Energy level alignment at the interfaces in a multilayer organic light-emitting diode structure

Abstract

We use photoelectron spectroscopy to study the electronic structure and energy level alignment throughout an organic light-emitting diode. The structure under investigation is a state-of-the-art long-living red phosphorescent device composed of doped charge-injection layers, charge-blocking layers, and an emission layer. By
consecutively building up the whole device, the key parameters of every interface are measured. Our results show that the doped layers have a significant influence on the device energetics, especially in controlling the built-in potential, and that there are mostly only small dipoles present at the interfaces of the intrinsic organic
layers.

Full text

Page 1

Energy level alignment at the interfaces in a multilayer organic light-emitting diode structure
S. Olthof,* R. Meerheim, M. Schober, and K. Leo
Institut für Angewandte Photophysik, Technische Universität Dresden, D-01062 Dresden, Germany
?Received 2 March 2009; revised manuscript received 4 May 2009; published 11 June 2009?
We use photoelectron spectroscopy to study the electronic structure and energy level alignment throughout
an organic light-emitting diode. The structure under investigation is a state-of-the-art long-living red phospho-
rescent device composed of doped charge-injection layers, charge-blocking layers, and an emission layer. By
consecutively building up the whole device, the key parameters of every interface are measured. Our results
show that the doped layers have a significant influence on the device energetics, especially in controlling the
built-in potential, and that there are mostly only small dipoles present at the interfaces of the intrinsic organic
layers.
DOI: 10.1103/PhysRevB.79.245308PACS number?s?: 79.60.Fr, 85.60.Jb
I. INTRODUCTION
Current highly efficient organic light-emitting diodes
?OLEDs? consist of a large number of different organic lay-
ers such as doped injection layers, blocking layers, and ac-
tive emitting layer.1With rising complexity of the device
stack, the interfaces between these layers become increas-
ingly important for the OLED performance. For simplicity,
vacuum-level alignment is usually assumed so that the inde-
pendently measured highest occupied molecular orbital
?HOMO? and lowest unoccupied molecular orbital ?LUMO?
can be used to obtain the energy diagram. However, it is well
known from detailed spectroscopic studies on model systems
that interface dipoles originating from effects such as polar-
ization, chemical reactions, and bond formation very often
change the simple picture of vacuum alignment.2In our de-
vices, we use doped layers to increase the carrier density,
facilitate carrier injection, and ensure Ohmic contact behav-
ior; therefore the alignment is affected by the level bending
and the resulting depletion layer.3These effects can lead to
deviations as large as 2 eV from the commonly assumed
vacuum-level alignment. To obtain deeper insight into simu-
lation and optimization of a full OLED device, it is therefore
necessary to measure the energy alignment at all metal/
organic and organic/organic interfaces. This can be per-
formed, e.g., by ultraviolet photoelectron spectroscopy
?UPS? and x-ray photoelectron spectroscopy ?XPS?. Various
studies of organic/organic4,5and metal/organic6interfaces
can be found in literature. However, to understand the align-
ment in a complete device, it is not sufficient to look at
individual interfaces since the choice of the electrode
material,7pinning effects of organic layers,8polar-bond
orientation,9or doping-induced level bending can influence
the energetic position of the following layers and thereby the
interface formation.
In this paper, we present interface resolved studies of an
incrementally built complete OLED structure by UPS and
XPS. We show that significant interface dipoles and a
built-in potential are present in such a device structure. Also,
we show that, in particular, doped layers can determine the
energetic positions of several subsequent layers.
For our studies, we choose the structure of a long
living and highly efficient red phosphorescent device.10
In this stack, an indium tin oxide ?ITO? anode is contacted
bya hole-injection layer
tetrakis?4-methoxyphenyl?-benzidine
with 4 mol % of 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquino-
dimethane ?F4-TCNQ? and an electron-blocking layer
of
N,N?-di?naphthalen-1-yl?-N,N?-diphenyl-benzidine
??-NPD?. This is followed by a light-emitting layer com-
posed of ?-NPD coevaporated with 10 wt % of the phos-
phorescent redemitter
dibenzo?f,h?quinoxaline??acetylacetonate?
andahole-blocking layer
phenanthroline ?BPhen?. Finally, there is an electron-
injection layer of BPhen doped by cesium and a silver top
contact.
consisting of
N,N,N?,N?-
?MeO-TPD?
doped
iridium?III?
bis?2-methyl-
?Ir?MDQ?2acac?
4,7-diphenyl-1,10- of
II. EXPERIMENT
Measurements are performed with a Phoibos100 system
?Specs, Berlin, Germany? under ultrahigh vacuum ?UHV?
with a base pressure of 1?10−10mbar. For UPS measure-
ments, the He I line ?21.22 eV? from a discharge lamp with
an energy resolution of 130 meV and for XPS an Al K?
x-ray source ?1486.6 eV? with a resolution of 400 meV are
used. The experimental error of both methods can be esti-
mated from the reproducibility achieved in separate measure-
ments and is below 50 meV. The UHV evaporation tool with
a base pressure of ?1?10−8mbar for the deposition of the
organic layers is directly connected to the measurement
chamber and different chambers for n- and p-type doping as
well as for intrinsic layers can be used to avoid cross con-
tamination.
The OLED samples for the characterization of the
current-voltage curves were prepared in a different single-
chamber UHV system ?Kurt J. Lesker Co. Ltd., Hastings,
UK? at a base pressure of 1?10−8mbar. Here it is possible
to fabricate devices with different structures on the same
sample ensuring equal evaporation conditions for a high
comparability. After preparation, the devices are encapsu-
lated with cavity glass lids by an epoxy glue under the nitro-
gen atmosphere of a glovebox that is directly attached to the
vacuum system.
The organic materials MeO-TPD and ?-NPD ?both Sen-
sient, Wolfen, Germany? two times sublimated, BPhen ?Ald-
PHYSICAL REVIEW B 79, 245308 ?2009?
1098-0121/2009/79?24?/245308?7?
©2009 The American Physical Society245308-1

Page 2

rich, Munich, Germany? two times sublimated, and F4-
TCNQ and Ir?MDQ?2acac ?both TCI, Zwijndrecht, Belgium?
one time sublimated were deposited at room temperature
from heated crucibles. Cesium was purchased from Saes
?Milan, Italy? and is evaporated out of an alloy. Doping is
achieved by coevaporating host and dopand while the evapo-
ration rates are controlled independently by two quartz crys-
tals. As substrate, either sputter-treated ITO ?Thin Film De-
vices, Inc., Anaheim, USA? or sputter-cleaned silver foil
?99.995%, MaTecK, Juelich, Germany? is used. The metal/
organic and organic/organic interfaces are built incremen-
tally in the same order as in the OLED device and the mea-
surement steps in each layer range in thickness from 2 Å up
to 20 nm. For each step, the high-binding energy cutoff
?HBEC?, the hole-injection barrier ?HIB?, and characteristic
core-level peaks are measured. Several UPS measurements
are taken successively to ensure that the sample shows no
sign of charging.
III. RESULTS AND DISCUSSION
The first interface investigated is the one from the anode
ITO to the hole-injection layer MeO-TPD p doped with
4 mol % of F4-TCNQ. Before measurement, the ITO is
sputter cleaned using argon to remove residual carbon. This
is necessary since ITO commonly shows degradation under
UVexcitation dueto
contaminations.11After the treatment, the anode shows a
work function ?Wf? of 4.1 eV. On top of this, the organic
semiconductor is deposited in steps to observe interface and
level-bending effects. The development of the HBEC and
HOMO region of the UPS spectra taken at different layer
thicknesses is shown in Fig. 1?a?. Here, as in all following
spectra, the binding-energy scale is in reference to the Fermi
edge of the anode and the curves are vertically shifted for
clarity.
Within the first 0.5 nm, an interface dipole of −120 meV
is observed in the vacuum level that is followed by an up-
ward bending of 720 meV. The HOMO of the MeO-TPD can
be clearly observed after 1 nm deposition and shows the
same bending behavior as the vacuum level. After a deple-
tion region of 5 nm, the energy levels saturate, resulting in a
HIB of 0.45 eV and the ionization potential ?IP? of MeO-
TPD is 5.1 eV. The resulting energy diagram for this inter-
face is shown in Fig. 1?b? where a HOMO-LUMO gap of
3.2?0.3 eV is assumed, as determined from optical-
a reaction with surface
5 V
5.4 e
0.85eV
Yá-NPD
0.49eV
MeO-TPD:
F4-TCNQ
5.1eV
EF
Evac
HOMO
LUMO
(a)
(b)
181716
bindingenergy(eV)
153210 -1
5.0nm
2.0nm
1.0nm
0.5nm
intensity(arb. units)
pMeO-TPD
HBEC
HOMO
FIG. 2. ?Color online? Interface between doped MeO-TPD and ?-NPD. ?a? Development of the HBEC and HOMO region of the UPS
He I spectra as a function of increasing thickness of the ?-NPD layer. The dotted vertical line marks the HBEC position. ?b? Resulting
schematic energy level diagram.
EF
Evac
1
4. eV
5 1eV
.
0.45eV
ITO
MeO-TPD:
F4-TCNQ
HOMO
LUMO
0.72eV
5nm
19 18 17 16 153210
HOMO
0.12eV
intensity(arb. units)
0.72eV
HBEC
ITO
0.2nm
0.5nm
1.0nm
2.0nm
5.0nm
10nm
bindingenergy(eV)
20nm
(a)(b)
FIG. 1. Interface between ITO and MeO-TPD doped with 4 mol % of F4-TCNQ. ?a? Development of the HBEC and HOMO region of
the UPS He I spectra as a function of increasing thickness of the organic layer. The dotted lines mark the change in vacuum level and HOMO
position. ?b? Resulting schematic energy level diagram.
OLTHOF et al.
PHYSICAL REVIEW B 79, 245308 ?2009?
245308-2

Page 3

absorption edge measurements.12The larger error margin of
this value is due to the unknown exciton binding energy that
is estimated to be 0.3 eV.13Since ITO is less reproducible
and more difficult to handle compared to a metal foil, the
same experiment was repeated using a sputter-cleaned silver
foil with a Wfof 4.3 eV. Again after 5 nm the level bending
is completed and this contact results in the same energetic
position of the thick p-MeO-TPD layer ?data not shown?.
The equivalence of these substrates gets apparent when com-
paring the ionization potential and HIB of the 20 nm
p-MeO-TPD layer on ITO of Fig. 1 and the 6 nm
p-MeO-TPD layer on silver of Fig. 2. As the values coincide
one can assume equal growth modes in both cases.14There-
fore, for all the following measurements a silver foil is used
as a substrate.
For the investigation of the interface between MeO-
TPD:F4-TCNQ and the intrinsic ?-NPD blocking layer, a
layer of doped MeO-TPD ?6 nm? is prepared on a silver foil
and subsequently ?-NPD is evaporated. No dipole is ob-
served at the interface as can be seen from the constant
HBEC in Fig. 2?a?, so there is vacuum-level alignment be-
tween the two layers. The IP of the 5-nm-thick ?-NPD layer
is 5.45 eV and the HIB is 0.85 eV as is shown in Fig. 2?b?;
here a transport gap of 3.1 eV is assumed for ?-NPD that
was measured by inverse photoelectron spectroscopy.4
Using this interface, one can nicely show the strong influ-
ence of the doped layer on the energetic alignment within the
OLED device. If the deposition order of the two layers is
reversed as shown in Fig. 3 and ?-NPD is directly put on
silver, the HIB is 1.73 eV, about 1 eV larger compared to the
same layer deposited on the doped MeO-TPD. If then the
MeO-TPD:F4-TCNQ layer is deposited on top, a shift of 980
meV happens in the vacuum level as the HOMO of the
doped layer shifts close to the Fermi energy again. Fitting the
HOMO regions of the different thicknesses with peaks for
?-NPD and MeO-TPD shows that the HOMO of ?-NPD
gets pulled up by 960 meV due to the contact with the doped
layer. This shift is due to the electrostatic field introduced
between metal and ionized dopands.
On a further sample, the interface between intrinsic
?-NPD and the active layer where ?-NPD is coevaporated
with 10 wt % of the phosphorescent emitter Ir?MDQ?2acac
5.13eV
0.40eV
MeO-TPD:
F4-TCNQ
7
5.4 eV
1 3
.7 eV
Yá-NPD
98e
0.
V
EF
Evac
HOMO
LUMO
(a)(b)
1817 16
bindingenergy(eV)
153210
intensity(arb. units)
0.98eV
HBEC
10nm
0.2nm
5.0nm
2.0nm
1.0nm
0.5nm
?-NPD
HOMO
0.96eV
FIG. 3. ?Color online? Interface between ?-NPD and doped MeO-TPD. ?a? Development of the HBEC and HOMO region of the UPS
He I spectra as a function of increasing thickness of the doped MeO-TPD layer. The dotted vertical line marks the change in the position of
the HBEC. ?b? Resulting schematic energy level diagram.
1817 1615
210
HOMO
7.0nm
4.0nm
2.0nm
1.0nm
0.5nm
0.2nm
intensity(arb. units)
bindingenergy(eV)
??NPD
0.1eV
HBEC
.
5 41 V
e
0.9eV
Y á-
Ir(MDQ) acac
2
NPD:
0.79eV
.4 V
5 e
Yá-NPD
ID = 0.1eV
EF
Evac
HOMO
LUMO
(a)
(b)
FIG. 4. ?Color online? Interface between ?-NPD and ?-NPD coevaporated with 10 wt % Ir?MDQ?2acac. ?a? Development of the HBEC
and HOMO region of the UPS He I spectra as a function of increasing thickness of the coevaporated layer. The dotted vertical line marks
the change in the position of the HBEC. ?b? Resulting schematic energy level diagram.
ENERGY LEVEL ALIGNMENT AT THE INTERFACES IN A…
PHYSICAL REVIEW B 79, 245308 ?2009?
245308-3

Page 4

is investigated. Again the doped MeO-TPD ?6 nm? and in-
trinsic ?-NPD ?3 nm? are prepared on a silver sample. On
top of that, the mixed layer of ?-NPD:Ir?MDQ?2acac is de-
posited stepwise. The measurements in Fig. 4 show a small
downward shift of the vacuum level and ?-NPD HOMO
position in the beginning by −100 meV that saturates after 1
nm when approximately one monolayer ?ML? is closed. The
HIB of the mixed layer is 0.9 eV and the IP remains the same
as for intrinsic ?-NPD.
Since the electron transport in the device takes place in
the LUMO of the Ir?MDQ?2acac and not on the host
material,10it is important to know the position of the
Ir?MDQ?2acac HOMO and LUMO within the ?-NPD ma-
trix. The 10 wt % doping ratio used before does not produce
distinct HOMO features from the iridium complex mol-
ecules. Therefore two more samples were prepared, one with
a doping ratio ?-NPD:Ir?MDQ?2acac of 1:1 and a pure
Ir?MDQ?2acac sample. In Fig. 5?a?, the HOMO region of the
(a)
(b)
4321043210
4
bindingenergy[eV]
3210
?-NPD:Ir(MDQ)2acac(1:1)
?-NPD
intensity[arb. units]
intensity[arb. units]
UPSmeasurement
Curvefit
Ir(MDQ)2acac
UPSmeasurement
Curvefit
bindingenergy[eV]
UPSmeasurement
Fit ?-NPD
Fit Ir(MDQ)2acac
Combinedfits
FIG. 5. ?Color online? ?a? UPS HOMO regions of pure ?-NPD and pure Ir?MDQ?2acac; the two measured curves are fitted by multiple
Gaussian peaks after a polynomial background is subtracted. ?b? UPS HOMO region of a 1:1 ratio of ?-NPD mixed with Ir?MDQ?2acac after
a polynomial background is subtracted; here the fitted curves from the pure layers are used to match the measured data.
65432102019181716
bindingenergy[eV]
HOMO
1.28eV
HBEC
intensity(arb. units)
0.98eV
6664626058
Ir(MDQ)
BPhen:Cs
0.2nm
5.0nm
2.0nm
1.0nm
0.5nm
0.2nm
5.0nm
2.0nm
1.0nm
0.5nm
BPhen
??NPD:
Iridium4f peaks
0
layerthickness(nm)
2468 10
64.0
63.6
63.2
62.8
62.4
62.0
61.6
0
layerthickness(nm)
2468 10
4.0
3.6
3.2
2.8
2.4
2.0
1.6
0
layerthickness(nm)
2468 10
2.0
2.4
2.8
3.2
3.6
4.0
4.4
4.8
0
layerthickness(nm)
2468 10
3.2
2.8
2.4
2.0
1.6
1.2
0.8
Ir 4f peak(eV)
HIBBPhen(eV)
workfunction(eV)
BPhen BPhen:CsBPhen:Cs BPhen
HIB?-NPD(eV)
BPhen:CsBPhen
BPhen:CsBPhen
. 6 36eV
0
.9e
V
Y á-
Ir(MDQ) acac
2
NPD:
EF
Evac
>3.15 V
e
5.41eV
.38e
V
6
HOMO
LUMO
4. 3e
V
1
BPhen:Cs
~4 7e
V
.0
V
>2.2e
BPhen:g
3
6. 6e
V
eV
2.63
.41 V
e
5
eV
0.9
.9 e
0 8 V
EF
Evac
HOMO
LUMO
Y á-
Ir(MDQ) acac
2
NPD:
BPhen:g
9eV
1.4
0.2 V
e
(f)
(g)
(b)
(a)
(c)(d)(e)
FIG. 6. ?Color online? Interface from ?-NPD:Ir?MDQ?2acac to BPhen and to BPhen:Cs. ?a? Development of the HBEC and HOMO
region of the UPS He I spectra and Ir 4f XPS peaks with increasing thickness of the organic layers; the dotted vertical lines mark the change
in the position of the HBEC, HOMO, and Ir 4f7/2peak. Change in ?b? work function, ?c? ?-NPD HIB, ?d? Ir 4f7/2peak, and ?e? BPhen HIB
as a function of the increasing BPhen and BPhen:Cs layer thickness; the dashed vertical line marks the start of the doped BPhen layer. ?f?
Resulting schematic energy level diagram for BPhen deposited on ?-NPD:Ir?MDQ?2acac. ?g? Resulting schematic energy level diagram
when BPhen:Cs is deposited on ?-NPD:Ir?MDQ?2acac/BPhen.
OLTHOF et al.
PHYSICAL REVIEW B 79, 245308 ?2009?
245308-4

Page 5

pure ?-NPD layer and the pure Ir?MDQ?2acac layer are
shown, fitted by multiple Gaussian peaks after a polynomial
background is subtracted. These two fits are used in the
HOMO region of the 1:1 mixed layer ?Fig. 5?b?? and are
shifted in relative position and intensity to match the mea-
sured data. This fit shows that the HOMO cutoffs are at the
same position, so the HOMO of the ?-NPD and the iridium
complex molecule are aligned in the mixed layer. The posi-
tion of the Ir?MDQ?2acac LUMO can be estimated from the
2.6?0.2 eV transport gap measured by cyclic voltammetry
to be 0.5 eV below the LUMO of ?-NPD.
For the investigation
?-NPD:Ir?MDQ?2acac and the hole-blocking layer BPhen, a
silverfoilwithdoped
?-NPD:Ir?MDQ?2acac ?5 nm? is prepared. Here the intrinsic
?-NPD had to be left out to prevent a charging of the sample
during measurement by the photogenerated holes. This has
no effect on the alignment since it does not lead to a change
in the energetic position of the mixed layer, as can be seen in
of theinterfacebetween
MeO-TPD
?5nm?
and
the same HIB of ?-NPD:Ir?MDQ?2acac in Figs. 4?b? and
6?f?. The UPS measurements in Fig. 6?a? and the resulting
values in Fig. 6?c? show that upon BPhen deposition, the
HOMO of ?-NPD:Ir?MDQ?2acac exhibits a downward shift
by −590 meV; therefore the HIB of this layer increases to
1.49 eV. The HOMO of BPhen can be distinguished after 0.5
nm of coverage; for increasing thickness, it shows a down-
ward bending of −200 meV ?Fig. 6?e??. The energetic align-
ment of this interface is shown in Fig. 6?f?. The strong
vacuum-level shift when depositing an intrinsic layer is un-
usual and must be the result of a chemical reaction between
the dissociated Ir?MDQ?2
the interface creating a dipole layer.10
Since UPS has only a small probing depth, the change in
position of the iridium 4f core-level peaks is observed by
XPS as well ?right side of Fig. 6?a??, which yields additional
information on the behavior of the mixed layer. These core-
level peaks show a downward shift by −410 meV in Fig.
6?d?, following the UPS shift in shape but not in intensity.
This can have two different reasons: ?i? XPS probes deeper
into the sample, so the bending of a lower layer is observed
or ?ii? the reaction between the iridium complex molecules
and BPhen at the interface could induce a different energy
shift of the ?-NPD compared to the Ir?MDQ?2acac.
A total BPhen thickness of only 5 nm is chosen since on
the same sample, the measurement is continued with the
electron-injection layer of BPhen doped with cesium atoms.
A thicker intrinsic BPhen layer would increase the probabil-
ity of charging during the following measurement. Instanta-
neously after the deposition of 2 Å of doped BPhen, the
work function changes by −900 meV and the BPhen HOMO
moves downward by −920 meV as can be seen at the dotted
vertical line in Figs. 6?b? and 6?e?. Since at such low cover-
age, the HOMO signal originates mostly from the underlying
intrinsic BPhen, this must mean that at the interface, the
energy levels of the intrinsic BPhen layer shift downward to
achieve alignment with the doped BPhen. Upon further
deposition, the vacuum level moves downward even more,
resulting in a total shift of −1.28 eV while the HOMO
shows a total shift of −1.3 eV. The doping ratio can be es-
timated from the relative carbon and cesium XPS peak in-
tensities and accounts in this case for one cesium atom per
BPhen molecule.
In this measurement the HOMO position of the underly-
ing ?-NPD:Ir?MDQ?2acac layer cannot be distinguished
anymore. However, by following the XPS iridium peak shift,
the influence of the n-doped layer can still be observed.
Upon BPhen:Cs deposition, a sudden downward shift of
−420 meV occurs in the Ir peak position as shown in Fig.
6?d?; for a BPhen:Cs thickness of 2 nm, this value further
increases to −710 meV without showing saturation. It is thus
most likely the shifting continues but for thicker BPhen:Cs
layers, the XPS signal becomes too faint to be detectable as
well. If one assumes a similar shift of the iridium XPS peak
and HOMO of ?-NPD, then the HIB should have increased
to at least 2.2 eV due to the contact with the n-doped layer.
In Fig. 6?g?, the assumed band alignment from this measure-
ment is shown. The position of the HOMO of BPhen at the
interface to ?-NPD:Ir?MDQ?2acac cannot be derived from
the measurement since it will move downward by some un-
+fragments and BPhen molecules at
8
V
2. 4e
Yá-
2
NPD:Ir(MDQ) acac
5 3e
V
.3
V
2.36e
0.61eV
5nm
0.18eV
0.48eV
.3
6 8e
V
. 4
4 0 eV
BPhen:a
.36eV
6
4.24eV
BPhen:Cs
0.2eV
HOMO
LUMO
EF
Evac
0.22eV
(a)
(b)
(e)
2019 18
bindingenergy(eV)
176420
HOMO
0.61eV
intensity(arb. units)
0.22eV
0.2eV
HBEC
5.0nm
10nm
2.0nm
1.0nm
0.5nm
5.0nm
BPhen:Cs
2.0nm
1.0nm
0.2nm
0.5nm
BPhen
??NPD:Ir(MDQ)
(c)(d)
04 8 12 16 20
2.0
2.2
2.4
2.6
2.8
3.0
3.2
048 12 16 20
4.4
4.2
4.0
3.8
3.6
3.4
3.2
0 4 8 12 16 20
layerthickness(nm)
1.2
1.4
1.6
1.8
2.0
2.2
2.4
workfunction(eV)
layerthickness(nm)
BPhenBPhen ?-NPD:
Ir(MDQ)
HIBBPhen(eV)
layerthickness(nm)
HIB?-NPD(eV)
?-NPD:
Ir(MDQ)
BPhen
?-NPD:
Ir(MDQ)
FIG. 7. ?Color online? Interface from BPhen:Cs to BPhen and to
?-NPD:Ir?MDQ?2acac. ?a? Development of the HBEC and HOMO
region of the UPS He I spectra with increasing thickness of the
organic layers. Change in ?b? work function, ?c? BPhen HIB, and
?d?
?-NPDHIBasafunction
?-NPD:Ir?MDQ?2acac layer thickness; the dashed vertical line
marks the start of the ?-NPD:Ir?MDQ?2acac layer. ?e? Resulting
schematic energy level diagram.
oftheBPhen and
ENERGY LEVEL ALIGNMENT AT THE INTERFACES IN A…
PHYSICAL REVIEW B 79, 245308 ?2009?
245308-5

Page 6

known amount. A minimum HIB of 3.15 eV can be assumed
since otherwise an upward dipole from the mixed layer
would result. In the schematic energy diagrams of Fig. 6, a
transport gap of 4.4?0.2 eV is assumed for BPhen; this is
concluded from the UPS measurement of the doped BPhen
layer since here the LUMO is positioned close to the Fermi
energy.
More measurements are needed to determine the HOMO
positionof theintrinsic
?-NPD:Ir?MDQ?2acac and to learn more about the align-
ment between BPhen and BPhen:Cs. Therefore, the reversed
deposition sequence is investigated as well and is shown in
Fig. 7?a?. BPhen:Cs of 10 nm is deposited on a silver foil,
followed by a stepwise deposition of 10 nm BPhen. The
measured values in Fig. 7?c? reveal a small interface dipole
of approximately 200 meV that is created within the first ML
when the HIB changes from 4.24 eV for the doped layer to
4.04 eV for the intrinsic one. The schematic alignment is
shown on the left part of Fig. 7?e?. On top of this layer,
?-NPD:Ir?MDQ?2acac is evaporated. The HOMO of BPhen
gradually shifts upward by 180 meV upon the deposition of
the mixed layer and ?-NPD:Ir?MDQ?2acac shows a shift of
480 meV as can be seen in Fig. 7?d? starting from the dashed
line. Since the BPhen layer shows only a small change in
energetic position here, the backward measurement demon-
strates that most of the shifting seen in the forward measure-
ment ?Fig. 6? occurs in the mixed layer.
It would be interesting to know how the layers behave
when p-MeO-TPD is evaporated on top. This should pull the
HOMO of the mixed layer up to a HIB of 0.9 eV again and
might influence the energetic position of the intrinsic BPhen.
However, it was not possible to do this measurement as the
sample shows immediate charging. From the combined mea-
surements in forward and backward directions, the actual
energy alignment of these interfaces is derived and is shown
as a part of Fig. 9. As stated before, we cannot exclude that
the intrinsic BPhen layer would be shifted further by the
built-in potential created between the two doped layers. The
same holds true for the intrinsic ?-NPD layer that could not
be included in the stack for most of the measurements; some
of the built-in potential could drop across this layer.
BPhenat the contactto
As a last step, the interface to the silver top contact is
investigated. For that purpose, 15 nm of BPhen:Cs is evapo-
rated on a silver foil and afterward silver is stepwise depos-
ited on top. The shift of the HOMO of BPhen is difficult to
distinguish in the UPS spectra ?Fig. 8?a?? since the silver 4d
levels are located around the same binding energy. From the
change in the position of the XPS carbon 1s state, we can
derive that the organic shifts upward by 300 meV. The posi-
tion of the vacuum level changes by about 260 meV when 10
nm silver is evaporated on top, which results in a work func-
tion of 2.4 eV that differs from the value for bulk silver. This
effect is well known for metal top contacts on organics15,16
and is probably due to the fact that one monolayer of the
organic material remains on top of the metal layer. Therefore,
it is assumed in Fig. 9 that the silver top contact has a work
function of 4.3 eV even though it is not observable by UPS
because of the residual BPhen on top. The thickness of the
depletion layer cannot be derived from this measurement.
However, by changing the deposition sequence in a separate
experiment and evaporating BPhen:Cs stepwise on silver, it
is determined to be in the range of 3 nm.
19186420
intensity(arb. unit)
0.26eV
HBEC
bindingenergy(eV)
HOMO
288 286 284 282
Carbon1s
10nm
5.0nm
2.0nm
1.0nm
0.5nm
0.2nm
BPhen:Cs
0.3eV
2 2e
V
.4
6.38eV
.22e
V
4
Silver
BPhen:Cs
0.3eV
EF
Evac
HOMO
LUMO
(a)
(b)
FIG. 8. ?Color online? Interface from BPhen:Cs to silver. ?a? Development of the HBEC and HOMO region of the UPS He I spectra and
C 1s XPS peak with increasing thickness of the metal top contact; the dotted vertical lines mark the change in the position of the HBEC and
C 1s peak. ?b? Resulting schematic energy level diagram.
4 3eV
.
6.36eV
4.24eV
Silver
BPhen:Cs
0.3eV
6.38eV
Y á-
Ir(MDQ) acac
2
NPD:
0.79eV
5. eV
4
Yá-NPD
4.1eV
5. eV
1
0.45eV
ITO
MeO-TPD:
F4-TCNQ
HOMO
LUMO
EF
Evac
0.72eV
0.9eV
5.41eV
4.04eV
5nm
3nm
2.84eV
BPhen:A
0.18eV
Vbi=2.53eV
FIG. 9. Schematic energy diagram of the full OLED device as it
can be derived from the measurements presented in this paper. The
LUMO of the ?-NPD is shown only as a thin line since the electron
transport happens on the LUMO of the Ir?MDQ?2acac. The built-in
potential Vbiis calculated from the offset between the vacuum lev-
els of the doped layers.
OLTHOF et al.
PHYSICAL REVIEW B 79, 245308 ?2009?
245308-6

Page 7

In Fig. 9 the full alignment of the red phosphorescent
OLED stack is shown. Due to the induced large carrier con-
centration on the order of 1018cm−3, the doping shifts the
HOMO of the p-type layer and LUMO of the n-type layer
close to the Fermi energy, creating a depletion layer at the
interface to the metal contact. This improves the charge-
carrier injection since the electrons and holes can efficiently
tunnel through the narrow depletion region. In addition the
conductivity increases by several orders of magnitude.17This
position close to the Fermi energy is only determined by the
host material and the amount of doping but is independent of
the choice of metals used as contacts. Therefore the doped
layers decouple the organic stack from the metals and the
performance is mainly determined by the built-in voltage
created by the doped layers and not by the difference in the
metal work functions. The choice of metal will still have
some influence since it will affect the width and bending
direction of the depletion layer.
To relate this measured built-in voltage to the perfor-
mance of an OLED device, the current-voltage characteris-
tics of the investigated stack are shown in Fig. 10. Four
samples are compared where the width of the emission layer
?-NPD:Ir?MDQ?2acac is varied between 5 and 20 nm. The
current characteristic can be divided into three regions. For
voltages below 2 V, the leakage currents dominate. In the
range from 2 V to approximately the built-in voltage of 2.5
V, all curves coincide and in this regime the shape can be
explained by the ideal-diode equation for p-n junctions con-
sidering the drift-diffusion regions. Above this voltage, the
current gets dominated by recombination and/or Ohmic con-
duction and the curves show a significant dependence on the
thickness of the emission layer which indicates a consider-
able drop of bias across this region.
IV. SUMMARY
In conclusion, based on UPS and XPS measurements, we
present the full energetic alignment of an OLED as it is
summarized in Fig. 9. The data prove that there is no com-
mon vacuum level throughout the device. This is mainly due
to the interface dipoles at the contacts and the properties of
the doped layers. The adjacent intrinsic layers show nearly
vacuum-level alignment, with only small deviations due to
interface dipoles. In the doped layers, one can furthermore
observe the typical level-bending effects and the small dis-
tance between transport level and Fermi energy. The p and n
doping of the injection layers leads to a built-in voltage of
2.53 V in this device which mainly drops across the
?-NPD:Ir?MDQ?2acac layer.
ACKNOWLEDGMENT
This work was funded by the German BMBF under Con-
tract No. 13N 8855, project acronym “ROLLEX.”
*olthof@iapp.de
†www.iapp.de
1K. Walzer, B. Maennig, M. Pfeiffer, and K. Leo, Chem. Rev.
?Washington D.C.? 107, 1233 ?2007?.
2H. Ishii, K. Sugiyama, E. Ito, and K. Seki, Adv. Mater. 11, 605
?1999?.
3J. Blochwitz, T. Fritz, M. Pfeiffer, K. Leo, D. M. Alloway, P. A.
Lee, and N. R. Armstrong, Org. Electron. 2, 97 ?2001?.
4I. G. Hill, D. Milliron, J. Schwartz, and A. Kahn, Appl. Surf. Sci.
166, 354 ?2000?.
5J. X. Tang, C. S. Lee, and S. T. Lee, J. Appl. Phys. 101, 064504
?2007?.
6A. Kahn, N. Koch, and W. Gao, J. Polym. Sci., Part B: Polym.
Phys. 41, 2529 ?2003?.
7J. X. Tang, K. M. Lau, C. S. Lee, and S. T. Lee, Appl. Phys. Lett.
88, 232103 ?2006?.
8S. Braun, M. P. de Jong, W. Osikowicz, and W. R. Salaneck,
Appl. Phys. Lett. 91, 202108 ?2007?.
9I. Salzmann, S. Duhm, G. Heimel, M. Oehzelt, R. Kniprath, R.
L. Johnson, J. P. Rabe, and N. Koch, J. Am. Chem. Soc. 130,
12870 ?2008?.
10R. Meerheim, S. Scholz, S. Olthof, G. Schwartz, S. Reineke, K.
Walzer, and K. Leo, J. Appl. Phys. 104, 014510 ?2008?.
11R. Schlaf, H. Murata, and Z. H. Kafafi, J. Electron Spectrosc.
Relat. Phenom. 120, 149 ?2001?.
12G. He, K. Walzer, M. Pfeiffer, K. Leo, R. Pudzich, and J. Sal-
beck, Proc. SPIE 5519, 42 ?2004?.
13S. Krause, M. B. Casu, A. Schöll, and E. Umbach, New J. Phys.
10, 085001 ?2008?.
14S. Duhm, G. Heimel, I. Salzmann, H. Glowatzki, R. L. Johnson,
A. Vollmer, J. P. Rabe, and N. Koch, Nature Mater. 7, 326
?2008?.
15B. Jaeckel, J. B. Sambur, and B. A. Parkinson, Langmuir 23,
11366 ?2007?.
16I. G. Hill, A. Rajagopal, and A. Kahn, J. Appl. Phys. 84, 3236
?1998?.
17M. Pfeiffer, K. Leo, X. Zhou, J. S. Huang, M. Hofmann, A.
Werner, and J. Blochwitz-Nimroth, Org. Electron. 4, 89 ?2003?.
01234
1E-5
1E-4
1E-3
0.01
0.1
1
10
100
III
II
5nm?-NPD:Ir(MDQ)
10nm?-NPD:Ir(MDQ)
15nm?-NPD:Ir(MDQ)
20nm?-NPD:Ir(MDQ)
current density(mA/cm2)
voltage (V)
I
FIG. 10. ?Color online? Current-voltage characteristic of the red
phosphorescent OLED stack at different thicknesses of the emitting
layer ?-NPD:Ir?MDQ?2acac. The current can be divided intro three
regions: ?I? leakage-current, ?II? drift-diffusion-dominated, and ?III?
recombination/Ohmic-conduction-dominated region.
ENERGY LEVEL ALIGNMENT AT THE INTERFACES IN A…
PHYSICAL REVIEW B 79, 245308 ?2009?
245308-7