Comprehensive, nonintercepting electron-beam diagnostics using spontaneous-emission characteristics

Nuclear Instruments and Methods in Physics Research A296 (1990) 134-143 North-Holland

134

COMPREHENSIVE, NONINTERCEPTING ELECTRON-BEAM DIAGNOSTICS USING SPONTANEOUS-EMISSION CHARACTERISTICS Alex H. LUMPKIN

Los Alamos National Laboratory, Physics Division, P.O. Box 1663, Los Alamos, NM 87545, USA

Characterization and optimization of electron-beam parameters are important aspects of optimizing free-electron-laser (FEL) performance . The visible spontaneous emission (A = 650 nm) from the 5 m long undulator of the Boeing FEL experiment can be characterized in sufficient detail with a streak/spectrometer to deduce time-resolved electron-beam spatial position and profile, micropulse duration and energy.

1. Introduction

2. Experimental considerations

The optimization of free-electron laser (FEL) performance can depend critically on one'-- ability to characterize the electron beam in the wiggler region . The visible spontaneous emission (A = 650 nm) from the 5 m long undulator of the Boeing FEL [1] has been characterized in sufficient detail to deduce noninterceptively electron-beam spatial position and profile, energy and micropulse duration . Additionally, these measurements have been extended by utilizing the Los Alamos streak/ spectrometer [2] to provide time-resolved spectral and spatial information on the submacropulse (< 100 [.s) and submicropulse (< 10 ps) time scales . Although these measurements involved iiïe convolution of effects (electron beam and wiggler) and emission contributions extending over the entire wiggler length, they have proven very useful. Prior to lasing, we have routinely tuned and monitored the electron-beam micropulse duration (10-15 ps) to maximize peak current and the spontaneous emission spectrum (- 650 nm centroid) in a macropulse time average to verify electron-beam energy. We have observed the spectral distribution change with the match into the wiggler, the shift of the spectrum during the macropulse, and a shift of beam position during the macropulse (which impacts e-beam overlap with the optical mode). These effects as well as some second-harmonic radiation aspects and tapered-Wgger effects will be addressed in the following sections.

These experiments were performed on the Boeing/ LANL visible FEL experiment. The basic operations have been described elsewhere [3] . During these runs from April 1988 to May 1989, the electron beam-properties were - 109 MeV energy, 2-3 nC per micropulse, 258.3 ns micropulse spacing, and a macropulse length of 30-110 [,s. The initial experiments were all done with the uniform-period wiggler, or tapered hybrid undulator (THUNDER) [1], whose parameters are given in table 1. With a period A w = 2.18 cm, a field parameter K peak -- 1 .85, peak field of 1.02 T and the 109 MeV energy, one would obtain fundamental spontaneous emission at about 640-650 nm. Some of the early data were at 700 nm. Characterization of this radiation's location, spectrum and duration is directly relatable to e-beam location, energy and pulse length .

Work supported and funded by the US Department of Defense, Army Strategic Defense Command, under the auspices of the US Department of Energy . Elsevier Science Publishers B.V. (North-Holland)

Table 1 THUNDER operating specifications [11 Length Period Number of periods Magnetic gap Peak field Wiggler parameter (peak) Betatron period Optical wavelength Rayleigh length Electron energy Electron-beam radius

L, AW NW g

Bo Kpeak A ß = 2m/k ß

A, ZR

ym oc2 re

5m 2.18 cm 220

4.8 mm 1 .02 T

1 .85 5.6 m 0.6 j. m 2.4 m

110 MeV 350

Relationships of relevance : AA/AS =1/N = 0.4% (untapered, perfect e-beam) A s = (Aw /2y 2 yl + â K 2 + Y202)

~Lm

A . H. Lumpkin / Comprehensive,, nonintercepting electron-beam diagnostics TO DIAGNOSTIC TABLE

e-BE,v4 SPECTRO3RAPH

YURN-OUT END OUADRUPOLE MAGNET MIRROR TRIPLET SC 14 ,._ ;`

20 DETECTOR ~ E(t) PHOSPHOR 196

FROM ACCELERATOR LEGEND : CURRENT MONITOR CM SCREEN SC

135

TO CONTROL ROOM

END MIRROR POP IN MIRROR

NOT TO SCALE

Fig. 1. A schematic of the burst-mode oscillator cavity showing the end mirrors, wiggler, spectrometer and pop-in mirror .

A schematic of the oscillator cavity is shown in fig . 1 . The relativistic electron beam is brought onto the resonator axis via an 180* doubly achromatic, isochronous bend and matched into the 5 m wiggler by two quadrupole triplets. An electron-beam spectrograph after the wiggler bends the e-beam off the optical axis and provides both spectral information and a beam dump. The pop-in mirror just after this spectrometer can be inserted to intercept spontaneous-emission radiation (SER) from the wiggler after one pass, and that light is then transported to the control-room diagnostic table. Part of this path is shared with that of the laser beam outcoupled from the cavity-end mirror and window. We have previously reported time-resolved FEL beam characterizations in last year's Proceedings [2] as well as in this year's [4]. The principal diagnostic device that we used was the

streak/ spectrometer. For completeness, fig. 2 is included, showing a schematic layout of this system . The optical beam (information) was brought onto this station, and the beam was directed through a Jarrell-Ash â m monochromator (used as a spectrometer), and then the wavelength-dispersed information was focused onto the entrance slit of the Hamamatsu C1587 streak camera . In this mode, one can obtain time-resolved spectra on either the macropulse, submacropulse or micropulse time domains . When used in "focus" mode we are able to ob. :in the time-averaged spontai . ~ous emission spectrum from one macropulse. This system was significantly (X100) more sensitive th?n the z m Jarrel-Ash, read out by a silicon diode array (optical multichannel analyzer) on the same diagnostics table. With the slow-sweep unit in the streak-camera mainframe, one could monitor the time-evolution of the

FROM FEL SOURCES

VIDEO MONITOR

PERISCOPE TO SECOND LEVEL

LENS 2

ANALOGUE C2280 MICRO COMPUTER (Digital processing)

R~AMD AÛT

f ), COUPLiNG OPTICS

GRATINGS

M2S(Rernovat.*A

STREAK CAMERA HA A ATSU î. W S' I Î

LENS 1 i

PATH 1

J/ml

HORIZONTAL SLIT

Fast sweep STREAK PLUGIKS Slow sweep Synchroscan (10® MHz)

UPPER LEVEL OF CONTROL ROOM DIAGNOSTICS TABLE

Fig. 2. A schematic of the Los Alamos streak/spectrometer system layout . I. EXISTING EXPERIMENTS

A .H. Lumpkin / Comprehensive, nonintercepting electron-beam diagnostics

136

spectra (and hence e-beam energy) during the macropulse. Spectral breadth can be used to bound the e-beam energy spread since for a 220 period wiggler, the fractional optical spectral breadth is 1/N or about 0.4% for an ideal electron beam . Alternatively, the mirrors MlS and M2S would be removed from their kinematic mounts and the radiation would be directly focused on the horizontal entrance slit of the streak camera. In this mode, the spatial information is provided orthogonal to the time axis. The streak camera images were read out by the SIT camera, digitized by the 2280 microcomputer, and the video redisplayed on a monitor in the control-room area. 3. Results and discussion A number of experiments have been executed during the last year to test usefulness of the techniques and to identify areas for further work . For completeness and chronology it is interesting to start with the first measurements in March 1988. Of course, two of the early objectives prior to lasing were to set the cavity length close enough to exact synchro-

nism to be within the range of the remote actuators and to check that the micropulse length was in the 10-15 ps range. Fig. 3 shows one of the initial fast streaks of a single electron-beam micropulse . The image is quite interesting because we demonstrated with narrow-bandpass filters that the earlier (upper) portion of the image is from the fundamental at - 700 nm and the second portion is best attributed to the long-wavelength-shifted (off-axis) second harmonic. This initiated at about 350 nm on axis, but resulted in a continuum of wavelengths towards the red until apertured at the larger off-axis angles . The time tilt of this radiation was caused by the dispersion through the lenses in the transport from the accelerator vault to the control room station. The analysis window indicated by the two vertical white lines showed that these two harmonic portions were separated by about 40 ps. The intensity profile is shown to the left of the figure as time increases going downward. The second harmonic radiation does not extend as far to the left as the fundamental and this is probably part of the aperturing effect. The second-harmonic contribution was eliminated by adding a spatial filter (19 mm diameter located 6 m from the end of the wiggler) in later experiments. The pulse length of the fundamental

Fig. 3 . Streak image of a single e-beam micropulse that generated both fundamental and second-harmonic radiation.

A.H. Lumpkin / Comprehensive, nonintercepting electron-beam diagnostics 30 -025

® n=1, vs. POLAR

ANGLE

0 n=2, vs . POLAR

ANGLE

137

o n=2, vs . INTENSITY _

K v >- 20 l'

~.r

W

W 15

6 4

Z

4

O

2 CL 0 0.0

0.2

0.4

0.6

WAVELENGTH (Fcm)

Fig. 4 . Calculated spontaneous-emission radiation spectral content (for THUNDER) versus angle and intensity for the fundamental and second harmonic .

was about 10-12 ps so the e-beam bunching was as expected . Fig . 4 illustrates the contributions of X at different angles and the effective wavelength gap be-

tween the fundamental and second-harmonic radiation [51 that could translate into the observed temporal gap of fig. 3 .

CI) Z LLI Z

CHANNEL NUMBER (TIME) -Fig. 5 . Spontaneous-emission ring down through the end mirror for the detuned cavity length. 1 . EXISTING EXPERIMENTS

Lumpkin / Comprehensive, nonintercepting electron-beam diagnostics

Having determined that the two components in the images were at different wavelengths, a few hours later we were able to demonstrate that the cavity length was too short to be synchronous with the e-beam macropulse spacing . Fig. 5 shows the temporal profile (rotated 90* for analysis) on a macropulse time scale using the light which has gone through the end mirror (-- 0.46 transmission for X = 650 nm). The second-harmonic bluegreen radiation is highly transmitted and is the most intense peak observed. The red fundamental is indicated on this peak's left side (earlier in time) and several other peaks are seen at even earlier times with decreasing intensity. These are due to multiple reflections of earlier pulses . Their spacing of 145 ps was then used to deduce that the cavity was 22 mm too short. The cavity length was adjusted and we did not detect these ringdown pulses subsequently on this time scale. Since FEL lasing occurred about a week later, using a cavity length very close to this, the interpretation was confirmed. The ring-down losses of 20-30% were partly due to running at 700 nm where the mirror transmission was much higher than the 0 .49 at the nominal 630-650 nm wavelengths. After these issues were clarified, the spatial filter was added at the exit port on the pop-in mirror path and the rest of the discussion will be based on detection of the fundamental spontaneous emission . Generally, the data will be in the 630-650 nm region . As the lasing experi-

SPATIAL PROFILE

ments progressed to higher power levels, we noticed that lasing too near the 632 nm cavity stabilization laser wavelength resulted in breaking the end-mirror locking so we decided to tune to the 650 nm region . In these early studies, we demonstrated that we could track the average spatial position, profile and intensity of the macropulse within the macropulse . Fig. 6 illustrates -. 6 t,s sampling of a macropulse where the separated column of spots at the right are the individual micropulses with 258.3 ns spacing. The vertical analysis window (time) shows how the spots moved towards the left on the horizontal slit by about one spot radius during this time sample. This spatial motion can be attributed to the 180 ° bend transport which was not completely achromatic, and the e-beam energy slew from the linac. The intensity profiles show that the light collected from these micropulses was not of constant amplitude and this feature could be caused by charge transport variations related to the spatial shifts. This aspect of the technique needs correlated data from strip lines just before the wiggler and within the wiggler to qualify the interpretations. Fig. 7 shows a composite of 6 tLs span streak images which were time-delayed to different parts of the macropulse to explore effects within the macropulse. Micropulse temporal length measurements were in the original objective since we had planned to monitor the laser macropulse length anyway . The duration of the

D POSITION (x,y)

INTENSITY PROPORTIONAL TO CHARGE " MICROPULSE DURATION " ELECTRON SEAM ENERGY EFFECTS (streak/spectrometer)

.0- X

Example: 0

AC OPULSE AVERAGE

SPONTANEOUS EMISSION x(t)

Fig. 6. A 6 [,s sampling of the spatial position versus time of the e-bears.

A.H. Lumpkin / Comprehensive, nonintercepting electron-beam diagnostics

example is shown in fig. 8 where the operator had tuned only on FEL energy. A very complex spectrum resulted, and as one test we returned to the spontaneous emission pulse length . The upper ieft figure shows the doublet character in the micropulse temporal profile. It persisted in a 30 event average (lower left) of that particular micropulse selected from 30 different macropulses. Timejitter in the streak camera reduced the visibility of the doublet. However, we found that this particular doublet structure was insensitive to simple injector tuning, but controlled by the phase of accelerator #6, as shown in the phase and re-phase examples at the upper right and lower right, respectively . Since the spatial filter was in the pop-in path and the 1% energy collimator was in the 180 ° bend, we attribute this effect to different energies being delivered to the wiggler at the submicropulse level and maintained through the macropulse. The energy collimator selected those electrons within a 1%n wide energy window for the bend. The last facet of these diagnostic capabilities used the streak/spectrometer mode with slow streaks on the

Table 2 Summary of average intra-macropulse effects in Fall 1988 Macropulse time

Micropulse Al [Ps]

Early Middle Late

14 9 9

a) b)

a>

e-beam b) AE [%]

Lasing

1.2 0.9 0.8

rarely usually usually

139

From spontaneous-emission At. From spontaneous-emission spectral width (AA/A).

spontaneous-emission pulse should be directly traceable to the e-beam micropulse that entered the wiggler. This ability to tune the injector or verify that the e-beam was bunched properly has been reported previously [2]. Poor bunching of the e-beam by mistuning the bunchers in the injector altered the micropulse length from 8-10 to 15-16 ps. Table 2 shows a summary of the variation of micropulse length and energy spread during the macropulse as observed in the Fall of 1988. A more recent

W H

, o -- X ---o-

Ui

Fig. 7. A composite of three slow streak images (6

~Ls

span) of the spatial position during the macropulse. I. EXISTING EXPERIMENTS

140

Lumpkin / Comprehensive, nonintercepling electron-beam diagnostics

A. . Lum kin / Comprehensive, nonijercepling electron-beam "osW

Fig. 9.

30-macro ulse average of the time-resolved spontaneous-emission spectrum (upper) and the HeNe refenmse (lower) .

s ect first, fig. 9 shows the spontaneous-emission s ectru generated over a 63 ps span and a 30-macroulse verage . In the lower portion of fig . 9 the Helse ali ant laser for the cavity was transported on the e path, and the Unfiting resolution of about 2.2 nm e location of 632 nm in the 90 nn- span was fn fig. 10, we have taken horizontal (or wavecuts through the image at locations about 30 gs , .4 L 1111C . L V et to .4 M111 &PPIUJURIMILUlly L wavelength centroid shift which we attribute to a minor energy slur during this time interval . Additionally, the full width at half maximum of the spectrum of t 10 nm (11%) for the lower sample indicated an e-beam energy spread of about 0.8% . Table 2 sumarizes effects within the macropulse that are attributable to rf phase or amplitude slew. There are, of course, a er of other factors, such as wiggler errors, e-beam ttance, etc, that should be evaluated . Still another aspect of wiggler evaluation appeared wit the wiggler-period tapering (- 4% in resonant energy) ex eriments in Spring, 1989. Figs. 11 and 12 show . cenn the s apt nt the ck, specsr~pm in s rter w velengths as expected by about 25 nm or 4%. .1 -

*

IM

The Hewe reference laser at 632 nm in each figure can be used to show this effect . And finally, in fig. 13, we show that the tapered wiggler still supports lacing at 6 nrn but with increased extraction, as reported elsewhere in these Proceedings [4].

.

-0c5 ma"Hons-V MRMU

cones

In summary, spontaneous-emission radiation c acteristics have been, and can be, used as a comprehe sine, nonintercepting e-bearn diagnostic. Further exyrimants with coordinated strip-line, wal-l-curv-,nt monitor and e-bearn spectrometer data would be ve qualifying the techniques. Additionally, more comparisons of such experimental data with the address deconvolving e-bearn and wiggler effects or determining sensitivities in the harmonics could lea more quantitative information [6,71 . Developments i the techniques may include phase m&ching by the me wi er of viewing ni the amal dependence throu sectiLn with a the source profiles and an add-on L E ISTING MPEMMENTS

A.H. Lumpkin / Comprehensive, nonintercepting electron-beam diagnostics

142

a

1 .0-

k=632nm

Z 0.5ta FZ 10Z

X

0.0 Lr 680

4" ' 1' \TCJ ra

FWHQA loam

W F Z

660 640 620 8011 ®- WAVELENGTH (nm)

F

0.01 680

'_ 680 640 620 600 ®- WAVELENGTH (nm)

Fig. 11. Comparison of the spontaneous-emission spectrum from the uniform wiggler at EB =109 MeV to the HeNe line: (a) HeNe reference laser, (b) single macropulse. 1 .01 -a-

X

Fig . 10. Selected spectra from time intervals 30 ws apart in fig. 9. A small centroid shift was detectable and attributed to e-beam energy slew.

propriate separation to exploit optical-klystron techniques. Finally, it is noted that these nonintercepting e-beam diagnostic techniques are amenable to streak/ spectrometers over the range A = 200 to 1300 nm, a range which includes a number of proposed experiments in the near UV and near JR.

Ac

owl

a

X-632nmfl FWH -2nm

z 0.5J,

0.0 680

660 640 620 600 -WAVELENGTH (nm)

mente

The author acknowledges useful discussions with John Goldstein (Los Alamos) and Dave Shemwell (Spectra Technology, Inc . of Bellevue, CIA) in the early aspects of these studies. He also acknowledges the support of the experimental teams at Boeing including .k.R. Lowrey, L. Tyson, K. Davis, D. Dowels, P. Johnson, . Murphy, R. Heeke and S. Ferrier for providing beam in

680

660

640 620 600 VELENGTH (nm)

Fig. 12. Comparison of the spontaneous-emission spectrum from the tapered wiggler at Ell = 109 MeV to the HeNe line: (a) HeNe reference laser, (b) single macropulse.

A.H. Lumpkin / Comprehensive, nonintercepting electron-beam diagnostics 1 .0,

143

the wiggler and the latter two for maintenance o¬ the

optical transport path . References

H

[1] K.E. Robinson, D.C. Quimby, J.M . Slater, T.L. Churchill 680

660

640

620

-WAVELENGTH (nm)

600

and A.S . Valla, Nucl. Instr. and Meth. A259 (1987) 62-71 and IEEE J. Quantum Electron. QE-23 (1987) 1497 . [2] A.H . Lumpkin, N.S.P. King, M.D. Wilke, S.P. Wei and K.J . Davis, Proc. 10th Int. Free Electron Laser Conf., 1988, Jerusalem, Israel, Nucl. Instr. and Meth. A285 (1989) 17. 13] D. Shoffstall, Proc. Int. Conf. Lasers '88 (STS Press, McLean, VA, 1989) p. 171. [4] A.H. Lumpkin, R.L. Tokar, D.H. Dowell, A.R. Lowrey, A.D. Yeremian and R.E . Justice, these Proceedings (11th Int. Free Electron Laser Conf., Naples, FL, USA, 1989) Nucl. Instr. and Meth . A296 (1990) 169. [5] J.C. Goldstein, private communication, (Los Alamos National Laboratory, June 1988). [6] W.B . Colson, private communication (Naval Postgraduate School, Monterey, CA, August 1989). [7] R. Barbini, F. Ciscii, G. Dattoli, L. Giannessi and A. Torre, synchrotron radiation from magnetic undulators as a prospective diagnostic tool, to be published.

"- WAVELENGTH (nm)

Fig. 13 . Comparison of the spontaneous-emission spectrum from the tapered wiggler at EB =109 MeV (a) to the lasing (macropulse) at _ 650 nm (b).

1. EXISTING EXPERIMENTS