Surface Science 571 (2004) L305–L310
www.elsevier.com/locate/susc
Surface Science Letters
Ion-induced electron emission from MgO by exciton decay
into vacuum
P. Riccardi
a,b
, M. Ishimoto
a,c
, P. Barone b, Raul A. Baragiola
a,*
a
b
Material Science and Engineering, Laboratory for Atomic and Surface Physics, Thornton Hall B-103,
University of Virginia, Charlottesville, VA 22904, USA
Dipartimento di Fisica, Laboratorio IIS, Universitá della Calabria, and INFM Unità di Cosenza, 87036 Arcavacata di Rende,
Cosenza, Italy
c
Fujitsu Laboratories Ltd., Akashi, Hyougo, 674-8555, Japan
Received 14 April 2004; accepted for publication 22 July 2004
Available online 8 August 2004
Abstract
We report observations of electron emission from MgO surfaces induced by impact of 100 eV–4 keV He+, Na+, Ne+,
Ar ions. The energy distribution of emitted electrons is nearly independent of ion type and energy, showing that it is
caused by the decay of an intrinsic electronic state of the solid excited by the ions. The similar yields and energy distributions for incident Na+ and Ne+ ions rule out the potential mechanism of electron emission. The results are consistent with a novel model in which excitation occurs when electrons centered at the oxygen anions are promoted
during a collision with the projectile (kinetic mechanism), and transferred to a surface exciton that can autoionize since
it lies above the vacuum level, as determined by combining measurements of electron energy loss and photoelectric
threshold.
Ó 2004 Elsevier B.V. All rights reserved.
+
Keywords: Electron emission; Magnesium oxides; Ion–solid interactions
Electron emission, a fundamental consequence
of the interaction of slow ions with solid surfaces,
results from kinetic and potential emission mecha*
Corresponding author. Address: Material Science and
Engineering, Laboratory for Atomic and Surface Physics,
Thornton Hall B-103, University of Virginia, Charlottesville,
VA 22904, USA. Tel.: +1 4349241059; fax: +1 4349241353.
E-mail address: [email protected] (R.A. Baragiola).
nisms [1], in which the electron excitation energy is
provided, respectively, by the motion of the incoming ion or by its potential energy. The vast majority of the studies on ion-induced electron emission
have used metal samples while insulator surfaces,
of high relevance in both basic research and technology, have received little attention [2,3] due to
experimental difficulties. Electron emission is most
interesting for ion energies below a few keV where
0039-6028/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2004.07.031
CE
RF
A
LE C E
TT S
E R C IE
S N
SU
L306
P. Riccardi et al. / Surface Science 571 (2004) L305–L310
kinetic emission becomes very small and potential
electron emission dominates if energetically allowed. In the Auger neutralization (AN) process
responsible for potential emission, an electron
from the solid neutralizes the projectile and the excess energy is taken by another electron, which
may escape the surface potential barrier. For ionic
insulators of large ionization energy I (sum of
band-gap Eg and electron affinity A), Auger neutralization cannot occur for ions like Na+ that
have a neutralization energy En < 2I + e, where e
is the interaction energy between the final two
holes in the valence band. However, the yield of
electrons ejected from these ionic solids by slow
ions has been found to be surprisingly larger than
for metals [2–5], even though less energy is required to remove an electron from metals (the
work function). In addition, strong electron emission from ionic insulators has been observed at impact energies below 100 eV [4–10], showing no
indication of a kinetic energy threshold as observed in metals at energies below 0.5–2 keV [1].
To examine the question of enhanced electron
emission in the interactions of slow ions with insulators, we measured the energy distributions of
electrons emitted from an MgO surface under the
impact of slow noble gas and sodium ions. The
study of MgO, a simple ionic crystal, is justified
not only from the point of view of fundamental
physics––interactions of ions with insulator surfaces are poorly understood––but also because of
advantageous properties of this material for use
in plasma display panels: a large ion-induced electron yield (and therefore a low discharge voltage)
[8,11,12] and high stability under ion bombardment [13].
Our results show that the energy distribution of
electrons emitted by slow He+, Na+, Ne+, Ar+, and
Xe+ are very similar. This remarkable behavior,
which was first reported by us before in a brief
form [8], is explained with a model in which oxygen-2p electrons are promoted in a close collision
between the projectile and the anion, populating
an exciton level that decays into vacuum due to
the negative electron affinity of MgO.
The experiments were performed in an ultrahigh vacuum (10 10 Torr) system used in previous electron emission studies [14]. Electrons
ejected from the MgO samples were energy analyzed with a double-pass cylindrical mirror spectrometer operated inside a magnetic shield at
constant pass energy of 50 eV and a resolution
of 0.2 eV. The surface of the samples was normal
to the ion beam and at 78° with respect to the spectrometer axis. Sodium ions were produced in a
thermal ionization source, and noble gas ions in
an electron impact source, operated with 58 eV
electrons. Similar results were obtained using 30
eV electrons in the ion source, indicating negligible
contamination of the ion beam with doubly
charged ions. The MgO films, about 100 nm thick,
were prepared by electron beam deposition on a
highly doped Si substrate. They are polycrystalline
with the grains oriented such that they present an
oxygen-terminated (1 1 1) surface, as determined
by X-ray diffraction. The samples were sputter
cleaned with 1 keV Ar+ and the cleaning was monitored by Auger Electron Spectroscopy (AES) and
Electron Energy Loss Spectroscopy (EELS). Weak
features in the EELS spectra due to band-gap
states disappeared after cleaning [13], also indicating negligible ion bombardment damage. AES and
EELS spectra for the sputter cleaned MgO surfaces were in excellent agreement with published
spectra [15–17] of clean MgO.
Normal methods for electron emission measurement are not possible with insulators such as
MgO since the surface electrostatic potential becomes positive as ions deposit their charge on the
surface and electrons are emitted into vacuum.
This positive surface potential produces a timedependent distortion in the energy spectra. To
minimize this problem we used low ion fluxes
(<100 nA/cm2) and neutralized the surface charge
immediately before the acquisition of each spectrum by flooding the surface with low energy electrons from a nearby hot W filament (an electron
flood gun used routinely for charge compensation
in photoelectron spectroscopy of insulators). The
neutralization procedure reduced the surface
potential below 0.2 V. This is demonstrated by
the quick rise in the energy distributions of electrons very close to the voltage at which the sample
is biased ( 4.9 V), and by the constancy of the
AES energies measured after exposing the sample
to the electron flood gun. Repeated energy spectral
SU
N
IE
SC RS
CE E
A TT
R F LE
CE
P. Riccardi et al. / Surface Science 571 (2004) L305–L310
scans showed that, at the ion current densities
used, neither surface charging nor damage had significant effects over the 20 s spectrum acquisition
time.
Fig. 1 shows N(E), the energy spectra of electrons emitted from MgO by 200 eV He+, Ne+,
and Ar+ ions, and 500 eV Na+ ions. The total electron yields c, shown in Fig. 2, were obtained from
the currents measured on the sample under positive and negative bias, with an uncertainty of
30%. The fact that the yields for He+ are larger
than for Ar+ (which sputters much more effectively) clearly indicates that emission of negative
ions has a negligible contribution to the yields
(the O yields are below 0.06 for 250 eV He+,
Ne+, Ar+ and other ions incident on oxidized
Mg) [18]. The increase of the total electron yield
above unity for high energy He+ signals the contribution of multiple excitation collisions during ion
penetration.
Fig. 1. Normalized energy distribution of electrons ejected
from a MgO surface under impact of 200 eV noble gas ions and
128 eV electrons. The sample was biased at 4.9 eV.
L307
Fig. 2. Energy dependence of the total electron yields from a
MgO(1 1 1) surface at normal incidence. Also shown are
representative statistical uncertainties.
Fig. 1 shows that the shape of the energy distributions is nearly the same for the different ions, in
remarkable contrast with the large variations observed for collisions with metal samples [1,19].
This finding allows us to exclude emission mechanisms that involve the electron levels of the projectile ions, such as Auger neutralization (AN) [1],
that have been often invoked as a source of electrons in slow ion interactions with MgO [20]. In
Auger neutralization, the maximum electron energy is En 2I [1,19], where the ionization energy
of MgO is I = 5.5 eV at the surface (Fig. 3), and En
the ionization potential of the parent atoms, varies
from 24.58 to 5.14 eV going from He to Na. If AN
does not occur outside the surface, the incident
projectile will penetrate and the energy levels will
be modified by the strong electric fields inside
MgO. The available energy for AN will increase
for ion trajectories close to Mg2+ cations but electron emission from such events is already improbable for Ar+ [10] and therefore even more unlikely
for Na+, due to its much lower potential energy.
The independence of the spectra on the type of
projectile also rules out other mechanisms, such
as double electron capture to a negative ion followed by electron detachment in a third collision
[21], and the decay of a transient autoionization
complex of the incident ion and two adjacent
target atoms [22].
CE
RF
A
LE C E
TT S
E R C IE
S N
SU
L308
P. Riccardi et al. / Surface Science 571 (2004) L305–L310
Fig. 3. Photoelectric threshold of a MgO(1 1 1) surface.
The electron energy spectra, which is nearly
identical for different ions, is different to that excited by electron impact (Fig. 1), indicating that
the shape of N(E) does not result from a peculiar
density of final electronic states. This suggests that
ion-induced electron emission results from the preferential excitation by ions of a state intrinsic to the
solid, i.e., not involving the electronic properties of
the projectile. An intrinsic mechanism was suggested recently by Matulevich et al. [10]. These
authors measured electron yields c induced by 40
eV noble gas ions on very thin (1–5 nm)
MgO(1 0 0) films grown on oxidized Mo, obtaining
c much higher than reported here and not finding
the independence of N(E) on the type of ion. They
proposed, in addition to AN for He+ and Ne+ on
the MgO surface, an intrinsic mechanism for He+
and Ar+, in which electrons are emitted by AN
of valence holes at the substrate and transported
to the surface. The valence holes, assumed to be
produced by neutralization of the projectiles at
the surface, are highly immobile in MgO but drift
to the interface in the high electric field produced
by electrostatic charging (the surface charges by
2 eV, as judged from their energy diagram). This
high electric field (4–20 MV/cm) alters electron
transport [23], and may even produce field emission at the interface; both effects enhance electron
yields, in particular in MgO thin films [24], and can
explain the higher yields obtained by these authors
compared to those reported here. Justification of
the model by Matulevich et al. on the base of their
experiments is diminished by the fact that two of
their nearly concurrent papers [8,10] show inconsistent energy distributions. On the other hand,
we cannot rule out that the differences between
their and our c and N(E) result from different
experimental conditions besides charging (e.g.,
ion energy, MgO growth conditions, crystalline
orientation, film thickness) or measurement techniques. From the point of view of the present discussion, we note that AN can be ruled out even for
He+ and Ne+ since experiments do not show the
3 eV shift in the high energy edge of N(E) expected (and calculated) [10] for those ions. The
strongest argument against the hypothesis [25] that
AN dominates electron emission from MgO is, of
course, the very similar electron yields for Ne+ and
Na+, which have vastly different neutralization
energies.
The electronic excitation mechanism that is
consistent with our results is the well-known electron promotion effect in atomic collisions, which
results from electron-electron interactions in the
interpenetration of the valence shells of the projectile and a target atom (oxygen anion) during a
close collision, forming a transient quasi molecule.
This description of ion-surface interactions as binary collisions is adequate in ionic solids because
the valence charge is localized at the anion sites.
Collisions with oxygen anions promote oxygen2p electrons along quasi-molecular orbitals (MOs),
above a threshold projectile energy which has been
determined to be 50 eV for the analogous case of
Na+ exciting oxidized Al [26]. The O-2p electrons
can be promoted by all the projectiles studied, as
follows from calculations for these and similar systems [27,28], and from Barat–Lichten MO correlation rules [29]. The promoted MO are 3dr(He–O),
3dp and 4fr(Ne–O, Na–O), and 4fr(Ar–O), where
nl indicates the atomic orbital of the united atom
in the limit of zero internuclear separation, and
r(p) indicate the projection, 0 (1) of the angular
momentum on the internuclear axis. The promoted MO cross empty levels, such as the 3sr,
4sr for He, Ne, and Na projectiles and the 3dd,
4sr for Ar, which correlating with O-3s and O-4s
excitons. We note that this excitation model is
somewhat similar to the idea of Vogan et al. [18]
SU
N
IE
SC RS
CE E
A TT
R F LE
CE
P. Riccardi et al. / Surface Science 571 (2004) L305–L310
that, in a collision with a MgO surface, electrons
result from excitation of a (MgO) pair, but we
do not need to invoke oxide molecules and it is
uncertain what role one or more Mg2+ ions play
except to shift the electrostatic potential at the
anion site. We disagree with Vogan et al. [18] in
their opinion that this process is not kinetic emission because of the low projectile energy. Electron
promotion is allowed by the kinetic energy of the
projectile and hence the electron emission process
it affords is indeed of the kinetic type.
Once promoted, the O-2p electrons can transfer
to excited states at the crossings of the promoted
MO with unfilled MOs that correlate to excitons
and conduction band states. The population of
the final levels will decrease with excitation energy
DE; and therefore will favor excitonic to continuum states. Electron promotion to continuum
states [7,8] and MO autoionization may give a significant contribution to the weak high-energy tail
of the electron spectrum.
We propose that the excitons populated by electron promotion contribute to electron emission because they decay into vacuum. Excitons are
normally considered to be bound states of solids
but, in MgO (and in LiF), the negative electron
affinity of the (1 1 1) surface causes the exciton to
be above the vacuum level [30]. The band gap of
MgO (7.8 eV in the bulk) [13] drops at the surface
due to the decrease in the Madelung potential, by
more than 1 eV, with variations among different
crystal faces [31,32]. EELS measurements show
a minimum surface excitation energy of 5.5 eV
[30], and a decrease in the exciton energy from a
bulk value of 7.7 to 6.2 eV, for MgO(1 0 0) [13].
The surface exciton exists in a region of changing
surface potential extending from outside the solid
to the first atomic layers [30]. To establish the location of the vacuum level, we determined the ionization energy of our samples (energy difference
between the top of the valence band and the vacuum level) by measuring the photoelectron threshold (Fig. 3). The value of 5.3 eV is nearly the
same as the minimum excitation energy measured
in EELS, showing that the surface exciton is above
the vacuum level, which can explain its unusually
large width. With this information we construct
the approximate energy level diagram for the
L309
Fig. 4. Schematic energy levels for a MgO(1 1 1) surface,
showing how the decay of an exciton, allowed by the negative
electron affinity, gives rise to the observed energy distribution of
emitted electrons. The block labeled Ôsurface excitonÕ represents
the range of surface energy losses measured in EELS, and its
extension from outside the solid to the first surface layer [30].
MgO surface, shown in Fig. 4, which includes
the decrease in conduction band energies and increase in valence band energies at the surface that
have resulted from detailed calculations [32,33].
The negative electron affinity of the surface (vacuum level below the bulk conduction band minimum) allows excitons to couple to the continuum
of states outside the solid, which explains the considerable width of surface excitons [30]. The decay
of the excitons into vacuum (or exciton break-up)
[34,35] then produces electrons with an energy distribution peaked at the observed low energies, due
to the small (positive) difference in energy between
the peak exciton energy and the vacuum level.
In conclusion, the electrons emitted from MgO
surfaces by slow ions have energy distributions
remarkably independent of the type of projectile,
suggesting that they originate from the decay of
an intrinsic excitation in the solid, i.e. excitons,
which are populated by promotion of O-2p electrons in binary projectile-oxygen collisions. The
position of the surface exciton above the vacuum
level, demonstrated by measurements of the photoelectric threshold, is likely the cause of the high electron yields from MgO compared to other oxides.
CE
RF
A
LE C E
TT S
E R C IE
S N
SU
L310
P. Riccardi et al. / Surface Science 571 (2004) L305–L310
The results show the importance of the surface electronic structure in electron emission from insulators, and suggest that modifications in the atomic
structure of the surface may enhance electron yields
even further.
References
[1] R.A. Baragiola, in: J.W. Rabalais (Ed.), Low Energy IonSurface Interactions, Wiley, New York, 1994 (Chapter 4).
[2] J. Schou, in: R.A. Baragiola (Ed.), Ionization of Solids by
Heavy Particles, Plenum, New York, 1993, p. 351.
[3] H. Winter, Prog. Surf. Sci. 63 (2000) 177.
[4] U.A. Arifov, R.R. Rakhimov, S. Gaipov, Radio Engineering and Electronic Physics 15 (1970) 1059.
[5] Y. Ushio, T. Banno, N. Matuda, Y. Saito, S. Bada, A.
Kinbara, Thin Solid Films 167 (1988) 299.
[6] M. Vana, F. Aumayr, P. Varga, HP. Winter Europhys.
Lett. 29 (1995) 55.
[7] P. Stracke et al., Nucl. Inst. and Meth. Phys. Res. B 125
(1997) 67.
[8] M. Ishimoto, R. Baragiola, T. Shimoda, Proceedings of
the Seventh International Display Workshop,, Society for
Information Display, 2000, p. 683.
[9] Y.T. Matulevich, T.J. Vink, L.F. Feiner, P.A. Zeijlmans van
Emmichoven, Nucl. Instr. Meth. Phys. Res. 193 (2002) 632.
[10] Y.T. Matulevich, T.J. Vink, P.A. Zeijlmans van Emmichoven, Phys. Rev. Lett. 89 (2002) 167601.
[11] H.S. Uhm, E.H. Choi, G.S. Cho, Appl. Phys. Lett. 78
(2001) 592.
[12] T.J. Vink, A.R. Balkenende, R.G.F.A. Verbeek, H.A.M.
van Hal, S.T. de Zwart, Appl. Phys. Lett. 80 (2002) 2216.
[13] V.E. Henrich, P.A. Cox, The Surface Science of Metal
Oxides, Cambridge Univ. Press, Cambridge, England, 1994.
[14] R.A. Baragiola, C.A. Dukes, Phys. Rev. Lett. 76 (1996)
2547.
[15] Handbook of Auger Electron Spectroscopy––Physical
Electronics Inc., Eden Praire, Minnesota, USA (1976).
[16] D.M. Roessler, W.C. Walker, Phys. Rev. 159 (1967) 733.
[17] V.E. Henrich, G. Dresselhaus, H.J. Zieger, Phys. Rev. B
22 (1980) 4764.
[18] W.S. Vogan, R.L. Champion, V.A. Esaulov, Surf. Sci. 538
(2003) 211.
[19] H. Hagstrum, in: N.H. Tolk, J.C. Tully, W. Heiland, C.
White (Eds.), Inelastic Ion Surface Collisions, Academic
Press, New York, 1977, p. 1.
[20] S.J. Yoon, I. Lee, J.-W. Lee, B. Oh, Jpn. J. Appl. Phys. 40
(2001) 809.
[21] P. Roncin, J. Villette, J.P. Atanas, H. Khemliche, Phys.
Rev. Lett. 84 (1999) 864.
[22] J.C. Lancaster, F.J. Kontur, G.K. Walters, F.B. Dunning,
Phys. Rev. A 64 (2001) 54901.
[23] R.A. Baragiola, M. Shi, R.A. Vidal, C.A. Dukes, Phys.
Rev. B 58 (1998) 13212.
[24] J. Lee et al., Appl. Surf. Sci. 174 (2001) 62.
[25] J.Y. Lim et al., J. Appl. Phys. 94 (2003) 764.
[26] J.C. Tucek, R.L. Champion, Surf. Sci. 382 (1997) 137.
[27] R. Souda et al., Surface Sci. 343 (1995) 104.
[28] R. Souda, T. Suzuki, K. Yamamoto, Surface Sci. 397
(1998) 63.
[29] M. Barat, W. Lichten, Phys. Rev. A 6 (1972) 211.
[30] P.A. Cox, A.A. Williams, Surf. Sci. 175 (1986) L782.
[31] C. Satoko, M. Tsukada, H. Adachi, J. Phys. Soc. Jpn. 45
(1978) 1333.
[32] S. Russo, C. Noguera, Surface Sci. 262 (1992) 245.
[33] P.V. Sushko, A.L. Shluger, C.R.A. Catlow, Surface Sci.
450 (2000) 153.
[34] C. Bandis, B.B. Pate, Phys. Rev. Lett. 74 (1995) 777.
[35] V.M. Asnin, I.L. Krainsky, Appl. Phys. Lett. 73 (1998)
3727.