Interaction of solid organic acids with carbon nanotube field effect transistors
ABSTRACT A series of solid organic acids were used to p-dope carbon nanotubes. The extent of doping is shown to be dependent on the pKa value of the acids. Highly fluorinated carboxylic acids and sulfonic acids are very effective in shifting the threshold voltage and making carbon nanotube field effect transistors to be more p-type devices. Weaker acids like phosphonic or hydroxamic acids had less effect. The doping of the devices was accompanied by a reduction of the hysteresis in the transfer characteristics. In-solution doping survives standard fabrication processes and renders p-doped carbon nanotube field effect transistors with good transport characteristics. Comment: 5 pages, 4 figures, 1 table
arXiv:cond-mat/0702198v1 [cond-mat.mtrl-sci] 8 Feb 2007
Interaction of solid organic acids with carbon nanotube field effect transistors
Christian Klinke,∗Ali Afzali, and Phaedon Avouris
IBM T. J. Watson Research Center, 1101 Kitchawan Road, Yorktown Heights, NY 10598, USA
A series of solid organic acids were used to p-dope carbon nanotubes. The extent of doping is shown
to be dependent on the pKa value of the acids. Highly fluorinated carboxylic acids and sulfonic acids
are very effective in shifting the threshold voltage and making carbon nanotube field effect transistors
to be more p-type devices. Weaker acids like phosphonic or hydroxamic acids had less effect. The
doping of the devices was accompanied by a reduction of the hysteresis in the transfer characteristics.
In-solution doping survives standard fabrication processes and renders p-doped carbon nanotube field
effect transistors with good transport characteristics.
Carbon nanotubes (CNT) have been shown to be high-
performance building blocks in electronic devices like
field effect transistors (FET) [1, 2]. They provide high
carrier mobility  and chemical sensitivity .
properties of such FETs depend very strongly on the
metals used as leads [5, 6] and the environmental con-
ditions [7, 8]. For example, due to absorption of oxygen
at the metal contacts, intrinsic CNTs can form p-type
devices . Recently, we reported chemical p-doping of
CNT by treatment with a one-electron oxidizing agent
to form air-stable p-type CNTFETs . Strong organic
acids, like trifluoroacetic acid have been shown to be
effective one-electron oxidants for electron rich organic
compounds , while Dukovic et al. showed that car-
bon nanotubes can be protonated by strong acids .
Thereby, the charge transfer depends on the chirality
and accordingly on the bandgap of the nanotubes .
CNTFETs are Schottky barrier transistors. Those bar-
riers appear at the interface between the semiconduct-
ing nanotubes and the metallic leads. The workfunction
alignment of the leads with the nanotube determines the
extent of the Schottky barriers [13, 14, 15]. The switching
of the CNTFETs is based on the bending of the nanotube
bands. The doping effect manifests mainly in the shift-
ing of the threshold voltage to more positive (p-doping)
or more negative (n-doping) values [9, 16, 17]. In this
communication we explore systematically the oxidation
effect (p-doping) of various strong solid organic acids on
CNTFETs and we show that the extent of doping de-
pends strongly on the acidity of the organic acids. Direct
doping of the devices with acids has the disadvantage of
increasing the leakage current to the gate electrode. This
could be prevented by doping the nanotubes already in
solution followed by the fabrication of the devices with
the functionalized CNTs.
∗Electronic address: firstname.lastname@example.org; Present ad-
dress: Institute of Physical Chemistry, Universtity of Hamburg,
20146 Hamburg, Germany.
Carbon nanotube FETs were fabricated by the de-
position of laser ablation carbon nanotubes from dis-
persion in dichloroethylene onto a silicon wafer with
20 nm thermally grown silicon dioxide.
transistors with channel lengths of 400 nm were defined
by e-beam lithography and successive e-beam deposi-
tion of a titanium adhesive layer (0.7 nm) and palla-
dium (25 nm). The substrates with CNTFETs were
then immersed in a 10 mM ethanolic solution of var-
ious acids, N-hydroxyheptafluorobutyramide (HFBA),
perfluorododecanoic acid (PFDA), 2-methyl acrylamido-
propane sulfonic acid (AMPS), aminobutyl phosphonic
acid (ABPA), and hexadecyl phosphonic acid (HDPA)
for 24 hours and then dried under a stream of nitrogen.
The pKa value of those acids are (HFBA) ∼ 4.5, (HDPA)
∼ 2.6, (PFDA) ∼ 1.0, and (AMPS) ∼ -2.0. The electric
transport behavior of the CNTFET devices was charac-
terized in a nitrogen-purged glove-box using an Agilent
4165C semiconductor Parameter Analyzer.
In order to evaluate the doping effect independently
from the transport measurements we employed absorp-
tion spectroscopy on purified HiPCO nanotubes specifi-
cally for AMPS. We dried an ethanolic solution of nano-
tubes on a quartz plate. The absorption of this sample
was then measured with a Perkin Elmer UV/Vis/NIR
spectrometer Lambda 9. A 10 mM solution of AMPS,
the strongest dopant, was then added dropwise to the
We measured the transfer characteristics of the CNT-
FETs before and after doping with the solid acids N-
hydroxyheptafluorobutyramide (HFBA), perfluorodode-
canoic acid (PFDA), and 2-methyl acrylamidopropane
sulfonic acid (AMPS) (Fig. 1). These compounds show
increasing acidity in this sequence. In most cases the
drive current Ion(upper saturation current) decreases by
less than one order of magnitude. The effect of doping,
i.e. the threshold voltage shift  increased with the
strength of the acid (Tab. I). The threshold voltage is
defined by the upper point where the drain current Id
FIG. 1: Typical transfer characteristics (Id vs. Vg) of CNT-
FET devices with a source-drain voltage of Vds = -0.5 V on a
20 nm gate oxide, demonstrating the doping effect of the solid
organic acids a) HFBA, b) PFDA, and c) AMPS. Arrows in-
dicate the direction of voltage sweep (black - pristine device;
red - doped device) and the threshold voltage shift (in green).
deviates from exponential increase in the transfer char-
acteristics. The threshold voltage was shifted to more
positive values, i.e. the devices showed a more p-type
FIG. 2: Absorption spectroscopy of purified and dried nano-
tubes produced by the HiPCO method. The sample was step-
wise doped with one droplet of a 10 mM AMPS solution. Ar-
rows indicate the most prominent peaks.
behavior after doping. Additionally, in all cases the gate
hysteresis decreased. Stronger acids decreased the hys-
teresis more. However, the leakage current (current to
the gate) increased with the strength of the acid and the
ON/OFF ratio decreased. The actual values are summa-
rized in Tab. I. The threshold shift vanished when the
devices were measured in vacuum, and could be restored
by exposing them to air - suggesting a cooperative effect
between the acid and oxygen.
The doping effect was confirmed independently by ab-
sorption measurements. First, the UV/Vis/NIR spec-
trum of purified HiPCO nanotubes dried on a quartz
plate was measured. By adding dropwise a solution of
10 mM AMPS in ethanol, the intensity of the absorp-
tion peaks decreased substantially (Fig. 2). Larger wave-
lengths, corresponding to tubes with smaller bandgap,
and thus tubes with larger diameters, were suppressed
more effectively. The reduction of absorption peaks is
typically attributed to electron depletion from or filling
specific bands due to doping . The doping is more
efficient for tubes with smaller bandgaps (larger wave-
lengths). AMPS itself is transparent in the range where
the spectra were recorded.
In other experiments, bifunctional aminobutyl phos-
phonic acid (ABPA) containing both amine and phos-
phoric acid and hexadecyl phosphonic acid (HDPA) were
used for doping experiments (Fig. 3). The doping effect
of amino groups on carbon nanotubes have been stud-
ied earlier, and were shown to be strong n-dopants .
Whereas HDPA with only a phosphonic acid group mod-
erately p-dopes the CNT, treatment of CNTFETs with
the bifunctional ABPA results in moderate n-doping.
This experiment shows that the electron donating affin-
ity of the amino group is higher than electron-accepting
affinity of the phosphonic acid. In order to test the effect
TABLE I: Comparison table.
HFBA 4.5 
HDPA 2.0 
PFDA 0.5 
AMPS -2.0 
pKa Threshold shift [V] ON/OFF ratio (before/after) Hysteresis (before/after) [V]
0.6 7.3E5 / 1.3E5
2.11.3E5 / 3.4E2
2.3 1.7E6 / 1.8E4
4.7 2.0E4 / 3.1E2
-1.6 6.6E5 / 2.0E5
4.7 / 2.8
6.3 / 2.5
4.0 / 2.6
4.6 / 0.7
4.2 / 4.3n/a
of the acid group we used HDPA. This molecule is sim-
ilar to ABPA but without the amino group at one end.
This molecule drives the device again to a more p-type
Since the in situ doping increased the OFF-current
(lower limit of Id) significantly we doped the nanotubes
with acids in solution and fabricated devices with these
already doped tubes. For this we used again the strongest
acid AMPS (Fig. 4). Under the consideration that our
pristine nanotube devices usually switch at Vth = -
3 V 1 V, we conclude that the devices fabricated from
solution-doped (ex situ) nanotubes switches at threshold
voltages which are comparable to the in-situ doped nano-
tube, namely at -0.6 V for in situ doped vs. -1 V for ex
situ doped devices.
The ability of highly acidic trifluoroacetic acid to ox-
idize electron-rich organic compounds to their radical
cation has been reported by Eberson et al. . The ox-
idizing effect of trifluoroacetic acid is attributed to first
protonation of electron-rich compounds followed by an
electron transfer to an oxygen molecule.
oroactic acid has very high vapor pressure and is des-
orbed easily upon moderate heating, we were prompted
to study the effect of higher molecular weight organic
acids on oxidation (p-doping) of carbon nanotubes. We
chose different classes of organic acids. PFDA with a pKa
of about 0.5 and AMPS with a pKa of about -2.0 proved
to be effective p-dopants and to shift the threshold volt-
age of CNTFETs to more positive voltages significantly
as shown in Fig. 1. Milder organic acids like phospho-
nic acids (HDPA) and hydroxamic acids (HFBA) with a
pKa value of about 2.5 and 4.5, respectively, were less
effective and the shift to more positive threshold volt-
ages were less pronounced. The contacts are not much
affected during the doping of the whole device since the
ON-current does not change much.
The effect is most likely due to protonation, followed by
oxidation to radical cation in the presence of oxygen [12,
20]. The presence of oxygen seems to be necessary for the
stabilization of the protonation, since we find that the
doping effect vanishes in vacuum and can be restored by
exposure to air. The level of doping is determined by the
strength of the acid i.e. the effectiveness of protonation of
demonstrating the doping effect of a) the bifunctional acid
ABPA, and b) the solid acid HDPA. Arrows indicate the di-
rection of voltage sweep (black - pristine device; red - doped
device) and the threshold voltage shift (in green).
Transport measurements with CNTFET devices
CNTs. Lower pKa value leads to stronger hole injection.
The hysteresis in the transfer characteristic of CNT-
FETs might be due to organic contamination from the
lithography process . The reduction of hysteresis is
attributed to ”cleaning” of the devices by the doping pro-
cedure (soaking the devices for 24 h in the corresponding
acid solutions). This treatment might remove organic
demonstrating the effect of ex-situ doping with AMPS.
Transport measurements with CNTFET devices
material around the carbon nanotubes. Similar cleaning
effects were observed before .
We confirmed the doping effect independently by ab-
sorption measurements. Furthermore, we showed that
the acids group of the molecules is responsible for the p-
doping effect by comparing the molecules with an amino
group containing acid. The known n-doping effect of
amino groups , competes with the acid group. The fi-
nal effect is an n-doping effect, whereas a similar molecule
without the amino group does not show this effect, but
a shift to more p-type behavior.
Direct doping of CNTFET devices leads also to higher
OFF-currents due to higher leakage currents to the gate,
and the subthreshold slopes become shallower.
higher leakage is due to the protonation of the oxide.
We were able to circumvent this disadvantage by dop-
ing the nanotubes first in solution and then integrating
the so-treated tubes into devices. This process restores
the high performance of pristine nanotube devices while
producing the desired threshold shift.
To conclude, we used a series of solid organic acids
in order to p-dope carbon nanotubes devices. The ex-
tent of doping was shown to be dependent on the pKa
value of the acids. Highly fluorinated carboxylic acids
and sulfonic acids were very effective in shifting the
threshold voltage of carbon nanotube field effect transis-
tors. Weaker acids like phosphonic or hydroxamic acids
had less doping effect. The doping effect was also con-
firmed by UV/Vis/NIR absorption spectroscopy. The in-
solution doping survives standard fabrication processes
and renders p-doped CNTFETs with good transport
characteristics. The influence of acids is not only in-
teresting form a doping point of view, but acids are also
present as contaminations in semiconductor processing
from etching steps or from rinsing with solvents.
We like to acknowledge gratefully the Alexander-von-
Humboldt Foundation for financial support and Bruce
Ek for technical assistance.
 S. Tans, A. Verschueren, C. Dekker, Nature (London)
393 (1998) 49.
 R. Martel, T. Schmidt, H. R. Shea, T. Hertel, Ph.
Avouris, Appl. Phys. Lett. 73 (1998) 2447.
 M. S. Fuhrer, B. M. Kim, T. Durkop, T. Brintlinger,
Nano Lett. 2 (2002) 755.
 P. G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science
287 (2000) 1801.
 R. Martel, V. Derycke, C. Lavoie, J. Appenzeller, K. K.
Chan, J. Tersoff, Ph. Avouris, Phys. Rev. Lett. 87 (2001)
 D. Mann, A. Javey, J. Kong, Q. Wang, H. J. Dai, Nano
Lett. 3 (2003) 1541.
 V. Derycke, R. Martel, J. Appenzeller, Ph. Avouris,
Appl. Phys. Lett. 80 (2002) 2773.
 W. Kim, A. Javey, O. Vermesh, O. Wang, Y. M. Li, H.
J. Dai, Nano Lett. 3 (2003)193.
 J. Chen, C. Klinke, A. Afzali, Ph. Avouris, Appl. Phys.
Lett. 86 (2005) 123108.
 L. Eberson, F. Radner, Acta Chemica Scandinavia 45
 G. Dukovic, B. E. White, Z. Zhou, F. Wang, S. Jockusch,
M. L. Steigerwald, T. F. Heinz, R. A. Friesner, N. J.
Turro, L. E. Brus, J. Am. Chem. Soc. 126 (2004) 15269.
 M. J. O’Connell, E. E. Eibergen, S. K. Doorn, Nature
Mat. 5 (2005) 412.
 M. P. Anantram, F. Leonard, Rep. Prog. Phys. 69 (2006)
 S. Heinze, J. Tersoff, R. Martel, V. Derycke, J. Appen-
zeller, P. Avouris, Phys. Rev. Lett. 89 (2002) 106801.
 P. Avouris, Acc. Chem. Res. 35 (2002) 1026.
 K. Bradley, J. C. P. Gabriel, M. Briman, A. Star, G.
Grner, Phys. Rev. Lett. 91 (2003) 218301.
 C. Klinke, J. Chen, A. Afzali, Ph. Avouris, Nano Lett. 5
 M. Radosavljevic, J. Appenzeller, Ph. Avouris, J. Knoch,
Appl. Phys. Lett. 84 (2004) 3693.
 S. Kazaoui, N. Minami, R. Jacquemin, H. Kataura, Y.
Achiba, Phys. Rev. B 60 (1999) 13339.
 M. S. Strano, C. B. Huffman, V. C. Moore, M. J.
O’Connell, E. H. Haroz, J. Hubbard, M. Miller, K. Ri-
alon, C. Kittrell, S. Ramesh, R. H. Hauge, R. E. Smalley,
J. Phys. Chem. B 107 (2003) 6979.
 H. Shimauchi, Y. Ohno, S. Kishimoto, T. Mizutani, Jap.
J. Appl. Phys. 45 (2006) 5501.
 C. Klinke, J. B. Hannon, A. Afzali, Ph. Avouris, Nano
Lett. 6 (2006) 906.
 Estimated from the pKa of non-fluorinated hydroxamic
acids: Oskar N. Ventura, Jose B. Rama, Laszlo Turi, J.
J. Dannenberg, J. Am. Chem. Soc. 115 (1993) 5754.
 L. D. Freedman, G. O. Doak, Chem. Rev. 57 (1957) 479.
 E. T. McBee, O. R. Pierce, D. D. Smith, J. Am. Chem.
Soc. 76 (1954) 3722.
 A. K. Covington, R. Thompson, J. Solution Chem. 3