Trap States - Università degli Studi di Roma "Tor Vergata"

Download Report

Transcript Trap States - Università degli Studi di Roma "Tor Vergata"

Influence of carrier mobility and interface trap states on the
transfer characteristics of organic thin film transistors.
A. Bolognesi, A. Di Carlo
INFM and Dept. of Electronic Engineering University of Rome “Tor Vergata”
INFM
Via del Politecnico, 1 - 00133 Rome, Italy
Field Dependent mobility: Monte Carlo results
Abstract
In order to calculate the field dependent mobility we apply
the model first introduced by Bassler1:
We have used two dimensional drift diffusion simulations to calculate the electrical properties of
bottom contact pentacene based organic thin film transistor, taking into account field-dependent
mobility and interface or bulk trap states.
Currently there is significative interest in organic TFT due to the possibility of realizing low cost, large
area devices on flexible substrate. The main limitation comes from scarce knowledge of microscopic
charge transport mechanism and the nature of metal/organic and organic/insulator interfaces. In
order to derive from basic principles the transport properties of the organic semiconductor we have
developed a Monte Carlo simulator to calculate the field dependent mobility. The results are
consistent with the well-known behavior which is proportional to the exponential of square root of
electric field, for field below 2x106 V/cm, while a saturated mobility is obtained for higher field. This is
consistent with the Marcus theory of the inversion region. We show the importance of the mobility
model on the output and transfer characteristics of organic thin film transistor. Moreover we show
that the extraction of mobility from experimental results using the analytic theory of the silicon
MOSFET leads to wrong values.
Concerning the metal/organic interfaces we show that the contact barrier height strongly influences
the electrical characteristics of the device.
The last point we analized is the presence of trap states at the interface between organic material
and gate insulator. We show the influence of trap states combined with the effect of field-dependent
mobility on the transfer characteristic with particular attention to subtreshold region where the trap
states influence the inverse subtreshold slope. We compare the transfer characteristics obtained with
and without trap states. Numerical calculations are also made for the output characteristics of the
device.
A very good agreement between available experimental results and simulated curves is obtained.
• The sample is described as a cubic lattice of
170x170x20 hopping sites.
• Energies of the sites are chosen randomly from
a Gaussian density of states (DOS).
• Under the influence of external field the mean
energy is given as
E ( x )  E0 ( x )  qFx
Rate of hopping is described by Miller-Abrahams
expression
 E j  Ei  E j  Ei
i , j  v0 exp 2Rij exp 

2kT

where




Rij
Distance between sites

Localization Radius
E j , Ei
Energies of the sites
1Ref.
   0e
E / E0
H.Bassler, PRB 1999, vol 59, n. 11, 7507
Simulation Model and Experimental Device
Drift Diffusion Equation
We simulate organic devices using an industry standard
device simulation tool, namely ISETCADTM, a package able
to resolve the standard drift-diffusion equations coupled
with Poisson’s equation in two and three dimensions.
ISETCADTM is not capable to handle charge transport in
organic material, so we have implemented a new model
based on the following properties:
•Organic/Inorganic Band Alignment
•Density of States (equal to molecule density)
•Monte Carlo extracted Field dependent Mobility (see panel
above)
J n  q n  qDnn
Mobility
Use mobility value extracted from experimental saturation
current leads to an overestimation of drain currents
expecially at low drain voltage:
Measured output
characteristic
Geometric Characteristics
J p  q p  qD pp
n 1
   J n  (G  R )
t q
p
1
    J p  (G  R )
t
q
All measurements were carried out in
vacuum
atmosphere
at
room
temperature with Agilent Semiconductor
Analyzer 4155C.
Pentacene Band Structure
Vacuum
Level
2.6 eV
Source/Drain
Metal
Square root of
saturation corrent
From square root of IDSS is possible
to calculate pentacene mobility
using the expression:
LOMO
0.1 eV
2.4 eV
HOMO
Effect of contact barrier heigth
Trap States
The subtreshold slope is strongly influenced by trap states
at the interface between oxide and pentacene. We model
this effect using acceptor single level trap states near the
valence band of pentacene.
Transfer characteristics (@
Vds=30 V) obtained by:
The Schottky barrier height (B) of metal/organic interface
at the drain and source contact plays a major role1 on the
electrical characteristics of organic thin film transistor. We
performed simulation with field dependent mobility varying
B between 0.1 eV and 0.4 eV in order to account for
different metal of the contacts.
The figure below shows the mobility extracted from the
simulations using the same method used for the extraction
from the experimental results.
1. Varying the concentration
(0.15 eV above VB)
For this device (which has
good injecting contact so
B=0.1 eV ) the
overestimation for the field
dependent mobility model
is around 20% which is
consistent with the
comparison between
experimental mobility
(4x10-3 cm2/Vs) and low
field mobility (3.3x10-3
cm2/Vs )
2. Varying the level (for
concentration of 1012 cm-2)
By using our proposed
model of field
dependent mobility we
obtain perfect
agreement with:
Measured transfer
characteristic @ Vds=-30 V
  ( 0 r  )  q( p  n  N D  N A )
•Trap states at the interface between organic material and
silicon oxide (see panel below).
We apply this model to the simulation of pentacene organic
thin film transistor. In particular we simulate a device with a
channel length of 12 m and oxide thickness of 250 nm.
The same device was realized and measured from our
group.
Experimental Thin Film Transistor
Perfect agreement is
obtained for single
acceptor trap level 0.15
eV above VB with
concentration of 1012 cm-2
1Ref.
A. Bolognesi, Appl. Phys. Lett., 81, 4646 (2002)