 4 and for this arrangement, the SBDdetector s referred o Detector A.Detector A is limited to viewing energies below 9 MeVdue to the maximum field obtainable in the electronsspectrometer. To extend the electron energy measurements oenergies beyond 9 MeV, a new port was added to the electronspectrometerat about 5x the radius of curvature shown by thelong-dashed curve in Fig. 4. In this case, two SBDs werecollecting data for each shot: Detector B at the low energyposition and Detector C at the high energy position. Asummary of the number of detected electrons (SBD signal inmV times electrons/mV sensitivity of the detector/preampcombination, corrected for any limiting lead apertures placedover the detectors) vs. the energy location of the detectors.Although this data was accumulated over many laser shots,wecan see that the numbers are falling off rapidly with energy,going out to 20 MeV. This rapid fall-off with energy can beexpected for two reasons. First, the electrons are not pre-bunched and therefore occupy all phases n the accelerationbuckets. Secondly, energy-bunching is expected to occur onlyfor longer acceleration engths where transient or start-up phase

2554PAC 1993


5/5

slippage is not important. In this experiment, however, theelectrons executea large swing in phaseas hey accelerate romy= 5 (well below the Lorentz fac tor for the wave, yph) to y=20 (slightly above yph).

o...Detector B

5 10 15 20 25Energy (MeV)

Figure 7: Number of detected electrons vs. energy location ofthe detectors. Data is from many shots. Detectors B and C wereon-line together. Triangle point is from a gaseous Cherenkovtube coupled to a photomultiplier.The maximum energy seen in these shots was 20 MeV.This is an energy gain of 18 MeV over the nominal 1 cm ofinteraction length which corresponds o an averageacceleratinggradient of 1.8 GeV/m. From the engineering formula givenin the introduction, this corresponds o a wave amplitude of E= 20%. This is within a factor of two of the best one wouldexpect for our experimental parameters. The maximum energy

gain may be limited by the dynamics discussed n connectionwith Fig. 6.CONCLUSION

To summarize, the acceleration of an externally injectedbeam of electrons in a relativistic electron plasma wave hasbeen demonstrated. Acceleration is only observed when thelinac injector, the two-frequency laser, and the plasma arc allon simultaneously. The numbers of accelerated electrons iscorrelated with independent optical diagnostics. Energies outto 20 MeV have been observed implying an average gradient(over = 1 cm) of about 1.8 GeV/m, only about 2 times lowerthan the optimum expected or our experimental parameters.To conclude, it appears that the plasma beat waveacceleratorconcept can be relied upon to provide energy gainsnearly in accordance with theory. If one chooses a differentparameter space--for example, replacing the CO;! laser with atwo-frequency laser operating at a wavelengths around 1 pm--one can find theoretical energy gains of several hundred MeVfrom gradients approximately 10x higher than demonstrated nthis experiment.* Such an experiment would essentially be anextension of the current CO2 laser experiment in that itrequires no new physics. Thus, with todays glass laser

technology, one could envision building a small plasmaaccelerator based on the beat wave concept which wouldapproach he GeV energy range.ACKNOWLEDGMENTS

The authors would like to acknowledge useful discussionswith Drs. W. B. Mori and P. Mora, and Professors J. M.Dawson, T. Katsouleas,and R. Williams and thank D. Gordonfor his technical assistance. This work is supported by the U.S. Department of Energy under grant no. DE-FG03-92ER40727.

REFERENCES1. C. E. Clayton, K. A. Marsh, A . Dyson, M. Everett, A. Lal, W.P. Leemans, R. Williams, and C. Joshi, Ultrahigh-gradientacceleration of injected electrons by laser-excited relativisticelectron plasma waves, Phys. Rev. Lett. a, 37 (1993).2. See, for example, Advanced Accelerator Concepts, ed. by C.Joshi. AIP Conf. Proc. No. 193 (Amer. Inst. Phys., NewYork, 1989).3. See. for example, the Special Issue on Plasma-Based HighEnergy Accelerators, IEEE Trans. Plasma SC., ps-15, (1987)

and C. Joshi. W. B. Mori. T. Katsouleas, J. M. Dawson, J. M.Kindel, and D. W. Forslund, Ultrahigh gradient particleacceleration by intense laser-driven plasma density waves,Nature 311. 525 (1984).4. M. Rosenbluth and C. S. Liu, Excitation of plasma waves bytwo laser beams, Phys. Rev. Lett. z, 701, (1972).5. C. E. Clayton, K. A. Marsh, W. P. Leemans, M. Everett, and C.Joshi, A terawatt, short-pulse, multiline CO2 laser facilitybased on optical free-induction decay (to be published).6. W. P. Leemans, C. E. Clayton, W. B. Mori, K. A. Marsh, A.Dyson, and C. Joshi, Plasma physics aspects of tunnel-ionized gases, Phys. Rev. Lett. 68, 321 (1992) and W. P.

Leemans, C. E. Clayton, W. B. Mori, K. A. Marsh, P. K. Kaw,A. Dyson and C. Joshi, Experiments and simulations oftunnel-ionized plasmas, Phys. Rev. A 46, 1091 (1992).7. C. E. Clayton and K. A. Marsh, A 2 MeV, 100 mA electronaccelerator fo r a small laboratory environment, Rev. Sci.Instr., &, 728 (1993) and C. E. Clayton and K. A. Marsh, Aversatile 2 MeV, 200 mA compact x-band linac, (thisproceedings).8. C. Joshi, C. E. Clayton, K. A. Marsh, A. Dyson, M. Everett,A. Lal, W. P. Lccmans, R. Williams, T. Katsouleas, and W. B.Mori, Acceleration of injected electrons by the plasma beatwave accelerator, in AIP Proceedings of Workshop onAdvanced Accelerator Concepts, Port Jefferson, Ed. J.Wurtele. to be published (1992).9. P. 13. Corkum, N. H. Burnett and F. Brunei, Above-thresholdionization in the long-wavelength limit, Phys. Rev. Lett.,vol. 62, pp. 1259-62, March 1989.10. C. M. Tang, P. Spranglc, and R. N. Sudan, Dynamics ofspace-charge waves in the laser beat wave accelerator, Phys.Fluids 28, 1974 (1985).

11. A. Langsdorf, Jr., A continuously sensitive diffusion cloudchamber, Rev. Sci. Inst. lo, 91 (1939).12. C. E. Clayton, K. A. Marsh, A. Dyson, M. Everett, A. Lal, W.P. Leemans, R. Williams, and C. Joshi, ExperimentalDemonstration of Laser Acceleration of Electrons viaRelativistic Plasma Waves, Proc. of the SPIE OE-LASE 93Conf., Jan. 16-23, 1993. Los Angeles, CA.

2555PAC 1993

c. e. clayton et al- demonstration of plasma beat wave acceleration of electrons from 2 mev to 20...

Documents