Electric field and charge distribution imaging with sub-micron resolution in an organic Thin-Film Transistor
ABSTRACT Here we show how Stark spectroscopy, coupled with confocal microscopy, is able to directly map the electric field in an n-type Copper-Fluorinated Phthalocyanine Thin-Film Transistor (TFT) under different operating conditions. To this extent, we locally probe Electro-Reflectance, with a nominal spatial resolution better than 500 nm, exploiting the fact that the detected signal is directly proportional to the square of the local field on the probe volume. This electric field imaging technique has unique advantages because it is non-invasive, since it exploits low incident power and because it probes the existing field in the bulk rather than the surface. Combining the experimental data with numerical modeling, it is possible not only to reconstruct the space charge profile in the few-nanometer thick accumulation layer, but also to extract the AC electron mobility.
Electric field and charge distribution imaging with sub-micron resolution
in an organic Thin-Film Transistor
Calogero Sciasciaa,⇑, Michele Celebranob, Maddalena Bindaa, Dario Natalia,c,
Guglielmo Lanzania, Juan R. Cabanillas-Gonzalezd
aCNST IIT@Politecnico di Milano, via Pascoli 70/3, 20133 Milano, Italy
bPhysics Department, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy
cElectronics Department, Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy
dMadrid Institute for Advanced Studies (IMDEA) in Nanoscience, Facultad de Ciencias, Av. Tomas y Valiente 7, Cantoblanco, 28049 Madrid, Spain
a r t i c l ei n f o
Received 21 July 2011
Accepted 23 September 2011
Available online 20 October 2011
Stark shift spectroscopy
a b s t r a c t
Here we show how Stark spectroscopy, coupled with confocal microscopy, is able to
directly map the electric field in an n-type Copper-Fluorinated Phthalocyanine Thin-Film
Transistor (TFT) under different operating conditions. To this extent, we locally probe Elec-
tro-Reflectance, with a nominal spatial resolution better than 500 nm, exploiting the fact
that the detected signal is directly proportional to the square of the local field on the probe
volume. This electric field imaging technique has unique advantages because it is non-
invasive, since it exploits low incident power and because it probes the existing field in
the bulk rather than the surface. Combining the experimental data with numerical model-
ing, it is possible not only to reconstruct the space charge profile in the few-nanometer
thick accumulation layer, but also to extract the AC electron mobility.
? 2011 Elsevier B.V. All rights reserved.
Many applications envisioned in organic electronics,
like sensoristics, displays, smart packaging, etc. , require
the development of performing Thin-Film Transistors
(TFTs). However, the optimization of performances re-
quired for such applications demands a deep understand-
ing of the device physics that we are still lacking in many
While for inorganic semiconductors the equations
describing charge injection and transport are completely
set , the same does not hold true for carbon-based mate-
rials [3–6]. Intrinsic complexity of disordered organic sys-
tems makes them not theoretically accessible as their
crystalline counterparts. In fact, organic semiconductors
are chemically and morphologically inhomogeneous caus-
ing fluctuations at the microscopic level in both average
and variance of electronic levels distribution . As far as
the charge transport, no consensus exists on how to model
its dependence on charge density, electric field and disor-
der [6,8–11]. Nor is the interpretation of charge injection
straightforward. In most cases, metallic–organic interface
cannot be simply described in term of Mott-Schottky bar-
rier because interface dipoles, still not easy to be predicted,
arise in consequence of pillow effect, Fermi level pinning,
chemical reactions, etc. [12,13].
The relatively poor knowledge and engineering of me-
tal–organic interfaces often results in oTFTs suffering of
contact resistances that in addition can show a non-linear
dependence from applied electric field .
In this framework it appears evident the importance of
a tool for detecting the electric field in the channel of a
working device. The major techniques used for electric
field detection in organic electronics are Electrical Force
Microscopy (EFM) and Kelvin Probe Microscopy (KPM).
1566-1199/$ - see front matter ? 2011 Elsevier B.V. All rights reserved.
⇑Corresponding author. Tel.: +39 02 23999896; fax: +39 02 23996126.
E-mail addresses: email@example.com (C. Sciascia), michele.celeb-
firstname.lastname@example.org (M. Celebrano), email@example.com (M. Binda), dar-
firstname.lastname@example.org (D. Natali), email@example.com (G. Lanzani),
firstname.lastname@example.org (J.R. Cabanillas-Gonzalez).
Organic Electronics 13 (2012) 66–70
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/orgel
The KPM (EFM) methods are sensitive to the electric poten-
tial (or its gradient) at the probe. The remote field sensing
of the AFM-based methods may lead to problems in reso-
lution and interpretation of the signals . In particular,
in bottom-gate oTFT the most intriguing physics plays at
the buried interface between the semiconductor and the
gate insulator where solid probes cannot usually penetrate.
The AFM methods may detect buried charges but such ver-
tical sensitivity is paid in terms of lateral resolution. Opti-
cal methods such as confocal microscopy are instead
sensitive to the electrical field in the volume of the optical
beam which has a typical spatial lateral resolution of k/
2?NA, with k the light wavelength and NA the numerical
aperture of the optical objective . This leads to consid-
erable advantages in terms of localization and interpreta-
tion of the signal. Being techniques of not invasive nature
they do not modify the system of study or its boundary
conditions. Moreover high operation bandwidths can be
easily implemented which enables for instance fast data
In the present work, we map the electric field distribu-
tion of the active area of an oTFT in operating conditions
with an all-optical technique which overcomes some of
the previous limitations. The basic idea consists of measur-
ing the Electro-Reflectance (ER) response coming from the
organic layer – proportional to the square of the electric
field module – taking advantage of submicron spatial res-
olution of a confocal microscope . For this purpose
we use a non-perturbative optical field, with considerably
low incident laser intensity (average power <100 nW) and
a reduced risk of optical damaging. In fact, all our measure-
ments were performed in ambient conditions and we do
appreciate good signal stability over hours of continuous
irradiation. The good lateral resolution (better than
500 nm) combined with sensitivity at bulk level allows
us to monitor electric field profile coming from few-nano-
meters thick accumulation layer. The paper is organized as
follows: in Section 2 the experimental set-up is outlined;
in Section 3 the Electro-Reflectance data are presented;
in Section 4 measurements and simulations are combined
to extract electrical field and charge profile and finally the
conclusions are given.
2. Materials and methods
The far field ER measurements are performed in a stan-
dard set-up using a halogen lamp (ASB-W-30, Spectral
Products), filtered from single monochromator (CM 112,
Spectral Products with 0.3 mm slits) and detected through
a Si photodiode (SM1PD1A, Thorlabs). The signal is then
processed with a lock-in amplifier from Stanford Instru-
ments. The electrical sources are an amplified Hewlett–
Packard function generator (mod. 3310A) and a power sup-
ply (E3648A from Agilent). For the microscopic ER set-up,
the photo-detector and the acquisition system is the same
as the far field system, but the probing light is provided by
a Ti:Sapphire tunable laser with power below 100 nW to
minimize photo-degradation of the organic film. The focus-
ing apparatus is composed by a confocal microscope (with
0.75-NA long-working distance objective) in reflection
configuration. Images are obtained by raster-scanning the
sample with respect to the beam through a 3D piezo stage
(P-517, Physik Instrumente) in the (x, y) plane, with the z
axis used for focus fine optimization. As far as oTFT, the
gate electrode consists of highly doped silicon underneath
a 130 nm silicon dioxide layer, which acts as insulator. On
top of silicon dioxide, a set of interdigitated gold electrodes
(with chromium adhesion layer) are deposited by photoli-
thographic technique. For our measurements, we used de-
vices with channel length of 12 lm with a typical gold
thickness of 100 nm. As final step, we evaporate a 40 nm
thick film of CuPcF16at maximum pressure of 10?5mBar
and with deposition rate of 0.2 nm per second. The final
thickness of organic layer is kept smaller compared with
electrodes in order to avoid interferometer artefacts on
top of the electrodes, known as dark electrode effect ,
as well as to avoid charge broadening over the metallic fin-
gers . The material is provided by Sigma–Aldrich and
we did not perform any further purification or target sur-
face functionalization. The device behaves as an electron
only transporter, with charge mobility le, evaluated from
trans-characteristic curve , around 5 ? 10?7cm2/Vs in
air and 10?5cm2/Vs in vacuum. In spite of this modest
mobility (likely attributed to the purity degree of the mate-
rial and to the absence of functional groups on the oxide
surface) , this material is suitable for ER confocal
experiments owing to its high photochemical stability in
air as well as large ER response at the 780 nm, tuned to
the fundamental wavelength of the Titanium:Sapphire la-
ser [24,25]. Simulations were performed using PDETOOL
in MATLAB 7.4.0 environment.
As starting point for the microscopic analysis we mea-
sured the ER response from the interdigitated array area
taken with a standard ‘‘macroscopic’’ set-up. ER is a well-
established technique measuring reflectance variation as
function of an applied voltage. In order to remove artefacts
due to fluctuations in optical density, spectra are normal-
ized to the zero field reflectance as (Ron–Roff)/Roff, where
the apexes on and off stay for the presence and absence
of applied field and R for reflectance. ER is widely used in
solid state physics, and both the intensity and spectral
shape of the signal are ruled by the external electric field
According to perturbation theory, the electric field leads
both to an energetic shift of the electronic levels (DE) and
to a re-distribution of the oscillator strength among differ-
ent states (Stark shift). Such variations affect the dielectric
constant and in ultimate analysis the reflectance.
In case of an isotropic distribution of molecules:
with a the absorption coefficient, ~ m the permanent dipole
of the molecule and p
to Stark shift, usually dominant in organic molecules, opti-
cally-active field-induced species, such as polarons and bi-
polarons can also affect and play a role in determining the
$the polarizability tensor. In addition
C. Sciascia et al./Organic Electronics 13 (2012) 66–70
ER signal . To confirm that reflectance modulation
arises exclusively from Stark shift and to rule out spectro-
scopic features of charged species, we repeated the exper-
iment under different oTFT bias conditions. We compared
spectra acquired when carriers accumulation occurred (ob-
tained modulating the gate to source voltage and keeping
the drain grounded), to spectra acquired without carriers
accumulation (obtained modulating the drain to source
voltage while keeping the gate grounded). Fig. 1 shows
that the same spectral features over the 600–900 nm spec-
tral range are obtained, thus proving the Stark-effect origin
of the signal.
Furthermore the out-of-phase component (not reported
here), related to the presence of long-living species, is zero
for frequencies up to 3 kHz. Since polaron absorption is of-
ten delayed with respect to the field phase due to the inter-
play of charge trapping – recombination kinetics, this
supports a negligible role of charge induced absorption in
defining the ER. Moreover the results are in agreement
with spectroscopic studies performed on related phthalo-
Fig. 2 reports the profile of confocal R and ER profiles at
780 nm along two channels within the interdigitated area
when VDS= 0 V and VGS= 15 V(1 + sin(2pft)), being the fre-
quency f set at 1 kHz. The R profile depends on the geom-
etry and the materials composing the oTFT rather than its
electrical polarization. It highlights the topography of elec-
trodes, appearing as well-defined bumps emerging from a
flat background due to the difference in reflectance be-
tween gold and silicon dioxide. From the analysis of the
leading edges, it is possible to infer an optical lateral reso-
lution better than 500 nm.
The effect of the drain voltage modulation on the ER
distribution is displayed in Fig. 3. The 1D scans represent
ER distribution under VDS= 15 V(1 + sin(2pft)) when VGS
is 22 and 0 V (panel A and B respectively). In both cases
the ER profile is higher close to the drain side, the peak
being more shallow when VGS= 22 V, and more sharp and
intense when VGS= 0 V.
VDS= 0 Vand
VGS= 15 V(1 + sin(2pft)), reported in Fig. 2. This electrical
situation corresponds to the simplest one, where no signif-
icant net current is expected to flow since the source and
drain are at the same potential, while a large density of
charge is expected to be accumulated at the oxide/semi-
conductor interface due to VGS. The obtained ER profile dis-
plays two symmetric sharp peaks, close to drain and source
electrodes, gradually decreasing towards the center of the
channel and abruptly vanishing in correspondence of elec-
trodes. The symmetry is a consequence of source and drain
equipotentiality, while the absence of ER on electrodes
comes from the electric shielding from underneath metal.
From a qualitative point of view, the constant (DC) bias
VGS= 15 V induces a spatially uniform accumulation layer
within the organic film. The alternate (AC) voltage
VGS= 15 Vsin(2pft) injects and sweeps away carriers every
1/f period. Efficient electron drift would lead to fast carrier
re-distribution giving a spatially uniform time-dependent
charge sheet. On the contrary if the mobility of electrons
is modest, as in our case, short penetration from source
and drain into the channel is expected even at moderate
modulation frequency. The result is a space-dependent
charge density, which in turn is responsible for the mea-
sured non-constant ER profile.
We now turn to a quantitative analysis, and we com-
pare measured ER data with numerical solution of Laplace
equation with static field in order to extract the charge dis-
tribution. It is to be stressed that the numerical model
being purely electrostatic, it does not postulate any mech-
anism about charge injection or transport, and conse-
quently the nature of these two phenomena does not
influence the accuracy of the simulation. According to the
geometrical symmetry, we set our differential problem in
a two dimensional frame corresponding to a transverse
section of the device (the longitudinal dimension can be
rigorously neglected because there is no voltage gradient
along electrodes). In order to determine the space charge
distribution we use an optimization iterative procedure
described hereafter. It is to be noted that, at the reference
frequency f, since the Stark shift is proportional to the
square of the field, the ER signal is given by the product be-
tween the DC and the AC electrical field amplitude . As
to the DC term, it can be readily computed imposing a con-
stant charge density that completely shields the electric
field from the gate under Dirichlet boundary conditions
VGS= 15 V and VDS= 0 V. As to the AC term, we set the AC
injected charge as the unknown variable to be adjusted in
order to minimize the square of the difference between
simulated (?FDC?FAC) and experimental ER data.
Fig. 4A reports the simulated ER compared with the
measured one. Panel B reports the corresponding calcu-
lated charge density profile, shown as sum of DC and AC
terms. As expected, the maxima and minima are located
close to the electrodes, where the injection occurs, and
Fig. 1. Normalized ER spectra measured modulating the Gate (black filled
squares) or the Drain (white open circles) with respectively Drain and
Gate at 0 V. Considering that charge injection is expected to occur in the
former case only, the fact two spectra have identical shape rules out
polaronic features in this wavelength range. Hence the signal is purely
due to Stark shift.
C. Sciascia et al./Organic Electronics 13 (2012) 66–70
the profile flattens toward the middle of the channel
(Fig. 4B), where it attains the value of a capacitor with volt-
age difference (DC) VGS= 15 V and specific capacitance
er?eo/t (eothe vacuum permittivity, erthe relative permit-
tivity and t the thickness of SiO2layer). The electron pene-
tration depth from the contacts into the channel is about
2 lm upon the 1 kHz modulation. Combining this informa-
tion with the longitudinal electric field (?60 kV/cm) result-
ing from the calculated charge profile, we can estimate the
carrier mobility le(le= 2?d/F?s, being d the electron pene-
tration depth, F the magnitude of the longitudinal electric
field and s the period of the modulation). Interestingly en-
ough, the mobility turns out to be 0.6 ? 10?5cm2/Vs, thus
more similar to the value obtained in vacuum from transfer
characteristic curves, than to value obtained in ambient
atmosphere from analogous measurements (remind that
ER measurements were performed in ambient atmo-
sphere). This can be explained considering that ER mea-
surements give access to the AC mobility, filtering out
processes with characteristic long relaxation times. Thus
if the modulation frequency is high enough, it can bypass
deep traps often associated to the presence of oxygen or
moisture. In addition, we can also deduce that trapping/
2530 3540 45
Fig. 2. The red line corresponds to Reflectance (R) profile, the blue line is the Electro-Reflectance (ER) profile when a modulated gate voltage is applied;
VGS= 15 V(1 + sin(2pft)) and VDS= 0 V. ER signal is proportional to the squared module of the local electric field. R is expressed as the voltage read out by the
photo-detector during the scan, while ER is the ratio between DR and R, with DR is the variation of Reflectance induced by the modulated applied voltage. R
signal resembles the electrodes topology and gives indication about the lateral spatial resolution of the system (?500 nm). The probing light for both R and
DR is the emission from a mode-locked Titanium:Sapphire laser tuned at 780 nm. (For interpretation of the references to colour in this figure legend, the
reader is referred to the web version of this article.)
ER @ VDS= V0+V0.sin(ωt); VGS=22V
ER @ VDS= V0+V0.sin(ωt); VGS=0V
Fig. 3. ER maps of the transistor when an alternated voltage is applied
between drain and source and: (A) applied Gate bias VGSis 22 V or (B) VGS
is constantly kept at 0 V. Letters S and D locate respectively the position of
the source and the drain electrodes. We observe a reversible and
reproducible ER dwindling and flattening when the Gate voltage is
Fig. 4. (A) Measured (red squares) and simulated (blue stars) ER signal at
VDS= 0 V and VGS= 15 + 15 V?sin(2pft). (B) The black solid line is the
steady state charge induced by DC VGSbias, red and blue dashed lines are
the charge densities due to the AC VGScomponent. (For interpretation of
the references to colour in this figure legend, the reader is referred to the
web version of this article.)
C. Sciascia et al./Organic Electronics 13 (2012) 66–70
detrapping occurs in air on a time scale larger than the
modulation period (few hundreds of microseconds) .
This electro-optical technique for the measurements of
the AC mobility in oTFT, which basically relies on the
choice of a modulation period shorter that the carrier tran-
sit time, is advantageous with respect to all-electrical tech-
niques since it avoids the non-trivial management of stray
and device capacitances .
VDS= 15 V(1 + sin(2pft)) and DC gate voltage VGS= 22 V,
shown in Fig. 3A. From a qualitative point of view, the
DC terms of VGSand VDSinduce a non-uniform accumula-
tion layer that is periodically shaped by the AC drain–
source component. Similarly to the case under zero
drain–source field, modest electron mobility will likely
limit carrier re-distribution. The region of the channel close
to the drain actually contains information about the mo-
bile charge injection/extraction, while the region close to
the source is almost insensitive to this process (as the gate
to source voltage is fixed). The influence of carrier accumu-
lation on ER profile can be visualized when comparing
Fig. 3A and B. Fig. 3B depicts the case where drain–source
bias is modulated whereas the gate is grounded, i.e. negli-
gible injection. Under these conditions the ER profile re-
flects a sharp peak close to the drain due to the so-called
blade effect which intensifies the electric field in proximity
of sharp geometry, i.e. electrode edge [33,34]. Compared to
Fig. 3A, the absence of accumulation layer in Fig. 3B is re-
flected on a sharper ER profile solely dependent on the
boundary conditions. In this case electrostatic shielding
from accumulation layer is absent and all field lines end
into drain edge.
We image with sub-micron resolution the electric field
across the channel of an oTFT. The Stark signal coming
from probe areas along the channel is influenced both by
the electrostatic boundary conditions as well as charge dis-
tribution inside the transistor channel. Combining micro-
mathematical model of the device, we are able to extract
the electric field profile and charge carrier distribution
(both static and dynamic) within the active organic mate-
rial. In addition, exploiting the electron penetration depth
lower than channel length, we develop an electro-optical
technique for the measurement of AC mobility. We demon-
strate that for phthalocyanine oTFT, AC mobility is insensi-
tive to environmental-induced traps, which usually limit
the continuous operations at room conditions.
J.C.-G. Acknowledges financial support from the Spanish
Ministry of Science and Innovation through Programa Ra-
mon y Cajal (RYC-2009-05475) and POLYDYE Project
(TEC2010-21830-C02-02). The authors thank Dr. Mario
Caironi for useful discussions.
 D. Braga, G. Horowitz, Adv. Mater. 21 (2009) 1–14.
 S.M. Sze, K.K. Ng, Physics of Semiconductor Devices, third ed., Wiley
 N.C. Greenham, S.C. Moratti, D.D.C. Bradley, R.H. Friend, A.B. Holmes,
Nature 365 (6447) (1993) 628–630.
 G. Yu, J. Gao, J.C. Hummelen, F. Wudl, A.J. Heeger, Science 270 (5243)
 H. Sirringhaus, P.J. Brown, R.H. Friend, Nature 401 (6754) (1999)
 S.R. Forrest, Nature 428 (6986) (2004) 911–918.
 H. Bassler, Phys. Status Solidi B 175 (1993) 15.
 V. Podzorov, E. Menard, J.A. Rogers, M.E. Gershenson, Phys. Rev. Lett.
95 (226) (2005) 601.
 J. Takeya, J. Kato, K. Hara, M. Yamagishi, R. Hirahara, K. Yamada, Y.
Nakazawa, S. Ikehata, K. Tsukagoshi, Y. Aoyagi, T. Takenobu, Y. Iwasa,
Phys. Rev. Lett. 98 (2007) 196804.
 P.G. Le Comber, W.E. Spear, Phys. Rev. Lett. 25 (1970) 509.
 M.C.J.M. Vissenberg, M. Matters, Phys. Rev. B 57 (12) (1998) 964.
 H. Vazquez, R. Oszwaldowski, P. Pou, J. Ortega, R. Perez, F. Flores, A.
Kahn, Europhys. Lett. 65 (2004) 802.
 S. Braun, W.R. Salaneck, M. Fahlman, Adv. Mater. 21 (2009) 1450–
 M. Caironi, C. Newman, R. Moore, D. Natali, H. Yan, A. Facchetti, H.
Sirringhaus, Appl. Phys. Lett. 96 (2010) 183303.
 D.S.H. Charrier, M. Kemerink, B.E. Smalbrugge, T. de Vries, R.A.J.
Janssen, ACS Nano 2 (2008) 622.
 C. Sciascia, N. Martino, T. Schuettfort, B. Watts, G. Grancini, M.R.
Antognazza, M. Zavelani-Rossi, C.R. McNeill, M. Caironi, Adv. Mat.
 T. Manaka, E. Lim, R. Tamura, M. Iwamoto, Nat. Photonics 1 (2007)
 Z.Q. Li, G.M. Wang, N. Sai, D. Moses, M.C. Martin, M. Di Ventra, A.J.
Heeger, D.N. Basov, Nano Lett. 2 (2006) 224–228.
 M. Celebrano, C. Sciascia, G. Cerullo, M. Zavelani-Rossi, G. Lanzani, J.
Cabanillas-Gonzalez, Adv. Funct. Mater. 19 (2009) 1–6.
 J.H. Lee, C.C. Liao, P.J. Hu, Y. Chang, Synth. Met. 144 (2004) 279.
 J.D. Slinker, J.A. De Franco, M.J. Jaquith, W.R. Silveira, Y.-W. Zhong,
J.M. Moran-Mirabal, H.G. Craighed, H.D. Abruna, J.A. Matohn, G.G.
Malliaras, Nat. Mater. 6 (2007) 894–899.
 D. Natali, L. Fumagalli, M. Sampietro, J. Appl. Phys. 101 (2007)
 L. Li, Q. Tang, H. Li, W. Hu, J. Phys. Chem. B 112 (2008) 10405–10410.
 Z. Bao, A.J. Lovinger, J. Brown, J. Am. Chem. Soc. 120 (1998) 207.
 J. Cabanillas-Gonzalez, H.J. Egelhaaf, A. Brambilla, P. Sessi, L. Duò, M.
Finazzi, F. Ciccacci, G. Lanzani, Nanotechnology 19 (2008) 424010.
 G. Weiser, Á. Horváth, Electro-absorption spectroscopy of p-
publ/book/Chapter12.pdf> (Chapter 12).
 S.A. Whitelegg, Experimental Investigations into the Physics of Light
Emitting Conjugated Polymers, Ph.D. thesis, Cambridge University
 P.J. Brown, H. Sirringhaus, M. Harrison, M. Shkunov, R.H. Friend,
Phys. Rev. B 63 (2001) 125204.
 K. Yamasaki, O. Okada, K. Inami, K. Oka, M. Kotani, H. Yamada, J.
Phys. Chem. B 101 (1997) 13–19.
 The applied electric field has two contributions: FAppl(r) = FDC(r) +
FAC(r)?sin(2pft),consequentlytheERsignalwillbe:Signal ? FApp2(r) =
FDC2(r) + FAC2(r) + 0.5?FDC(r)?FAC(r)?sin(2pft) ? 0.5 FAC2(r) cos(2?2pft).
 M. Kitamura, T. Imada, S. Kako, Y. Arakawa, Jpn. J. Appl. Phys. 43
 M. Caironi, Y.-Y. Noh, H. Sirringhaus, Semicond. Sci. Technol. 26
 J.D. Jackson, Classical Electrodynamics, John Wiley & Sons, New York,
 N. Majumdar, S. Mukhopadhyay, v1 (2006).
C. Sciascia et al./Organic Electronics 13 (2012) 66–70