Electron emission during grazing impact of atoms
on insulator and metal surfaces
H. Winter
Institut für Physik der Humboldt-Universität zu Berlin
Invalidenstr. 110, D-10115 Berlin, Germany
Abstract. The coincident detection of projectile energy loss with the number of emitted electrons for the scattering of
atoms from atomically clean and flat insulator and metal surfaces under grazing angles of incidence allows one to
identify the relevant electron excitation and emission processes. From detailed studies for the scattering of H and He
atoms from a LiF(001) as well as from an Al(111) surface we find evidence for clearly different interaction mechanisms.
For not too high projectile velocities with respect to kinematic thresholds, electron emission for the insulator target is
understood by an electronic promotion mechanism with the formation of H- ions as dominant precursor, whereas for the
metal target electronic excitations are governed by direct energy transfer in binary collisions with the atomic projectiles.
Based on these features we reveal a microscopic understanding for the well established, but so far poorly understood
property that electron emission induced by atom or ion impact on insulator targets is more efficient than for metals.
In this respect we also mention that for insulator
targets no threshold effects of projectile stopping are
observed down to rather low projectile energies
despite the presence of a wide energy band gap [4, 5]
INTRODUCTION
For impact of atoms or ions on solid surfaces one finds
the interesting feature that electron emission is much
more efficient for many insulators than for metal
targets [1–3]. In particular for ionic crystals, this
observation comes somewhat surprising, since binding
energies of valence band electrons for e.g. LiF are
larger than 12 eV, whereas the work function of a
metal surface amounts to typically 4 to 5 eV (cf.
Fig.1).
Recently we have started to investigate the emission of
electrons during impact of neutral atoms (no
contributions of potential electron emission) on
insulator and metal surfaces under a grazing angle of
incidence. In this regime of scattering projectiles are
steered by atoms of the topmost surface layer in a
sequence of small angle scattering, so called (surface)
“channeling”, with well defined trajectories in front of
the solid. An important feature of our studies is the
application of concepts of translation energy
spectroscopy which allows us to obtain information on
the overall inelastic processes. In addition, the
projectile energy loss is recorded in coincidence with
the number of emitted electrons so that we can relate
electronic excitations of the target to a specific number
of emitted electrons. This experimental method
provides a detailed investigation of the interaction
mechanisms, and we will summarize here recent
progress
achieved
towards
a
microscopic
understanding of the electronic excitation and emission
phenomena for insulator as well as for metal targets.
FIGURE 1. Sketch of energy diagram for electronic
structure of metal and insulator target (CB = conduction
band, VB = valence band).
CP680, Application of Accelerators in Research and Industry: 17th Int'l. Conference, edited by J. L. Duggan and I. L. Morgan
© 2003 American Institute of Physics 0-7354-0149-7/03/$20.00
82
excitons, the local excitation of valence electrons in
the vicinity of a halogen lattice site. 3. A fair number
of electrons are emitted under these conditions (see
below results obtained with an Al target).
EXPERIMENT AND RESULTS
In the experiments H and He atoms are scattered under
a grazing angle of incidence of typically 1o from a
LiF(001) surface, an ionic crystal with a band gap of
14 eV that extends to vacuum energies, as well as from
an Al(111) surface, the prototype of a free-electron gas
metal. The projectile energy loss is measured with a
time-of-flight (TOF) setup with an overall time
resolution of some ns (corresponds to some eV for the
projectile energy), and neutral atoms are produced by
(near) resonant charge transfer in a gas target
positioned behind an electric beam chopper. The
number of emitted electrons per projectile impact is
derived from the pulse height of a surface barrier
detector (SBD) biased to 25 kV [6]. For each TOF
event the coincident pulse height of the SBD is
recorded and stored in a 2D-array of a memory unit.
Corresponding plots, i.e. TOF spectra (energy loss) vs.
SBD pulse height (number of electrons) gives a
detailed overview of the inelastic interaction and
electron emission processes.
.
FIGURE 3. Same plot as in Fig. 2, but for scattering from
Al(111).
In Fig.3 we display a 2D-spectrum obtained under the
same conditions of scattering for an Al(111) target. A
pronounced difference of the two spectra obtained for
the insulator and metal targets (cf. Figs. 2 and 3) is
evident. A clearly smaller total electron yield is
observed for the metal in comparison to the insulator
(see small signal in Fig. 3 representing events for the
emission of one electron). Furthermore the overall
energy loss is much larger and does not show discrete
structures.
12
0
H - LiF(001) , Φin=1,8°
0
electron yields
10
FIGURE 2. 2D-plot of projectile energy loss vs. electron
number for scattering of 1 keV H atoms from LiF(001) under
an angle of incidence Φin = 1.8o .
H - Al(111)
8
6
4
2
As a first prominent example we show in Fig. 2 a 2Dspectrum obtained for grazing scattering of 1 keV
hydrogen atoms from a LiF(001) surface [7] which
shows a number of interesting features: 1. The energy
loss spectra show discrete structures which are
attributed to specific excitations of valence band
electrons. 2. The discrete patterns are also observed for
the emission of no electron which points to internal
excitations of the target; Roncin et al. [8] have
attributed this feature to the production of surface
0
0,0
0,1
0,2
0,3
0,4
0,5
0,6
projectile velocity (a.u.)
FIGURE 4. Total electron emission yields for scattering of
H atoms from LiF(001) and Al(111) under Φin = 1.8o as
function of projectile energy.
In Fig. 4 we present a direct comparison of total
electron yields derived from the relation
83
∞
γ
=
of about 2 eV which is interpreted by the (internal)
excitation of surface excitons with a binding energy of
typically 1 eV with respect to vacuum, whereas
emitted electrons result from initial valence band states
with a mean energy of about 13 eV and about 1 eV
kinetic energy after ejection.
∞
∑n W ∑W
n
0
(1)
n
0
with Wn being the total probability (projection of 2Dspectra on electron number axis) for the emission of n
electrons. Note that the experimental method provides
direct information also on those events which are
related to the emission of no electrons, since in this
case energy loss data are referred to the noise spectrum
of the SBD. This feature is the basis for accurate
(small) experimental total electron emission yields
obtained by this method [7]. The total electron
emission yields for H atoms scattered from LiF(001)
and Al(111) under Φin = 1.8o plotted in Fig. 4 as
function of projectile velocity show for both targets a
monotonic increase with velocity. The data reveal the
same gross feature as found in previous studies using
near normal impact, where clearly higher γ are
observed for insulator than for metal targets. From the
data shown one concludes on a smaller impact velocity
for the onset of kinetic electron emission for the LiF
target than for the metal.
For metal targets the energy loss spectra are clearly
different (cf. Fig. 3). We observe a much higher
energy loss than for a LiF target, and a comparison of
data for the emission of no and one electron indicates
that contributions of electron emission to energy loss
are small; i.e. most of the dissipated projectile energy
goes into internal low energy excitations of the metal
target.
DISCUSSION
From the data presented here for interactions of fast
atoms with insulator and metal targets one expects
clearly different electronic excitation and emission
mechanisms. In a simple classical picture of energy
transfer in binary elastic collisions, one finds for a free
electron gas with maximum kinetic energy Emax (Fermi
energy for metals) and minimum binding energy Ebind
(work function for metals) a threshold of projectile
energy for electron emission [9]
E bind 
M 
E th =
 E max − E max ( E max + E bind ) +
 (2)
2 me 
2 
which amounts for H atoms (mass M) and Al (Emax =
EF = 10.6 eV; Ebind = W = 4.3 eV) to 168 eV, whereas
for LiF (Emax ≈ 4 eV and Ebind = 12 eV) we have
Eth = 1836 eV. This is in sheer contrast to the
experimental observations with the LiF target.
Interesting information on the interaction mechanisms
is also obtained from the energy loss spectra. We show
in Fig. 5 TOF-spectra for 1.2 keV H atoms scattered
from LiF separated with respect to events for specific
number of emitted electrons.
counts
0 electron
1 electron
2 electrons
3 electrons
zmin
0
20
40
60
80
100
energy loss / eV
FIGURE 5. Energy loss spectra for 1.2 keV H atoms
scattered from LiF(001) under Φin = 1.8o related to the
emission of 0, 1, 2, and 3 electrons.
F°+H-
C
vacuum
(F°+H°)
F - * +H°
B
F°+H -
The discrete structure of the spectra is the direct
consequence of the wide band gap of ionic crystals
(and a sufficient experimental energy resolution). This
allows one to extract the contributions of internal and
external excitations to the overall inelastic interactions.
In detail, comparison of the second peak for the 0electron data with the first peak of the spectrum related
with the emission of one electron reveals a small shift
F - +H°
A
F-
F-
F - +H°
FIGURE 6. Upper panel: scattering geometry for grazing
ion surface collisions; lower panel: sketch of initial and final
potential energy curves for production of excitons (F-*),
emission of electrons (e-), formation of negative ions (H-).
84
derived from Eq. 2 using bulk values for the Al target.
We interpret this finding by the fact that projectiles
interact under conditions of surface channeling with
the selvage of conduction electrons at the surface with
reduced electron density. The observed vth (Eth)
corresponds to an effective Fermi momentum kF* =
0.67 kF which relates via kF = (3π2 ne)1/3 to an electron
density ne of about 25 % of the bulk value. This
density is reached at a distance of about 3 a.u. from the
surface, in fair agreement with trajectory calculations
for the present system. The inset of Fig. 7 shows the
same data on an enhanced vertical scale and reveals an
onset of electron emission at about the expected
threshold for an electron gas of bulk density. This is
explained by small fractions of projectiles with
trajectories affected by surface defects which probe
higher electron densities up to bulk values.
An important aspect for an understanding of the
efficient emission of electrons for atom impact on
insulators is based on the recent observation of large
fractions of negative ions after grazing scattering of
atoms/ions from alkali halide surfaces [10]. This
feature is the basis of an interaction model sketched in
Fig. 6 where the formation of a negative ion formed in
binary collisions with halogen ions (“active sites”) is
the common precursor for electron excitation (surface
exciton) and emission. Note that the energy defect in
the first part of the interaction sequence (labelled A in
Fig. 6) is substantially reduced by the Coulomb
interaction of the hole remaining at the active site with
the negative ion state (~ Madelung potential). Then
transition probabilities are enhanced (estimated by
Demov theory) and lead to an efficient formation of
transient negative ions and finally to emission of
electrons. The branching ratio of internal excitations
and electron emission results from the curve crossing
(labelled B) of the negative ion and (surface) exciton
diabatic potential curves (estimated by Landau-Zener
theory). Detailed measurements of electron emission
and excitation yields as well as statistics for impact of
hydrogen atoms on LiF(001) can be consistently
described by this approach [11].
ACKNOWLEDGMENTS
The fruitful collaboration with K. Maass, S.
Lederer, Dr. A. Mertens (Berlin) and Profs. HP.
Winter and F. Aumayr (Wien) is gratefully acknowledged. This work is supported by the Deutsche
Forschungsgemeinschaft under contract Wi 1336.
0,04
o
He - Al(111)
0,002
electron yield
0,03
0,02
REFERENCES
[1]
0,001
[2]
[3]
0,01
0,000
0,00
0,00
0,00
0,05
0,10
0,05
0,15
0,10
0,20
0,15
[4]
[5]
0,20
projectile velocity (a.u.)
[6]
FIGURE 7. Total electron emission yields for scattering of
He atoms from Al(111) under Φin = 1.8o as function of
projectile velocity. Insert: scale of ordinate enhanced by
factor of 20.
[7]
Fig. 7 shows total electron emission yields for grazing
scattering of He atoms from Al(111) under Φin =1.8o
as function of projectile velocity. The data reveal a
threshold behaviour which is consistent with a model
of energy transfer in binary encounter of projectiles in
a free-electron gas [12]. From this classical model we
derive a functional dependence (v – vth)2 for electron
yields near threshold and obtain from a best fit to the
data (solid curve) vth = 0.112 a.u. or Eth = 1.25 keV.
This value is larger than vth = 0.082 or Eth = 0.67 keV
[8]
[9]
[10]
[11]
[12]
85
D. Hasselkamp, in “Particle induced electron
emission II”, Springer Tracts in Mod. Phys. 123, 1
(1992).
J. Schou, Scan. Microscopy 2, 607 (1988).
P. Varga and HP. Winter, in “Particle induced
electron emission II”, Springer Tracts in Mod.
Phys. 123, 149 (1992).
K. Eder et al., Phys. Rev. Lett. 79, 4112 (1997).
C. Auth, A. Mertens, H. Winter, and A.G. Borisov,
Phys. Rev. Lett. 81, 4831 (1998).
F. Aumayr, G. Lakits, and HP. Winter, Appl. Surf.
Sci. 47, 139 (1991).
A. Mertens, K. Maass, S. Lederer, H. Winter, H.
Eder, J. Stöckl, HP. Winter, F. Aumayr, J. Viefhaus, and U. Becker, Nucl. Instr. Meth. B182, 23
(2001).
P. Roncin, J. Vilette, J.P. Atanas, and H.
Khemliche, Phys. Rev. Lett. 83, 864 (1999).
R.A. Baragiola, E.V. Alonso, and A. Oliva-Florio,
Phys. Rev. B19, 121 (1979).
C. Auth, A.G. Borisov, and H. Winter, Phys. Rev.
Lett. 75, 2292 (1995).
H. Winter, S. Lederer, K. Maas, A. Mertens, F.
Aumayr, and HP. Winter, J. Phys. B: At. Mol. Opt.
Phys. 35, 3315 (2002).
H. Winter and HP. Winter, Europhys. Lett.,
submitted for publication.