IEEE ELECTRON DEVICE LETTERS, VOL. 27, NO. 7, JULY 2006555
Low-Frequency Noise Measurement and Analysis
in Organic Light-Emitting Diodes
Lin Ke, Xin Yue Zhao, Ramadas Senthil Kumar, and Soo Jin Chua
Abstract—Low-frequency noise characteristics of organic light-
emitting diodes are investigated. Two noise components were
found in experimental low-frequency noise records, namely: 1) 1/f
Gaussian noise from device bulk materials and 2) an excessive
frequency-related part of noise related to device interfaces or
defects and traps. 1/f noise is said to be related to carrier mobil-
ity. Degradation, especially photo-oxidation of the electrolumines-
cence polymer, is a possible reason that affects carrier mobility.
The excessive part of noise is believed to be related to the carrier
numbers and could come from the interface deterioration, defects
and traps generation and furnish. The excessive part of noise
increases much faster during device stress. This shows that the
degradation related interface defects and traps is much faster.
Index Terms—Low-frequency noise, organic light-emitting
diode (OLED), 1/f noise.
to those of light-emitting diodes (LEDs) based on inorganic
materials and have a lot of advantages over traditional liquid-
crystal display (LCD) , ; however, long-term stability re-
mains one of the critical issues hindering practical applications.
Conventional techniques for studying device degradation
are mainly based on lifetime tests under accelerated stress
conditions, and they are well suited for collecting statistical
information on the expected lifetime of a given set of de-
vices. Low-frequency noise is a sensitive diagnostic tool to
examine the internal mechanisms of electrical devices , .
For example, it has been shown that devices with identical
current–voltage (I–V ) behavior can exhibit very different low-
frequency noise characteristics. This is primarily because the
I–V behavior represents a macroscopic description of the de-
vice characteristics, whereas low-frequency noise is a sensitive
probe of defects, nonuniformities, surface velocity fluctuations,
etc., due to incomplete bonding or defect sites at surfaces or
interfaces . Furthermore, a correlation between the device
morphological changes that occurred during the degradation
process and the low-frequency noise changes is observed. This
HE BRIGHTNESS and efficiency of organic light-
emitting diodes (OLEDs) are now considered comparable
Manuscript received March 15, 2006; revised April 20, 2006. The review of
this letter was arranged by Editor P. Yu.
L. Ke and R. S. Kumar are with the Institute of Materials Research and
Engineering, Singapore 117602 (e-mail: firstname.lastname@example.org).
X. Y. Zhao is with the Center of Optoelectronics, National University of
Singapore, Singapore 119260.
S. J. Chua is with the Institute of Materials Research and Engineering, Singa-
pore 117602 and also with the Center of Optoelectronics, National University
of Singapore, Singapore 119260 (e-mail: email@example.com).
Digital Object Identifier 10.1109/LED.2006.877283
suggests that sampling of noise during device operation can be
and explore the inner device degradation mechanism of LEDs
in real-life applications, which could not be easily obtained by
macroscopic method .
There have been several publications dealing with low-
frequency noise in both semiconductor materials and devices.
However, there are relatively very few studies on the low-
frequency noise of OLEDs. In this letter, a low-frequency
voltage noise investigation across OLEDs was reported.
Indium–tin–oxide (ITO)-coated glass with a sheet resis-
tance of 20 Ω/sq was used as a substrate for OLED device
fabrication. The routine cleaning procedure includes sonica-
tion in acetone and methanol followed by oxygen plasma
treatment. A naphthyl-substituted benzidine derivative (NPB)
hole-transport layer measuring 75 nm and an aluminum tris
(8-hydroxyquinoline) (Alq3) electroluminescence (EL) layer
measuring 75 nm are deposited in high-vacuum 2 × 10−5Pa. A
5 Å lithium fluoride (LiF) and a 200-nm-thick aluminum (Al)
are deposited as cathode.
The device is kept in a shielded metal box and powered by
a constant current supply, which is powered by batteries. The
photodiode is also put in the metal box for light detection. The
voltage across the OLED and current from the photodiode were
measured using a National Instrument PXI-4070 digital multi-
meter. Both of the noises from the OLED voltage and from the
photodiode current were measured using a National Instrument
PXI-5122 two-channel high-performance high-speed digitizer,
which can perform both spectrum and time-domain analysis.
The optical noise of the OLED device is subjected to another
publication. A verification of the system noise measurement is
done using a commercial metal-film resistor.
III. RESULTS AND DISCUSSION
Fig. 1 shows the frequency dependence of the noise power
spectrum density (PSD) Sv(f) at five different stress times
of 4 min, 12, 35, 56, and 74 h. The device is driven un-
der a constant current of I = 2 mA, which corresponds to a
current density of 0.05 A · cm−2at a room temperature of
T = 295 K in dark shielded metal box without encapsulation.
The device lifetime has been continuously tested for about
80 h. The inset shows device lifetime curves. With increasing
stress time, the luminescence is decreasing and driving voltage
0741-3106/$20.00 © 2006 IEEE
556IEEE ELECTRON DEVICE LETTERS, VOL. 27, NO. 7, JULY 2006
versus time and luminescence versus time).
Frequency dependence of PSD Sv(f) at five different stress time of
is increasing. As shown in Fig. 1, all the Sv(f)s show generic
1/fαbehavior. The 1/f curve is also shown in the figure (solid
line) for comparison. It can be observed that the absolute value
of the low-frequency noise slope indicated in Fig. 1 in log
scale is the frequency exponent α in Hooge’s equation. The
calculated slope α is from 1.4 to 1.8. From the figure, it shows
that with increasing stress time, the total noise and noise slope
are increasing. This phenomenon was observed in every device
measured, and it means that as a device degrades, the total low-
frequency noise is increasing.
1/f noise is found in all conductors, and their spectral inten-
sity is of the form 1/fαwith α close to unity. The contribution
of the total noise of the OLED structure under test is a total
contribution from bulk of EL materials, bulk of metal contacts,
the interfaces, and other sources such as traps and defects,
etc. If the 1/f noise is from bulk of materials and metal, the
slope should be strictly unity according to Hooge’s theory .
frequency noise slope is not unity, must be from the interfaces
and other resources. Assume the total spectral density of the
low-frequency noise observed in the studied devices can be
described in general as
Sv(f) = Sv,1/f+ Sv,e
where Sv(f) is the total PSD, Sv,1/findicates the total PSD of
1/f noise, and Sv,eshows the total PSD of the excessive part
of noise from the interfaces. In order to analyze and compare
the strength of 1/f noise and the excessive part of noise, an
extraction of 1/f noise and the excessive noise is done through
spectrum separation. Fig. 2 shows the total 1/f noise component
of Sv,1/fand the total Sv,ecomponent for the five stressing
times indicated in Fig. 1. As shown in Fig. 2, both of the
1/f noise and excessive noise are increasing as stress time
increases. The excessive part of noise increases much faster
with a slope of 7.68 × 10−10V2/Hz · s compared with the 1/f
noise, which has an increasing rate of 4.24 × 10−14V2/Hz · s.
It can be observed that the absolute value of the low-
frequency noise slope indicated in Fig. 1 in log scale is the
Sv,Icomponent for the five respective stressing times indicated in Fig. 1.
The increase rate of the excessive part of noise is 7.68 × 10−10V2/Hz · s,
whereas that of the 1/f noise is 4.24 × 10−14V2/Hz · s.
Total 1/f noise component of Sv,1/fand total excessive part noise
frequency exponent α versus stress time for sample point no. 6.
Frequency exponent α change rate versus device lifetime. Inset:
frequency exponent α in Hooge’s equation. α can be obtained
foreach 1/fαnoisePSDateachspecific time.TheinsetofFig.3
plots the frequency exponent α at every specific time. dα/dt
is obtained by linear fitting the plot in the inset of Fig. 3. A
set of NPB/Alq3OLED devices with the same structure and
process are fabricated and subjected to the lifetime test under
either light or dark. Devices with a lifetime larger than 100 h
are encapsulated devices. dα/dt versus the different devices’
corresponding lifetimes are then shown in Fig. 3. It shows that
the slower the slope increases, the longer the device lifetime;
the change trend follows the exponential rule.
Hooge proposed that 1/f noise is associated with carrier
mobility . Therefore, one explanation for the increase of the
1/f noise of OLED during its degradation is that the mobility
of carriers inside the OLED decreases. Hence, high mobility
of carriers produces less noise, and low mobility of carriers
produces more noise. This is also reasonable in the case of EL
materials because electrons move in EL materials by hopping.
Low mobility makes electrons hard to hop forward and make
KE et al.: LOW-FREQUENCY NOISE MEASUREMENT AND ANALYSIS IN OLED 557 Download full-text
mobility is the degradation of EL materials. It is well known
that the critical drawback of the EL materials is the rapid rate
of photo-oxidation under ambient conditions , . The
oxygen precipitate can act as scattering centers to cause the
change of mobility of the charge carriers, which in all cases
cause the carrier mobility decreases, further degrading device
performance and ultimately limiting device lifetime. The mo-
bility related to scattering centers such as oxygen precipitate is
also reported in other papers .
The excessive part of low-frequency noise, which leads to an
abnormally high level of 1/f noise, can be explained by the fluc-
tuations in the number of charge carriers. Song et al.  used
the tunneling model with a direct application of McWhorter’s
potential  fluctuation mechanism and explained the higher
amplitude of noise PSD in the low-frequency range. They sug-
gested that the energy barriers resulting from abundant defects
should cause fluctuations in the number of charge carriers,
OLEDs, the degradation is under extensive studies and largely
attributed to the formation of nonemissive spots caused by the
degradation of the EL materials,the delamination at the cathode
interface, cathode oxidation, and electrochemical reactions at
the organic/electrode interfaces. The interface deterioration has
been proved to play an important role in the device degradation
process , which definitely increases and fluctuates the en-
carriers. The other possible reasons, which also lead to the
fluctuations of the carrier number, are widely existing material
defects and flaws. The EL material defects could come from the
synthesis and purification process, etc., and could decrease the
carrier injections and cause the much-increased low-frequency
Comparing the two aforementioned low-frequency noise
mechanisms, as shown in Figs. 2 and 3, a higher increase rate
of frequency exponent dα/dt shows that the faster the noise
increase, the shorter the device lifetime and that the excessive
part of the noise increases much faster than the 1/f noise
part. The low-frequency noise analysis hints that in order to
increase the device lifetime, measures have to be adopted to
suppress the low-frequency noise level and its increase rate.
The excessive part of noise, which increased faster than the 1/f
noise part, shows that carrier-number-increase-related interface
or defected degradation increased much faster than mobility-
decrease-related bulk degradation during device lifetime stress.
Low-frequency noise characteristics of OLEDs are investi-
gated. Spectrum extraction techniques are used to obtain two
different noise spectra. Two noise components were found in
experimental noise records, namely: 1) 1/f Gaussian noise from
device bulk materials and 2) an excessive frequency-related
part of noise related to device interfaces or defects and traps.
1/f noise is said to be related to carrier mobility. Degradation,
especially photo-oxidation of the EL active material, is a pos-
sible reason that affects carrier mobility. The excessive part of
noise is believed to be related to the carrier numbers, which
the fluctuation could be from, the interface deterioration, and
defects and traps generation and furnish. The excessive part
noise increase much faster during device stress, which shows
that the degradation related to interface, defects and traps is
much faster. These two noises have different physical origins
in LEDs. The shorter the lifetime, the faster the noise slope
 D. D. C. Bradley, “Electroluminescent polymers: Materials, physics and
device engineering,” Curr. Opin. Solid State Mater. Sci., vol. 1, no. 6,
pp. 789–798, Dec. 1996.
D. Roitman, and A. Stocking, “Organic electroluminescent devices,”
Science, vol. 273, no. 5277, pp. 884–888, Aug. 1996.
 S. M. Bezrukov and J. J. Kasianowicz, “Current noise reveals protonation
kinetics and number of ionizable sites in an open protein ion channel,”
Phys. Rev. Lett., vol. 70, no. 15, pp. 2352–2355, Apr. 1993.
 B. D. Ursutis and B. K. Jones, “Low-frequency noise used as a lifetime
test of LEDs,” Semicond. Sci. Technol., vol. 11, no. 8, pp. 1133–1136,
 M. Fukuda, T. Hirono, and F. Kano, “Degradation behavior of narrow-
spectral-linewidth DFB lasers,” IEEE Photon. Technol. Lett., vol. 5,
no. 218, pp. 1165–1167, 1993.
 G. Letal, S. Smetona, R. Mallard, J. Matukas, and V. Palenskis, “Reliabil-
ity and low-frequency noise measurements of InGaAsP multiple quantum
well buried-heterostructure lasers,” J. Vac. Sci. Technol. A, Vac. Surf.
Films, vol. 20, no. 3, pp. 1061–1066, May 2002.
 F. N. Hooge, T. G. M. Kleinpenning, and L. K. J. Vandamme, “Experi-
mental studies on 1/f noise,” Rep. Prog. Phys., vol. 44, no. 5, pp. 479–532,
 M. Kroutvar, “Detecting Si-defects with diodes,” M.S. thesis, Dept. Phys.,
Univ. Missouri, St. Louis, Jul. 12, 1999.
 B. H. Cumpston and K. F. Jensen, “Photo-oxidation of polymers used
in electroluminescent devices,” Synth. Met., vol. 73, no. 3, pp. 195–199,
 Y. Kaminorz, E. Smela, O. Inganas, and L. Brehmer, “Sensitivity of poly-
thiophene planar light-emitting diodes to oxygen,” Adv. Mater., vol. 10,
no. 10, pp. 765–769, 1998.
 K. L. Wang, “Low noise microwave devices: GaAs/InAs pseudomor-
phic high electron mobility transistor (pHEMTs),” Electr. Eng. Dept.,
Univ. California, Los Angeles, CA, Final Rep. 1997-98 for MICRO
 Y. Song, A. Misra, P. P. Crooker, and J. R. Gaines, “1/f noise and mor-
phology of YBa2Cu3O7-δ single crystals,” Phys. Rev. Lett., vol. 66, no. 6,
pp. 825–828, Feb. 1991.
 S. L. Rumyantsev, Y. Deng, S. Shur, M. E. Levinshtein, M. Asif Khan,
G. Simin, J. Yang, X. Hu, and R. Gaska, “On the low frequency noise
mechanisms in GaN/AlGaN HFETs,” Semicond. Sci. Technol., vol. 18,
no. 6, pp. 589–593, Jun. 2003.
 L. Ke, S. J. Chua, K. R. Zhang, and N. Yakolev, “Degradation and failure
of organic light-emitting devices,” Appl. Phys. Lett., vol. 80, no. 12,
pp. 2195–2197, Mar. 2002.