JOURNAL OF APPLIED PHYSICS 103, 124509 共2008兲
Contact effects and extraction of intrinsic parameters
in poly„3-alkylthiophene… thin film field-effect transistors
M. Jamal Deen,1,a兲 Mehdi H. Kazemeini,1 and S. Holdcroft2
1
Electrical and Computer Engineering Department (CRL 226), McMaster University, Hamilton, Ontario
L8S 4K1, Canada
2
Department of Chemistry, Simon Fraser University, Vancouver, British Columbia V5A 1S6, Canada
共Received 20 December 2007; accepted 15 April 2008; published online 20 June 2008兲
We report on contact effects in polymeric thin film transistors based on poly共3-octylthiophene兲 and
poly共3-hexadecylthiophene兲 with gold contact electrodes and in the bottom contact configuration. A
method to extract the intrinsic channel mobility from the measured extrinsic mobility over a broad
range of gate voltage is presented. This method uses the I-V characteristics of the transistor in its
reverse mode operation. The results show that the intrinsic mobility in the channel is gate voltage
dependent and increases almost linearly with voltages at biases above the threshold voltage. By
applying a model based on the theory of space-charge-limited conduction, the dependence of the
threshold voltage on the contacts and the shifts observed in this parameter with different polymer
film thicknesses are explained. We also apply this model to explain the effects of light in reducing
the contact effects and changing the device parameters from extrinsic in the dark to intrinsic under
illumination. © 2008 American Institute of Physics. 关DOI: 10.1063/1.2942400兴
I. INTRODUCTION
Polymeric and organic thin film transistors 共PTFTs/
OTFTs—hereafter we use OTFTs to mean both organic and
polymeric transistors兲 have attracted considerable interest in
recent years due to the significant improvement in their electrical performance and their high potential for applications
such as low-cost, large-area flexible electronic systems.
However, more efforts are still needed to better understand
the influence of the source and drain contact metals on the
transistor’s performance since they present large gate voltage
dependent contact resistances that result in electrical characteristics of the OTFTs deviating from the standard silicon
metal oxide semiconductor field effect transistor 共MOSFET兲
model predictions.1–11 This is in addition to the gate-voltage
influence on the intrinsic mobilities in OTFTs, which is different from that in silicon MOSFETs, as reported in many
publications, for example, Refs. 12–15.
Several studies 共for example, Refs. 1, 2, and 13兲 have
shown that OTFTs with the bottom contact 共BC兲 design demonstrate inferior performance, compared with those with the
top contact 共TC兲 design, because of higher parasitic series
resistances and lower extrinsic mobilities. Also, it has been
observed that the extrinsic mobility decreases with the channel length in both TC TFTs and BC TFTs due to the contact
resistance effect. This can create a serious problem when
using short channel length 共⬍10 ␮m兲 transistors needed as
drivers for large-area display applications, for example, since
the contact resistance may become comparable to or even
exceed the channel resistance.3,15–17 However, BC TFTs are
preferred for fabrication of organic electronic circuits since
they are easier to fabricate on a flexible substrate.1
The detection of voltage drops at the source/drain contacts by the method of scanning potentiometry has demona兲
Electronic mail: jamal@mcmaster.ca.
0021-8979/2008/103共12兲/124509/7/$23.00
strated that the transport properties of the transistor’s channel
can be separated from those of the electrode contacts in a
simple model.2,5,6 In TFTs with non-Ohmic contacts, such as
those based on poly-9 , 9⬘ dioctyl-fluorene-co-bithiophene
共F8T2兲 with Au contacts, a large contact resistance may appear at the source contact due to a Schottky barrier for hole
injection, while for Ohmic contacts such as those in Au/
P3HT TFTs, the source and the drain contact resistances are
almost identical and are governed by bulk transport through
a low-mobility region close to the contact metal.2 Therefore,
a general approach to modeling of OTFTs is to divide the
total source/drain voltage into an intrinsic channel component and another component for the contacts.5,6
In using the standard MOSFET equations for OTFTs and
obtaining a better fit for the measured data, some methods
consider that the mobility and/or contact resistance are gate
voltage dependent.5,13,14 For Au-sexithiophene TFTs, a
method assuming a constant contact resistance and a gatevoltage dependent mobility was developed to extract the intrinsic mobility from the transfer characteristics and to estimate values for the threshold voltage and contact
resistance.14,16 For Au-pentacene TFTs, a model incorporating a gate voltage dependent mobility and highly nonlinear
drain and source contact resistances was proposed to simulate the transport characteristics.13 In Ref. 5, polymer TFTs
based on F8T2, which makes a non-Ohmic contact with Au,
were studied and assumed that the mobility is constant and
the nonlinear current-voltage relation at the source depends
on the gate voltage. The contact resistance was extracted
from the output characteristics measured with transistors of
different channel lengths, and the contact current was found
to be exponentially related to the contact voltage.5
With the channel length scaling technique, it is possible
first to obtain the contact resistance via length scaling measurements and then to remove it from mobility measure-
103, 124509-1
© 2008 American Institute of Physics
Downloaded 27 Apr 2009 to 131.175.136.39. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
124509-2
J. Appl. Phys. 103, 124509 共2008兲
Deen, Kazemeini, and Holdcroft
ments. This technique has been used for different TFTs based
on different organic materials.6,7,15,16 In BC-TFTs based on
poly共3-hexylthiophene兲 共P3HT兲, when Au contacts are used,
it has been shown that both the mobility and the contact
resistance are gate voltage dependent.6,15,17 Two difficulties
with this technique of measurement are the need for a series
of devices with different channel lengths and the assumption
that the contact and channel transport properties in each device of different lengths are identical.3
Contact resistance effects on the threshold voltage in
short channel OTFTs have been investigated experimentally
and with analytical models. It has been found that the higher
the contact resistance, the larger is the threshold voltage, and
therefore the extracted mobility becomes lower.17–19 In Ref.
17, a model for Au-pentacene TFTs has been proposed showing that the extrinsic threshold voltage has two
components—an intrinsic component and a contribution
from the contacts that depends of the film thickness.
In this paper, we present a detailed study of the electrical
properties of OTFTs in the forward and reverse modes of
operation, with an emphasis on the influence of the contact
resistance on the field-effect mobility. We introduce a
method to extract the intrinsic mobility and contact resistance, using reverse mode I-V characteristics. The method
was studied for OTFTs based on poly共3-octylthiophene兲
共P3OT兲 and poly共3-hexadecylthiophene兲 共P3HDT兲 with gold
contacts and in a BC configuration. We also propose a model
based on the theory of space-charge-limited current 共SCLC兲
for the disordered region in the polymer near the contact to
explain the contact effect in shifting the threshold voltage.
The proposed model is also used to explain the effects of
light in reducing the contact effects and changing the device
parameters from extrinsic in the dark to intrinsic under illumination.
II. EXPERIMENTAL DETAILS
PTFTs with a channel of width W = 1.5 cm 共multifinger
configuration兲 and length L = 10 ␮m were fabricated. Two
identical gold electrodes 共20 nm thick兲 were deposited onto a
thermally grown silicon dioxide film 共⬃200 nm兲 on a degenerately doped n-type Si substrate to form the source and
drain electrodes. Then, two different polymers of P3HDT
and P3OT with a regioregular 95% head-to-tail ratio were
deposited by spin coating at various rpm from 2 mg/ ml solutions. After the formation of the polymer film, the devices
were annealed in a nitrogen atmosphere at 118 ° C in the
absence of light and under N2 gas for 3 min. Then, the transistors were cooled very slowly 共overnight兲 in a pure nitrogen atmosphere to room temperature and were stored in
vacuum. The thickness of the polymer films was determined
with an atomic force microscope 共Thermo Microscope, explorer兲 in noncontact mode.
The I-V characteristics, in the normal mode and reverse
mode of biasing, were measured in the dark with a HewlettPackard 4145B-semiconductor parameter analyzer. Also, the
photoinduced current in the PTFTs under broadband whitelight illumination from a halogen lamp through a microscope
were measured and the I-V characteristics were obtained.
III. CHARGE TRANSPORT MODEL
A. Normal mode of transistor operation
Since polythiophene-based TFTs are p-channel transistors, they operate in the normal mode when the drain and
gate voltages are both negative with respect to the source
共VDS and VGS兲. To consider the effects of contact resistances,
we use a simplified low-frequency model, where the conducting path between the source and drain is divided into a
series of three resistive elements: the source contact resistance RS, the drain contact resistance RD, and the gatemodulated channel resistance Rch. The resistance due to the
injection process at the source is considered to be negligible
since the Schottky barrier is low when Au electrodes are
used.2 Therefore, RS and RD arise only from bulk transport
through defect-rich regions near the contacts where the density of the trapped carriers is high and no-accumulation layer
forms. The length of this low-mobility region is around
100 nm and its resistance is controlled by the gate bias
voltage.2 Also, the channel resistance, which is related to the
accumulation layer of charges at the interface with the insulator, is controlled by the gate bias voltage but the contact
resistance RC 共where RC = RS + RD兲 is independent of the
channel resistance.
With the above model, there is a voltage drop of 共VDS
− VC兲 across the main channel, where VC is the voltage
dropped across a contact resistance RC. By using the gradualchannel approximation and after integration of the potential
along the channel, the following equation is obtained for the
current-voltage behavior:5,6
ID,nor = −
W␮Ci
关共VGS − VAT兲共VDS − VC兲
共L − d兲
2
− VC2 兲兴,
− 0.5共VDS
共1兲
where L is the total channel length, d is the length of the
depleted region at the contact, W is the channel width, Ci is
the gate capacitance per unit area, ␮ is the intrinsic mobility,
and VAT is the apparent threshold voltage that is expected to
contain a contribution from the contact. Since the channel
length is long, L = 10 ␮m in our devices, then we can justifiably neglect d 共⬃100 nm兲 in Eq. 共1兲. To simplify this equation, we will replace VC by ␣VDS, where ␣ 关⬇RC / 共Rch + RC兲兴
is defined for devices with Ohmic contacts. Note that VAT
and VC in Eq. 共1兲 are negative values because they correspond to the signs of VGS and VDS, respectively.
B. Reverse mode of transistor operation
The polymer-based TFTs operate in the reverse mode if
the drain voltage VDS is positive and the gate voltage VGS is
negative. This means that the drain in the normal mode
共VDS ⬍ 0兲 becomes the source in the reverse mode 共VDS
⬎ 0兲 and the effective gate voltage changes to VGD = 共VGS
+ VSD兲, where VGS and VSD are negative. Therefore, the gate
voltage is not fixed as VDS changes. The current-voltage
characteristics in the reverse mode can be described by the
following equation:
Downloaded 27 Apr 2009 to 131.175.136.39. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
124509-3
J. Appl. Phys. 103, 124509 共2008兲
Deen, Kazemeini, and Holdcroft
FIG. 1. 共Color online兲 Proposed model for the reverse mode of TFT operation 共VDS ⬎ 0兲 at VGS = 0 V. The hole transport is in the reverse direction
共from the drain to the source兲 and passes through an accumulation layer and
a low-mobility region in front of the source contact. The low-mobility region near the drain is not shown since its traps are filled by the induced
charges from the gate-drain bias voltage. Since the gate-source bias voltage
is zero, the trap filling at the source only happens by the electric field from
VDS.
ID,rev = −
WCi
␮共1 − ␣兲共VGS − VAT兲VDS
L
+ 0.5
WCi
2
␮共1 − ␣兲2VDS
.
L
共2兲
This equation is obtained and simplified from Eq. 共1兲 after
replacing VGS with the effective gate voltage 共VGS + VSD兲 and
VC with ␣VSD, where VSD = −VDS is negative. Using Eq. 共2兲,
two different apparent mobilities at low and high VDS biases
can be defined as ␮共1 − ␣兲 and ␮共1 − ␣兲2, respectively. Since
␮ is independent of ␣, the parameters 共1 − ␣兲 and 共1 − ␣兲2 are
the device parameters that make the field-effect mobility different from the intrinsic mobility ␮. Additionally, it has been
shown that the intrinsic mobility is not constant in organic
semiconductors since the dependence of conductivity on carrier concentration is nonlinear.20 This is because in organic
semiconductors, there is a high density of trapped carriers
and it is the hopping of carriers between the localized states
that constitutes the means for conduction. The mobility has
been described by effective carrier mobility at a reference
carrier concentration using the generalized concept of mobility edge and exponential band tails in disordered
semiconductors.20 Therefore, our model agrees with this report and several reported results that the field-effect mobility
in OTFTs changes strongly with the bias conditions.21–23 In
this paper we present a method to extract the two apparent
field-effect mobilities ␮共1 − ␣兲 and ␮共1 − ␣兲2 and separate ␮
from the device parameter 共1 − ␣兲.
C. Turn-on voltage at VGS = 0 V
In the reverse mode of operation 共VDS ⬎ 0 V兲, when
VGS = 0 V, the ID − VDS characteristics can be explained using
our proposed model in Fig. 1. As shown in this figure, the
drain electrode is now acting as a source. This is because the
gate, with respect to the drain, is biased by VSD = −VDS, while
the gate to source is zero. Therefore, all traps in the lowmobility region close to the drain can be filled by the induced
charges due to the gate bias, while the low-mobility region
near the source remains the same unless the filling of the
traps is achieved by a high electric field from VDS. Then, the
trap-filled voltage VTF determines a current at which the conduction becomes Ohmic, but the SCLC at VDS ⬍ VTF is proportional to 共VDS兲n, where n is much greater than 1 at low
VDS and decreases to 2 at VDS = VTF. This is mainly due to
trap charging of the low-mobility region that increases the
conduction strongly with VDS. Therefore, for the SCLC, a
turn-on voltage Von共0兲 Ⰶ VTF can be defined and extracted by
fitting the measured current with ID ⬀ 共VDS − Von共0兲兲2 or an apparent threshold voltage can be defined and extracted by fitting the data with a linear I-V characteristic, i.e., ID ⬀ 共VDS
− 兩VAT兩兲. As a result, it can be shown that 兩VAT兩 = Von共0兲 + 兩VT兩,
where 兩VT兩 is the intrinsic threshold voltage.24
It should be emphasized that at VDS = VTF, the organic
material is expected to be destroyed 共breakdown兲 by a high
applied electric field if the breakdown occurs inside the
whole space between the source and drain.25 Moreover, for
this case, VTF would be very large. However, in our model
VTF is not large since the high electric field appears across a
narrow region with the length of d, as shown in Fig. 1, and
also the current is limited by the channel resistance.
At VDS ⬎ Von共0兲, holes are mainly transported through the
accumulation layer and the contact resistance in order to
reach from the drain electrode to the source. To find the
trap-filled voltage, we need to apply a model of SCLC.26–28
We use the model of gap type introduced in Ref. 28 since it
is developed for thin semiconductor layers between two
knifelike contacts. As Fig. 1 shows, the defect-rich region
with length d has a knifelike contact with the channel 共hole
accumulation layer兲 in one side and a plate metal contact 共for
which the area is determined by the thickness of polymer兲 on
the other side. Changing the thickness affects mainly the
total traps in this region and has no effect on the capacitance
across d since the thickness of the hole accumulation layer is
independent of the polymer film thickness. According to the
model in Ref. 28, the current-voltage characteristics is given
by
ID,sub =
V2
2W
␮C␧ DS
,
d2
␲
共3兲
where ␮C Ⰶ ␮ is the low mobility of the contact region and ␧
is the permittivity. At low VDS, mobility ␮C increases due to
VDS since at VGS = 0 V, trap filling in the low-mobility region
near the source is achieved only by a high electric field from
VDS, according to the model in Fig. 1. When this region
contains empty traps of surface density NS at zero electric
field 共VDS = 0 V兲, the theory yields a trap-filled voltage VTF
given by
VTF =
␲qNSd
,
4␧
共4兲
beyond which the current become Ohmic.28 Equation 共3兲 can
be used to determine the current at VDS 艋 VTF and it gives the
bulk mobility at VDS = VTF, using the data obtained from measurements.
Downloaded 27 Apr 2009 to 131.175.136.39. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
124509-4
Deen, Kazemeini, and Holdcroft
J. Appl. Phys. 103, 124509 共2008兲
IV. RESULTS AND DISCUSSION
FIG. 2. 共Color online兲 Transfer normal mode characteristics in P3HDTTFTs with t ⬇ 7 nm and L = 10 ␮m measured for various VDS. The inset
figure shows the corresponding normal mode transport characteristics. The
reverse mode transport curve at VGS = 0 does not overlap the normal mode
transfer curves at saturation, as expected from TFTs with no contact resistance and constant mobility.
The value of NS can be extracted from Eq. 共4兲, when VTF
is measured. NS depends on the film thickness t with NS
= NBt, where NB is the density of empty traps in the defectrich region. Therefore, it is expected 兩VAT兩 = Von共0兲 + 兩VT兩 to
increase with the polymer film thickness.
Figure 2 shows the transfer characteristics of P3HDTTFTs at various VDS measured in the normal mode of operation. The inset figure shows the corresponding output characteristics at various VGS. To extract the apparent threshold
voltage VAT and the field-effect mobility ␮FET, it is common
to fit the data measured in the saturation by the standard FET
equation that explains that the saturation current is ⬀␮共1
− ␣兲共VGS − VAT兲2. For best fitting, we find approximately
VAT ⬇ −6 V and ␮FET = 5.5⫻ 10−5 cm2 / V s. However, these
results are not accurate since the apparent mobility ␮共1 − ␣兲
is gate voltage dependent. In Fig. 2 we have also plotted the
reverse mode I-V curve at VGS = 0 V. The current data can be
fitted by ID ⬀ ␮共1 − ␣兲2共VDS − Von共0兲兲2. Since the device is
symmetric, these two curves is expected to overlap if ␣ = 0.
Therefore, there is no complete overlap due to ␮共1 − ␣兲2
⬍ ␮共1 − ␣兲. This is due to the differences in trap charging by
VGS 共in the saturation兲 and that by VDS 共in the reverse mode
biasing with VGS = 0 V兲.
Figure 3共a兲 shows the output I-V characteristics for
P3HDT-TFTs in the reverse mode of operation for various
negative VGS. In this mode, the current continuously increases with the drain voltage, and no saturation regime is
observed. As shown, for small negative gate voltages below
−6 V, the characteristic is nonlinear and have different
turn-on voltages depending on the value of VGS. However, at
large negative gate voltages 共above −6 V兲, a transition from
linear to nonlinear behavior occurs, as expected from FET
currents in the reverse mode of operation.
The curves in Fig. 3共a兲 are redrawn in Fig. 3共b兲 with
respect to 共VGS
⬘ − 6兲, where VGS
⬘ = 兩VGS兩 + VDS is the effective
FIG. 3. 共Color online兲 Reverse mode
transport characteristics of Au-P3HDT
TFTs. 共a兲 The I-V characteristics are
linear at low VDS and nonlinear at
large VDS. 共b兲 Redrawn curves vs VDS
+ 兩VGS兩 − 6 results in overlapping
curves at 兩VGS兩 艋 6 V since 兩VAT兩
⬇ 6 V and crossing curves at large
VGS
⬘ = VDS + 兩VGS兩 because of gatevoltage dependent mobility. 共c兲 Redrawn I-V curves in log scale to show
the details.
Downloaded 27 Apr 2009 to 131.175.136.39. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
124509-5
Deen, Kazemeini, and Holdcroft
gate voltage and VAT = −6 is chosen. As a result, the I-V
curves from VGS = 0 to − 6 V are aligned on top of each
other. This is described as a pure shift of the curves with the
turn-on voltages. However from VGS = −8 V to VGS = −18 V,
the I-V curves at low VDS become separated since the slope
of the linear term increases with 兩VGS兩, but at larger values of
VDS and 兩VGS兩, the curves are closer and cross each other
since the mobility increases with VGS
⬘ .
In Fig. 3共c兲, the curves of Fig. 3共a兲 are redrawn in logarithmic scale. By using this figure and Eqs. 共3兲 and 共4兲, we
explain the electrical behavior of the low-mobility region at
low bias voltages based on the theory of SCLC. As the figure
shows, at VGS = 0 V, the current is proportional to 共VDS兲n at
low VDS biases, where n ⬇ 6. The slope decreases to n ⬇ 2 at
VDS ⬇ 12 V and then increases to a slope higher than 2. This
curve, up to VDS ⬇ 12 V, demonstrates the I-V characteristics
of the low-mobility region at the source contact since the
depletion at the drain disappears at low VDS according to the
model in Fig. 1. Therefore, a trap-filled voltage of VTF
⬇ 12 V is found from the I-V curve at VGS = 0 and the slope
of 2 at this point is due to the nonlinear conduction of the
device in the reverse mode of operation at large VDS. Figure
3共c兲 shows that the bulk current at this point is ⬃32 nA. By
using Eq. 共4兲, a value of 1.3⫻ 1012 cm−2 is calculated for NS
when VTF = 12 V. This is equivalent to a bulk trap density of
NB = 1.3⫻ 1018 cm−3 at zero field 共i.e., VDS = 0 V兲. Also, by
measuring the current at VDS = 12 V and VGS = 0 and using
Eq. 共3兲, the mobility near the contact ␮C ⬇ 1
⫻ 10−5 cm2 / V s is obtained. This mobility is expected to
drop to much lower values at VDS Ⰶ VTF since ␮0 = ␪␮C,
where ␪ is the fraction of the total charges free to move and
␮0 is the mobility at zero field 共i.e., VDS = 0 V兲.26
Figure 3共c兲 shows that by increasing the negative bias of
VGS, the conductivity in the low-mobility region of the
model, shown in Fig. 1, increases and therefore VTF drops.
For example, VTF drops to 6 V when VGS changes from
0 to − 6 V. The slope at this point is ⬃1, showing that the
linear conduction is reached. Note that at larger VDS, the
conduction becomes nonlinear again as expected by the nonlinear conduction of the device in the reverse mode of operation and thus the slope increases to 2. In Fig. 3共c兲, the
curve at VGS = −18 V demonstrates the characteristics of the
MOSFET in the reverse mode biasing, i.e., the slope is 1 at
low VDS and 2 at large VDS.
To extract ␮ vs VGS
⬘ = 兩VGS兩 + VDS, we use the data in Fig.
3共a兲 and obtain the solution of Eq. 共2兲 for ␮共1 − ␣兲 and ␮共1
− ␣兲2. At a small and fixed VDS 共e.g., 2 V兲, the second term
of Eq. 共2兲 is eliminated and ␮共1 − ␣兲 vs VGS
⬘ = 共VGS − 2兲 is
extracted from the linear term. The results show that the
apparent mobility, ␮共1 − ␣兲, continuously increases with VGS
⬘ .
The second term in Eq. 共2兲 is used to extract ␮共1 − ␣兲2 vs
VGS
⬘ = VDS by choosing VGS = 0 V. The variations in ␮共1 − ␣兲
and ␮共1 − ␣兲2 with VGS
⬘ are shown in Fig. 4. Two curves for
ID vs VGS
⬘ are also drawn in this figure for the two different
bias combinations: VDS = 2 V with variation of VGS
⬘ = 兩VGS兩
+ 2 and VGS = 0 V with variation in VGS
⬘ = VDS. It is noted that
the two parameters ␮共1 − ␣兲 and ␮共1 − ␣兲2 are extracted from
J. Appl. Phys. 103, 124509 共2008兲
FIG. 4. 共Color online兲 Two different apparent mobilities ␮共1 − ␣兲 and ␮共1
− ␣兲2 extracted from two different I-V characteristics at VGS = 0 V and VDS
= 2 V, with respect to effective gate voltage VGS
⬘ .
the different part of the graph in Fig. 3. However, both are
related to the same range of the effective gate voltage and
thus to same ␮.
Figure 5 shows ␮ and 共1 − ␣兲 as functions of VGS
⬘ . It is
concluded that the intrinsic field-effect mobility is gate voltage dependent and behaves almost linear at above the threshold voltage. This is similar to the results reported in Ref. 14
for OTFTs based on oligothiophene and the result in Ref. 12
for pentacene TFTs where long channel devices have been
used to eliminate the effect of contact resistance. Also, it
agrees with the result in Ref. 20 for an effective mobility in
disordered semiconductors described by generalized transport band definition and its extension to hopping transport.
It should be noted that in our method, the intrinsic mobility ␮ is assumed to be voltage independent. In the case of
a voltage dependent intrinsic mobility, this method cannot
separate ␮共VDS兲 from ␮共VGS兲 since the effective gate voltage
consists of both VDS and VGS. Also it should be noted that our
FIG. 5. 共Color online兲 Intrinsic field-effect mobility ␮ and contact resistance
attribute 共1 − ␣兲 are deduced from the data in Fig. 4. ␮ increases continuously with the gate voltage at the above threshold and becomes almost linear
at large gate voltage.
Downloaded 27 Apr 2009 to 131.175.136.39. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
124509-6
J. Appl. Phys. 103, 124509 共2008兲
Deen, Kazemeini, and Holdcroft
FIG. 6. 共Color online兲 Effect of film
thickness on diodelike characteristics
of P3OT-TFTs at VGS = 0 V. 共a兲 Von共0兲
and therefore VAT shift almost linearly
with the polymer film thickness. 共b兲 At
VGS = −10 V, the total current 共channel
current+ body
current兲
becomes
Ohmic and the three different slopes
for the three film thickness are due to
differences in VTF 共or Von共0兲兲 and ␣
共=RC / Ron兲.
method is valid for devices with Ohmic contacts due to the
assumption of a linear relationship between contact resistance and channel resistance. Simplified Eq. 共2兲 共used in our
method兲 is obtained by assuming that RC is Ohmic and the
Schottky barrier at the contact is negligible. In the case of
presence of Schottky barriers at the contacts, the accuracy of
our method would deteriorate.
Figure 6共a兲 shows the diodelike characteristics of three
devices based on P3OT-TFTs in the reverse mode of operation at VGS = 0 V, for three different polymer film thicknesses: t1 = 7 nm, t2 = 13 nm, and t3 = 38 nm. We conclude
from Eq. 共2兲 that ID ⬇ 共WCi / L兲␮共1 − ␣兲共VDS − 兩VAT兩兲VDS explains the electrical behavior at low VDS, where 兩VAT兩
= Von共0兲 + 兩VT兩 and VT = 1.5 V. We extract Von共0兲 from the
square root of 共冑ID − VDS兲 and then calculate 兩VAT兩. The results are 2.5 V for t1 = 7 nm, 3.5 V for t2 = 13 nm, and 7.5 V
for t3 = 38 nm in P3OT-TFTs, and it is concluded that 兩VAT兩
increases with the film thickness due to the change in trapfilled voltage. This is similar to the result obtained for the
threshold voltage in the normal mode in pentacene TFTs.17 It
is also concluded that the shift of 兩VAT兩 共or the shift of VTF兲
with the polymer film thickness is accompanied by some
changes in the surface trap density NS given by Eq. 共4兲. Table
I shows the results for the three different polymer film thicknesses. NS is calculated based on a bulk trap density of NB
⬇ 1018 cm−3.8
At VGS = −10 V, the linear relation of ID共VDS兲 at low VDS
in Fig. 6共b兲 shows that RC is Ohmic. Based on Eq. 共2兲, ID
⬇ 共WCi / L兲␮共1 − ␣兲共10− 兩VAT兩兲VDS and the differences in
slopes, for the three different thicknesses, are mainly due to
the differences in their threshold voltages.
Figure 7 shows the light effect on the I-V characteristics
of P3HDT-TFTs in the reverse mode of operation when the
device is under illumination of white light with an irradiance
of 1.5 mW/ cm2. The I-V characteristics in the dark are also
given for comparison. We conclude that the light mainly affects the contact region conduction, similar to the effect of
VGS in reducing the contact resistances, while the light has no
effect on the intrinsic mobility in the channel. As discussed
in Ref. 9, the defect-rich regions close to the Au contacts are
the photoactive sites and it is the photogeneration of charges
and the trap charging in these regions that decrease the contact resistance and its effect on mobility due to 共1 − ␣兲. It
reduces the trap-filled voltage, which in turn leads to the
reduction of the apparent threshold voltage to the intrinsic
value. Therefore, the large increase in current under an irradiance of 1.5 mW/ cm2 共Fig. 7兲 can be explained by using
Eq. 共2兲, in which 兩VAT兩 drops from 7.5 V in the dark to
⬃1.5 V under the illumination and ␣ is reduced significantly.
In Fig. 7, the two I-V curves at VGS = 0 V 共one in the dark
and one under illumination兲 can be approximated as two quadratic equations ID ⬇ 共WCi / L兲␮共1 − ␣兲共VDS − 7.5兲VDS and ID
⬇ 共WCi / L兲␮共VDS − 1.5兲VDS, respectively. At VGS = −10 V,
these two curves become linear. In the dark, we have ID
⬇ 共WCi / L兲␮共1 − ␣兲共10− 7.5兲VDS and under illumination we
have ID ⬇ 共WCi / L兲␮共10− 1.5兲VDS. Therefore, it is concluded
that light changes the device parameters from apparent in the
dark to intrinsic under illumination.
TABLE I. Effect of polymer film thickness on VAT or Von共0兲 in P3OT-TFTs.
NS is calculated based on bulk trap density NB = 1018 cm−3.
t
共nm兲
NS
共cm−2兲
兩VAT兩
共V兲
7
13
38
0.7⫻ 1012
1.3⫻ 1012
3.8⫻ 1012
2.5
3.5
7.5
FIG. 7. 共Color online兲 P3HDT-TFTs under an irradiance of 1.5 mW/ cm2,
demonstrating large increases in current due to reduction in RC and VAT.
Light has no effect on ␮ but affects the contact low-mobility region. It
increases the contact resistance attribute 共1 − ␣兲 to 1 and reduces the trapfilled voltage VTF to zero 共equivalent to reducing VAT to VT兲.
Downloaded 27 Apr 2009 to 131.175.136.39. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp
124509-7
V. CONCLUSIONS
Detailed experimental results and analysis are reported
for OTFTs, with an emphasis on the influence of contact
resistance on the extracted mobility from measurement data.
We use the I-V measurement of devices based on P3OT and
P3HDT in the BC configuration and introduce a method of
extracting the intrinsic mobility and contact resistance using
the MOSFET model. For this method, we need to analyze the
data from each transistor over a wide range of bias voltages
in the reverse mode of operation. The results show that the
field-effect mobility is strongly dependent on the applied
gate bias. We also propose a model based on the theory of
SCLC to investigate the threshold voltage and bulk current
characteristic. It is concluded that the apparent threshold
voltage has a contribution from the low-mobility region near
the contacts. This model is used to explain the shifts observed in the threshold voltage of the devices with different
polymer film thicknesses. The effect of light in reducing the
contact resistance and changing the device parameters from
apparent in the dark to intrinsic under illumination are also
explained.
ACKNOWLEDGMENTS
This research was supported in part by grants from the
Natural Sciences and Engineering Research Council
共NSERC兲 of Canada and the Canada Research Chair 共CRC兲
program.
1
J. Appl. Phys. 103, 124509 共2008兲
Deen, Kazemeini, and Holdcroft
I. Kymissis, C. Dimitrakopoulos, and S. Purushothaman, IEEE Trans.
Electron Devices 48, 1060 共2001兲.
2
L. Burgi, T. J. Richards, R. H. Friend, and H. Sirringhaus, J. Appl. Phys.
94, 6129 共2003兲.
3
D. Gundlach, L. Zhou, J. Nichols, P. V. Necliudov, and M. Shur, J. Appl.
Phys. 100, 024509 共2003兲.
4
P. V. Necliudov, M. S. Shur, D. J. Gundlach, and T. N. Jackson, SolidState Electron. 47, 295 共2003兲.
5
R. A. Street and A. Salleo, Appl. Phys. Lett. 81, 2887 共2002兲.
6
B. H. Hamadani and D. Natelson, Proc. IEEE 93, 1312 共2005兲.
7
H. Klauk, G. Schmid, W. Radlik, W. Weber, L. Zhou, C. D. Sheraw, J. A.
Nichols, and T. N. Jackson, Solid-State Electron. 47, 297 共2003兲.
8
M. J. Deen, M. H. Kazemeini, Y. M. Haddara, J. Yu, G. Vamvounis, S.
Holdcroft, and W. Woods, IEEE Trans. Electron Devices 51, 1892 共2004兲.
9
M. J. Deen and M. H. Kazemeini, Proc. IEEE 93, 1312 共2005兲.
10
M. S. A. Abdou, F. P. Orfino, Z. W. Xie, M. J. Deen, and S. Holdcroft,
Adv. Mater. 共Weinheim, Ger.兲 6, 838 共1994兲.
11
M. S. A. Abdou, X. Lu, Z. W. Xie, M. J. Deen, and S. Holdcroft, Chem.
Mater. 7, 631 共1995兲.
12
K. Ryu, I. Kymissis, V. Bulovic, and C. G. Sodini, IEEE Electron Device
Lett. 26, 716 共2005兲.
13
P. V. Necliudov, M. S. Shur, D. J. Gundlach, and T. N. Jackson, J. Appl.
Phys. 88, 6594 共2000兲.
14
G. Horowitz, M. E. Hajlaoui, and R. Hajlaoui, J. Appl. Phys. 87, 4456
共2000兲.
15
B. H. Hamadani and D. Natelson, J. Appl. Phys. 95, 1227 共2004兲.
16
E. J. Meijer, G. H. Gelinck, E. van Veenendaal, B. H. Huisman, D. M. de
Leeuw, and T. M. Klapwijk, Appl. Phys. Lett. 82, 4576 共2003兲.
17
R. Schroeder, L. A. Majewski, and M. Grell, Appl. Phys. Lett. 83, 3201
共2003兲.
18
R. A. Street, J. E. Northrup, and A. Salleo, Phys. Rev. B 71, 165202
共2005兲.
19
L. Torsi, A. Dodabalapur, and H. E. Katz, J. Appl. Phys. 78, 1088 共1995兲.
20
P. Servati, A. Nathan, and G. A. J. Amaratunga, Phys. Rev. B 74, 245210
共2006兲.
21
M. C. J. M. Vissenberg and M. Matters, Phys. Rev. B 57, 12964 共1998兲.
22
R. A. Street, J. E. Northrup, and A. Salleo, Phys. Rev. B 71, 165202
共2005兲.
23
C. D. Dimitrakopoulos, S. Purushothaman, J. Kymissis, A. Callegari, and
J. M. Shaw, Science 283, 822 共1999兲.
24
M. J. Deen and Z. X. Yan, Solid-State Electron. 33, 503 共1990兲.
25
G. Horowitz and P. Delannoy, J. Appl. Phys. 70, 469 共1991兲.
26
D. Chirvase, Z. Chiguvare, M. Knipper, J. Parisi, V. Daykonov, and J. C.
Hummelen, J. Appl. Phys. 93, 3376 共2003兲.
27
O. Marinov, M. J. Deen, and B. Iniguez, IEE Proc.: Circuits Devices Syst.
153, 189 共2005兲.
28
J. A. Geurst, Phys. Status Solidi 15, 107 共1966兲.
Downloaded 27 Apr 2009 to 131.175.136.39. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp