Full text of DOI:10.1063/1.115031

Evidence of carrier number fluctuation as origin of 1/f noise in polycrystalline silicon
thin film transistors
A. Corradetti, R. Leoni, R. Carluccio, G. Fortunato, C. Reita, F. Plais, and D. Pribat
Citation: Applied Physics Letters 67, 1730 (1995); doi: 10.1063/1.115031
View online: http://dx.doi.org/10.1063/1.115031
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/67/12?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Characterization of traps in the gate dielectric of amorphous and nanocrystalline silicon thin-film transistors by 1/f
noise
J. Appl. Phys. 108, 106103 (2010); 10.1063/1.3506527
Origin of low-frequency noise in polycrystalline silicon thin-film transistors
J. Appl. Phys. 91, 9919 (2002); 10.1063/1.1481964
Hot-carrier-induced modifications to the noise performance of polycrystalline silicon thin-film transistors
Appl. Phys. Lett. 71, 1216 (1997); 10.1063/1.119855
Conduction and 1/f noise analysis in amorphous silicon thinfilm transistors
J. Appl. Phys. 79, 923 (1996); 10.1063/1.360873
1/f noise in MOS transistors due to number or mobility fluctuations
AIP Conf. Proc. 285, 345 (1993); 10.1063/1.44663
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.105.215.146 On: Sat, 20 Dec 2014 09:31:17

Evidence of carrier number fluctuation as origin of 1/f noise
in polycrystalline silicon thin film transistors
A. Corradetti, R. Leoni,a) R. Carluccio, and G. Fortunato
IESS-CNR, Via Cineto Romano 42, 00156 Roma, Italy
C. Reita, F. Plais, and D. Pribat
Thomson-CSF, LCR, 91404 Orsay Cedex, France
͑Received 14 April 1995; accepted for publication 17 July 1995͒
A systematic study of the noise performances of polycrystalline silicon thin film transistors is
presented. The drain current spectral density of these devices shows an evident 1/ f behavior and
scales, when operating in the linear regime, with the square of the mean value of the drain current.
The origin of the noise can be ascribed to carrier number fluctuations related to the dynamic
trapping and detrapping of the oxide traps. © 1995 American Institute of Physics.
Polycrystalline silicon ͑polysilicon͒ thin-film transistors
͑TFTs͒ are currently investigated for application in static ran-
dom access memories ͑SRAMs͒ and for addressing and driv-
ing in active matrix liquid crystal displays ͑AMLCD͒. In
these applications there is a common demand for high per-
formances, high stability, and low noise devices. A number
of technological efforts have been devoted in order to in-
crease the field effect mobility, now exceeding 400 cm2/V s
in polysilicon TFTs recrystallized by excimer laser.1,2 The
assessment of the long term reliability of the electrical sta-
bility has been the subject of a number of recent
investigations,3,4 while the noise characterization of such de-
vices has received little attention and only preliminary data
exist in the literature.5–7 On the other hand the noise is the
main limitation to the use of polysilicon TFTs in large area
image sensors. In fact, the performances of such applications
would greatly improve using a local amplification of the sig-
nal, but the equivalent noise charge has to be lower than
about 1000 electrons.8,9
In the present letter we have studied the noise perfor-
mances of polysilicon TFTs fabricated with a four mask se-
quence. The active layer ͑100 nm thick͒ was deposited amor-
phous by pyrolysis of Si2H6 in an UHVCVD reactor10 and
subsequently furnace crystallized at 580 °C. After definition
and reactive ion etching ͑RIE͒ of the active polysilicon is-
lands, the gate SiO2 was deposited to a thickness of 120 nm
by distributed electron cyclotron resonance ͑DECR͒ plasma
enhanced CVD ͑PECVD͒.10 The gate electrode was defined
by RIE of an in situ doped amorphous silicon layer deposited
by LPCVD at 560 °C. The self-aligned source and drain con-
tacts were formed by a phosphorous implant, which was ac-
tivated by a furnace anneal at 580 °C. This thermal treatment
also converted the amorphous gate material to low resistivity
degenerated polycrystalline silicon. The contacts were de-
fined by lift-off of an evaporated aluminum film and the
devices were then annealed at 450 °C under flowing forming
gas ͑10% H2 in N2͒. Finally, the samples were covered with
a SiO2 passivation layer. Field effect mobility in these de-
vices, without posthydrogenation treatment and with a maxi-
mum processing temperature of 580 °C, is around 50
cm2/V s.
The noise measurements were performed biasing the de-
vice with constant source-drain voltage Vds , and measuring
the fluctuations of the drain current Id , through a low noise
current amplifier ͑PAR 181͒ connected to a spectrum ana-
lyzer ͑HP 3561A͒. The noise contribution of the source/drain
contact resistance was found negligible. Polysilicon TFTs
with channel width 40 ␮m (W) and channel lengths; 5, 10,
20, 30, and 40 ␮m (L) were studied. Figure 1 shows the
transfer characteristics for different channel lengths, where
an high on/off ratio ͑more than 5 orders of magnitude͒ is
observed. Figure 2 shows a typical drain current spectral
density SI , measured in a device operating in the linear re-
gime: Vdsϭ2 V and gate voltage, Vgϭ14 V. As can be noted
the noise shows a 1/f behavior, commonly observed in crys-
talline Si ͑c-Si͒ MOSFETs.11–14 When the device is biased in
the linear regime with a constant gate voltage the drain cur-
rent spectral density increases as V2ds, or, equivalently, as the
square of the mean value of Id , in agreement with the 1/f
theory,11 as confirmed by Fig. 3, where SI /I2d is plotted ver-
sus the mean value of Id .
The origin of the low frequency noise in MOSFETs has
been related to either carrier number fluctuation or carrier
mobility fluctuation. In the carrier number fluctuation model,
based on the McWorther theory,12,15,16 the fluctuations of the
FIG. 1. Transfer characteristics measured at Vdsϭ0.2 V for different channel
lengths: continuous line Lϭ5 ␮m dashed line Lϭ20 ␮m, and dotted line
Lϭ40 ␮m.
a͒
Electronic mail: roberto@iess.rm.cnr.it
1730
Appl. Phys. Lett. 67 (12), 18 September 1995
0003-6951/95/67(12)/1730/3/$6.00
© 1995 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.105.215.146 On: Sat, 20 Dec 2014 09:31:17

FIG. 2. Drain current spectral density for a TFT with channel length L
ϭ10 ␮m Vdsϭ2 V and Vgϭ14 V.
FIG. 4. Normalized drain current spectral density, SI /I2d, at 20 Hz vs the
mean value of the drain current Id , for TFTs with channel length Lϭ5 ␮m
͑᭹͒, Lϭ20 ␮m ͑᭿͒, and 40 ␮m ͑᭝͒. Continuous lines are the best fit to Eq.
͑1͒.
drain current are induced by fluctuations of the interfacial
oxide charge due to the dynamic trapping and detrapping of
free carriers into slow oxide traps. In other words, due to the
interface charge fluctuations, the flatband voltage Vfb fluctu-
ates and, therefore, also the charge induced in the semicon-
ductor fluctuates. In the mobility fluctuation model,17 the
fluctuations of the drain current arise from the fluctuations of
the carrier mobility possibly through a fluctuation of the scat-
tering cross section. This results in a 1/f noise whose inten-
sity is inversely proportional to the total number of carriers
in the system. In order to discriminate between the two
mechanisms the normalized drain current spectral density
can be analyzed. In fact, according to the carrier number
fluctuations, at low values of the drain current, SI /I2d shows a
plateau before decreasing as IϪd 2 as the TFT is turned on.16
Vice versa, in case of mobility fluctuations, SI /I2d is expected
to be inversely proportional to Id .16
the channel, Cox is the gate oxide capacitance per unit area,
gm is the device transconductance ͑evaluated from the data
shown in Fig. 1͒, and SVfb is the flatband voltage spectral
density. The continuous lines shown in Fig. 4 represent the
best fit of the data with Eq. ͑1͒ and, as can be easily recog-
nized, a very good agreement is observed. The mean value of
the fitting parameter ␣ was found 1.5 104 V s/C, close to that
reported for c-Si MOSFETs ͓104 V s/C ͑Ref. 16͔͒, while the
value of the fitting parameter SVfb was around 1.2 10Ϫ8
V2/Hz. The noise in polysilicon TFTs appears much higher
than in c-Si MOSFETs7,13,14 and such an higher noise level
could be related to fluctuations of the barrier heights present
at the polysilicon grain boundaries7 or to an higher density of
traps in the oxide close to the interface. Assuming a predomi-
nance of the latter mechanism and according to the conven-
tional tunneling mechanism,12,15 the oxide trap state density
Nt can be evaluated from the following expression:
Figure 4 shows SI /I2d versus Id for different channel
lengths. The measurements were performed biasing the TFTs
in the linear regime ͑Vdsϭ0.2 V͒ and changing Vg . From the
shape of the SI /I2d the noise the polysilicon TFTs can be
ascribed to carrier number fluctuations. In order to confirm
this point we have fitted the experimental data with the theo-
retical expression for the SI /I2d in the case of carrier number
fluctuations:16
SVfbϭ q2KT␭N ͒/ WLC2 f ͒,
͑2͒
͑
͑
t
ox
where ␭ is the tunnel attenuation distance ͓around 0.1 nm
͑Ref. 13͔͒, q is the unit charge, and KT is the thermal energy.
From Eq. ͑2͒ we obtain, by using the fitted values for SVfb , a
Nt value around 1020 states/eV cm3, which is about three
orders of magnitude higher than the Nt value generally found
for CMOS technology.13,14
2
2
S /I ϭ 1ϩ␣␮ I /g ͒ g /I ͒2S
,
͑1͒
͑
͑
m
I
eff d
m
d
Vfb
d
where ␣ is a constant, ␮eff is the effective carrier mobility in
In conclusion, we have, for the first time, systematically
studied the noise performances in polysilicon TFTs. These
devices are characterized by the presence of a strong 1/f
noise. From the analysis of the drain current spectral density
we have identified the origin of the low frequency noise in
the carrier number fluctuations. The noise level appears to be
much higher than in c-Si MOSFETs. This increased noise
level could be related to several effects including the pres-
ence of localized states at the polysilicon grain boundaries7
or an higher oxide trap density. In the latter case we have
estimated an oxide trap density around 1020 states/eV cm3,
which is three order of magnitude higher than typical values
for c-Si MOSFETs.13,14 Although our DECR PECVD oxide,
deposited at low temperature ͑Ͻ100 °C͒, contains an higher
density of traps if compared to thermal oxide, the estimated
value of Nt is not compatible with the quality of our oxide.18
FIG. 3. Normalized drain current spectral density, SI /I2d, at 20 Hz vs the
mean value of the drain current Id , for a TFT operating in the linear regime
with Lϭ10 ␮m and Vgϭ14 V.
Appl. Phys. Lett., Vol. 67, No. 12, 18 September 1995
Corradetti et al.
1731
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.105.215.146 On: Sat, 20 Dec 2014 09:31:17

Therefore, further investigations are needed to clarify the
microscopic mechanisms causing the excess noise observed
in polysilicon TFTs.
9 L. E. Antonuk, J. Boudry, Y. El-Mohri, W. Huang, J. Siewerdsen, J. Yor-
ston and R. A. Street, Proc. SPIE 2163, 118 ͑1994͒.
10 D. Pribat, F. Plais, P. Legagneux, T. Kretz, R. Stroh, O. Huet, C. Walaine,
and M. Magis, Rev. Tech. Thomson-CSF 26, 73 ͑1994͒.
11 A. van der Ziel, in Noise in Solids State Devices and Circuits ͑Wiley, New
York, 1986͒, p. 145.
We wish to thank G. Torrioli for many useful discussion.
12 S. Christensson, I. Lundstrom, and C. Svensson, Solid State Electron. 11,
797 ͑1968͒.
1 T. Sameshima, Solid State Phenomena 37–38, 269 ͑1994͒.
2 H. Kuriyama, T. Kuwahara, K. Wakisaka, S. Kiyama, S. Tsuda, and S.
Nakano, Electrochem. Soc. Proc. 94-35, 77 ͑1995͒.
13 G. Ghibaudo and O. Roux-dit-Buisson, Proceedings of the European Solid
State Device Research Conference, edited by C. Hill and P. Ashburn ͑Edi-
tions Frontieres, France, 1994͒, p. 693.
3 G. Fortunato, A. Pecora, G. Tallarida, L. Mariucci, C. Reita, and P. Miglio-
rato, IEEE Trans. Electron Devices 41, 340 ͑1994͒.
14 P. Morfouli, P. McLarty, V. Misra, J. Hauser, J. J. Wortman, T. Ouisse, and
G. Ghibaudo, in Ref. 13, p. 313.
4 J. R. Ayres and N. D. Young, IEE Proc. Circuits Devices Syst. 141, 38
͑1994͒.
15 A. L. McWorther, in Semiconductor Surface Physics ͑University of Penn-
sylvania Press, Philadelphia, 1957͒, p. 207.
5 A. G. Lewis, T. Y. Huang, R. H. Bruce, M. Koyanagi, A. Chiang, and I. W.
Wu, Proc. IEDM 1988, 264 ͑1988͒.
16 G. Ghibaudo, O. Roux, Ch. Nguyen-Duc, F. Balestra, and J. Brini, Phys.
Status Solidi A 142, 571 ͑1991͒.
6 C. Reita and S. Fluxman, IEE Proc. Circuits Devices Syst. 141, 60 ͑1994͒.
7 T. Hashimoto, M. Aoki, T. Yamanaka, Y. Kamigaki, and T. Nishida, Sym-
posium on VLSI Technology 1994, p. 87.
17 F. Hooge, Physica ͑Utrecht͒ B 83, 14 ͑1976͒.
18 R. Stroh, F. Plais, T. Kretz, P. Legagneux, O. Huet, M. Magis, D. Pribat,
N. Jiang, M. C. Hugon, and B. Agius, IEE Proc. Circuits Devices Syst.
141, 9 ͑1994͒.
8 L. E. Antonuk, Y. El-Mohri, W. Huang, J. Siewerdsen, J. Yorkston, and R.
A. Street, Mater. Res. Soc. Symp. Proc. 336, 855 ͑1994͒.
1732
Appl. Phys. Lett., Vol. 67, No. 12, 18 September 1995
Corradetti et al.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
129.105.215.146 On: Sat, 20 Dec 2014 09:31:17