Determination of interface dipole and band bending at the Ag/tris (8-hydroxyquinolinato) gallium organic Schottky contact by ultraviolet photoemission spectroscopy

Article abstract of DOI:10.1016/S0039-6028(00)00232-6

A tris(8-hydroxyquinolinato) gallium (Gaq₃) thin film was grown in several steps on a previously in situ evaporated Ag thin film. Ultraviolet photoemission spectroscopy (UPS) measurements carried out prior to growth and after each growth step allowed the determination of the alignment of the highest occupied molecular orbital (HOMO) relative to the Fermi level of the Ag substrate. The deposition of ultra-thin (submonolayer) initial Gaq₃ films allowed us to distinguish between band bending and interface dipole related high binding energy cutoff (secondary cutoff) shifts, which is necessary to determine the interface dipole with high precision. In order to determine the band bending with high accuracy it was necessary to identify the HOMO position of the submonolayer Gaq₃ films. This was accomplished by removing the Ag related emission background in the low coverage spectra using the Fermi edge intensity as a measure for the Ag related emission. Our results demonstrate that the interface dipole builds up during the growth of the first one or two monolayers during which the HOMO position remains constant. The offset between the HOMO cutoff (low binding energy cutoff of the UP spectra) and the Ag Fermi edge was determined to be 1.67 eV, while the interface dipole amounted to 0.69 eV. In order to find an estimate for the alignment of the lowest unoccupied molecular orbital (LUMO) relative to the Ag Fermi edge, the HOMO/LUMO gap (2.70 eV) of Gaq₃ was determined by optical absorption measurements. The LUMO offset was estimated to be -1.04 eV.

Full text of DOI:10.1016/S0039-6028(00)00232-6

Surface Science 450 (2000) 142–152
www.elsevier.nl/locate/susc
Determination of interface dipole and band bending at the
Ag/tris (8-hydroxyquinolinato) gallium organic Schottky
contact by ultraviolet photoemission spectroscopy
*
R. Schlaf a, , P.G. Schroeder b, M.W. Nelson b, B.A. Parkinson b, C.D. Merritt c,
L.A. Crisafulli c, H. Murata c, Z.H. Kafafi c
a State University of New York, Department of Physics, Binghamton, NY 13902, USA
b Colorado State University, Department of Chemistry, Ft. Collins, CO 80523, USA
c US Naval Research Laboratory, Code 5615, Washington, DC 20375, USA
Received 4 November 1999; accepted for publication 22 December 1999
Abstract
A tris (8-hydroxyquinolinato) gallium (Gaq ) thin film was grown in several steps on a previously in situ evaporated
3
Ag thin film. Ultraviolet photoemission spectroscopy (UPS) measurements carried out prior to growth and after each
growth step allowed the determination of the alignment of the highest occupied molecular orbital (HOMO) relative
to the Fermi level of the Ag substrate. The deposition of ultra-thin (submonolayer) initial Gaq films allowed us to
3
distinguish between band bending and interface dipole related high binding energy cutoff (secondary cutoff ) shifts,
which is necessary to determine the interface dipole with high precision. In order to determine the band bending with
high accuracy it was necessary to identify the HOMO position of the submonolayer Gaq films. This was accomplished
3
by removing the Ag related emission background in the low coverage spectra using the Fermi edge intensity as a
measure for the Ag related emission. Our results demonstrate that the interface dipole builds up during the growth
of the first one or two monolayers during which the HOMO position remains constant. The offset between the
HOMO cutoff (low binding energy cutoff of the UP spectra) and the Ag Fermi edge was determined to be 1.67 eV,
while the interface dipole amounted to 0.69 eV. In order to find an estimate for the alignment of the lowest unoccupied
molecular orbital (LUMO) relative to the Ag Fermi edge, the HOMO/LUMO gap (2.70 eV) of Gaq was determined
3
by optical absorption measurements. The LUMO offset was estimated to be −1.04 eV. © 2000 Elsevier Science B.V.
All rights reserved.
Keywords: Growth; Photoelectron emission; Semiconducting films; Semiconductor–semiconductor interfaces; Semiconductor–semi-
conductor thin film structures; Surface electronic phenomena (work function, surface potential, surface states, etc.); Visible and
ultraviolet photoelectron spectroscopy; Work function measurements; X-ray photoelectron spectroscopy
* Corresponding author. Fax: +1-607-777-2546.
E-mail addresses: rschlaf@binghamton.edu (R. Schlaf ), bruce_parkinson@mail.chm.colostate.edu (B.A. Parkinson),
kafafi@ccsalpha2.nrl.navy.mil (Z.H. Kafafi)
0039-6028/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S0039-6028 ( 00 ) 00232-6

R. Schlaf et al. / Surface Science 450 (2000) 142–152
143
1. Introduction
dipole between the materials in contact. Since band
bending shifts the whole spectrum while the occur-
rence of an interface dipole only affects the high
binding energy cutoff, additional measurements
are needed, allowing the determination of the band
bending independently from the interface dipole.
As was recently demonstrated, additional X-ray
photoelectron spectroscopy (XPS) measurements
of core levels of the organic or inorganic materials
in contact allow the precise and independent deter-
mination of the band bending as it develops during
the film growth process [8–11]. This combined use
of UPS and XPS is a well-established method in
the field of inorganic semiconductor heterointer-
faces, and has been used successfully for many
years [12–16]. Once the magnitude of the band
bending related shift is known, the interface dipole
related part of the shift of the high binding energy
cutoff can be determined. If XPS is not available
an attempt can be made to measure the band
bending from the shifts of the HOMO peak or
any other feature of the UP spectra that can be
assigned uniquely to the overlayer. Unfortunately,
this can be problematic when relying on UPS only
measurements. Since most materials show signifi-
cant emission from valence states close to the
Fermi edge, the emissions from the substrate and
the overlayer usually overlap. This makes it very
difficult to find emission features related to only
one of the materials in contact, which are needed
to measure band bending shifts with high precision.
However, as we will show in the experimental
determination of the electronic structure of the
In recent years the characterization of the
electronic structure of organic semiconductor
Schottky contacts by ultraviolet photoemission
spectroscopy (UPS) has become one of the major
areas of research in the field of organic semicon-
ductor devices [1–7]. In these measurements either
the organic material is grown by vacuum depos-
ition on a metal substrate or the metal is grown
onto a previously grown thin film of the organic
material. Usually, these films are deposited in a
multi-step growth procedure during which the
electronic structure of the sample surface is charac-
terized by UPS after each growth step. This results
in a series of spectra from which the development
of the work function and the position of the
highest occupied molecular orbital (HOMO) can
be obtained as a function of the film thickness. In
principle, if the work function and the HOMO
position are known as a function of the film
thickness, the orbital alignment and the interface
dipole can be determined, and the electronic struc-
ture can be drawn.
This experimental procedure using UPS meas-
urements only, however, has severe limitations
resulting mainly from the fact that the observed
changes and shifts in UP spectra represent a convo-
lution of several phenomena occurring at the same
time, resulting in difficulties in interpreting the
spectra. The most prominent phenomenon is the
occurrence of band bending due to the contact
potential difference between the materials in con-
tact, which can result in strong shifts of the entire
photoemission spectrum [8]. In this context we
need to note that the origin of the observed band
bending is not well understood yet, since photo-
emission spectroscopy (PES) is not sensitive
enough to detect the minute amounts of dopants
needed to account for the effects observed. The
quantification and incorporation of the observed
shifts into orbital line-up measurements, however,
are crucial to the success of these measurements.
One of the problems is to discriminate between
interface dipole related and band bending related
shifts of the high binding energy cutoff (or ‘second-
ary edge’) of the HeI spectra. The high binding
energy cutoff is used to determine the interface
Ag/Gaq Schottky contact presented below, in
3
some special cases it is possible to circumvent these
problems and to determine the electronic orbital
line-up including the magnitude of the interface
dipole with high accuracy by UPS alone.
We primarily chose the Ag/Gaq Schottky con-
3
tact as a candidate for an UPS alone experiment
since Ag has only weak emission features in the
range where the Gaq HOMO peak appears and
3
due to its chemical inertness. Gaq was chosen for
3
the present investigation not only due to its conve-
nient HOMO position for the UPS only experi-
ment, but also since it was suggested to be a
superior emitter material compared with the other
metal quinolates for use in organic light emitting

144
R. Schlaf et al. / Surface Science 450 (2000) 142–152
diodes (OLEDs) [17]. This makes it interesting to
Three equivalents of 8-hydroxyquinoline (12.36 g,
85.17 mmol, 0.5 M) was dissolved in 170.34 ml of
ethanol. Under N , one equivalent of GaCl (5 g,
28.39 mmol, 0.5 M) was dissolved in 56.8 ml
of ethanol and added to the pale yellow
8-hydroxyquinoline solution dropwise via a drop-
ping funnel. The solution was stirred at room
investigate the Gaq frontier orbital alignment
3
relative to metal electrode materials. In general,
2
3
organic Schottky contacts represent ideal systems
for UPS only measurements since the metal Fermi
edge represents an emission feature that is unique
to the metal. The organic semiconductors do not
possess significant emissions in this range due to
their HOMO/lowest unoccupied molecular orbital
(LUMO) gap. This allows one to use the Fermi
edge intensity to determine the contribution of the
metal emissions to the spectra during the first
growth steps where the emissions from the metal
and the organic overlayer are superimposed. We
will demonstrate below that the HOMO peaks of
the very thin overlayers can be resolved with good
resolution using a background removal procedure
utilizing the Fermi level intensity as a feature on
which the spectra can be normalized.
temperature under N for 24 h. Then an excess of
2
ammonium hydroxide (3.9 ml, 100 mmol, conc.)
was added dropwise and stirred for 1 h at room
temperature. The bright yellow solid precipitate
was vacuum filtered and washed with water and
ethanol several times to remove ammonium chlo-
ride and unreacted 8-hydroxyquinoline and gal-
lium chloride. The product was dried under
vacuum and purified twice by vacuum train subli-
mation. The reaction gave an 84% yield (11.9 g)
prior to sublimation. The optical gap of the
Gaq compound was determined from UV–vis
3
measurements on a 315 nm thick, vapor deposited
film on silica using a Perkin-Elmer l-9 spec-
2. Experimental
trophotometer.
The Gaq films were grown at room temper-
3
˚
The experiments were carried out using a com-
mercial Omicron XP ultra-high vacuum (UHV)
apparatus consisting of an analysis chamber equ-
ipped with a HIS13 high current UV lamp and a
VSW EA 125 electron energy analyzer. An
interconnected growth chamber which holds two
water-cooled effusion cells (Oxford Instruments)
and a Leybold Inficon quartz-crystal thickness
monitor (QCM) allowed in situ sample prepara-
tion and characterization. The base pressure of
both chambers was 5×10−11 mbar. The effusion
cells were controlled by proportional (PID) con-
trollers (Oxford Technology) allowing source tem-
ature at deposition rates of about 2 A/min (source
temperature 260°C). The pressure during the
deposition of the Gaq films was about
5×10−10 mbar. After each growth step the sample
3
was characterized by UPS (HeI, 21.21 eV; 1 mm
circular entrance slit, 5 eV pass energy) measure-
ments. A −5 V bias was applied to the sample
during the UPS measurements in order to distin-
guish between analyzer and sample high binding
energy cutoff. The samples were mounted about
5 mm elevated from the Omicron standard sample
plate in order to avoid the collection of stray
electrons from the surrounding manipulator/
sample plate parts which can give rise to incorrect
high binding energy cutoff position measurements.
This effect occurs when the sample is smaller than
the UPS lamp spot, which results in the excitation
of photoelectrons from the surrounding sample
plate area. It seems that the −5 V bias is capable
of attracting part of the very slow electrons at the
high binding energy cutoff from the sample holder
into the analyzer, which leads to a superposition
of sample and sample holder cutoff signals. This
results in an erroneous work function determina-
tion. By elevating the sample plate the surrounding
perature stability better than
1°C. The Ag
substrate was prepared by evaporating about
˚
400 A Ag on a Cu sample plate. The Cu plate was
heated for 2 h at about 600°C prior to film growth
in order to remove residual contamination. The
pressure during deposition of the Ag film was
2.5×10−9 mbar at a source temperature of 975°C
˚
and the deposition rate was 20 A/min.
Gaq was synthesized following a recipe given
3
in Refs. [18,19]: a 500 ml three-necked round-
bottomed flask equipped with a magnetic stirrer
and 250 ml dropping funnel was flushed with N .
2

R. Schlaf et al. / Surface Science 450 (2000) 142–152
145
sample holder area is moved out of the vertical
analyzer focus, eliminating its contribution to the
UP spectra. The spectrometer was calibrated using
the Fermi edge position of the initial Ag film,
which was defined as binding energy zero of the
spectra. All data evaluation was carried out using
Igor Pro software (Wavemetrics).
HOMO cutoff positions are determined by adding
0.1 eV. The complete set of spectra taken after the
deposition of the Ag film and after each Gaq
deposition is shown in the center part of Fig. 2.
3
The graph on the left shows the high binding
energy cutoff of the normalized spectra for easier
comparison. On the right, the HOMO emission
range is shown, magnified to demonstrate the
evolution of the HOMO peaks in more detail. The
bottom spectrum was taken right after the depos-
ition of the Ag film. The emission bands between
3.5 and 8 eV are mostly related to the Ag d-bands,
while the weak features towards the zero energy
are mainly connected to the Ag s- and p-bands.
Unfortunately, one of these weak features is
located at the binding energy range (2–3 eV) where
3. Results
In order to determine the interface dipole and
Gaq HOMO cutoff alignment relative to the
3
Fermi edge of the Ag substrate, the resolution of
the analyzer needs to be determined first since the
positions of high binding energy cutoff and HOMO
cutoff need to be corrected for the analyzer broad-
ening. Fig. 1 shows the Fermi edge of the Ag film
measured with a high signal-to-noise ratio at the
analyzer settings used throughout the experiment.
The total width of the edge can be used as a
measure for the analyzer broadening (including
thermal broadening) which was determined graph-
ically as indicated by the lines drawn into the
graph. From the intersections of the line drawn
into the edge with the horizontal lines, the width
was determined to be 0.2 eV. Therefore, the high
binding energy cutoff positions of the spectra need
to be corrected by subtracting 0.1 eV from the
values determined from the spectra, while the true
the Gaq HOMO peak initially appears. As the
3
Gaq film grows thicker the Ag d-band emissions
3
are effectively attenuated, giving way to the five
distinct Gaq peaks which dominate the spectrum
3
˚
after the 16 A deposition step. This fast decrease
of the Ag related emissions indicates good layer-
by-layer growth of the Gaq film. Immediately
3
˚
after the deposition of the 1 A Gaq film, a strong
3
initial shift (0.38 eV) of the high binding energy
cutoff can be noticed in the spectrum. This shift
continues to increase towards higher binding
˚
energy up to the final 256 A Gaq coverage. This
3
shift is partly related to the formation of an
interface dipole, but is also due to the development
of band bending in the Gaq layer and, possibly,
3
to charging due to the photoemission process. The
sample work function after each growth step was
determined from the high binding energy cutoff
position by fitting a line into the respective cutoff
and determining the intersection point with the x-
axis. This position was then corrected for the
analyzer resolution by subtracting 0.1 eV. This
procedure is indicated schematically by the 0.1 eV
shifted line and the two arrows drawn next to the
Ag film cutoff in the left graph of Fig. 2. The work
function was then calculated by subtracting these
corrected cutoff positions from the HeI excitation
energy (21.21 eV). A more detailed description of
this procedure can be found in Refs. [10,16,20].
The so obtained work function development vs.
Fig. 1. HeI Fermi edge measured on a freshly deposited Ag film.
The analyzer broadening was estimated to be 0.2 eV, as is evi-
dent from the intersections between the line drawn into the
Fermi edge and the horizontal lines.
Gaq layer thickness is plotted in the bottom graph
3
of Fig. 3. It is evident that the work function

146
R. Schlaf et al. / Surface Science 450 (2000) 142–152
Fig. 2. HeI UP spectra of the Gaq multi-step growth sequence. The full spectra are shown in the center part. On the left the
3
normalized high binding energy cutoff regions of the spectra are shown. In the bottom spectrum the fitted line for the determination
of the Ag film work function (0.1 eV shifted to compensate for the analyzer resolution) is shown. On the right the development of
˚
the HOMO peak with increasing thickness is shown, magnified for more detail. After the 3 A growth step it becomes evident that
the HOMO peak begins to develop at about 2.4 eV. As the Gaq film grows thicker the HOMO peak starts to shift towards higher
3
binding energy, in accordance with the high binding energy cutoff. This indicates that part of the total shift of the high binding
energy cutoff is related to band bending or charging effects.
˚
strongly decreases from the initial value of 4.85 eV
of the Ag film to the final value of 3.18 eV of the
the Ag related peaks from the 1–16 A Gaq cover-
3
age spectra. This is possible since the Fermi edge
˚
final 256 A Gaq coverage, corresponding to a
of the Ag film represents an emission feature that
is exclusive to the Ag film and which is not
superimposed by any significant contribution from
Gaq emissions. The only Gaq related feature
which is superimposed to that region is the ‘echo’
of the HOMO peak caused by the HeI(b) satellite.
However, this phenomenon only becomes signifi-
cant at higher coverages where the HOMO peak
position can be determined without problems from
the original spectra. Using the Fermi edge intensity
as a measure for the Ag related contributions to
3
total change of −1.67 eV. From the simultaneous
development of the HOMO peak and its shifts,
the band bending and/or charging related part of
the shifts can be determined independently from
the interface dipole related effects. It is immediately
evident, even from the unprocessed HOMO spectra
shown in the right graph of Fig. 2, that the HOMO
peak starts to develop at about 2.40 eV binding
energy (indicated by the dashed line) and remains
3
3
˚
at this position until the 8 A deposition step (while
the high binding energy cutoff shows the strongest
the UP spectra, it is possible to extract the Gaq
HOMO related emission peak for the low coverage
3
shifts during the Gaq growth steps). Starting at
3
˚
the 16 A step the HOMO peak shifts to higher
deposition steps from the original UV spectra.
This was accomplished by normalizing the initial
binding energy as band bending or charging
occurs. In order to better resolve the HOMO peak
position at the very low coverages, we subtracted
˚
Ag spectrum and the 1–16 A Gaq coverage
3
spectra to the intensity of their Fermi edges. This

R. Schlaf et al. / Surface Science 450 (2000) 142–152
147
Fig. 3. Top graph: Gaq HOMO maximum position shift rela-
3
tive to the Ag Fermi edge depending on the Gaq layer thick-
3
ness. Bottom graph: work function change of the sample surface
vs. Gaq layer thickness. The difference between work function
3
change and HOMO maximum shift equals the interface dipole
between the Gaq layer and the Ag surface.
3
˚
Fig. 4. Original HOMO spectra of the 64 to 256 A growth steps:
resulted in a set of spectra with Ag related emission
contributions of the same magnitude. The HOMO
related emission features were then obtained by
subtracting the normalized spectrum of the initial
(a) shown for comparison with the Gaq HOMO peak spectra
of the 1 to 16 A growth steps; (b1, b2) after removal of the Ag
related superimposed emissions. The Ag contributions to the
spectra were removed by normalizing the original spectra using
the Fermi edge intensity as normalization feature and subse-
quent subtraction of the plain Ag spectrum from the spectra
taken after Gaq deposition. It is clearly evident that the peak
remains stable at 2.4 eV up to the 6 A growth step, after which
the peak starts to shift to higher binding energy (indicated by
the dashed lines). The 1 and 2 A spectra shown in (b2) are
3
˚
˚
Ag film from each of the normalized 1–16 A
coverage spectra. The result of this procedure is
shown in Fig. 4b1 and 2. Graph b2 is magnified
by about a factor of three relative to the spectra
shown in b1 in order to allow for a better compari-
son. We need to point out that this background
removal procedure worked satisfactorily only for
the HOMO peak emission region. The peaks at
higher binding energy became distorted after the
subtraction process, which is probably related to
3
˚
˚
magnified by a factor of three relative to the spectra shown
in (b1).
binding energy features due to the strong increase
of the background towards the high binding energy
cutoff. That the subtraction process works well for
the HOMO emission range is demonstrated by the
fact that the weak Ag related feature between 1.5
and 2.0 eV is completely removed from the 1 and
2 A spectra shown in Fig. 4b2, while leaving only
a smooth HOMO peak shape. The spectra shown
in graph a of Fig. 4 are the original data which
the influence of the Gaq overlayer on the inelastic
3
electron scattering processes that cause changes in
the background signal. A change in the inelastic
background can significantly change the absolute
intensity of the underlying Ag emission features
(especially the very prominent d-band related
peaks). This effect will be strongest for higher
˚

148
R. Schlaf et al. / Surface Science 450 (2000) 142–152
were added to show the full development of the
HOMO peak over the entire growth procedure.
The two vertical dashed lines indicate the total
(1.0 eV) shift of the HOMO peak from the 2.40 eV
position of its initial appearance (which remains
˚
unchanged up to the 6 or 8 A spectra) to the
˚
position of the final 256 A peak at 3.40 eV. The
shift of the HOMO maximum position relative to
the Ag Fermi level position vs. the Gaq layer
3
thickness is plotted in the top graph of Fig. 3.
In order to determine the offset between the
HOMO cutoff and the Ag Fermi edge, the HOMO
cutoff to maximum (HCM) distance needs to be
determined. This was done using the spectrum
˚
measured after the 128 A growth step (we did not
˚
choose the 256 A spectrum since it seems that it is
somewhat blurred by charging or lateral inhomo-
geneous band bending effects (see Discussion)
which would result in the determination of a too
large HCM value). Fig. 5 shows the 0–13 eV range
˚
of the original 128 A spectrum. The background
curve was determined by fitting the integral of the
spectrum to the background of the spectrum [21].
This procedure produces an acceptable result for
the HOMO peak as shown in the bottom spectrum
of Fig. 5. The determination of the HOMO maxi-
mum position was accomplished by fitting a
Gaussian–Lorentzian lookalike function [22] to
the HOMO peak. This fitting procedure yielded
3.10 eV for the maximum position. In order to
determine the HCM distance, the cutoff position
was determined to be 2.36 eV by fitting a straight
line into the onset, determination of the intersec-
tion with the x-axis and adding 0.1 eV to compen-
sate for the analyzer resolution. The HCM value
was then obtained by subtracting the cutoff posi-
tion from the maximum position resulting in
HCM=0.74 eV.
With regard to UPS measurement errors, it
should be noted that usually an error margin of
about 0.1 eV is assigned to photoemission meas-
urements. This error margin refers to the determi-
nation of the absolute values for ionization energy,
work function, HOMO cutoff and HOMO peak
position. These errors are easily located in the case
of the position determination of the cutoff features.
Their evaluation strongly depends on the way the
lines are fitted into the edges. HOMO peak posi-
˚
Fig. 5. Background removal procedure of the 128 A spectrum
which was applied in order to determine the width of the
Gaq HOMO peak. Top graph: original spectrum with fitted
background determined by the integral method [29]. Bottom
graph: background removed spectrum. Inset: HOMO peak with
fitted Gaussian–Lorentzian lineshape and fitted straight line for
the determination of the HOMO cutoff position. The correction
for the analyzer resolution is indicated by the parallel line
shifted by 0.1 eV to higher binding energy.
3
tions, however, can be determined more accurately
due to the relatively well-defined peak shapes in
combination with peak fitting procedures. We esti-
mate that for the case of good signal-to-noise ratio
an accuracy better than 0.03 eV can be obtained.
Therefore, the main source of errors in the absolute
HOMO peak positions results more likely from
calibration differences between electron energy
analyzers. Since for the determination of the
orbital offsets, band bending and interface dipole
only differences of absolute values are used, the
errors that have to be considered for these quanti-
ties are probably much smaller. We estimate that
the overall measurement errors in orbital offsets
and interface dipoles are of the order of 0.05 eV.
UV–vis absorption was measured on a 315 nm
thick Gaq vapor deposited film on silica. The
3

R. Schlaf et al. / Surface Science 450 (2000) 142–152
149
spectra. This results not only from the better
signal-to-noise ratio available from thick layer
emissions, but also from the elimination of uncer-
tainties ensuing from HOMO spectral changes as
the bulk electronic structure develops. Despite the
only weak van der Waals interaction between
organic molecules, it is possible that narrow bands
develop as the deposited film grows thicker, similar
to the bands found in layered semiconductors
perpendicular to their (0001) surfaces, which are
also only bonded by van der Waals forces [23].
Our results indicate that in selected organic systems
the full-width at half-maximum (FWHM) of the
HOMO increases by about 50% between the first
monolayer coverage and final film thicknesses of
Fig. 6. Optical absorption spectrum of a film of Gaq vacuum
deposited on a silica substrate. The thickness of the film was
315 nm. The optical gap was determined from the intersection
of the line tangent to the long wavelength side of the band and
the corrected base line.
3
˚
the order of 100–200 A [24]. Since the goal of the
presented measurement is the determination of the
orbital alignment between thick films of Ag and
result of this measurement is shown in Fig. 6. The
corrected base line of the absorbance spectra is
drawn horizontally. In order to determine the
absorption edge a straight line was fitted into the
right side of the lambda maximum peak (located
at approximately 400 nm) and its intersection with
the corrected base line was determined. This inter-
section was found to be at 457.5 nm, which corres-
ponds to an optical gap of 2.7 eV.
Gaq as they might be found in organic light
emitting devices, it is important that the HOMO
width of a bulk (thick layer) peak is used in the
above evaluation.
However, we need to point out here that this
procedure still has a source of potential error
based on the necessary assumption that the
HOMO peak only changes its intensity but not its
shape as the film grows in thickness. If the peak
were to change its shape, the maximum position
of the peak relative to its cutoff position would
also change. This would result in an error when
the cutoff position at the interface is calculated
3
4. Discussion
Due to the well-separated positions of the
using the HCM value of the thick Gaq film and
the HOMO maximum position of the thin film.
From Fig. 4, however, it appears that in the case
of the measurements presented in this study the
peak shape does not change significantly between
the initial coverages and the final thick overlayer.
3
Gaq HOMO peak and the prominent d-band
3
emission peaks of the Ag substrate, the barrier
height between the HOMO maximum position and
the Ag Fermi level W
can be determined
bh(max)
directly from the initial position of the HOMO
peak at low coverages. The HOMO initially
appears at 2.40 eV binding energy (compare Fig. 4)
Using the optical gap of Gaq (2.70 eV) as an
3
estimate [the exciton binding energy which can be
which directly equals W
is defined by the Fermi edge. The hole injection
barrier height between the HOMO cutoff and Ag
since the energy zero
significant (~0.5 eV) is neglected in this approach]
bh(max)
for its electronic HOMO/LUMO gap E , the
g
electron injection barrier W can be determined
be
Fermi edge, W , is obtained by subtracting the
by subtracting
E
from
W
which yields
W =1.66−2.70 eV=−1.04 eV. The minus sign
indicates that W points upwards from the Fermi
bh
g
bh
HCM value of 0.74 eV from W
, which yields
bh(max)
be
W =2.40−0.74 eV=1.66 eV.
bh
be
This method is more accurate compared with
level in the opposite direction to W .
bh
the direct determination of the cutoff position from
one of the HOMO peaks of the initial 1–6 A
The determination of the interface dipole
˚
requires the precise quantification of peak shifts

150
R. Schlaf et al. / Surface Science 450 (2000) 142–152
that affect the entire spectra, since these shifts will
also affect the high binding energy cutoff. If such
shifts were absent, the interface dipole would
directly equal the shift of the high binding energy
cutoff between substrate and final overlayer UP
spectra. Shifts that affect the entire spectra can
result from the development of band bending due
to the necessary equilibration of the different bulk
Fermi level positions relative to the vacuum level
of the two materials in contact or due to charging
caused by the photoemission process. We need to
point out here that the term ‘band bending’ is
derived from inorganic semiconductor theory and
assumes the presence of freely moving carriers and
immobile oppositely charged ions. This usually
requires the existence of electronic bands, which is
still a subject of research and controversy in the
organic semiconductor field [5]. We only use the
term ‘band bending’ since it seems to describe our
experimental findings fairly well, but we need to
keep in mind that the process observed might have
a different origin. Another possibility for shifts
affecting the entire spectrum is charging which can
occur if materials with relatively poor conductivity
Fig. 7. Schematic of the electronic structure of the Ag/Gaq
3
interface as determined from our measurements. Y is the work
Ag
function of the initially deposited Ag film; V is the band bend-
b
ing in the Gaq layer; W and W are the electron and hole
3
be
bh
injection barriers between the Ag Fermi edge and the LUMO
and HOMO cutoff positions; W is the hole injection bar-
bh(max)
rier between Ag Fermi edge and the HOMO maximum posi-
tion; E is the ionization energy of the Gaq film; e is the
ion
3
D
interface dipole; E is the HOMO/LUMO optical gap of
g
Gaq ; HCM is the HOMO cutoff to maximum distance.
3
are investigated, which is the case for Gaq . Since
3
the samples always become charged positively
during the measurement process, the spectra are
always shifted towards higher binding energy if
charging occurs (due to the attractive Coulomb
forces between charged sample and ejected
dipole it is sufficient to know the magnitude of the
non-dipole related part of the shift of the high
binding energy cutoff, whether it is band bending
related or not. In our case this non-dipole related
shift is easily determined from the HOMO shift as
shown in the top graph of Fig. 3: the difference
electrons). Unfortunately, since Gaq has a smaller
3
work function than Ag, band bending related shifts
˚
in the experiment presented here are also expected
to point towards higher binding energy, which
makes it difficult to distinguish between the contri-
butions of the two effects on the peak shifts
observed. From the spectral shapes of the spectra
between the initial HOMO position and the 128 A
layer amounts to 0.70 eV. The total change of the
work function (equaling the total high binding
energy cutoff shift) between Ag substrate and
˚
128 A overlayer was determined to be 1.39 eV. The
˚
at higher coverage it occurs that the 256 A
interface dipole related part of the total work
function change can now be determined by
subtracting the HOMO shift from the total work
function change, which yields an interface dipole
spectrum appears somewhat wider and blurred
˚
compared with the 128 A spectrum. This is an
indication for the onset of charging phenomena in
˚
the 256 A layer. This conclusion is in excellent
of e =1.39−0.70 eV=0.69 eV.
D
agreement with recent results obtained on
At this point the orbital line-up of the
Au/Gaq [25] and Pt/Gaq [26] interfaces using
Ag/Gaq interface can be drawn, which is shown
3
3
3
the same radiation intensity and apparatus as in
this experiment, where it was demonstrated that
in Fig. 7. The ionization energy E =5.82 eV
ion
given for Gaq was calculated by adding the work
3
˚
no charging occurs in layers up to 128 A.
function (3.46 eV) and HOMO cutoff position
For the accurate determination of the interface
[2.36 eV; calculated by subtracting HCM=0.74 eV

R. Schlaf et al. / Surface Science 450 (2000) 142–152
151
from the maximum position (3.10 eV)] of the
128 A layer film. We believe that the ionization
Fermi level of the Ag substrate. By using the Fermi
edge intensity as a normalization feature to deter-
mine the amount of the Ag substrate contribution
˚
energy of organic semiconductors is defined as the
energy required to remove an electron from the
HOMO cutoff energy level [analog to the valence
band maximum (VBM) in inorganic semiconduc-
tors] instead from the (frequently assumed)
HOMO maximum energy level. We need to point
out that this definition requires the necessary
assumption of the formation of narrow HOMO
‘bands’ as the organic molecules form the solid
state. This assumption is supported by some of
our results [24] that indicate that the HOMO peak
FWHM indeed increases with increasing layer
thickness in systems where the HOMO is observ-
able in sufficient resolution and unobstructed by
superimposed substrate emissions. Also the obser-
vation of relatively strong band bending in organic
semiconductor junctions [8–10,27] is an indicator
by itself. However, as discussed more extensively
in Refs. [8,26], the origin of the observed band
bending could also lie in the radiation induced
creation of trap states or in mobile ionic molecules.
If band bending shifts were observed due to the
separation of mobile ions as a result of the Fermi
level difference (contact potential) between the
materials in contact, then no bands need to exist
in order to explain the occurrence of shifts behav-
ing exactly as band bending. Such contact potential
dependent shifts were also recently observed in
insulating LiF layers grown on substrates with
strongly differing work functions [28], confirming
the possibility of such effects in poorly conductive
materials. The results presented in this study,
however, do not allow a final conclusion on this
matter, and further experiments in order to investi-
gate these phenomena more closely are currently
under way.
to the low Gaq coverage spectra, the HOMO
position during the very first growth steps was
3
determined with high accuracy. This allowed the
determination of the band bending related shifts
with high precision, which is important for the
calculation of the interface dipole related part of
the total high binding energy cutoff shift. The
measurements yielded an interface dipole of
0.69 eV while the hole injection barrier between
the Ag Fermi level and the Gaq HOMO cutoff
3
was determined to be 1.66 eV. Using the optical
HOMO/LUMO gap of Gaq the electron injection
3
barrier between the LUMO cutoff and the Ag
Fermi level was estimated to be −1.04 eV. Our
results demonstrate that the electronic structure of
organic Schottky contacts can be measured with
good accuracy by UPS HeI measurements alone,
if the HOMO peak of the organic material and
strong emission features of the metal are non-
overlapping and the metal background emissions
are removed from the spectra of the initial
growth steps.
Acknowledgements
Funding from ONR and DARPA is greatly
appreciated. R.S. gratefully acknowledges support
by a ‘Forschungsstipendium’ fellowship granted
by the Deutsche Forschungsgemeinschaft (DFG).
References
[1] I.G. Hill, A. Rajagopal, A. Kahn, Y. Hu, Appl. Phys. Lett.
73 (5) (1998) 662.
[2] H. Ishii, K. Seki, IEEE Trans. Electron Dev. 44 (8)
(1997) 1295.
5. Conclusions
[3] K. Seki, E. Ito, H. Ishii, Synth. Met. 91 (1997) 137.
[4] Y. Hirose, A. Kahn, V. Aristov, P. Soukiassian, V. Bulovic,
S.R. Forrest, Phys. Rev. B 54 (19) (1996) 13 748.
[5] H. Ishii, K. Sugiyama, E. Ito, K. Seki, Adv. Mater. 11 (8)
(1999) 605.
[6] Y. Gao, Acc. Chem. Res. 32 (3) (1999) 247.
[7] V.-E. Choong, M.G. Mason, C.W. Tang, Y. Gao, Appl.
Phys. Lett. 72 (21) (1998) 2689.
A tris (8-hydroxyquinolinato) gallium (Gaq )
thin film was grown in several steps on a previously
in situ evaporated Ag thin film. UPS measurements
carried out prior to growth and after each growth
step allowed the determination of the alignment
between the HOMO and LUMO relative to the
3
[8] R. Schlaf, P.G. Schroeder, M.W. Nelson, B.A. Parkinson,

152
R. Schlaf et al. / Surface Science 450 (2000) 142–152
P.A. Lee, K.W. Nebesny, N.R. Armstrong, J. Appl. Phys.
86 (3) (1999) 1499.
[19] H.M. Stevens, Anal. Chim. Acta 20 (1959) 389.
[20] G. Ertl, J. Kuppers, Low Energy Electrons and Surface
¨
Chemistry, VCH, Weinheim, 1985.
[21] X. Li, Z. Zhang, V.E. Heinrich, J. Electron Spectrosc.
Relat. Phenom. 63 (1993) 253.
[22] I. Kojima, M. Kurahashi, J. Electron Spectrosc. Relat.
Phenom. 42 (1987) 177.
[23] A. Klein, O. Lang, R. Schlaf, C. Pettenkofer, W. Jae-
germann, Phys. Rev. Lett. 80 (2) (1998) 361.
[24] R. Schlaf, unpublished results.
[25] R. Schlaf, L.A. Crisafulli, H. Murata, C.D. Merritt, Z.H.
Kafafi, P.G. Schroeder, M.W. Nelson, B.A. Parkinson,
P.A. Lee, K.W. Nebesny, N.R. Armstrong, In: Proc. SPIE
44th Annual Meet., Denver, USA (1999), in press.
[26] R. Schlaf, C.D. Merritt, L.A. Crisafulli, Z.H. Kafafi,
J. Appl. Phys. 86 (10) (1999) 5678.
[27] T. Shimada, K. Hamaguchi, A. Koma, F.S. Ohuchi, Appl.
Phys. Lett. 72 (15) (1998) 1869.
[28] R. Schlaf, B.A. Parkinson, P.A. Lee, K.W. Nebesny, G.
Jabbour, B. Kippelen, N. Peyghambarian, N.R. Arm-
strong, J. Appl. Phys. 84 (12) (1998) 6729.
[9] R. Schlaf, B.A. Parkinson, P.A. Lee, K.W. Nebesny, N.R.
Armstrong, Appl. Phys. Lett. 73 (8) (1998) 1026.
[10] R. Schlaf, B.A. Parkinson, P.A. Lee, K.W. Nebesny, N.R.
Armstrong, J. Phys. Chem. 103 (1999) 2984.
[11] R. Schlaf, B.A. Parkinson, P.A. Lee, K.W. Nebesny, N.R.
Armstrong, Surf. Sci. 420 (1999) L122.
[12] K. Horn, Appl. Phys. A 51 (1990) 289.
[13] J.R. Waldrop, R.W. Grant, Phys. Rev. Lett. 43 (22)
(1979) 1686.
[14] R. Schlaf, T. Loher, O. Lang, A. Klein, C. Pettenkofer,
¨
W. Jaegermann, Mater. Res. Soc. Symp. Proc. 448
(1997) 469.
[15] R. Schlaf, O. Lang, C. Pettenkofer, W. Jaegermann, N.R.
Armstrong, J. Vac. Sci. Technol. A 15 (3) (1997) 1365.
[16] R. Schlaf, O. Lang, C. Pettenkofer, W. Jaegermann,
J. Appl. Phys. 85 (5) (1999) 2732.
[17] P.E. Burrows, L.S. Sapochak, D.M. McCarthy, S.R. For-
rest, M.E. Thompson, Appl. Phys. Lett. 64 (1994) 2718.
[18] F.E. Lytle, D.R. Storey, M.E. Juricich, Spectrochim. Acta
29A (1973) 1357.
[29] M. Cardona, L. Ley, Photoemission in Solids, Springer,
Berlin, 1979.