Full text of DOI:10.1016/S0009-2614(00)01407-X

26 January 2001
Chemical Physics Letters 333 (2001) 432±436
www.elsevier.nl/locate/cplett
Inhibition of dark spots growth in organic
electroluminescent devices
M.K. Fung, Z.Q. Gao, C.S. Lee, S.T. Lee *
Department of Physics and Materials Science, Center of Super-Diamond and Advanced Films (COSDAF),
City University of Hong Kong, 83 Tat Chee Avenue, Kowloon, Hong Kong, People's Republic of China
Received 18 October 2000; in ®nal form 28 November 2000
Abstract
Humidity-induced oxidation has been considered as a major cause for the growth of dark spots in organic elec-
troluminescent devices. In this Letter, we found that a layer of 100 nm thick calcium deposited on top of the mag-
nesium:silver (Mg:Ag) cathode can slow down the growth rate of dark spots. Further experiments veri®ed that calcium
not in direct contact with the cathode could not protect the dark spots against growing. This excludes the possibility
that the protective function of calcium derives solely from its action as a moisture-resistive agent. In contrast, calcium
functions as a sacri®cial metal forming a galvanic couple with the magnesium electrode such that calcium would oxidize
in preference to the Mg:Ag. Ó 2001 Elsevier Science B.V. All rights reserved.
1. Introduction
interdi€usion and crystallization of organic layers
might not be necessary factors for the failure of
OELD. Most studies today share a common belief
that the formation of dark spots and their growth
are related to the reaction of the OELD with water
or oxygen species [6±10]. Among these studies, the
defects related to cathode oxidation are likely the
most probable sites for the formation and propa-
gation of dark spots [6±8]. Mg:Ag is often used as
a cathode in OELD because of its low work
function. Protecting the Mg:Ag cathode against
oxygen and moisture is therefore a necessary task.
In the present Letter, we deposited a thin layer
of calcium (Ca) directly on top of the Mg:Ag
cathode to protect the underlying cathode against
oxygen and moisture absorption. The fabricated
OELDs were exposed to di€erent relative humid-
ities and the corresponding growth rates of dark
spots were monitored. It was found that calcium,
being the second highest metal in the electro-
Formation and growth of dark spots is a major
degradation cause in organic electroluminescent
devices (OELD). This has long been an obstacle
for practical applications of OELD. This degra-
dation mechanism has been widely reported since
the ®rst high-eciency OELD was demonstrated
by Tang and Van Slyke [1] in 1987. Di€erent
causes have been suggested to account for the
formation and growth of dark spots. Fujihira et al.
[2] reported that the dark spots were caused by
interdi€usion of organic layers. Cumpston et al. [3]
proposed that the electromigration of cathode was
responsible for these non-emissive spots. Lee et al.
[4] and Gao et al. [5], respectively, showed that the
* Corresponding author. Fax: +852-2784-4696.
E-mail address: apannale@cityu.edu.hk (S.T. Lee).
0009-2614/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 2 6 1 4 ( 0 0 ) 0 1 4 0 7 - X

M.K. Fung et al. / Chemical Physics Letters 333 (2001) 432±436
433
dark-spot monitoring were taken from time to
time with a Leica DMLM ¯uorescence microscope
attached with a charge coupled device camera. All
non-encapsulated devices were stored in enclosed
containers with di€erent levels of humidity (50%,
70% and 95%) unless taken out for photography.
Changes in the diameter of representative dark
spots were measured throughout the storage time
and the areas of dark spots were counted within a
region of 2 mm  1:5 mm.
chemical series (E.C.S.), can protect the Mg:Ag
from oxidation.
2. Experimental
Indium tin oxide (ITO) coated glass with a sheet
resistance of 50 X=Ã was used as the anode and
substrate for OELD. They were ultrasonically
cleaned in a commercial detergent and then rinsed
in deionized water. Before depositing the organic
layers, the substrate was treated with an oxygen
plasma for 1 min with an AC power of 100 W.
Double-layer devices were fabricated by sequential
3. Results and discussion
7
The emissive area of all devices was approxi-
mately 3:0 mm  3:3 mm. Current±voltage±lumi-
nance (IVB) measurements showed that the
luminance of the freshly prepared devices averaged
high vacuum ꢀ10 mbar† vapor deposition of a
0
ꢀ
hole transport layer (HTL) of 600 A thick N,N -
bis-(1-naphthyl)-N; N0-diphenyl-1,10-biphenyl-4,40-
diamine (NPB) and an electron transport layer
above 550 cd=m2 at
a
current density of
ꢀ
(ETL) of 600 A thick tris-(8-hydroxyquinoline)
aluminum (Alq3). The Mg:Ag (10:1 by mass)
20 mA=cm2 and a voltage of 7.3 V. The charac-
teristics were similar to those in our previous
Letter [5] and hence the IVB curves were not
plotted here. This implied that the extra ®lms on
top of the Mg:Ag do not have any observable in-
¯uences on the performance of the freshly fabri-
cated devices.
ꢀ
cathode with a thickness of about 2000±3000 A
was then co-evaporated on top of the Alq3 layer.
The deposition rate of the organic layers (both the
HTL and ETL) and the cathode was controlled at
ꢀ
2±3 and 10 A/s, respectively. To study the e€ect of
the Ca over-coating, a device of the same con®g-
uration was fabricated but with an additional layer
Fig. 1a demonstrates an optical image of a
freshly prepared type A device which was con-
nected to a 9 V battery. It can be seen that only
a small fraction of the device area was occupied
by dark spots. The formation of these dark spots
is inevitable. It was claimed that either the
presence of residual moisture in the device [9] or
reaction of the device with contamination intro-
duced during fabrication [6] played an important
role. The sequence of the growth of dark spots
in type A devices is shown in Figs. 1b and c.
These images correspond to devices which have
been stored in an environment with a relative
humidity of 50%. In addition to a direct rela-
tionship of the growth rate of dark spots with
storage time, it is also interesting to note that
each dark spot was surrounded by a bright ring.
The mechanism of this bright ring formation is
still not clear. Localized current due to defects
[2,6] or cathode delamination due to crystalliza-
tion of Alq3 upon moisture absorption [10] may
be possible reasons.
ꢀ
of 1000 A thick Ca evaporated on top of the
Mg:Ag. A third type of device was made with an
ꢀ
extra layer of 500 A NPB or Alq3 deposited in
between the Mg:Ag and Ca. This layer completely
isolated both the metals and veri®es if any changes
of the growth rate of dark spots were due to
mechanisms other than the sacri®cial protection of
Ca. Alq3 of the same thickness was also prepared
on the Mg:Ag for comparison. The con®gurations
of the four types of devices were as follows:
ꢀ
Type A: ITO/NPB/Alq3 /Mg:Ag (3000 A),
ꢀ
Type B: ITO/NPB/Alq3 /Mg:Ag (2000 A)/Ca
ꢀ
(1000 A),
ꢀ
Type C: ITO/NPB/Alq3 /Mg:Ag (2000 A)/NPB
or Alq3 /Ca,
Type D: ITO/NPB/Alq3 /Mg:Ag (2000 A)/Alq3.
For the ease of comparison, the total thickness of
ꢀ
ꢀ
the metals in type A and B devices was 3000 A.
The luminance of the devices after deposition was
measured with a computer-controlled DC source
and a PR 650 SpectroScan. Optical EL images for

434
M.K. Fung et al. / Chemical Physics Letters 333 (2001) 432±436
Fig. 1. Optical EL images of: (a) as-prepared type A devices; (b) type A devices after a storage time of 24 h; (c) type A devices after a
storage time of 180 h. All devices were stored at a humidity of 50%.
The EL images of type B devices are shown in
Fig. 2. The density of the dark spots in the as-
evaporated device (Fig. 2a) was comparable to
that in type A device (Fig. 1a). Figs. 2b and c in-
dicated that dark-spot area was visibly reduced,
compared to the corresponding devices (Figs. 1b
and c) without the Ca layer. The relationship be-
tween the areal growth rate ꢀGA† of dark spots and
storage time for type A and B devices in di€erent
storage humidity (h ˆ 50%, 70% and 95%) is
shown in Fig. 3. Since it is dicult to count the
number of dark spots at a particular storage time
and not all the dark spots were of the same di-
ameter, GA was obtained from measuring the total
areas that were occupied with the dark spots.
Therefore, it is not an average areal growth rate of
a speci®c dark spot. At a given storage time, Ca
always retards the growth rate of dark spots to
some extent. Since the propagation of dark spots is
well known to be a humidity-induced and humid-
ity-accelerated process, there is then no doubt that
Ca can protect the devices against the moisture
e€ect. The standard electrode potentials of Ca, Mg
and Ag are )2.87, )2.34 and +0.8 V respectively
[11]. Ag is suciently inert and is less likely in-
volved in the oxidation process. Ca, having a more
negative potential than Mg in the E.C.S., is anodic
with respect to Mg. The Ca then corrodes prefer-
entially when galvanically coupled to the Mg:Ag.
In other words, Ca acts as a sacri®cial metal and is
®rst consumed resulting in the protection of the
Mg:Ag structure.
Fig. 3 shows that the GA increases linearly with
storage time for the device without the Ca layer, or
that the radial growth rate ꢀGR† of dark spots is
proportional to the square root of the storage
time. This relation suggests that the growth of
dark spots is a di€usion-controlled process. How-
ever, the situation is di€erent for devices with the
Ca layer. The best ®tted curve is not linear. Fig. 3
Fig. 2. Optical EL images of: (a) as-prepared type B devices; (b) type B devices after a storage time of 24 h; (c) type B devices after a
storage time of 180 h. All devices were stored at a humidity of 50%.

M.K. Fung et al. / Chemical Physics Letters 333 (2001) 432±436
435
The inhibition of the growth of dark spots for
devices with Ca deposition at higher storage hu-
midity ꢀh ˆ 95%† is shown in the inset of Fig. 3.
The growth rate of dark spots was at least 2 orders
of magnitude higher than that for devices exposed
to a humidity of 50%. Nevertheless, Ca still func-
tions as a sacri®cial metal.
To further arm the meritorious role of Ca in
OELD, type C devices with an additional Alq3 in
between the Mg:Ag and Ca were fabricated. Type
D devices as a reference were also fabricated so as
to identify any changes of the amount of dark
spots in type C devices that were not due to the
extra Alq3 layer. The fraction of dark spots in
devices B, C and D was measured and EL images
were taken after a storage time of 24 hours at a
relative humidity of 70%. We noted that the size
of dark spots in type C and type D devices was
indistinguishable (Fig. 4b and c) while the non-
emissive areas observed in type C devices (Fig.
4b) were larger than that in the type B devices
(Fig. 4a). Accordingly, Ca not in contact is less
e€ective in protecting the OELD against dark
spots growing. Hence, it is clear that the protec-
tive function of Ca involved in redox reaction is
more signi®cant than that as a moisture-barrier
layer. We also used NPB instead of Alq3 as the
inter-layer. The situation was even worse as the
dark spots grew faster than that in the reference
devices (type A). Further pinholes created in the
Mg:Ag/NPB interface due to the strong surface
tension of NPB may provide an additional path
for oxygen and water.
Fig. 3. Areal growth rate of dark sports ꢀGA† for devices ex-
posed to di€erent humidity (h ˆ 50%, 70% and 95%) as a
function of storage time. The inset shows the relationship of
growth rate GA with storage time at a humidity 95%. Open
symbols correspond to divices with Ca deposition.
shows that the growth rate of dark spots normally
increases slowly at the beginning, followed by a
progressive increase at the end of the curves. We
suggest that Ca rapidly formed a chemically inert
CaO passive ®lm immediately upon exposure to air.
This passive ®lm primarily acted as a barrier and
hence momentarily protected the Ca against oxi-
dation. Ultimately many corrosive sites were
formed on the surface of this sacri®cial metal, thus
accelerating the oxidation rate until the Ca was
completely consumed. Nevertheless, the nonlinear
e€ect was not so apparent for devices stored at 50%
humidity. As more than one mechanism may con-
tribute to the growth of dark spots, we suspect the
growth rate of dark spots in the above case may be
dominated by other factors rather than humidity.
Fig. 4. Optimal EL images of (a) type B devices after a storage time of time of 24 h. All devices were stored at a humidity of 70%.
Approximately 10% and 15% of dark spots were counted in a selected area of 1:5 mm  2 mm for case (a), (b) and (c), respectively.

436
M.K. Fung et al. / Chemical Physics Letters 333 (2001) 432±436
4. Summary
References
[1] C.W. Tang, S.A. Vanslyke, Appl. Phys. Lett. 51 (1987) 913.
[2] M. Fujihira, L.M. Do, A. Koike, E.M. Han, Appl. Phys.
Lett. 68 (1996) 1787.
Growth of dark spots in OELD was inhibited
through covering the surface of Mg:Ag with a
ꢀ
layer of 1000 A thick Ca. The presence of Ca did
not a€ect the luminance and eciency of the de-
[3] B.H. Cumpston, K.F. Jensen, Appl. Phys. Lett. 69 (1996)
3941.
ꢀ
vices. With a 500 A thick NPB or Alq3 deposited
[4] S.T. Lee, Z.Q. Gao, L.S. Hung, Appl. Phys. Lett. 75 (1999)
1404.
in between the Mg:Ag and Ca, the protective e€ect
of Ca disappeared. Ca is thus considered as a
sacri®cial metal. When Ca is in contact with Mg,
Ca is preferentially oxidized, thus protecting Mg
from oxidation. We suppose that utilization of this
technique, together with the conventional encap-
sulation, can further prolong the lifetime of
OELD.
[5] Z.Q. Gao, W.Y. Lai, T.C. Wong, C.S. Lee, I. Bello,
S.T. Lee, Appl. Phys. Lett. 74 (1999) 3269.
[6] P.E. Burrows, V. Bulovic, S.R. Forrest, L.S. Sapochak,
D.M. McCarty, M.E. Thompson, Appl. Phys. Lett. 65
(1994) 2922.
[7] J.R. Sheats, H. Antoniadis, M. Hueschen, W. Leonard, J.
Miller, R. Moon, D. Roitman, A. Stocking, Science 273
(1996) 884.
[8] J. McElvain, H. Antoniadis, M.R. Hueschen, J.N. Miller,
D.M. Roitman, J.R. Sheats, R.L. Moon, J. Appl. Phys. 80
(1996) 6002.
[9] H. Aziz, Z. Popovic, S. Xie, A.M. Hor, N.X. Hu, C. Tripp,
G. Xu, Appl. Phys. Lett. 72 (1998) 756.
Acknowledgements
[10] H. Aziz, Z. Popovic, C.P. Tripp, N.X. Hu, A.M. Hor,
G. Xu, Appl. Phys. Lett. 72 (1998) 2642.
This work was supported by the Research
Grants Council of Hong Kong (project number
9040430 and 8730009). M.K. Fung would like to
thank Prof. L.S. Hung for his comments.
[11] Clark, Hawley, Encyclopedia of Chemistry, Electrochem-
istry, Reinhold, New York, 1957, p. 338.