Caracterisation of interface traps in submicronic
MOSFET’s using Spectroscopic charge pumping technique
N. Guenifi and F. Djahli
Department of Electronics, Faculty of Engineering,
University Ferhat Abbas, Sétif, 19000, Algeria.
E

mail:
guenifi_2000@yaho
o.fr
Abstract
We have simulated the experimental spectroscopic charge pumping technique by the
implementation of a model in the electrical simulator SPICE 3F4. The model takes into
account the temperature effect on the geometrical and electrical paramete
rs of the studied
transistor.
Our result, can provide many interesting results concerning the parameters of the
considered MOSFET.
Key Words
Spectroscopic
Charge Pumping, Modeling, SPICE 3F4, MOS transistor, Interface traps.
1.
Introduction
Since the demonst
ration of the existence of surface states at the silicon/silicon dioxide
interface, several techniques have been proposed for the determination of the density of these
states and their energy distribution in the forbidden energy gap of silicon. Most of the
se
techniques are based on measurements on MOS capacitors and have been studied in great
detail. However, none of these techniques, can be accepted for MOS transistors submicronic .
The first method, developed for MOS transistors, extracts informations on
the surface
states from the transistor behavior in a weak inversion region. Then, it is only applicable for
the long

channel devices. The second one is the deep

level transient

spectroscopy technique
(DLTS). It yields informations on the surface state dens
ity and captures cross sections, from
the measurement of the capacitance transients resulting to the electrons and holes emission
from the interface states to the conduction and the valence band. Another used technique is
based on the relation between nois
e (1/f) and surface state density.
The charge pumping (CP) technique has been successfully applied, in the past years,
to characterize the interface traps in the MOS transistors. It is a well

established technique for
the characterization of interface trap
s in MOS transistors. The CP effect in MOS transistor
was reported for the first time by Brugler and Jespers in 1969 [1,4], when applying
rectangular pulses to the gate of a MOSFET, while source and drain are kept at a small
reverse voltage.
By adjusting
the signal parameters, we can calculate the average density of the
interface state <D
it
> and their average cross

section <
> and we also again access to the
energy distribution of the state D
it
(E) and to their spatial distribution D
it
(x).
Using the spectroscopic technique similar to the DLTS it is possible to define an
energy window by using a differential signal. D
it
(E) an
d
(E) are then accessible by
performing a temperature sweep. In [5], this technique has been coupled with static
characteristics of the transistor.
The MOSFET’s models of SPICE3F4 do not consider the charge pumping
phenomena. The implementation of the ch
arge pumping technique, which is purely
experimental, in the electrical simulators becomes enough required to study the components
reliability.
In this work, we have implemented the spectroscopic charge pumping technique in
SPICE 3F4. We have used the M
OSFET level 3 of SPICE, to which we added a generator of
current between drain and substrate for the measurement of the charge pumping current [6].
This current, measured versus the different temperature.
The model takes into account the
temperature effect
on the geometrical and electrical parameters of the studied transistor.
Our
result, can provide many interesting results concerning the parameters of the considered
MOSFET.
2.
Spectroscopic charge pumping
The
charge pumping
effect in MOS transistor was rep
orted for the first time by
Brugler and Jespers, when applying rectangular pulses to the gate of a MOSFET, while source
and drain are kept at a small reverse voltage. At the substrate contact, a recombination current
proportional to the SiO
2
interface trap
density is measured. It is based on the exploitation of a
repetitive process whereby majority carriers coming from the substrate recombine with
minority carriers previously trapped in interface states. Considering the emission processes,
which control the
exchange of charges at the interface, the information concerning the capture
cross

section and the energy distribution of the interface states can be obtained.
The MOSFET’s models used in all the electrical simulators (Spice, Smart

Spice,Esacap…) do not
take into account the weak substate current involved in the charge
pumping phenomenon. Indeed, the charge pumping technique being purely experimental, it is
not yet considered in the commercial electrical simulators.
In this work we propose an equivalent
schema of the MOSFET taking into account the
substrate current. This model has been implemented in the In the electrical simulator SPICE.
We have used an modified the level

3 SPICE MOSFET Model.
3.
Model description
The spectroscopic CP method consists in m
onitoring and subtracting the CP currents
obtained with a trapezoidal signal for two distinct and consecutive t
r1
and t
r2
values of the
E
em,h
(t
r1
)
E
em,h
(t
r2
)
E
em,e
(t
f
)
E
F,acc
E
F,inv
E
C
E
V
E
or
T
D
it
(E
or
)
density of the
interface state
BV
Si

O
Si

Si
BC
Figure 2
(a)
(b)
t
r2
t
f
t
r1
V
gh
V
T
V
fb
V
gl
V
g
Figure 1
Waveforms used in spectroscopic charge pumping
Fig 2
distribution of interface states (a) and domain of energy
band gap which can can be explored by the SCPtechnique using
the signal using ni (b)
rising edge of the gate pulse, while keeping the falling edge unchanged; the other
parameter
s are kept unchanged (Fig. 1). The difference between the two CP currents is due to
the fact that hole emission stops earlier (i.e. E
em,h
(t
r1
)<E
em,h
(t
r2
)) when the rising edge is
sharper (i.e. t
r1
<t
r2
). (where E
em,h
and E
em,e
correspond to the end of the n
on

steady hole
emission and to the end of the non

steady

state electron emission)
By subtracting the two signals, we obtain a third signal whose magnitude depends on
the portion of energy band gap thus scanned. The lower half of the band gap can thus be
sc
anned by an energy window (defined by t
r1
and t
r2
for t
f
constant) forced to move through
the band gap by varying the sample temperature (likewise, the upper half can be scanned by
using two distinct values while maintaining t
r
constant and by varying T as
shown in Fig. 2.
The spectroscopic signal (S
r
) corresponds to the energy window defined by t
r1
and t
r2
for constant t
f
, the differential CP current is given by:
(1)
where q is the electron charge, A
eff
is the channel area
of the transistor (cm

2
), , E
or
is the
mean energy level corresponding to the window (t
r1
,t
r2
), D
it
(Eor) is the interface trap density
at an energy level Eor(cm

2
eV

1
), K is the Boltzmann constant,T is the temperature and f is the
frequency of pulse.
Likew
ise, the spectroscopic signal (S
f
) corresponds to the energy window defined by
t
f1
and t
f2
for constant t
r
, the differential CP current is given by:
(2)
E
of
is the mean energy level corresponding to the window (t
f1
,t
f2
).
We note the
Spectroscopique charge pumping technique has been simulated by a electrical
simulator called
SPICE [6]
where our previously developed mathematical model concerning
the pumping current equation in function of temperature is given by:
(3)
With
Where
v
th
is the thermic velocity of the carriers, n
i
the intrinsic concentration, t
em,e
and t
em,h
are the
emission duration of electron and holes [6]
4.
Results and discussion
Many models of the M
OSFET, with different complexities and sensitivities, are
included in SPICE 3F4. To develop our model, we have used the MOSFET level3 of SPICE
3F4, and we have added to it a current source to simulate the charge pumping current. We
have considered the thre
e following effects on the threshold tension V
T
:
1.
the channel length modulation [6],
2.
the reduction of V
T
(due to interface states and oxide charges) [6],
3.
the effect of V
R
on V
T
[6],
4.
the model takes into account the temperature effect on the geometrical and
electrical
parameters of the studied transistor.
Fig. 3 shows a number of spectroscopic signals (S
r
) plotted on nMOSFET, in a temperature
range between 60 and 380 K. The gate was pulsed with a frequency of 50 KHz, between
–
3V
and +3V and for three differe
nt energy windows. Four rise times (250, 500, 1000 and 2000
µs) were applied at a fixed fall time of 1µs.
Fig. 4 shows spectroscopic signals (S
f
) plotted on nMOSFET, for a temperature between 60
and 380 K, the gate was pulsed with a frequency of 50 KHz, be
tween
–
3V and +3V and for
one energy window. Three rise times (250, 500, 1000 µs) were applied at a fixed fall time of
1µs. S
r
and S
f
, expressed by (1) and (2) vary linearly with respect to
Thus, for a given
temperature set for t
r
(
or t
f
) values and device parameters, we can extract D
it
(E) using the S
r
(or S
f
) plots.
the model takes into account the temperature effect on the geometrical and
electrical parameters of the studied transistor
The simulations results plottted in this pape
r have been obtained on conventional n

channel
MOSFET’s. The oxide thichness T
ox
is of 32nm, the channel length L and width W on mask
are respectively, 0,1 and 100µm.
For the validation of our models, we have used the measurements carried out usin
g the
different models an (BISIM1, BISIM2, MOS6, CPM that we have plotted) we then compared
the caracteriristics with those simulated by our model. Generally, we follow the evolution of
I
cp
, with the same parameters given.
5
Conclusion
In this work, we have
developed a spectroscopic charge

pumping model that we
implemented in SPICE 3F4. We have plotted the spectroscopic signals versus the temperature
for various energy windows. Our simulated results are in good agreement with other recent
experimental results
, which proves the efficiency of the developed model.
The technique studies allows a direct characterization that is sensible and consiste of
the interfacial zone in submicronic components.Nevertheless, it does not permit to attain an
energetic distributio
n in efficient sections of interface traps capture in satisfactory manner.
Fig. 3 Spectroscopic signals (S
r
) plotted for n

channel
transistor with different emission windows and t
r
=1µs.
Fig. 4.
Spectroscopic signal (S
f
) plotted for
n

channel transistor with one emission window
That is why new charge pumping technique ( a three Level Charge Pumping,…) are being
developed.
6. References
[1]
J. S. Brugler and P. Jespers,
Charge pumping in MOS devices
, I
EEE Trans. Electron
Devices, vol. 16, pp. 297

302, March 1969.
[2]
G. Groeseneken, E. M. Herman, N. Beltran and R. Keersmaecker,
A Reliable Approach
to Charge

Pumping Measurements in MOS Transistors
, IEEE Trans.
Electron Devices,
vol.
ED

31, pp. 42

53, January 1984
[3]
P. Hermans, J. Witters, G. Groeseneken and H. Maes, "Analysis of the charge pumping
technique and its application for evolution of MOSFET degradation", IEEE, Trans.
Electron Device, vol. 36, pp. 1318

1335, july 1989.
[4]
H. E. Maes a
nd G. Groeseneken, "Determination of spatial surface state density
distribution in MOS and SIMOS transistors after channel electron injection", Electronics
letters, 1998.
[5]
F. Djahli,
Mise aux point d’un dispositif expérimental pour l’étude des structur
es MOS
:
Application à l’étude de vieillissement des TMOS micronique par la technique de
pompage de Charge
, Thèse de doctorat, Lyon, septembre 1992.
[6]
N. Guenifi and F. Djahli,
A study of the degradation of short

channel TMOSs using
SPICE3F4
,
Semi
cond. Sci. Technol. Vol. 17, N°3 pp. 219

226,
March 2002
[7]
Autran J. L., Balland B. and Barbottin G. “ Charge pumping techniques their use for
diagnostic and interface states studies in MOS transistors”
I
nstabilities in Silicon,
Elsevier science publ
ishers B.V North Holland, (1999) (Editors Vapaille A.).
Σχόλια 0
Συνδεθείτε για να κοινοποιήσετε σχόλιο