THE SECONDARY ELECTRON YIELD OF TECHNICAL MATERIALS

AND ITS VARIATION WITH SURFACE TREATMENTS

V. Baglin, J. Bojko1 , O. Gröbner, B. Henrist,

N. Hilleret, C. Scheuerlein, M. Taborelli CERN, Geneva, Switzerland

Abstract effectiveness discussed in the context of a possible use in

an accelerator.
Secondary electron emission of surfaces exposed to
oscillating electromagnetic field is at the origin of the 2 EXPERIMENTAL SET UP AND
multipacting effect that could severely perturb the PROCEDURE
operation of particle accelerators. This contribution tries
to illustrate by measurement results, the origin of the SECONDARY ELECTRON
COLLECTOR (cage)
secondary electron emission as well as the main reasons
for the discrepancies between technical materials and pure
metals. The variation of the secondary electron yield with PRIMARY
SAMPLE
ELECTRONS
the incident electron energy will be discussed for various
ELECTRON GUN
types of technical surfaces. The influence of a gas (0-3 kV)
condensation on these surfaces will also be addressed in
the context of the LHC accelerator. Various treatments SECONDARY
aiming at a permanent reduction of the secondary electron ELECTRONS
yield will be presented. A special attention will be paid to
the decrease of the secondary electron yield under electron
or photon impact and to its possible beneficial Figure 1: The experimental set-up
consequences for the processing of devices prone to
multipacting The measurement principle consists in recording
simultaneously the current from a sample and at a
1 INTRODUCTION secondary collector called cage while bombarding the
sample by primary electrons with a variable energy
The electron multiplication on surfaces exposed to an between 60 and 3000 eV. A scheme of the set up is
oscillating electromagnetic field causes the phenomenon shown on Figure 1. The measurement assembly is
of multipacting, which can degrade significantly the mounted in a bakeable, all metal U.H.V. system pumped
performance of particle accelerators. Such phenomena by a 260 l/s turbomolecular pump and equipped with
have been described in many accelerators, especially in calibrated Bayard Alpert gauge and residual gas analyser. A
accelerating cavities and more recently in the SPS [1] leak valve is used to feed various gases in the system in
where measurable pressure increases have been observed
during the circulation of LHC type proton beams. In this
case the electric field generated by the passing bunches can
SECONDARY ELECTRON

2.5
accelerate electrons which, at their impact with the COPPER AS RECEIVED
COPPER BAKED TO 300 C
vacuum chamber create secondary electrons and stimulate 2.0
COPPER + A.G.D.
neutral molecular desorption. Among other parameters,
YIELD

1.5
the multiplication of the incident electrons depend on the
secondary electron yield (S.E.Y.) of the bombarded surface 1.0

which has been measured for most pure metals before 0.5
1940 [2], [3], [4], [5], [6]. However these data are not
applicable to the real environment of accelerators built 0.0
0 500 1000 1500 2000 2500 3000
from technical materials e.g. stainless steel, aluminum ENERGY (eV)
alloys or copper covered with their natural oxide and Figure 2: The S.E.Y. of copper for various surface
contaminants. In this contribution we will present the treatments
experimental set up and the procedure used to study the
order to modify the studied surface using ion
S.E.Y of technical materials then show some results
bombardment. The samples (maximum 13) are mounted
concerning these. Lastly, possible methods used to limit
on a rotatable sample holder.
the secondary electron yields will be addressed and their
A very important feature for the measurement of the
S.E.Y. is to control carefully the electron dose needed for
1
Present address: MPI-Institut für Plasmaphysik Wendelsteinstrasse 1 D-17491 Greifswald Germany

Proceedings of EPAC 2000, Vienna, Austria 217

a measurement. It will be shown later in this paper that baking under vacuum (e.g. to 300ºC), a process leading
the S.E.Y. is strongly dependent on the dose of primary mainly to the removal of the water vapor adsorbed on
electrons. For this reason the measurements are made their surface.
using short pulses (typically 30 ms) of low primary The two following experiments illustrate the
electron current (some nano amperes). Under these effect of the oxide layer and of the adsorbed water on the
S.E.Y. :
conditions, the total dose of electrons for a measurement
The influence of the natural oxide layer is shown on the
between 60 and 3000 eV is less than 10 nC/mm2.
Figure 4, displaying the evolution of the S.E.Y. of an
3 THE DIFFERENCE BETWEEN aluminum alloy (6061) as a function of the dose of argon
PURE METALS AND TECHNICAL 2.4
SURFACES

SECONDARY ELECTRON YIELD

2.2 Clean sample
The difference between pure metals and technical surfaces 2 45 monolayers of water

is illustrated in the Figure 2 which shows the variation of 1.8

85 monolayers of water
200 monolayers of water
the S.E.Y. measured for a copper sample in the as 1.6

1.4

4 1.2
SECONDARY ELECTRON

Aluminium 99.5%
3.5 Titanium 1

3 Copper OFHC
0.8
Stainless steel 0 500 1000 1500 2000 2500
2.5
TiN
YIELD

ENERGY (eV)
2
Figure 5: The S.E.Y. of copper covered
1.5 with condensed water
1
ions impinging on its surface. Starting at an initial value
0.5
of 3, at 300 eV, the S.E.Y. decreases with increasing
0
0 500 1000 1500 2000 bombardment close to the value obtained for pure
ENERGY (eV) aluminum: 0.95. The dose of argon ions (3.5 1017
Figure 3: S.E.Y. of various as received ions/cm2) needed to reach this value is equivalent to the
technical materials sputtering of a 30 nm thick oxide layer, a value close to
received state and after two different in-situ treatments: a that obtained for a similar layer by Auger analysis [7]. If
300 ºC bake-out and an argon glow discharge. In this the sample is re-exposed to air, the S.E.Y. is increased to
latter case the S.E.Y. of the pure material is obtained: a value higher than 2.5 which is subsequently decreased to
maximum yield 1.3 at 600 eV primary energy. The the bulk value after sputtering the equivalent of a 6 nm
highest yield is obtained for the as received sample thick aluminum oxide layer.
(greater than 2). A 300 ºC bake out decreases this value to The S.E.Y. of water condensed at liquid nitrogen
3.5
temperature on a baked copper surface has been measured
SECONDARY ELECTRON

3 for various water thicknesses. In Figure 5 the variation of

the S.E.Y. with the primary electron energy is plotted for
2.5
a clean sample and for coverages corresponding to 45, 85
YIELD

2 and 200 monolayers of condensed water.

INITIAL
TREATMENT
1.5
4 METHODS TO DECREASE THE
1
PURE ALUMINIUM
SECONDARY ELECTRON YIELD
nd TREATMENT
0.5 FTER AIR EXPOSURE
0 2E+17 4E+17 6E+17 8E+17 1E+18 The preceding measurements have shown the
ARGON ION DOSE (IONS/CM2) strong influence of an air exposure on the S.E.Y. of
technical materials. For example the argon ion glow
Figure 4: The S.E.Y. of aluminum alloy 6061
discharge treatment can reduce drastically the S.E.Y. of
as a function of the argon ion dose
aluminum but the effect is almost completely lost after an
1.8. The S.E.Y. of various technical materials is shown air exposure. In the case of accelerator components or
on Figure 3 in the as received state: Aluminum alloys vacuum chambers, it is almost excluded to treat the
have the largest S.E.Y. (greater than 3). For primary surfaces in-situ with the possible exception of bake- out.
energies lower than 300 eV, copper, stainless steel and Hence the efficiency of all attempts to decrease the S.E.Y.
titanium are equivalent. by modifying the surface composition is limited by the
The origin of the S.E.Y. difference between pure unavoidable subsequent exposure to air and to water vapor
materials and technical materials is due to the presence of during installation. Another possible way to lower the
a surface layer (oxide and contaminants) which can be S.E.Y. is to change the surface roughness. This causes
removed by an argon ion bombardment. Furthermore, the indirectly a reduction of the emissivity of the surface since
S.E.Y. of as received samples is significantly reduced by the solid angle for electrons to escape without further

218 Proceedings of EPAC 2000, Vienna, Austria

interaction with the vacuum chamber can be significantly
decreased. A third type of process will be described later
1.8
4.1 Changing the surface composition

SECONDARY ELECTRON YIELD

As received
1.6
2 hours 120°C
The argon ion glow discharge treatment is a powerful 1.4 2 hours 160°C
way to modify surfaces by ion bombardment that has been 2 hours 200°C
1.2
applied previously to the 2 km vacuum system of the 2 hours 250°C

former ISR at CERN [8]. By changing the gas used 1.0

during this treatment, it is possible to produce various 0.8

surface layers lowering permanently the S.E.Y. even after
0.6
0 500 1000 1500 2000 2500 3000
2.5 PRIMARY ELECTRON ENERGY (eV)
AGD AO2GD
MAXIMUM SECONDARY

+AIR +AIR
Figure 8: S.E.Y. of a TiZrV getter as a function
ELECTRON YIELD

2 BO N2GD
300ºC +AIR of the activation temperature
1.5 Titanium nitride is known to produce a reduction
of the S.E.Y. [9] that depends largely on the deposition
1
condition of the films. TiN coatings from various sources
show a large scatter in the maximum yields (between 2.5
0.5
and 1.5). The best layers can have a maximum yield
0 significantly lower than the S.E.Y. of any known metal
AS N2GD AGD AO2 G.D.
RECEIVED in the as received state. After a subsequent 150ºC bake
TREATMENTS out, the maximum yield is close to 1.4, baking to 300 ºC
Figure 6: Maximum S.E.Y. for niobium after lowers the yield to 1.2.
various surface treatments The activation of a getter layer [10] is another elegant
an exposure to air. This is illustrated in Figure 6 showing way to eliminate the oxide layer. In figure 8 the variation
in the case of niobium the maximum S.E.Y. after various of the S.E.Y. of a Ti Zr getter layer is plotted as a
glow discharge treatments using argon (A.G.D.), argon + function of the baking temperature. A marked decrease of
10% oxygen (AO2GD), nitrogen (N2GD). If most of the the S.E.Y. is visible when the layer is heated above its
beneficial effect of the discharge is lost after anair activation temperature. After a 300 ºC bake out, the
exposure following an AGD., a significant improvement highest yield is less than 1.2.
subsists after a 24 hours exposure to air following the
AO2GD. or an N2GD. In both these cases, the final yield
after exposure to air is close to 1.4, a value that can only 1.4

3 1.3
AS RECEIVED
SECONDARY ELECTRON

B-O 1.2 Cu after oxidation

2.5 N2GD in air at 20°C and
24h AIR 1.1 350°C bakeout
2 N2GD +24h AIR+BAKED 100C
1
YIELD

1.5
0.9 Cu after oxidation
in air at 350°C and
1 0.8 350°C bakeout

0.5 0.7

0.6
0
0 500 1000 1500 2000 0 500 1000 1500 2000 2500 3000
PRIMARY ELECTRON ENERGY (eV) primary electron energy in eV

Figure 9 : SEY of copper after a 24 hours 350°C bakeout

Figure 7: S.E.Y. of copper after various surface treatments compared to the SEY of copper after 5 minutes air
exposure at 350°C and 6 hours bakeout at 350°C under
vacuum
be obtained after an in-situ bake-out (B.O.) to 300 ºC.
In the case of copper, the curves giving the 4.2 Changing the surface roughness
evolution of the S.E.Y. following various treatments are
given in Figure 7. The permanent reduction of the S.E.Y. As explained above, the creation of a surface layer with
obtained after an N2GD is also visible. A mild in-situ increased roughness can permanently decrease the S.E.Y.
bake out to 100ºC restores most of the effect of the This modification can be produced either by a chemical
treatment when the system has been exposed for 24 hours reaction with the substrate or by the deposition of a
to air. strongly dendritic layer.

Proceedings of EPAC 2000, Vienna, Austria 219

 The first approach was applied to the case of copper These methods seem to be able to produce surfaces with
[11] which was oxidized at elevated temperature in air. very low secondary electron yields (lower than 1) but are
difficult to apply on a large scale as in an accelerator.
4.3 The "dose" effect
When a surface is exposed to the impact of an electron
beam, its S.E.Y. decreases [14] . In Figure 12 the

2.0

Cu #A as received

SECONDARY ELECTRON YIELD

1.8
UNBAKED COPPER 1.1E-7 C/mm2
1.22E-6 C/mm2
1.6
7.89E-6 C/mm2
1.15E-4 C/mm2
1.4
1.12E-3 C/mm2

1.2

1.0

0.8

Figure 10: Photograph of a textured copper surface 0.6

0 500 1000 1500 2000 2500

The stresses produced during the oxide growth often lead ENERGY(eV)

to cracks in the oxide film resulting in an increased Figure 12: Variation of the S.E.Y. of copper
surface roughness. It is also important that the formed with the incident electron dose
oxide has a low S.E.Y. as it is typically the case for variation of the S.E.Y. for an unbaked copper sample is
semi-conducting oxides like Cu2O. An in-situ bake out to plotted as a function of the electron dose. The S.E.Y.
350ºC at atmospheric pressure for 5 minutes followed by decreases for doses larger than 10-6 C/mm 2 and its
a 6 hours vacuum bake-out creates a surface with a low maximum stabilizes for doses greater than 1x 10-3 C/mm2
S.E.Y. (1.05) as shown on the Figure 9. That treatment at a value close to 1.2. This effect is permanent when the
causes an increase of surface roughness measured by surface is kept under vacuum (pressures < 1x10-5 Pa). The
xenon adsorption at 77K (B.E.T. method) [12] from 1.4 dose effect decreases the S.E.Y. for all primary electron
(as received state) to 6.5 (after air bake). energies as can be seen in figure 13 but the effect is more
The creation of a strongly dendritic surface on a pronounced for the lower primary electron energies. XPS
small copper sample, as shown on the electron analysis have shown that after irradiation, the carbon
microscope picture in figure 10 produces a surface which
has a S.E.Y lower than 1, even after an air exposure as it 1
is shown on the figure 11. Similar results have been ORMALISED DOSE VARIATION
(COPPER AS RECEIVED)
NORMALISED SECONDARY

published by A.N.Curren et al [13]. 0.9

ELECTRON YIELD

125 eV
SECONDARY EL ECTRON YIELD 300 eV
0.8 500 eV
500 eV
COPPER AS-RECEIVED 1000 eV
2.0 2000 eV
0.7
2500 eV

COPPER AFTER 350 ºC

1.5 25 eV
BAKE-OUT 0.6
1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02
DOSE (C/mm2)
1.0
AFTER 42 HOURS AIR Figure 13: Decrease of the secondary electron yield
EXPOSURE with the electron dose normalised to 1 before
0.5 PROCESSED COPPER bombardment
content on the sample surface is increased from 40% to
ENERGY (eV)
60% exclusively in the zone of the electron impact. The
500 000 1500 origin of the effect might be linked with the formation,
stimulated by the electron bombardment, of a carbon rich
surface layer. The origin of the carbon could be the
residual gas (e.g. CO, CO 2, hydrocarbons) or molecules
Figure 11: Secondary electron yield of a textured desorbing from the sample during the electron irradiation (
copper surface CO, CO2) [15]
Although not completely understood and difficult
to investigate, as the electron doses involved are too small

220 Proceedings of EPAC 2000, Vienna, Austria

 for the usual analysis techniques, this effect has been used electron yields measured under these various conditions are
since very long time in accelerators and is called: displayed in figure 15 which shows also that the decrease
processing! is faster for higher bias voltages (i.e. higher incident
As this effect could be a remedy for the electron electron energies).
cloud effect in the LHC, it has been studied in EPA ( a
low energy electron accumulator) at CERN, using a 5 CONCLUSIONS
The secondary electron emission is a surface dependent
2.4 A1 0 ph/m
phenomenon, more influenced, for technical metals, by
NORMALISED SECONDARY ELECTRON

2.2 A2 1.7e19 ph/m

BIAS: 820 V
2.0
A3 1.9e19 ph/m the surface preparation than by the material itself. Any
1.8
A5 3.5699e19 ph/m
insulating layers (e.g. oxides as well as adsorbed water)
A11 3.1871e20 ph/m
1.6 A13 2.2013e21 ph/m
significantly enhance the emissivity of surfaces. Various
YIELD

1.4 A20 1.1945e22 ph/m surface treatments involving a modification of the surface
1.2 (e.g. ion bombardment) or coatings (e.g. titanium nitride)
1.0 can reduce the S.E.Y. although a subsequent ambient air
0.8 exposure reduces significantly their efficiency. The dose
0.6 effect (processing) is a well-established and very powerful
0.4
0 500 1000 1500 2000 2500
method to circumvent the problems related with electron
ENERGY(eV) multiplication. Nevertheless, this method can only be
Figure 14: Variation of the S.E.Y. of with the photon dose applied under two conditions:
-A significant electron flux should be available to reach
remotely controlable system to measure the secondary a dose close to 10-3 C/mm 2 within a reasonable time and
electron yield of a copper sample. The synchrotron light the surfaces involved must be able to withstand the
radiated by the EPA electron beam ( 194 eV critical electron bombardment.
energy) travels along a tangential photon beam line. The -The electron multiplication must be controlled in order
sample is mounted at 90 º from the plane of the to avoid a runaway and possible damage.
synchrotron light and thus is not exposed to direct Further studies to better understand the physical
photons. Three dose experiments have been performed mechanisms at the origin of the dose effect could result in
using different bias applied to the sample: -45 V ( i.e. the an improved efficiency of the conditioning procedures in
2.5 accelerators.
nitial Normalised Max SEY
MAXIMUM SECONDARY ELECTRON

ample at - 45 V

2.0
ample at 0 V ample at + 350 V
REFERENCES
ample at + 100 V
[1] G. Arduini, K. Cornelis, O. Grobner, N. Hilleret, W.
1.5 Hofle, J.M. Jimenez, J.M. Laurent, G. Moulard, M. Pivi,
YIELD

100 V
350 V K. Weiss, this conference, Vienna, 2000
ample at + 825 V
1.0 -45 V [2] H. Bruining Physics and Application of Secondary
0 V puis 350 V ample at + 350 V

825 V
Electron Emission, Pergamon Press Ltd, London, 1954
~ 1 year of nominal
0.5 LHC operation [3] H. Bruining, J.H. de Boer, Physica, 5, 17, 1938.
[4] H. Bruining, Philips Tech. Rev. 3, 80, 1938
0.0
1.E+18 1.E+19 1.E+20 1.E+21 1.E+22 1.E+23 1.E+24
[5] L. Warnecke, J. Phys. radium, 7,270, 1936
PHOTON DOSE (ph/m)
[6] L.R.G. Treloar, D.H. Landon, Proc. Phys. Soc.
(London), B50, 625, 1938[
Figure 15: The variation of the copper maximum
[7] A.Grillot, technical note ISR-VA/AG/sm, 12/7/82
secondary electron yield as a function
[8] R. Calder, A. Grillot, F. Le Normand, A.G.
of the photon dose in EPA
Mathewson, Proc. 7 th Intern. Vac. Congr. , 231-234,
sample was only conditioned by diffused photons), Vienna 1977
+100 V, +350 V and 820 V (i.e. the sample was [9] E.L. Garwin, F.K. King, R.E. Kirby, O Aita, J.Appl.
bombarded by photoelectrons with approximately 100 eV Phys. 61, 1145, 1987
and 350 eV energy and 820 eV). The variation of the [10] C. Benvenuti et al., Vacuum, 53, 219, 1999
S.E.Y. measured at 820 eV is given in Figure 14 as a [11] I. Bojko, N. Hilleret, C. Scheuerlein, J. Vac. Sci.
function of the primary photon dose. When the sample is Technol. A, 18(3), 972-979, 2000
positively biased, the curves are very similar to those [12] V. Baglin Technical note LHC-VAC 97-03, 1997
obtained in the laboratory and saturate at a value close to [13] A.N.Curren , K.A. Jensen and R.F. Roman, NASA
1.2. When the sample is only bombarded by diffused report TP-2967, 1990
photons, the effect is much less pronounced and seems to [14] M. Lavarec, P. Bocquet, A. Septier C.R. Acad. Sc.
saturate at a S.E.Y. of 1.6. The maximum secondary Paris, t.288, B277, 1979
[15] M. Andritschky, Vacuum, 39,7/8,649-652, 1989

Proceedings of EPAC 2000, Vienna, Austria 221