Room-temperature negative differential conductance in carbon nanotubes
Transcript of Room-temperature negative differential conductance in carbon nanotubes
could not be analysed in pure polar solvents since fulle-
renes are almost completely insoluble in polar and H-
bonding solvents like acetone, tetrahydrofurane, aceto-
nitrile, the only exception being NMP in spite of its very
high dielectric constant (e = 32.2). This unusual solubil-
ity of fullerenes in such a polar solvent appears to be dueto fullerene-specific interactions involved in fullerenes
solubilization in NMP which are also responsible for
the observed spectroscopic changes. Aggregation in pure
NMP is probably due to the difference of polarity be-
tween fullerene molecules and NMP which forces the
molecules of fullerenes to aggregate in order to minimize
the contact with the solvent by means of weak p–p inter-
actions between fullerenes. In other words solute-soluteinteractions are favoured with respect to solute-solvent
interactions, resulting in the fullerenes aggregation. A
much larger number of interactions occur in solvent
mixtures like toluene/NMP that specifically are: NMP-
toluene, fullerene-fullerene, NMP-fullerene and tolu-
ene-fullerene interactions [5,6]. The larger dimensions
of the aggregates in toluene-NMP mixed solvents have
to be due to the predominance of NMP-toluene interac-tions which favours the fullerenes-fullerenes interaction.
This is the first time, on our knowledge, that fullerene
aggregation has been studied by DLS and TRFPA
methods measuring the size of aggregates in NMP and
toluene/NMP mixtures solvents. The use of NMP as dis-
persant/solvent for carbon-rich materials like fullerenes,
soot, etc., is promising for their structural analysis pro-
vided that the interactions of these materials with NMP
are recognised.
References
[1] Frackowiak E, Beguin F. Interaction between electroconducting
polymers and C60. J Phys Chem Solids 1996;57(6–8):983–9.
[2] Yevlampieva NP, Biryulin YuF, Melenevskaja EYu, Zgonnik VN,
Rjumtsev EI. Aggregation of fullerene C60 in N-methylpyrrolidi-
none. Colloids Surf A 2002;209:167–71.
[3] Brandelik DM, McLean DG, Brandt MC, Sutherland RL, Frock
LR, Crane RL. Complex formations in C60 solutions. Mat Res Soc
Symp Proc 1995;374:299–304.
[4] Apicella B, Ciajolo A, Barbella R, Tregrossi A, Morgan TJ, Herod
AA, et al. Size exclusion chromatography of particulate produced
in fuel-rich combustion of different fuels energy fuels
2003;17(3):565–70.
[5] Nath S, Pal H, Sapre A. Effect of solvent polarity on the
aggregation of C60. Chem Phys Lett 2000;327:143–8.
[6] Nath S, Pal H, Sapre A. Effect of solvent polarity on the
aggregation of fullerenes: a comparison between C60 and C70.
Chem Phys Lett 2002;360:422–8.
[7] Liu S, Lu Y, Kappes MM, Ibers JA. The structure of the C60
molecule: X-Ray crystal structure determination of a twin at 110K.
Science 1991;254:408–10.
Room-temperature negative differential conductancein carbon nanotubes
Jingqi Li *, Qing Zhang
Microelectronics Center, School of Electrical and Electronic Engineering, Nanyang Technological University, S1-B2c-20, Singapore 639798, Singapore
Received 27 August 2004; accepted 13 October 2004
Available online 25 November 2004
Keywords: A. Carbon nanotubes; D. Electrical (electronic) properties
Single-wall carbon nanotubes (SWNTs) have at-tracted much attention due to their unique electronic,
thermal and mechanical properties. The one-dimen-
sional (1D) structure with nanometer diameter and long
length (of the order of micrometer) makes it possible to
realize on-tube electronic devices, especially quantum
devices. Nanoscale intratube junctions and devices on
SWNTs have been prepared by chemical doping [1–3]
and mechanical bending techniques [4,5]. The fabricateddevices exhibit unique characteristics, such as single elec-
tron transistor (SET) characteristics [2,4,5], low temper-
ature negative differential conductance (NDC) [1] and
intratube rectifying characteristics [3], etc. In this paper,
we report a significant NDC phenomenon of a bundle of
SWNTs at room temperature. This room temperature
SWNT-NDC could be quite useful for practical applica-
tions of fast switching, oscillator and amplifying devices,etc. Possible mechanisms for the NDC phenomenon are
discussed.
* Corresponding author. Tel.: +65 679 045 57; fax: +65 679 333 18.
E-mail address: [email protected] (J. Li).
0008-6223/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2004.10.019
Letters to the Editor / Carbon 43 (2005) 651–673 667
The samples used here are back-gated carbon nano-
tube (CNT) based field effect transistors (CNTFETs)
[6–8]. P-type silicon with 500nm thick thermally grown
silicon dioxide layer is used as the back gate. Source
and drain electrodes are made of 20nm thick Ti fol-
lowed by 40nm thick Au prepared by standard UVlithography and lift-off processes. The gap between the
source and drain electrode is 2lm.
Commercial SWNTs with a high purity of �95% are
used in the experiment. The diameter of the SWNTs is
around 1.4nm. SWNTs are bridged across the source
and drain electrodes using an AC dielectrophoresis
method [8]. From AFM images, the diameter of the
bridging SWNT bundle is about 20nm. The drain cur-rent as a function of the drain-source voltage (Id–Vds
curve) is measured using a HP-4156B semiconductor
analyzer at ambient atmosphere. Fig. 1(a) shows Id verse
Vds as a function of the gate voltage Vg from �16 to
16V. One can see that when Vg > 0V, Vg has little influ-
ence on the Id–Vds curves. We suggest that Id (for
Vg > 0V) is dominated by those metallic SWNTs in
the bundle. To burn off the metallic SWNTs using elec-trical burning method [8,9], a gate voltage of +6V is ap-
plied to switch off and protect the semiconducting
SWNTs while Vds is scanned from zero to a positive
value, Vpds which is increased gradually from a lower
value, say, 5V, upto a higher one until some m-SWNTs
are destroyed. Two apparent current drops at
Vpds = 10.2V (Curve 2 in Fig. 1(b)) and 12.2V (Curve 4),
respectively, suggest the breakdown of some metallic
SWNTs.
After the above tube-burning process, the resultantId–Vds curves show significant NDC characteristics at
room temperature. It can be seen from Fig. 2(a) that
the current peak and valley occur at Vds = 9.2V and
9.8V, respectively, under Vg = 6V. The peak-to-valley
current ratio is �2. The NDC characteristics are repeat-
able at a given Vg as shown in Fig. 2(a) (Curves 2 and 3).
Fig. 2(b) shows the NDC two days after measuring the
Id–Vds curves in Fig. 2(a). At Vg = 6V, the peak voltageof the NDC is exactly at the same voltage value
although the peak current changes slightly. More inter-
estingly, the peak voltage gradually decreases from 9.2V
to 7.8V with Vg decreasing from 6V to �6V.
It is important to note that the NDC phenomenon
was not observed at the first round of the tube-burning
process even if Vds was higher than 9.2V (the peak volt-
age of the NDC), see Curves 1 and 2 in Fig. 1(b). There-fore, the bundle structure and imperfect contacts
between the tube bundle and electrodes should not be
responsible for the NDC. We suggest that it is caused
by a quantum well in the semiconducting CNT as the
appearance of the NDC is consistent with the hole reso-
nant tunneling through a quantum well [10]. At Vds = 0,
holes in the well are confined to discrete energy levels,
say En, in thermal equilibrium condition, as shown inFig. 3(a). As Vds is increased, an accumulation region
forms near the barrier by the drain side and a depletion
region forms near the barrier by the source side. How-
ever, very few holes can tunnel through the energy bar-
rier until Vds reaches a value (for example 9.2V) at
-6
-3
0
3
6
-1.5 -0.5 0.5 1.5Vds (V)
Id(x
10-7
A) (a) Vg = -16V~16V
Step: 4V
0
2
4
6
0 8 12Vds (V)
Id(x
10-5
A) (b)
1
234
54
Fig. 1. (a) Id as a function of Vds measured for Vg = � 16–16V in steps
of 4V. (b) Id as a function of Vds during the tube burning process.
Curve 1 is the one before burning. Curves 2 and 3 are sweepings from 0
to 10.2V and 11V, respectively. Curve 4 is sweeping from 0 to 12.2V
with a large conductance decrease at 11.8V. Curve 5 is the last
sweeping.
0
3
6
9
12
0 3 6 9 12 15Vds (V)
Vds (V)
I d (x
10-6
A)
A
B
C123
(a)
4
6
8
10
12
6 8 10 12
I d (x
10-6
A)
(b) Vg (V)
63
0-3
-6
Fig. 2. (a) Id as a function of Vds at Vg = 6V. Curves 1–3 were
measured under the same conditions. For clarity, Curves 2 and 3 for
the repeated measures are shifted up by 2.5 · 10�6A and 4 · 10�6A,
respectively. (b) Id as a function of Vds at Vg = 6, 3, 0, �3 and �6V.
668 Letters to the Editor / Carbon 43 (2005) 651–673
which the Fermi level of the drain side is aligned with En
in the well (Fig. 3(b)). When this happens, holes could
tunnel through the right barrier into the well at a much
high tunneling rate and subsequently through the left
barrier into unoccupied states in the valence band of
the source side. As Vds is increased further, the va-lence-band edge on the right is below En (Fig. 3(c)), so
that the tunneling rate of holes is significantly reduced.
As a result, the current begins to decrease with increas-
ing the voltage, resulting in an apparent NDC
phenomenon.
The energy levels in the well can be approximately
expressed as
En ��h2
2m�npl
� �2
; ð1Þ
where n is the quantum number labeling the energy lev-
els, l is the width of the well, and m* is the effective massof holes. When the energy difference between the levels,
E2 � E1, is larger than the thermal energy kBT =
0.026eV, i.e.,
3� �h2
2m�pl
� �2
P 0:026� 1:6� 10�19; ð2Þ
the NDC phenomenon could be observed at room tem-
perature. Using m* = 9.31 · 10�32 [11], we have
l � 8nm. The obtained well width is consistent with
the dimension of Leonard and Tersoff’s simulated car-
bon nanotube devices which show a strong room tem-
perature NDC [12].The shift of the peak voltage as a function of Vg can
be attributed to the variation of the conductance of the
SWNT bundle and the shift of the En with respect to the
Fermi level in the quantum well. Vds should drop on
the contacts between the source and drain electrodes
and SWNTs (Vc), the quantum well (Vp) and the
SWNTs beyond the quantum well (Vs), i.e., Vds = Vc +
Vp + Vs. Because of a large contact resistance [7] and
large SWNT (channel) resistance, Vp could be quite
small. Since the channel resistance decreases with
decreasing gate voltage from 6V to �6V for a p-type
CNTFET, more voltage will drop on the quantum well
at a given Vds. Therefore, Vds for triggering the resonant
tunneling decreases with the gate voltage changing from6 to �6V, resulting in the down shift of the peak volt-
age, as observed in Fig. 2(b). The peak current could
be constrained by the tunneling process so that it may
not be significantly influenced by the variation of the
channel resistance or Vg. On the other hand, the up-shift
of En with respect to the Fermi level induced by decreas-
ing Vg [13] would also lead to a decrease in the peak
voltage.Taking the burning process into account, we suggest
two possibilities for the formation of the quantum well.
Local depletion of adsorbed oxygen on the s-SWNT
walls could be one of the reasons. The significant drops
of Id in Fig. 1(b) indicate burning-off of the m-SWNTs
in the bundle [9,14]. We suggest that the tube-burning
mainly occurs in a small area far from the source and
drain so that the rest parts of the broken m-SWNTs stillkeep contact with the metal electrodes, as observed by
Collions et al. [9,14]. At the burning sites, oxygen that
initially adsorbed on s-SWNTs could react with carbon
atoms during the tube burning process and some rem-
nants [15] of the burned m-SWNTs might leave behind
on the surface of s-SWNTs. The remnants prevent oxy-
gen in air from being adsorbed again. Theoretic study
[16,17] has suggested that the chemisorption of oxygencould drastically decrease the energy band gap from
0.99 to 0.38eV for a (8, 4) SWNT. A reasonable specu-
lation is that the band gap of a SWNT could be strongly
associated with the amount of adsorbed oxygen on the s-
SWNT wall. Therefore, the s-SWNTs covered with the
remnants could form local energy barriers for hole
transport. In addition, variation of the diameter of the
SWNT bundle could be another reason. According toReich et al. [18], interaction between SWNTs in the bun-
dle could lead to a decrease in the energy gap of the
semiconducting SWNTs. At the tube-burning sites, the
tube number should be small due to the burning-off of
the m-SWNTs (the local bundle diameter should be
thinned, as observed by Collions et al. [9]) and the en-
ergy gap at the sites could be larger than the rest of
the bundle, resulting in energy barriers for a quantumwell.
In summary, room-temperature negative differential
conductance phenomenon has been observed in a bun-
dle of SWNTs in a CNTFET. The peak-to-valley cur-
rent ratio is about 2. This phenomenon is attributed to
a quantum well between two adjacent energy barriers
formed on semiconducting SWNTs due to the burning
off metallic SWNTs in the bundle. The conductance var-iation of the SWNTs and the shift of the energy level in
the quantum well are suggested to be responsible for the
EF (a)
(b)
(c)
E1
E1
E1
S D
S
S
D
D
Fig. 3. (a)–(c) Bias voltage dependent band diagrams for the points A,
B, and C on the Id–Vds curve in Fig. 2(a). Only the valence band is
shown here. The dashed lines are the Fermi level EF. S and D represent
the source and drain, respectively.
Letters to the Editor / Carbon 43 (2005) 651–673 669
peak voltage shifts from 9.2V to 7.8V with decreasing
the gate voltage from 6V to �6V.
References
[1] Zhou C, Kong J, Yenilmez E, Dai H. Modulated chemical doping
of individual carbon nanotubes. Science 2000;290:1552–5.
[2] Kong J, Cao J, Dai H, Anderson E. Chemical profiling of single
nanotubes: Intramolecular p–n–p junctions and on-tube single
electron transistors. Appl Phys Lett 2002;80(1):73–5.
[3] Kong J, Dai H. Full and modulated chemical gating of individual
carbon nanotubes by organic Amine compounds. J Phys Chem
2001;105:2890–3.
[4] Postma H, Teepen T, Yao Z, Grifoni M, Dekker C. Carbon
nanotube single-electron transistors at room temperature. Science
2001;293:76–9.
[5] Bozovic D, Bockrath M, Hafner J, Lieber C, Park H, Tinkham
M. Electronic properties of mechanically induced kinks in single-
walled carbon nanotubes. Appl Phys Lett 2001;78(23):3693–5.
[6] Tan S, Verschueren A, Dekker C. Room temperature transistor
based on a single carbon nanotube. Nature (London)
1998;393:49–52.
[7] Martel R, Schmidt T, Shea HR, Hertel T, Avouris Ph. Single- and
multi-wall carbon nanotube transistors. Appl Phys Lett
1998;73(17):2447–9.
[8] Li J, Zhang Q, Yang D, Tian J. Fabrication of carbon nanotube
field effect transistors by AC dielectrophoresis method. Carbon
2004;42(11):2263–7.
[9] Collins PG, Arnold MS, Avouris Ph. Engineering carbon nano-
tubes and nanotube circuits using electrical breakdown. Science
2001;292:706–9.
[10] Mendez EE, Wang WI, Ricco B, Esaki L. Resonant tunneling of
holes in AIAs-GaAs-AIAs heterostructures. Appl Phys Lett
1985;47(4):415–7.
[11] Bradely K, Jhi S, Collions PG, Hone J, Cohen ML, Louie S, Zettl
A. Is the intrinsic thermopower of carbon nanotubes positive?
Phys Rev Lett 2001;85(20):4361–4.
[12] Leonard F, Tersoff J. Multiple functionality in nanotube transis-
tors. Phys Rev Lett 2002;88(25):258302-1–258302-4.
[13] Luryi S, Capasso F. Resonant tunneling of two-dimensional
electrons through a quantum wire: A negative transconductance
device. Appl Phys Lett 1985;47(12):1347–9.
[14] Collions PG, Hersam M, Arnold M, Martel R, Avouris Ph.
Current saturation and electrical breakdown in carbon nanotubes.
Phys Rev Lett 2001;86(14):3128–31.
[15] Yao N, Lordi V, Ma SXC, Dujardin E, Krishnan A, Treacy
MMJ, Ebbesen TW. Structure and oxidation patterns of carbon
nanotubes. J Mater Res 1998;13(9):2432–7.
[16] Barone V, Heyd J, Scuseria GE. Effect of oxygen chemisorp-
tion on the energy band gap of a chiral semiconducting
single-walled carbon nanotube. Chem Phys Lett 2004;389:
289–92.
[17] Chan S, Chen G, Gong XG, Liu Z. Oxidation of carbon
nanotubes by singlet O2. Phys Rev Lett 2003;90:086403-1–
086403-4.
[18] Reich S, Thomsen C, Ordejon P. Electronic band structure of
isolated and bundled carbon nanotubes. Phys Rev B
2002;65:155411-1–155411-11.
Sonication assisted deposition of Cu2O nanoparticles onmultiwall carbon nanotubes with polyol process
Ying Yu a,b, Li-Li Ma a, Wen-Ya Huang a, Fei-Peng Du a, Jimmy C. Yu c,Jia-Guo Yu c, Jian-Bo Wang d, Po-Keung Wong b,*
a College of Physical Science and Technology, Central China Normal University, Wuhan 430079, Chinab Department of Biology, The Chinese University of Hong Kong, Shatin, NT, Hong Kong, SAR China
c Department of Chemistry, The Chinese University of Hong Kong, Shatin, NT, Hong Kong, SAR Chinad Center for Electron Microscopy, Wuhan University, Wuhan 430072, China
Received 23 July 2004; accepted 18 October 2004
Available online 25 November 2004
Keywords: A. Carbon nanotubes; B. Chemical treatment; C. Electron microscopy; D. Microstructure
Due to the special structure, the extraordinary
mechanical and unique electronic properties and the po-
tential applications, carbon nanotubes (CNTs) have
attracted considerable attention since they were discov-
ered. To generate new functionalities, it is necessary to
attach other nanostructures to the surface of CNTs.
The combination of carbon nanotubes and nanocrys-
tals are useful in field emission displays, nanoelectronic
devices, novel catalysts and polymer or ceramic rein-
forcement [1]. The coating of CNTs with metals, inor-
ganic compounds and polymers has been reported
0008-6223/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2004.10.031
*Corresponding author. Tel.: +852 2609 6783; fax: +852 2603 5767.
E-mail address: [email protected] (P.-K. Wong).
670 Letters to the Editor / Carbon 43 (2005) 651–673