Suppression of the Surface Charge Limit in
Strained GaAs Photocathodes*
T. Maruyamaa, A. Brachmamf, J.E. Clendennf, T. Desikaif, E.L. Garwina,
R.E. Kirbya, D.-A. Luha, C. Y. Prescotta, J. Turner3, and R. Prepostb
"StanfordLinear Accelerator Center, Stanford, CA 94309
Department of Physics, University of Wisconsin, Madison, WI53706
Abstract. Single strained, medium-doped (5x1018 /cm3) GaAs photocathodes show the surface charge
limit (SCL). The SCL poses a serious problem for operation of polarized electron sources at future
linear colliders such as the NLC/JLC. A high-gradient-doping technique has been applied to address
this problem. A 5 -7.5 nmp-type surface layer doped to 5xl019/cm3 is found sufficient to overcome the
SCL, while maintaining high beam polarization. This technique can be employed to meet the charge
requirements of the NLC with a polarization approaching 80%.
INTRODUCTION
Strained, medium-doped (5xl018 cm"3) GaAs photocathodes were introduced for
the SLAC polarized electron source in 1993. These photocathodes are capable of
producing up to IxlO11 electrons in a 2 ns pulse with 75-85% polarization. However,
they are susceptible to the surface charge limit (SCL). The SCL was attributed to a
photovoltage effect in the band bending region, which momentarily increases the
surface work function and thus suppresses emission. The SCL poses a serious problem
for operation of polarized electron sources at future linear colliders such as the Next
Linear Collider (NLC) and the Japan Linear Collider (JLC). The NLC is presently
designed to operate with a train of 190 micro-bunches, spaced 1.4 ns apart. Each
micro-bunch is required to have as much as 1.4xl010 e" per pulse in 0.5 ns at the gun,
totaling 2.7xl012 e" per train. The standard SLAC strained photocathodes saturate at a
level of SxlO11 e" for a 300-ns pulse, well below the space charge limit of the gun. The
SCL problem can be reduced significantly by increasing the /?-type doping
concentration to at least 2xl019 cm"3. However, increasing the doping level leads to
some depolarization of the electron spin. To increase the doping level without spin
depolarization, a high-gradient-doping technique has been explored. While a standard
SLAC photocathode is uniformly doped at 5xl018 cm"3, we have reduced the active
layer dopant density to 5xl017 cm"3 while increasing the density of a 10-nm surface
layer to 5xl019 cm"3. The lower-than-standard doping level in the active layer avoids
depolarization, while the high surface doping addresses the charge limit problem [1].
* This work was supported in part by Department of Energy Contract No. DE-AC03-76SF00515
CP675, Spin 2002:15th Int'l. Spin Physics Symposium and Workshop on Polarized Electron
Sources and Polarimeters, edited by Y. L Makdisi, A. U. Luccio, and W. W. MacKay
© 2003 American Institute of Physics 0-7354-0136-5/03/$20.00
1083
DOPANT CONCENTRATION DEPENDENCIES
A series of measurements were performed using unstrained 100 nm thick GaAs to
study the SCL dependence on the doping concentration [2]. The p-type doping
concentrations were 5xl018 cm"3, IxlO19 cm'3, 2xl019 cm'3, and 5xl019 cm'3 for a set
of four samples. The experiments were performed using the 121 kV diode gun in the
SLAC Gun Test Laboratory (GTL). Figure l(a) and l(b) show the temporal profiles
of the emission current pulses measured for a number of light pulse energies for the
5xl018 cm"3 sample and for the 2xl019 cm"3 sample, respectively. As seen in Figure
l(a), the emission current follows the rectangular laser shape when the laser energy is
low. As the laser energy is increased, the emission peak at the start of the light pulse
reflects retardation in the build-up of the photovoltage due to the finite time of the
charging of the surface states. As the photovoltage builds up, the emission current
decreases exponentially. The suppression of the later portion of the emission pulse
manifests the decrease of the surface escape probability with the growth of the
photovoltage and its relaxation to the steady state. As the laser energy is increased, the
photovoltage builds up more quickly and the emission current suppression is more
pronounced. For the 1x1019 cm"3 sample (not shown), the charge limit behavior is
much reduced. As seen in Figure l(b), the 2xl019 cm"3 sample does not show the
charge limit effect.
100
so
200
lime {01}
100
150
lime (ns)
FIGURE 1. The temporal profiles of the emission pulses for a) the 5xl018 cm"3
sample and b) the 2xl019 cm"3 sample.
HIGH GRADIENT DOPED STRAINED GaAs
While high doping concentration is necessary to suppress the SCL, electron
polarization will be significantly reduced as the doping level is increased. One way to
achieve the high doping level without suffering from polarization loss is to dope only
-10 nm of the surface layer at >2xl019 cm"3. To test this technique, polarization
measurements were performed at the SLAC Cathode Test System (CTS) using high
gradient doped GaAs. While the active layer is 100 nm thick GaAs, the surface layer is
doped at 4xl019 cm"3 and the rest at 3xl017 cm"3. Two samples with a highly doped
1084
layer thickness of 25 nm and 40 nm were used. The thickness of the highly doped
surface layer was reduced by anodization/stripping, and the polarization and quantum
efficiency (QE) measurements were repeated. Figure 2 shows the peak polarization as
a function of the highly-doped layer thickness. The polarization increased significantly
as the highly-doped layer thickness was reduced, reaching 80% with about 7.5 nm of
the highly-doped layer.
90
80
o
EL
80
50
0
10
20
30
40
Highly Doped Layer Thickness {nm}
FIGURE 2. The peak polarization as a function of the highly-doped layer thickness.
The next question is whether 7.5 nm of highly-doped layer is thick enough to
overcome the SCL problem. The GTL was used to study the charge performance.
Figure 3 shows the temporal profiles of the emission current pulses measured using
varying laser energies. As the laser energy was increased, the temporal profile simply
scaled without developing the leading edge spike typically observed in the surfacecharge-limited photoemission. Figure 4 shows the charge output as a function of the
laser energy. The observed charge output increases linearly with the laser energy,
reaching SxlO11 e'/pulse.
Although the sample did not show any charge limit behavior, the QE was lower
than expected. The low QE was thought to be due to the energy barrier in the
conduction band.
1085
2.0
€
•S
«
1.0
0.0
100
200
300
400
40
80
Laser Energy (p.J)
Time (ns)
120
FIGURE 3. The temporal profiles of the FIGURE 4 The photoemission charge
electron emission current.
per pulse as a function of the laser energy
HIGH GRADIENT DOPED STRAINED GaAsP
To remove the energy barrier in the conduction band, the active layer was changed
to GaAso.95Po.o5. The 5% phosphorus raised the conduction band by 58 meV
compensating the energy difference due to the high gradient doping. A QE of 0.2%
and peak polarization of 80% were measured at room temperature at 805 nm. The
charge output was measured using two laser systems, a flashlamp-pumped Ti:Sapphire
laser (Flash-Ti) with 270 ns pulse and a Nd:YAG pumped Ti:Sapphire laser (YAG-Ti)
with 4 ns pulse. The laser spot size of both lasers was set to 14 mm. Figure 5 shows
the charge output as a function of the Flash-Ti laser energy with and without the
YAG-Ti laser. The YAG-Ti laser by itself produced 2.3xlOne7pulse, equivalent to 9.2
A peak current, twice the NLC requirement. The charge output was linear up to the
maximum laser energy, producing a maximum charge of 2.2xl012 e" in 270 ns. If one
assumes a linear scaling with spot size, this charge is equivalent to 4.5xl012 e" (2.6 A)
from a 20 mm diameter cathode.
A SLAC high energy experiment, E158, to measure parity violation in Moller
scattering (e~e~ —> e"e") will use a 48 GeV polarized electron beam scattering off
unpolarized electrons in a liquid hydrogen target. The experiment requires a beam
intensity of SxlO11 e" in a 370 ns pulse at the gun. Since the beam intensity
requirement is difficult to meet using the standard SLAC photocathodes, it was
proposed to use the newly developed high-gradient-doped GaAsP photocathode
described above. A sample was installed in the SLAC polarized electron source and
heat-cleaned and activated twice. The initial QE was 0.4% at 805 nm. Figure 6 shows
the charge output as a function of the laser energy for a 100-ns long pulse. The charge
output was linear up to the maximum laser energy, producing a maximum charge of
2.3xl012 e" in 100 ns (3.7 A), which is nearly equal to the peak current required for the
NLC. Finally the charge output was observed to scale with the laser pulse length,
reaching more than 8xl012 e" for 370 ns. The charge output corresponds to ten times
the E158 requirement and more than twice the NLC-train charge.
1086
I
liT
I
2.5
T
2.0 _ • Long + Short Pulse
o Long Pulse Only
• QE = 0.31
o
J2
£1.5
•5
o
«N
D_
'0 1.5
'o
x 1.0
<D
s>
0.5
00.5
0.0
2-2002
8630A16
0.0
50
100
150
200
Flash-Ti Laser Energy (jiJ)
2-2002
8630A19
0
50
100
150
200
Laser Energy (uJ)
FIGURE 5. The photoemission charge FIGURE 6. The photoemission charge
per pulse as a function of the laser
per pulse as a function of the laser energy
energy for Flash-Ti only (open circles) measured at the SLAC injector,
and Flash-Ti and YAG-Ti overlaid (solid
circles).
CONCLUSIONS
The SCL effect has a strong doping concentration dependence, decreasing with
increased doping and diminishing to zero at a doping level of 5xl019 cm"3. The charge
capabilities of high-gradient-doped strained photocathodes have been determined. A
charge as large as 2.2xl012 e" per pulse was produced from a 14-mm diameter laser
spot in 270 ns. By overlaying a short pulse laser, the peak current capability was also
determined. A peak current as high as 9.2 A was extracted. There were no indications
of a surface charge limit even when the QE decreased, and the charge output was
limited by the laser energy. A cathode of this type is being used in a current high
energy polarization experiment at SLAC, for which the charge performance has been
consistent with, if not superior to, the GTL results. This is the first demonstration that
a NLC compatible beam with polarization approaching 80% is achievable.
REFERENCES
1. T. Maruyama, A. Brachmann, J.E. Clendenin, T. Desikan, E.L. Garwin, R.E. Kirby, D.-A. Luh, J.
Turner, and R. Prepost, Nucl Instrum. Meth. A 492, 199 (2002).
2. G.A. Mulhollan, A.V. Subashiev, J.E. Clendenin, E.L. Garwin, R.E. Kirby, T. Maruyama, R.
Prepost, Phys. Letters A 282, 309 (2001).
1087