Recent results from the Los Alamos free-electron laser

In this paper, we review the most recent experimental results of the Los Alamos free-electron laser program. Three major efforts will be described: lasing at improved efficiency over that previously attained, electron beam improvement, and energy recovery. An extraction efficiency of 2 percent was achieved with a wiggler having a 12 percent wavelength taper. The beam has been improved so that limits to its quality are now caused, not by injector performance, but by wake fields related to the high peak currents achieved. Limits to optical power are set by mirror damage. Experiments are described that demonstrate the successful operation of an energy-recovery system.

current and better quality and to instali the extra bearnline and the decelerate-s required by the energy-recovery experiment.
In previous experiments, the efficiency of the Los Alamos FEL for converting electron beam e,~ergy to laser light at 10 pm was l.0% for a uniform-period wigglerl and 1.3% for a tapered wiggler. The intracavity peak optical power was about 800 MW, corresponding to a peuk output power of 2 W. Ideally, we had expected to achieve efl'lciencies of 1.2% and 3 to 470 for the uniform-period and tapered wigglers, respectively, and correspondingly higher power levels, but we were limited by several factors related h accelerator performance.
In particular: (1) the electron beam emittance and energy spread were excessive, (2) the maximum charge we could accelerate was limited, (3) the temporal shape of the electron micropulse included extensive nonlasing wings, and (4)  The number of subharmonic bunchem was reduced from two to one for simplicity; the position was changed to improve emitta.nce and micmpulse shape, and the buncher was fabricated of stainless steel rather than copper to damp beamdriven cavity modes. achromatic '%enda-have two effkcti in addition to their intended one of bending the qlectron beam by 60°into coincidence with the optical beam. Because of their nonieochronous nature, the bends have a larger tranait time for low-energy electrons than for highenergy electrons. This property can be ueed ta bunch the qlectronrn in each mic.mpulue. We normally take advantage of this bunching to shorten the micropulses by a %ctor of 2 and thereby increase the peak current accordingly. Conversely, the nonisochronous nature of the bends causes a second, unfavorable effect i.e., fluctuations in the arrival time of the micropulses tit the wiggler whenever thsir energy fluctuates. This is our most common source of noise. It causes nonuniform lasing intensity and can reduce the efficiency.
We built a second magnet system that provides the 60°bend along with isochronous properties but, so far, have not chosen to employ this system and make the trade-off of noise improvements for lowei peak current.

B. Improvements Realized
These changes have caused the following impro /ements in accelerator performance: Peak accelerated cbrge is no longer limited by clipping the beam at the entrance to the accelerator because of the beam's excessive diametir.
The limiting charge baa been raiaed fkom 3 to 6 IIC, and this limit is now imposed by available rfpower.
Emittance, energy spread, and m.icropulee shape have been improved, eapacially at low charge, but are still excessive (see Fig. 2)ab0w-3nC. Micropulae length haa been rnewwed with a streak camera. At the wiggler, the rnkmpulae is shorter than 16 pa =d the bmrn current emeeda 900A.

C. Lsser Performance
We have uaea our impmved accelerator to drive our 10-pm ML millabr, using both a uniform-period wiggler and a wiggler with ti 12% wavelength t+er.
The measurement procedure and diagnostic &olB wem similar to those reported previously.'~As expcwd, extraction eff~ciency and power levels were increaaed. We achieved an efficiency of 2.0% with the tnpered .tiggler, with intiacavity peak vwerle~e~s ando~ltputof 2000 and 40 MW, respectively. Mom extensive synchmhn sidebands were seen than before, covering a spectral range as large as 10%. The FWHM of the detuning curve was 40 pm for the uniform-period wiggler and 4 prn for the tapered one.
Strong fluctuations in efficiency, optical power, and spectra are still seen from shot h shot and within a single macropulse because of accelerator noise.
Lasiag occurred when the micropulse charge was varied from 6 nC to as little as 0.25 nC. The maximum extraction e~lciency of 2.0% was attxined with 3 nC. With a larger charge, tie emittince and energy spread (we Fig. 2  (After this series of calculations, the streakcamem measuremata indicated that tha pulaea probably had been shorter and the peak curranta correspondingly higher. We have not yet performed calculations under such conditions.) The four theoretical approximations are as follows:    The krga incmaaae in emittance and energy spread placed severe limits orIthe UDSoftlm electron beam for king and dkoumged us fmm attempting to achieve till higher chargea. We have employed aeverul diagnostic deviea tn imla~tie source of the emittance and ene~spread. To meamme emittame, we have uaad various combinations of G&enkov radiation screens and fmuaing quadrupolee w dete~ne electron-beam spot si-t lleee meuauremenm gave the emit~nce avtmged over the 2000 micropuhws in n macropulse.

D. Comparison of M~easured and Calculated Extraction Efficiency
We established three emittance measuring stations in the bearnline shown by @ to @ in Fig. 1. Measurements were made at two micropulse charges, 1.1 and 5.5 nC, and under two conditions of magnetic bunching in the 60°bends. The peak current I was estimated fmm the charge and representative strea~[-camera pulse-width measurements. Table II summarizes the measurement supplemented by the following observations: q The emittance of the electron bem at station 0 was (within measurement error) independent of charge and well below our critical value of3 n.
q There was a large growth in emittance through the 60°bends and a significant growth in the subsequent straight section of beamline.
q The growth in emittance at stations @ and (9 increased with micropulse charge and depended on how well the charge was bunched in the magnetic buncher, i.e., peak current.

Similar measurements
were made of energy spread as a function of charge and bunching at two well-sep-ted stations, @ and @ shown in Fig. 1.
The spread was an average @en over the entire macmpulse. The results shown in Table III   The proportionality constant used in Fig. 3(d), 1.4%/100 A, was chosen to agree with the measurements, and was found to be consistent with theoretical estimatzs. Figure 3(e) shows a real measurement.
Our explanation for the energy droop is wake fields,e i.e., interactions of each micropulse with tie wall curren~that follow behind it. These interactions drain energy from the micropulse and are most serious at wall discontinuities such as bellows. Wake fields can also deflect an elecmn beam transversely in nonaxiaily symmetric systems or when the beam is allowed to move off-center in axially symmetric systems. The deflection is proportional to the instantaneous current. Such a deflection, when averaged over a micropulse, would be diagnosed a.a emittice growth.
13eamline d.iscontinuities that occur in a bend or that are not radially symxnewic, or an electron beam that is not well aligned, can also cause this kind of emittsnce growth.
Calculations have been performed7 using the computer codes TBCIe and LTRACK9 to estimate the magnitude of the longitudinal and transverse wakefield effkcta. Within a factor of 2, the mqjor discontinuities we have identified provide an explanation for dl the growth of exnittance and energy spread that wc have seen.

F. Summary
Irnprovementa hsve &en made in the i~ecmr region of the Los Alarnos rf linear accelerator allowing lxmms to be produced with higher charge, higher current, and better quality. The quality then suffsrs degradation further down the beamline because of wake-field effects. In spite of this degradation, a wiggler efficiency of 2.0%, and peak intracavity optical-power levels in excess of 1 GW have been achieved. At this power level, significant damage to the mirrors was obsexwed. Without this damage, even higher efllciencies and power levels could have been obtained. Efforts are now being made ta reduce wake-field effects by eliminating unneceriary discontinuities in the beamline and by ensuring that the beam remains centered in the beamline throughout its length. Modifications to the optical cavity to prevent mirror damage will include use of copper mirrors and increased cavity length.

A. Rationale for Energy Recovery
After passing through the FEL, the electron beam will have more than 95% of iti original energy and will retain intact its charge, emi~tance, and The reverse-path scheme has an advantage in that the accelerating and decelerating beams have the same energy at any point in thei~common structure. This coiifiguration eases beam-transport problems.
The scheme has an important disadw-ntage: when tie distance between pulses is !ess than the length of the accelerator, i.e., long accelerates and high average current, counterpropagating pulses will overlap at some point and space-charge effecti will defocus the beam and make beam transport diiWcult. This defocusing effect would make the reverse-path scheme diflicult tm scale up to large, highcurreiit machines.
The racetrack configuration will, in general, have beam-transport difficulties because accelerating and decelerating beams are copropagating

Co Results
J, Deceloratio~The resonant bridge couplem are a~uatable over the range of deceleration, -15%-100%, which is in reasonable agreement with ksta on prelirnhiary models. As expected, the n/2 mode of operation of the couplers is very stable. We changed the temperature of one of the acceleratma by 10" C, enough to alter i~resonant frequency by several line widths. In a weakly coupled system, this would produce a phase change of 90°between the accelerator and decelerator and a large amplitude difference between the two 'anks. In the present case, the phase difference between the two tanks was less than 1°and the amplitude difference less than 4%.
The energy of the beam after traversing the decelerators will vary with the time of arrival of the micropulses at the decelerators. The energy is expectid to depend on the position of the 180°bend in a sinusoidal fashion. The profile of the beam as it propagates around the bends and through the decelerators is qualitatively similar to that found by simulations. The lowest energy beam transported was -3.5 MeV. At large decelerations, we were forced to add trim coils along the decelerators to compensate for the fringe fields from the solenoids around the injector and accelerator.
The maximum charge that we have transported is 4,6 nC (equivalent to 0.1 A average current). We were able to transport the full charge from the scraper in the 60°bend through the decelerators down to an energy of -5 MeV. Between 5 and 3.5 MeV, we were able to keep the beam focused through the decelerators, but steering became progressively more difficult because of solenoid fringe fields, and charge was lost. The stability of the system in the computer simulation is somewhat less than that q xperimentally observed, but this difference may be due to the simplifications of the theory noted above.

Power Flow
Our resul~indicate that these instabilities can be a very serious consideration in future designs unless care is taken in design and implementation to avoid beam loss in those parta of the transport system that afhct the energy-recovery process.

E. Conclusions
We have determined that the component of the Lnsing at about 1% extraction efficiency did not degrade the performance of the system, but high-extraction efficiency wigglem will impact system design. On the basis of our measurement to date, we can conclude that energy-recovery could be usefully applied to large FEL systems.  The emittance is unnormalized and corresponds to the FWHM of the electron spatial distribution.