Вы находитесь на странице: 1из 12


Nuclear Instruments and Methods in Physics Research A 340 (1994) 219-230 & METHODS
North-Holland IN PHYSICS
Section A

Initial measurements of the U C L A rf photoinjector *

S.C. Hartman a,**, N. Barov a, C. Pellegrini a S. Park a j. Rosenzweig a G. Travish a,
R. Zhang a, C. Clayton b, p. Davis b, M. Everett b, C. Joshi b, G. Hairapetian b
a Department of Physics, Uniuersity of California, Los Angeles, CA 90024, USA
b Electrical Engineering Department, University of California, Los Angeles, CA 90024, USA

The 1.5 cell standing wave rf photoinjector has been operated for the past several months using a copper cathode. The
photoinjector drive laser produces sub 2 ps pulses of UV (A = 266 nm) light with up to 200 p~J/pulse which generates up to 3 nC of
charge. The emittance of the photoinjector was measured as a function of charge, rf launching phase, and peak accelerating field.
Also, the quantum efficiency and pulse lengths of the laser beam and the electron beam were measured.

1. Introduction This signal is multiplied 75 times to produce the

klystron operating frequency of 2.856 G H z and is then
We report the initial results of the operation of the sent to a 1 kW solid state amplifier. The amplifier
U C L A 4.5 M e V photocathode rf gun [1-3]. This elec- signal in turn feeds the klystron. This ensures the
tron source is part of a 20 M e V compact electron linac timing and a feed back loop stabilizes the laser pulse to
described before. It will be used for studies of the rf jitter to less than 4 ps.
interaction of relativistic beams, plasmas, and the gen-
eration of coherent radiation. All the components of
2.Z R f photoinjector
this system have been built and tested. Full assembly
will be completed during the fall of 1993. Our initial
work has been dedicated to a characterization of the The photocathode rf gun is based on the Brookhaven
photocathode rf gun, which is the electron source for design. It consists of a one and a half cell standing
the system. As part of this work we have measured the wave accelerator producing a beam with an energy up
electron b e a m emittance and the quantum efficiency of to 4.5 MeV. Accelerating gradients of up to 100 M V / m
a copper cathode under different conditions. W e will are achieved.
summarize the overall performance of the gun and our
plans for future development. The photocathode rf gun 2.3. Laser system
has also been used to study a thin plasma lens. This
experiment demonstrated electron beam focusing and For single photon photoemission, the photon energy
confirmed theoretical expectations. must exceed the work function of Cu (4.65 eV). The
photoinjector drive laser was designed to produce < 2
ps laser pulses at 266 nm (4.66 eV) [4] with up to 200
2. Accelerator system description ixJ / p u l s e . This is accomplished using chirped pulse
amplification and compression of a mode-locked Y A G
2.1. R f system laser (A = 1.064 p.m) and frequency upconverting using
two K D * P doubling crystals. The drive laser is a mode
The gun is powered by a S L A C XK5 type klystron locked N d - Y A G oscillator cavity. To compress the
producing 24 M W of rf power with a pulse duration of pulse the laser is matched into a 500 m fiber to
4 ixs. The rf system is driven by a signal produced by a produce a frequency chirp. The chirped pulse is then
master oscillator clock at a frequency of 38.08 MHz. amplified a million times by a regenerative amplifier
and sent to a grating pair where it is compressed. Once
it leaves the compression stage the laser beam is fre-
* Work supported by SDIO/IST through ONE Grant No. quency upconverted to green by a K D * P crystal. T h e n
N00014-90-J-1952 and US DOE Grant DE-FG03-92ER- it is doubled again to UV, 266 nm, by a second fre-
40493. quency doubling crystal. The upconversion efficiency is
** Corresponding author. typically 10%.

0168-9002/94/$07.00 © 1994 - Elsevier Science B.V. All rights reserved II. CONTRIBUTIONS
SSDI 0 1 6 8 - 9 0 0 2 ( 9 3 ) E 0 9 8 1 - W
220 S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230



Fig. l. Beamline configuration.

2.4. Diagnostics 0.2 . . . . . . . . . . . . . . . . . . .

T h e electron b e a m diagnostics are as follows. T h e ~=0.067%
main diagnostics are the p h o s p h o r screens which moni- 0.1
• gig
tor the spot size. A n integrating c u r r e n t transformer, 0.05 • • go
ICT, is used to m e a s u r e the electron b e a m charge C • •
along with F a r a d a y cups at various locations. Some of -0.05 • oo
the p h o s p h o r screens are floated so that they can
double as Faraday cups. A dipole is used to m e a s u r e -0.1 •
the energy and energy spread of t h e b e a m (Fig. 1). -0.15 •
-0.2 . . . . . . . . . . . . . . . . . . .
0.00 5.00 10.00 15.00 20.00 25.00
3. E l e c t r o n b e a m e n e r g y a n d e n e r g y s p r e a d Shot Number
Fig. 2. 8 p / p versus shot number (o- = 0.067%).
3.1. R f induced jitter

T h e time jitter of the laser pulse a n d the fluctua- I I I I

tions in the rf voltage are of deep c o n c e r n for future o =0.3%

F E L experiments. In o r d e r to get a m e a s u r e of the rf ~ = 4 psec
induced pulse-to-pulse jitter we m e a s u r e d the fluctua-
tions in the peak dark c u r r e n t electrons. Fig. 2 shows a
plot of g p / p ( % ) versus shot n u m b e r and has a o-=
0.067%. ;, • ,,
• .;...." :,.. • . -,.'.,
3.2. Timing jitter •: . "%: • .........
ooo •0
• • • oo •
In Fig. 3 we show the jitter in peak m o m e n t u m of
the photoelectrons; this gives a m e a s u r e of not only the
rf jitter but also the jitter b e t w e e n the laser and the rf.
O n e can see t h a t the resulting rms m o m e n t u m fluctua- ' 2 ; 4 ; "
tion is 0.25% and is due mainly to the residual laser Shot Number (in order of ascending Q)
pulse jitter. Fig. 3. Jitter in peak momentum of the photoelectrons.
S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230 221

We measured the dependence of the beam energy malisms such as matrix transformations do not apply.
versus rf phase to calibrate the absolute phase, Fig. 7. This means that the usual quadrupole scan technique
This dependence shows that the accelerating field in is not valid for the U C L A photo-electron beam. To
the two cells is unbalanced with a larger field in the overcome this problem a pepper pot emittance mea-
full cell. Introducing an imbalance of 50 M V / m in the surement apparatus has been designed and imple-
half cell and 100 M V / m in the full cell the data can be mented. The pepper pot is designed such that upon the
fit using the analytical theory of Kim [5]. From the electron beams passage through the pepper pot it is
beam energy measurements we can also determine the transformed from a space charge dominated beam to
gun shunt impedance which is evaluated to be 36 an emittance dominated beam. Once the electron beam
M £ t / m ; this is smaller than the value expected from is in the emittance dominated regime one can use
the Superfish calculations. The beam energy spread linear transform theory to calculate the emittance.
measured at low current, 50 pC, is less than 0.2%. Because of pulse-to-pulse fluctuations in the beam
However, due to the large space charge effects it charge the measurement must be made single shot. All
increases rapidly with charge, when using the damaged the beam parameters are measured every shot via fast
cathode. The space charge growth was modeled with sample and hold electronics.
P A R M E L A [6] and is discussed later in section 7.
4.1. Pepper p o t considerations

4. Emittance measurements The pepper pot used in these experiments is actu-

ally an array of eight slits. After the electron beam
The U C L A photoinjector [7] produces a very high traverses the slits it drifts approximately 20 cm onto a
peak current beam, up to 250 A, at relatively low phosphor screen. The image then gives the full phase
energy, 4.5 MeV. The consequence of this is that space, emittance, and other beam parameters. The
electron beam transport is in the highly non-linear width of the slit was chosen by considering the enve-
space charge dominated regime. Since the beam is lope equation, Eq. (1). For a round beam, A is the
space charge dominated, linear beam transport for- normalized beam transverse dimension, /3 is the nor-

I . . . . I . . . . I . . . . I




I . . . . I . . . . I . . . . I

0.1 0.2 0.3 0.4 0.5

Charge (nC)
Fig. 4. Emittance versus charge.

222 S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230

malized velocity, 7 is the normalized energy, and • is 1o ...I .... I .... I .... I .... I .... ~ .... I ....
the normalized emittance • /

2 8-
A"- flyk %
~- A3 0 where A = a [ f l y ] 1/2. ~, 6-
(1) 4-
Taking the ratio of the space charge to emittance term • O
2a21p Z 7o qo 10o 11o ~2o 13o
5~' = ( / 3 y ) l / 2 • ~ i A , (2) RF INJECTION PHASE (Degrees)
Fig. 5. Zero intercepts of rf injection phases.
where I v is the peak electron beam current and I A is
the Alfven current of 17 kA. When this ratio is greater
than unity, space charge forces dominate the electron
parameters fixed one expects the emittance to be lin-
beam and when the ratio is less than unity emittance
ear in charge as in the following equation [5]
effects dominate the electron beam transport. By
choosing the slit widths wisely we can go from the -rr 1 1 I
• xs c - (7)
space charge dominated regime to the emittance domi- 4 aK sin(Oo) IA/x"
nated regime. Now if we scale the transverse dimen-
Each emittance versus charge plot takes approximately
sion by rl,
30-60 s to complete, Fig. 4. So one can adjust on-line
t !
Ip = .q2/p; a' = r/a; • n = ~'/En, (3) the beam parameters to minimize the electron beam
emittance. This is a very powerful diagnostic.
the ratio of the space charge term to the emittance
term becomes 4.4. R f induced emittance vs laser injection phase
,~' =.~'r/2 . (4)
To measure the rf induced emittance versus laser
Using this expression and plugging in the numbers for
injection phase many emittance versus charge data
the U C L A experiment we get *7 = 0.01 which corre-
runs were taken with different laser injection phases.
sponds to a slit size of 50 Dxm.
The procedure for extracting the rf induced emittance
is as follows. For each laser injection phase the data is
4.2. Definition of emittance plotted and fit to a line. The slope and intercept of this
line are then extracted. The zero charge intercept gives
Electron beams do not typically have sharp bound- the rf induced emittance. Fig. 5 is a plot of the zero
aries so the definition of emittance is ambiguous at intercepts of rf injection phases. The data is then fit to
times. Some laboratories define emittance for a certain the equation
percentage of the beam particles enclosed within an
rf 2 I 2
ellipse, commonly 63% or 90% [8]. For the following •x )lcos< )t. (8)
measurements we define our emittance as the root-
Notice that the data have the correct scaling as one
mean-square emittance [8] given by
would expect.
• rms=((X2)(X'2)--(X'X')2) 1/2, (5)
4.5. Emittance vs peak accelerating field
and the normalized emittance is given by
To explore the effect of the variation of the peak
• n = flT~rms" (6)
electric field the rf power into the photoinjector was
varied, Fig. 6. Since the final beam energy is propor-
4.3. Emittance vs charge tional to the peak accelerating field we recover the
expected linear decreasing of the emittance.
The electron beam emittance in all of the following
was done with a beam m o m e n t u m of 3.5 M e V / c . The 4.6. R f injection phase calibration
first step in measuring the emittance is to measure the
linear increase as a function of charge. The laser The beam energy versus rf injection phase was
energy delivered to the cathode was varied and the measured in order to calibrate the absolute phase, Fig.
emittance was measured as a function of charge trans- 7. The equations of motion were then integrated and
ported through the ICT and to the slits. For all other compared to the experiment. This data highlights the
S.C. Hartman et al. / Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230 223

16 • . . I . . . I .. .I ..
~ 2 - I I I . • . I -
15- e, i
14- '\\
\ 4~ 1.5.
13- "e\ \ o
10- Q
N.1 0.5.
8~'''I . . . . . . . . . .
© 2.4 2'.8 3'.2 3'.6 0~' ' ' I' " " : " " I ' ' ' ', ' ' '
z © 0.02 0.04 0.06 O.Og 0. I

BEAM Momentum (MeV/c) Z Z (m)

Fig. 8. Gun balance measurement.
Fig. 6. Effect of variation of peak electric field.

T h e e m i t t a n c e values given are thus a set of worse case

fact t h a t the rf gun was u n b a l a n c e d . T h e field in the
full cell was 1.8 t i m e s t h a t in t h e half cell, Fig. 8. This
was c o n f i r m e d with a b e a d pull e x p e r i m e n t after the 5. Pulse length measurements
gun was r e m o v e d for a p o s t m o r t e m cold test.
T h e e m i t t a n c e m e a s u r e m e n t s p r e s e n t e d above are 5.1. Experimental setup
the first r o u n d results a n d suffer from a few problems.
T h e first is that u p o n installation of t h e p h o t o i n j e c t o r A 250 ~ m thick fused silica etalon [9] served as the
the m i c r o m e t e r which holds the c a t h o d e in place was C e r e n k o v radiator. T h e etalon side w h e r e the electron
b u m p e d . This caused the gun to b e c o m e detained. T h e b e a m e n t e r e d was s a n d e d forming a diffuse surface to
gun was t h e n r e t u n e d on-line by adjusting the c a t h o d e p r e v e n t b u n c h l e n g t h e n i n g due to multiple reflections•
position until the s h u n t i m p e d a n c e was maximized. F u r t h e r m o r e , the b e a m e n t r a n c e side of the e t a l o n was
Since the s h u n t i m p e d a n c e of the full cell is larger t h a n covered with a 0.005 in. thick a l u m i n u m foil which
the half cell the p e a k fields in t h e full cell were provides g r o u n d i n g of the C e r e n k o v r a d i a t o r a n d pre-
favored. T h e second artificial e m i t t a n c e growth m e c h a - vents any scattered laser light from e n t e r i n g the field
nism was t h a t the p h o s p h o r screens which were used to of view of the C e r e n k o v light. T h e e t a l o n was a t t a c h e d
m e a s u r e the spots were at 45 ° angles with respect to on a m o u n t which was externally r o t a t a b l e a b o u t an
the b e a m . This e n l a r g e d the spot size artificially a n d axis p e r p e n d i c u l a r to the beam. F o r a relativistic elec-
due to short d e p t h of focus i n d u c e d some parallax tron b e a m , the angle of C e r e n k o v radiation with re-
b r o a d e n i n g a n d blurring of the image. T h e s e p r o b l e m s spect to the b e a m axis is
were e l i m i n a t e d in a s u b s e q u e n t a t t e m p t to r e - m e a s u r e
t h e e m i t t a n c e b u t d u e to time limitations t h e m e a s u r e -
m e n t s were not c o m p l e t e d in time for this conference.
w h e r e n is the index of refraction of the m e d i u m . F o r
f u s e d silica n = 1.46 a n d t h e r e f o r e 0 c = 4 7 °. T h e
C e r e n k o v r a d i a t o r setup is depicted in Fig. 9.
This C e r e n k o v r a d i a t o r allows 1 ps time resolution.
~ 3.6] ,..I .... I .... I .... I .... I .... I .... I...
T h e limiting factor in the b u n c h l e n g t h m e a s u r e m e n t s
.- " - 0 °00
is the streak camera. W e use a H a d l a n d I m a c o n 500
00 °0 \\0 streak camera. A t the fastest sweep speed of 20 p s / m m ,
3.2- ee ".e the time resolution is n o b e t t e r t h a n 3.5 ps. T h e streak
c a m e r a c a n n o t b e installed directly viewing the
3- "" e
C e r e n k o v d e t e c t o r b e c a u s e of the high levels of radia-
2.8- 2 tion inside the lead shielding. T h e r e f o r e , the C e r e n k o v
light is t r a n s p o r t e d t h r o u g h a maze in the lead shield-
,,i . . . . i .... i . . . . l . . . . i . . . . i .... l,,

2"60 I'0 2~0 310 4'0 5'0 ~0 7'0 8~) ing. Because the light intensity levels are n e a r the
detection limits of the streak camera, the C e r e n k o v
RF Phase (Deg) image is r e d u c e d by a factor of 4 w h e n focused on the
Fig. 7. Beam energy versus rf injection phase. streak c a m e r a slit. To achieve 3.5 ps resolution, a 25

224 S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230

ixm slit is used at the entrance to the streak camera.

1000 . . - I • • • I • ~. I • • . I . • .
Accounting for the 400% reduction in image size, the 10mmdia.FWFMspotsize ~
25 ixm slit infers a 100 Ixm acceptance at the Cerenkov 532nml~serpulse ~
radiator. The maximum time delay possible between "~ 750.25 " 2 ps FWHM / /
any two Cerenkov photons created by an infinitely
small thin electron beam is 1 ps. When combined with .=
the resolution of the streak camera, the overall resolu-
tion of the Cerenkov streak system is 3.6 ps. The streak
camera resolution was verified by streaking the laser
pulses. This steak camera cathode is sensitive to green
532 nm light but not to U V 266 nm light. Therefore,
the laser pulses were streaked after the first doubling
crystal. A ~2 reduction in pulse width should occur ~. 250.75
when the green light is doubled to UV. In streaking
short pulses, care is taken to avoid space charge satura-
tion within the streak camera which can result in
erroneous measurements. Fig. 10 depicts a typical laser 1
streak and its corresponding integrated intensity plot. 0 4 8 12 16 20
Time (ps)
5.2. Streak measurements" Fig. 10. A typical laser streak and its corresponding integrated
intensity plot.
The laser pulse streak resulted in a pulse width of
3.9 ps F W H M . Since the resolution of the streak
camera is 3.5 ps, the actual laser pulse width is 1.7 ps. ments. The streak images appear spotted due to the
Auto correlation measurements of the cw mode-locked low light levels incident on the streak camera. Low
Y A G laser beam result in pulse widths of 2 ps F W H M . light levels are required to avoid space charge length-
After doubling to green light, the pulse width is ex- ening in the streak camera and to achieve 3.5 ps
pected to decrease by ~2, producing 1.4 ps pulses in resolution. In analyzing Cerenkov streaks, the images
reasonable agreement with the streak camera measure- are integrated along the space axis to provide better
statistics. Each array element corresponds to 0.56 ps.
Since the Cerenkov streak system resolution is 3.6 ps, a
smoothing algorithm is used to average over the six
nearest neighbors in the integrated array. The first
streak measurements were taken with 70 ° laser injec-
tion. Laser injection at 70 ° produces bunch lengthening
from a time delay across the cathode as the laser
wavefront strikes it. This produces an electron beam
with a linear space time correlation. This correlation is

'\ present in the Cerenkov streak shown in Fig. 11. A

laser spot size of 2 mm produces a time delay of
2 mm
~- ~- ~ 20 ps. (10)
c cos(70 °)
The measured delay from the streak in Fig. 11 was 25
ps. When the streak is corrected for the delay across
the wavefront, the measured pulse width becomes 13
200 mm Lens ps. At 2 ° injection, the time delay across the cathode is
"~ BG18 &VG9 Filters insignificant and the Cerenkov streaks indicate the
\ Center Frequency: 530nm electron bunch length directly. By focusing the laser
~ Bandwidth: 100nm FWHM spot to 2 mm, we were able to move the laser spot to
\ 50mm Lens
~ I-----
an undamaged portion of the cathode. A streak from
the undamaged cathode is depicted in Fig. 12. Streaks
from the damaged portion of the cathode resulted in
Mirror~ 2----~ , H / f ~ Streak Camera slightly longer bunches due to an elongated tail. Bunch
length measurements with the Cerenkov radiator and
Fig. 9. Cerenkov radiator setup. streak camera resulted in 9-15 ps F W H M for various
S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230 225

Cathode ,/
RF Gun

b 7
p0 A n g l e of I n c i d e n c e

/ t An.,eo,

Fig. 11. Cerenkov streak showing linear space time correla-
tion. Electric Field E
Fig. 13. Laser injection geometry.

charge levels from 0.2 to 2.6 nC, however no correla-

tion was found between the bunch length and the
charge output from the rf gun. for bunch lengthening is space charge. The quantum
All the streak camera measurements indicate elec- efficiency data indicated significant reductions in
tron bunch lengths longer than 8 ps. Measurements for charge production due to space charge. Microemitters
2 ° incidence on the undamaged portion of the cathode can further aggravate this problem, possibly accounting
provide the shortest bunch length measurements aver- for pulse lengthening and time structure of the streaks
aging 12 ps. However, the 2 ° measurements from dam- from damaged portions of the cathode. P A R M E L A
aged parts of the cathode clearly characterize longer [10] simulations were performed for various output
bunches with more time structure and long tails. These bunch charge levels from 100 pC to 1.3 nC. These
measurements contradict the assumption that electron simulations show a bunch lengthening as output charge
bunches will mimic the laser pulses in time. One source is increased. For typical charge levels between 0.4 and
1.0 nC P A R M E L A predicts bunch lengths from 9 to 15
ps. These bunch lengths predictions are within the
400 .... I .... I .... I .... I .... I .... I .... I .... scatter of the measured bunch lengths, however the
Beam Charge = .2 nC measurements did not indicate a correlation with
350- charge level.
6. Q u a n t u m efficiency
~. 250"

6.1. Laser geometry

I. 200"

.~ 15o- The laser injection geometry [11] is shown in Fig.

13. The polarization angle ~b is the angle the electric
~. 100 - field makes with respect to the plane of incidence.
Therefore, 0 ° corresponds to p-polarized light and 90 °
50- corresponds to s-polarized light. A 1 / 2 waveplate al-
lows continuous rotation of the laser polarization
.... I .... I .... I .... ', . . . . ', . . . . ','''I ....
through a full 360 °. The angle 0 is the angle of inci-
0 I0 20 30 40 50 60 70 80
dence of the laser beam with respect to the beam axis
Time (ps) of the electron gun. We are limited to only two angles
Fig. 12. Streak from the undamaged cathode. of injection, 70 ° and 2 °.

226 S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230

The electron charge is measured with two indepen- 2.1

dent diagnostics, the Faraday cup and the ICT. Both of \ /0
these diagnostics agree with each other to within 10%. 2 \ • /
Because the Faraday cup collects significant amounts
\ /
of dark current, the ICT was used to measure the 1.9 \ /
photo-induced charge per pulse free of the dark cur- /
rent background. A phosphor screen was used to en- /
sure the position of the electron beam was on axis near \
the center of our diagnostics. The beam energy was .~ 1.7 \ /
measured to be 3.5 MeV using a dipole spectrometer. \ /
\ /
The Cu cathode received no special surface prepara- 1.6 /
tion. After machining, the cathode was installed in the \,
rf gun and baked at 100°C with the gun under vacuum. 1.5 \ ~ j 3 mm dia. FWFM spot size
High power rf conditioning was used to reach a vac- • 266nm laser pulse
2 ps FWHM
uum level of 10 9 Torr. During operation of the gun, 1.4 I J
the laser spot was focused to sub mm spot sizes on the
0 45 90 135 180
cathode. The high intensity of the laser pulses dam-
Polarization Angle (deg)
aged the cathode surface, further complicating the
Fig. 15. P o l a r i z a t i o n d e p e n d e n c e of collected charge for a
emission process. The damaged spot is centered on the
l a s e r e n e r g y o f 100 IxJ.
cathode and therefore all quantum efficiency measure-
ments were made for photoemission from the damaged
ciency (7) corresponding to the slope of the line. From
6. 2. Experimental results these fits, an enhancement in quantum efficiency of
50% is observed for 70 ° p-polarized over 70 ° s-polarized
Measurements of collected charge vs laser energy light.
for three representative cases are shown in Fig. 14. Measurements of charge vs laser energy were ob-
From these measurements it is clear that saturation of tained for various polarization angles. For 2 ° injection,
the charge occurs at laser energies above 50 p,J. There- changing the polarization angle did not affect the
fore, values of quantum efficiency -q are taken in the charge collected. However, for 70 ° injection, measure-
low charge limit. Linear fits for laser energies below 25 ments resulted in curves of similar shape to those of
IxJ are presented and labeled with the quantum effi- Fig. 3 but which lie in between the 70 ° s-polarized and
70 ° p-polarized curves depending on the angle of polar-
ization. Fig. 15 shows the polarization dependence [12]
3.5 .... I .... I .... I .... I .... of collected charge for a laser energy of 100 ~tJ.
• T h e t a = 2 ° any ~ I Identical plots are found at different energies. The
T h e t a = 7 0 ° ' ~ = 1 0 ° (P) I [] functional form of this enhancement fits a cos2cb de-
o Thel pendence which implies that the enhancement is de-
pendent on the energy of p-polarized light rather than
2.5 its electric field.
For laser spot sizes less than 3 mm in diameter, it
was possible to inject the laser pulses at 2 ° incidence
without impinging directly on the damaged area of the
cathode. Careful quantum efficiency measurements
1.5 were not taken under these conditions, however a
factor of 3 decrease in the quantum efficiency was
1 observed from the undamaged portions of the cathode.
Q u a n t u m efficiencies of 1 × 10 -4 are surprisingly high
for photoemission from Cu using 266 nm light (4.66
eV) since the photon energy is very close to the work
function of Cu (4.65 eV). The quantum efficiency of Cu
has been measured to be 1.5 x 10 s using 248 [13] nm
50 100 150 200 250 light and 6 × 10 5 using 193 nm light. Typically the
L a s e r E n e r g y (p.J) quantum efficiency is higher for light of shorter wave-
Fig. 14. Measurements of collected charge versus laser energy length. Using the F o w l e r - D u b r i d g e [14] theory for
for three representative cases. photoemission, these previous measurements of quan-
S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230 227

tum efficiency can be scaled for our wavelength of 266 proportional to absorbed laser energy. A 90% increase
nm predicting r / = 10 8. The quantum efficiency can in absorption was measured for p-polarized light over
be increased in large electric fields ( > 10 M V / m ) s-polarized light, predicting a 90% improvement of the
through the Schottky effect. In order to increase r/ quantum efficiency for p-polarized injection. The mea-
from 10 -s to 10 -4, electric fields of almost 1 G V / m sured enhancement was only 50%. This discrepancy
are necessary at the cathode surface [15]. As was could be due to a difference in the relative reflectivi-
mentioned above, macroscopic fields at the cathode of ties of the damaged cathode in comparison to those of
only 50 M V / m were reached, however in the damaged the copper mirror.
area electric field enhancement factors of 20 are possi-
ble due to the surface roughness. The enhancement of
quantum efficiency from the damaged area over un- 7. Simulation of results
damaged area could explain the factor of 3 decrease in
quantum efficiency observed as the laser spot was One expects the charge measured at the exit of the
directed away from the damaged area. A more recent rf gun to be a linear function of applied laser energy to
measurement of quantum efficiency (1.4 × 10 4) for the photocathode. However, this is not the case for the
polished, clean Cu with low applied fields using 266 nm U C L A experiment thus far. The quantum efficiency
laser pulses was reported by Srinivassan-Rao. Al- data, which measures the total extracted charge out of
though this value agrees with our measurement, it the rf gun as a function of laser energy, is very non-lin-
implies that field enhanced emission did not occur and ear. From zero nC to approximately 0.5 nC the charge
does not explain the difference in photoemission be- extracted follows a linear trend. From 0.5 to 3 nC the
tween the damaged and undamaged parts of the cath- extracted charge falls off unexpectedly. In addition,
ode. The Fowler-Dubridge theory for the one photon streak camera measurements indicate that the beam's
photoelectric effect predicts a linear dependence of temporal length expands considerably from that of the
charge production on incident laser energy. However, laser. Because of the highly dynamical behavior of the
the measured charge vs laser energy manifests a satu- electron beam near the cathode, a series of P A R M E L A
ration of charge output for laser energies above 50 IxJ. simulations have been carried out to model the evolu-
The saturation can be explained by space charge ef- tion of the beam in this region, both in time structure
fects near the cathode surface. When the electrons are and in maximum allowed charge. These simulations
produced by the laser pulse, they are emitted as a thin give excellent agreement with the experimental data.
disk from the cathode: approximately 80 Ixm thick and Also, a simple model is put forth to explain the physi-
1 mm in diameter corresponding to the laser spot size cal process which limits the emission of charge.
on the cathode. The space charge electric field be-
tween the electron bunch and the cathode can be 7.1. Simple model
approximated by a surface charge density and its image
charge in the cathode. Using this simple model, a The emission of a short pulse of electrons from the
space charge field equaling the accelerating field of 50 cathode can be thought of in simple terms. The maxi-
M V / m results from only 0.25 nC of charge. This value mum number of electrons which can be emitted is a
agrees with the 0° data. The particle accelerator code, function of the electric field on the cathode and the
P A R M E L A , has been used to model this space charge space charge field generated by the electron beam
effect. This code calculates the space charge forces itself. For a given accelerating electric field strength
between a user specified number of test particles as there is an associated peak beam surface charge den-
the particles are accelerated from the cathode in an rf sity that can be extracted, ~m =6-oEz, which is an
gun. However, experimentally the space charge prob- analog, for transient beams, to the Child-Langmuir
lem is complicated by the cathode damage. Because of limit. To understand what is going on in our gun one
this damage, most of the electrons could be produced must think of the local characteristics of the charge
from microemitters on the cathode surface. At these emission and take into account emission from the tails
emitters the space charge could be worse. Despite of the distribution. Because of the damage to our
these limitations in the computer modeling, the cathode, and the small laser spot we use, the emission
P A R M E L A simulation showed saturation similar to is very local, with hot spots of unknown distribution
that of experimental data. The increase in quantum and size. These local emitters attempt to exceed the
efficiency for p-polarized vs s-polarized injection is maximum extractable current, and are therefore driven
probably due to the difference in the reflectivity of into saturation. However, the tails of the laser distribu-
copper at these polarizations. The reflectivity of a tion are not saturated and continue to emit until they
copper mirror was measured as a function of incident saturate as well, and so on. To this end we assume an
angle for both s- and p-polarized 266 nm light. Accord- effective Gaussian sigma o-r and then calculated the
ing to the generalized Fowler-Dubridge theory, r/ is emission profile as a function of charge deposited on

228 S.C. Hartman et al./Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230

Table 1 expect with no space charge depleted emission. The

Design and model parameters for the rf gun agreement is quite good, suggesting that emission from
Design Modeled a small damaged region coupled with a lower than
parameters parameters expected peak rf field in the first half cell produces the
o-~[psl 2 1 observed behavior.
o-r [mm] 3 0.3
Charge [nC] 0-3 0-3 7.3. A simple theoretical model
05[deg] 55 55
E~ 1/2 cell [MV/mJ 100 50 Consider a radial Gaussian emission region [6] which
El) full cell [MV/m] 100 100 produces a surface charge density of the form
Solenoid [A] 200 200
8 E / E [%] 0.1 3.5
~2r,y,n ["rr mm mrad] 1 10
MeV/c 4.5 3.5
where o-r is the effective rms spot size. Now, elemen-
tary considerations limit the maximum surface charge
density which can exist before electrons are re-ab-
the cathode. We then compare the experimental data,
sorbed by the cathode to be
simulation results, and theoretical curves.
~'m = eoEo sin(&0). (12)
7.2. Simulation
To find the total charge emitted, we must add up those
emitted in this saturated region and those in the unsat-
The rf gun when operating under the design param- urated tail. The total emitted in the saturated region is
eters in Table 1 behaves as expected according to
PARMELA simulations. No space charge suppression Q1 = "rra2. (13)
is seen, all of the charge deposited on the cathode is Here the saturated radius is simply
extracted. It has become apparent that the rf gun is not
operating under its design specifications. In short, we
believe that due to the damage on the cathode and the at=~[ 21nIl 2"rr°'r-~'~mJ] O'r" (14)
inherent laser spot size, that the effective spot sizes of
photoemission are sub-millimeter. Aggravating this is We can easily integrate the charge distribution emitted
the fact that it is also conceivable that the half-cell in the tail,
field is somewhat less than 100 M V / m . The rf gun was
tuned in place to maximize its shunt impedance, but Qt=2~r[~[,Q~°2]exp(-r2/2%2)rdr, (15)
since the shunt impedance in the half cell is less per Jar [ 2"rr°'r ]
unit length than in the full cell this is not a good which yields
technique to use. To model this effect the peak field in
the half cell was set to 50 M V / m and the full cell field Qt = Q0 exp(-aat/2~r,2). (16)
to 100 M V / m . Also, to model the small effective The total charge off the cathode is that given off in the
emission area on the cathode an effective Gaussian saturated region plus that given off in the tail
spot size of ~r = 0.3 mm was chosen. For comparison,
the design parameters and the modeled parameters are Q, = 32mrrat2, and Q2 = Qo e x p ( - a2/20"r 2)
given in Table 1. Fig. 1 shows the beamline configura- (17)
tion used. PARMELA simulations were then run with
with the total
the modeled parameters from the photocathode to the
first phosphor, where the extracted charge was mea-
sured. The simulation [6] was carried out from 0 to 2.3 Q~'sl = Q1 + Q2 = Q0 1 + In , (18)
nC in 5 pC steps. It is important to note that all of the
charge that is lost is lost in the first several time steps where
near the cathode region showing that they are forced
back into the cathode by longitudinal space charge. F.,, 2,rro.2 (19)
There are no particles lost transversely or by any other
means thereafter. Fig. 2 shows a plot of the experimen- is the peak surface charge density which would be
tal quantum efficiency data and the simulated data. emitted without the presence of space charge forces.
The experimental data was fit to a line for values Note that this expression is valid only when the emis-
below 0.5 nC and then it was extended to 2 nC. This sion is already in saturation 02 o > •m). Fig. 16 shows
was then plotted as a rcfcrence of what one would the calculated curve plotted with the experimental data
S.C. Hartman et al. / Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230 229

Pulse Lengthening due to Space Charge
E x p e r i m e n t a l Data 8=2 ° any ¢ Ii . . . j
1.4" 2.5 , i r , , I '
Simple T h e o r y [ t f ~
j l
1.2" . .... P a r m e l a Simulation I w'] laser injection tim, f
j ~inj= 1 nC

Qoxt = 0,74 nC

•" 15 ~x =0.6 mm
~y = 0.1 mm
0.s e
•~ 1 ,,5J o~,¢aser= 2 psec
1 ~,S aF B# 0 °fw rO °~''"''°° C

,~ 0.6
Note: beam FWHM = 10.7 psec

© 0.4' %
0 4
12 16 20

0.2 (psec)
/ 266nm laser pulse
•1 2 ps FWHM Fig. 18. L e n g t h of the b e a m during and just after emission.
0 ' ' ' ' ' I ' ' ' I ' ' ' ', " " "

0 0.4 0.8 1.2 1.6 2

Expected C h a r g e (nC) is about 0.32 mm, in close agreement with that used in
Fig. 16. C a l c u l a t e d curve plotted with the e x p e r i m e n t a l data the PARMELA runs.
and P A R M E L A data.

Pulse lengthening and energy spread

and PARMELA data. The agreement is again quite The presence of large longitudinal space charge
good, showing that the emission is governed by the forces which suppress the emission of the beam at the
space charge saturation of the emitters, and also by the cathode obviously must also lengthen the extent in
tails of the distribution. As subsequent measurements time of the beam. PARMELA was used to examine
showed that the accelerating field in the half-cell was this effect. In Fig. 18, the length of the beam during
actually 67 M V / m at the laser injection time, one can and just after emission is shown, with a very fine time
actually use the data in the saturated region to deduce resolution (20 fs). The beam, which should produce a
the emitter spot size in this model, by a single parame- nC, only manages 0.74 nC off of the cathode, while the
ter fit. Fig. 17 shows a fit to the data of the ratio of bunch time length expands (after only 20 ps of propa-
emitted to expected charge. The deduced emitter size gation from the beginning of the laser pulse), to 10.7 ps
FWHM, from the original laser pulse length of 4.5 ps
FWHM. Additional pulse lengthening due to longitudi-

Single parameter fit of QE data

. • • I • • . I . • I . . • I • • • I • • I • • .

- - R (Emit/Expect)
x~ x ×xX
Deduced ~=325 ~am x xx x
x x >~xx
(from fit parameter, with E sin(C0) = 67 M V / m ) 14.
x x
x ;"x
~' [2. xx x
x x

ml 039032438774 0 0019921
Chi 1 2636513315 NA
092657832632 N x

0 . . . . . . . ~ . . . . . . . . . ' . . . . , , ,

0.5 1 1.5 0.2 0.4 0.6 08 1 1.2

Expected Charge (nC) @ Cathode Charge @ psl (nC)

Fig. 17. R a t i o o f e m i t t e d to e x p e c t e d charge. Fig. 19. Pulse length as a function of b e a m charge.

230 S.C. Hartman et al. /Nucl. Instr. and Meth. in Phys. Res. A 340 (1994) 219-230

nal space c h a r g e - i n d u c e d energy spread, as well as the electron gun. F o r 70 ° laser injection, p-polarized
b e a m dynamics in the solenoid, produce an even longer light results in a 50% increase in q u a n t u m efficiency
b e a m at PS1. T h e pulse length as a function of b e a m over s-polarized light, probably due to their relative
charge is shown in Fig. 19. T h e s e calculated values are reflectivities.
in fairly good a g r e e m e n t with the observations. In the p r e s e n t experiments, the electron b u n c h
This straightforward model of the emission from length does not mimic the laser pulse length. T h e
d a m a g e d areas on the c a t h o d e with tails to the distri- shortest C e r e n k o v streak m e a s u r e m e n t s indicate 9 ps
bution explain many of the curious aspects of the b u n c h lengths. T h e physical m e c h a n i s m for this b u n c h
o p e r a t i o n of the rf gun to date. T h e spot sizes q u o t e d l e n g t h e n i n g is u n d e r investigation. P A R M E L A simula-
are all effectiL~e sigmas because we do not know the tions showed that b u n c h l e n g t h e n i n g on the o r d e r of 10
sigmas of the d a m a g e d regions, their n u m b e r s , or sym- ps can occur due to space charge, a l t h o u g h t h e b u n c h
metries. O n e r e c o m m e n d a t i o n is that the rf gun be r u n length m e a s u r e m e n t s did not show a correlation be-
at m a x i m u m rf power to obtain the highest peak accel- tween b u n c h length and charge o u t p u t from the gun.
erating field in the half-cell. T h e emission area is However, even with the b u n c h lengthening, b e a m cur-
nearly fixed, due to the d a m a g e on the cathode. In the rents of 100 A are produced.
near future the rf gun should be r e b a l a n c e d and a new
highly polished c a t h o d e installed. O n e should also be
extra careful not to o p e r a t e the gun in the explosive References
emission regime (which also is aggravated by small
laser spot size) or otherwise d a m a g e the cathode. T h e
[1] S.C. Hartman et al., Particle Accelerator Conf., San
extraction of charge and ultimate p r o p a g a t i o n of the Francisco, CA, 1991, p. 2967.
b e a m d e p e n d s on the c a t h o d e to a high degree, so [2] K. Batchelor et al., European Particle Accelerator Conf.,
great care must bc t a k e n not to d a m a g e the c a t h o d e in Rome, Italy, June 7-12, 1988.
any way. [3] C. Pellegrini et al., Particle Accelerator Conf., Washing-
ton, DC, 1993.
[4] Handbook of Chemistry and Physics.
8. Conclusion [5] K.-J. Kim, Nucl. Instr. and Meth. A 275 (1989) 201.
[6] S.C. Hartman and J.B. Rosenzweig, UCLA-CAA-
T h e U C L A p h o t o i n j e c t o r has b e e n o p e r a t e d suc- TECH-NOTE-internal, 1993.
[7] S.C. Hartman et al., Particle Accelerator Conf., Washing-
cessfully. T h e m e a s u r e m e n t s show t h a t the e m i t t a n c e
ton, DC, 1993.
scales as expected. T h e i m p o r t a n t thing to note is that [8] C. Lejeune and J. Aubert (eds.), Emittance and Bright-
for this set of e m i t t a n c e runs the rf p h o t o i n j e c t o r h a d a ness Definitions and Measurements (Academic Press,
field imbalance. T h e field in the full cell was 1.8 times New York, 1980).
that in the half cell. This c o n t r i b u t e d to e m i t t a n c e [9] G. Hairapetian et al., Particle Accelerator Conf., Wash-
blowup, as did the p h o s p h o r screens used to m e a s u r e ington, DC, 1993.
the emittance. T h e p h o s p h o r screens were placed o n t o [10] K.T. McDonald, IEEE Trans. Electron Devices ED-35
the b e a m l i n e at a 45 ° angle which created b r o a d e n i n g (1988) 2052.
of the line widths. T h e q u a n t u m efficiency measure- [11] J. Davis et al., Particle Accelerator Conf., Washington,
m e n t o f C u i n a rf gun resulted in ~ = l x 1 0 4. T h e DC, 1993.
[12] D.W. Decker, J.P. Waldron and R.J. Jaccodine, J. Opt.
photocmission from Cu u n d e r macroscopic electric
Soc. Am. 54 (1964) 216.
ficlds of 50 M V / m appears to be e n h a n c e d by d a m a g e [13] Y. Kawamura and K. Toyoda, Appl. Phys. Lett. 45 (1984)
on cathode. However, for charge levels g r e a t e r than 307.
0.25 nC the q u a n t u m efficiency is r e d u c e d by space [14] Appl. Phys. Lett. 49 (1986) 912.
charge n e a r the p h o t o c a t h o d e . Despite the limitations [15] T. Srinivasan-Rao, J. Ficher and T. Tsang, J. Appl. Phys.
imposed by space charge, up to 3 nC is p r o d u c e d from 69 (1991) 3291.