Microelectronic Engineering 46 (1999) 101-104

Manipulation of Carbon Nanotubes and Properties of Nanotube Field-Effect Transistors and Rings

IvIICE:)~-~’CTIROlt~

H.R. Shea, R. Martel, T. Hertel, T. Schmidt, and Ph. Avouris*

I B M Research Division, T.J. Watson Center, P.O. Box 218, Yorktown Heights, NY 10598, U.S.A.

Using the tip o f an atomic force microscope, we have manipulated individual carbon nanotubes on a patterned substrate, and have fabricated model nanodevices, including a room temperature field-effect transistor with a channel only 1.6 nm wide, as well as single-electron transistors. W e have also developed a technique to produce rings o f single-wall nanotubes with a very high yield.

1. I N T R O D U C T I O N

Carbon nanotubes (NT) are a novel class of nanostructures consisting of one or several graphene sheets rolled into a single seamless hollow cylinder, or several concentric cylinders, respectively. Depending on the width o f the graphene sheet and the angle at which it is rolled, the N T can be either a semiconductor or a metal. In addition to these intriguing electrical properties, nanotubes have a very high tensile strength and are extremely rigid. Because o f these characteristics, a great number o f applications for NTs have been proposed, such as 1D nanowires and switching elements in nano- devices [1].

In this paper we discuss first the manipulation o f nanotubes on a patterned substrate, then field-effect transistors (FETs) and single-electron transistors (SETs) in which nanotubes are the channel. Finally we demonstrate the capability o f making rings o f nanotubes.

2. M A N I P U L A T I O N

There is a strong Van der Waals attraction between nanotubes, and between nanotubes and the substrate they are deposited upon. Because o f the high binding energy with the substrate (~0.8 eV//~ for a 1 0 0 / ~ diameter multi-wall tube [2]) the tube can be pinned in a highly strained (bent) configuration despite its high Young’s modulus (~1 TPa). W e can manipulate the N T position at room temperature by applying lateral forces with the tip o f

an atomic force microscope (AFM). W e found the shear stress on surfaces such as H-passivated silicon is high, of the order o f 10 7 N/m, such that not only the position but also the shape o f the N T can be controlled [2].

a ) – ~ { ~ , ” b ) , , :i:: ::’

‘::ii~ :~ilili! ~i ~

Fig. 1: AFM manipulation o f a single multi-wall nanotube. Initially the N T is located on an insulating part o f the sample. In a stepwise fashion, it is dragged onto the 80 A high metal thin film wire, and finally stretched across the oxide barrier.

0167-9317/99/$ – see front matter 1999 Elsevier Science B.V. All fights reserved. PII: S 0 1 6 7 – 9 3 1 7 ( 9 9 ) 0 0 0 2 5 – 8

I02 H.R. Shea et al. / Microelectronic Engineering 46 (1999) 101-104

To perform the manipulation, we alternate between the non-contact mode of the AFM for imaging without moving the tubes, and the contact mode for displacing the tubes. While the forces in non-contact mode are very small (in the pN range), vertical loads of 10-50 nN are necessary to move the tube.

Fig. 1 is a sequence of manipulation steps to position a multi-wall NT over an oxide barrier. The tube is moved from an insulating substrate (SiO2) onto a tungsten thin film wire (~80 high) and then stretched across an insulating WOx barrier. Such a setup enables us to perform transport measurements of the NT.

3. F I E L D – E F F E C T T R A N S I S T O R

We have used both single-wall and multi-wall nanotubes (SWNTs and MWNTs) as the channel of an FET. The device consists of a single nanotube bridging two Au electrodes deposited on a 140 nm thick gate oxide film over a doped Si wafer, which is used as a back gate. The source-drain current Isd between the electrodes was measured as a function of source-drain voltage V~d and gate-source voltage Vg.

Nanotube

50

4O

20

10

0

– i | i i | i i I –

V ~ = ] O 0 m V 10-61 – , , , ‘ ‘ 1

i,~O~o ~” Io-~I- t 4

I0-11L , , , I J – ~ . . . ~ l b 2 ~ – 2 0 2 4 6

~ ~ vo {v)

-6 -4 -2 0 2 4 6 8

vQ (v)

Fig. 2: A on a single nanotube.

The top of Fig. 2 is a schematic diagram of the nanotube FET. The lower section shows the room temperature transfer characteristics Isd-V~ for several values of Vsa for a 1.6 nm diameter SWNT. We observe clear transistor action. For Vg<0 the Isd-V~d curves (not shown) are linear, but become increasingly non-linear for Vg>0, up to a point where the current becomes immeasurably small for Vg>>0. The inset in Fig. 2 shows the FET conductance vs. Vg: it can be modulated by over 5 orders of magnitude.

The nanotube FET behaves very much like a p- channel MOSFET, and transport is dominated by holes. The saturation at negative Vg is in part due to the large (~ 1 MD) contact resistance.

A key issue is the origin of the holes: one possibility is that the carrier concentration is inherent to the nanotube. Another possibility is that most of the holes are injected at the gold-NT contact because of the difference in work function between the two materials, as suggested by Tans et al. [3]. Our results favor the first explanation, and we determine the 1D hole density to be 9×106 cm l . This value corresponds to 1 hole per 250 carbon atoms, suggesting the tube is degenerate and/or highly doped, perhaps as a results of its processing [4].

Devices made with MWNTs rather than SWNTs typically exhibit no gate action. However, we have successfully used deformed MWNTs as the channel of a room temperature FET, although with a conductance modulation only of order 2 [4].

4. S I N G L E – E L E C T R O N T R A N S I S T O R

Because of their very small size, and the high contact resistance, our FET devices exhibit Coulomb blockade behavior at low temperatures, when the charging energy e2/2C to add an electron to the tube is much larger than kBT, where C is the capacitance of the tube, and T temperature [5]. Fig. 3 is a plot of Isa vs. Vg of a bundle of SWNT draped over two gold electrodes 0.4 gm apart at 290 K, 77 K, and 4 K for Vsa = 20 mV, 50 mV, and 5 mV, respectively. The insets are Isd-Vsa curves at selected gate voltages.

The 290 K data show FET behavior and linear I- V curves, with a minimum device resistance of 250 kD. At 4 K, the I-V curves have a pronounced Coulomb gap, and there is very sharp nonmonotonic structure in I~d vs. Vg corresponding to the lining up of states in the tube with the Fermi level of one of

H.R. Shea et al. I Microelectronic Engineering 46 (1999) 101-104 103

8 0 ‘ . ~ ;.’3 ~ ‘ ‘ ~’~’~’6~ 5 ~ , 4 , , ~ < . iTemperature = 290K Jl

8I-~’E :: / ‘ l ~ – , i

4 0 o 4 0 ~ , . . I V s d ( I m V ) , I I ” I I ~

-10 -5 0 5 10 Gate Voltage (V)

100J – I /~ ET’emperatur’e = 77K I ‘ l- ‘I . , – ” A i v 8 0 ~ 40

” -4o I 1-I 40~– I ~ / ‘ ~ – I’AI~ – , I , l [ H

, v11 ,o,oo 4o I 2 ~mV) O -20 -15 -10 -5 0 5

Gate Voltage (V) ‘ b I ‘ ,

, v , , , ,, 20 -40 o 40

‘ I I V~d ( m Y ) 0 I .U,,, -20 -16 -12 -8

Gate Voltage (V)

Fig. 3: Isd-Vg curves for a bundle o f S W N T device at 290 K, 77 K, and 4.2 K. The device behaves like an FET at 290 K, and like an SET at 4.2 K.

the leads. At 4 K, the device acts as a single-electron transistor. The 77 K data show features due to both FET and Coulomb blockade behavior.

Data at 4 K from other of our S W N T devices also show clear structure corresponding to adding one electron at a time to the tube. Each diamond in the grayscale dI/dVsd vs. Vg and Vsd plots, as shown in Fig. 4a, corresponds to changing the number o f electrons on the tube by one. For a single metallic island, the diamonds all have the same shape. In our nanotube SETs, the Coulomb gap oscillates aperiodically with Vg, which is suggestive of single- electron transport through multiple Coulomb islands formed within the N T bundle.

The depletion o f carriers in the tube, and the associated decrease in tube capacitance, are clearly visible in Fig. 4b as an increase in the width o f the blockade region at large positive gate voltage.

It has been observed by several groups that only a small section o f some nanotubes act as a Coulomb

island [6]. One can determine the charging energy o f a length o f N T by calculating its capacitance. Alternatively, one can infer the length o f the part o f the tube that is charging from the measured charging energy. We always observe that the length o f N T that is charging as determined from the measured charging energy is less than the spacing between the electrodes by a factor o f between 1.5 and 20. This behavior is probably attributable to multiple islands formed within a N T bundle, and possibly also to barriers formed along the tube due to bending o f the tube [6,7].

a)

>~

b)

3 –

2 –

1

0

– 1

-2

-3

-10 -5 0 5 10

V,o (mV)

>~

– 2 0 0 – 1 0 0 0 1 0 0 2 0 0

% (mY)

Fig. 4: Grayscale dI/dVsd vs. Vg and Vsd plots o f tWO different single S W N T devices at 4 K. Lighter gray corresponds to lower conductance.

5. R I N G S O F N A N O T U B E S

Because o f the strong van der Waals adhesion force between tubes, it is possible to form well defined rings (coils) o f single-wall nanotubes, where

104 H.R. Shea et al. / Microelectronic Engineering 46 (1999) 101-104

the strain energy to bend the tubes (proportional to 1/r 2, where r is the radius of curvature) is more than offset by the bonding energy o f the tube to itself (proportional to the length of overlap of the tube with itself). W e have developed a technique that forms rings with a ~50% yield starting with a solution of straight bundles o f SWNTs. The process is kinetically limited, and the average ring radius is 350 nm for bundles 3-4 g m in length. From TEM and AFM images, we find that the rings walls have a radius of between 5 and 30 nm.

Fig. 5: SEM micrograph of .

Fig. 6: AFM micrograph of a N T ring draped over three gold electrodes.

The SEM image in Fig. 5 illustrates the high yield o f rings produced by our procedure. The AFM micrograph in Fig. 6 shows an N T ring deposited over three gold electrodes. Preliminary measure- ments o f the temperature dependence o f the ring conductance indicate the ring is metallic, with resistance decreasing from 15 k D at 4 K to 10 kf2 at 290 K. We see no charging effects even at 4 K.

As shown in Fig. 7, we observe conductance fluctuations o f order e2/h in the ring at 4 K. Further analysis is required to confirm the universal nature o f these fluctuations. Should the conductance fluc- tuations be universal, the data then suggest that the phase coherence length o f the nanotubes is only slightly smaller than the ring dimension.

1.5

o?, 1.4

” o c 1.3 o o

1.2

. . . . ! . . . . i . . . . i . . . .

. . , a i . . . . i . . . . I .

-15 -10 -5 0 5

vg (v)

Fig. 7: Ring conductance in units of e2]h vs. Vg at 4 K for Vsd = 1 and 5 mV, showing conductance fluctuations o f a probably universal character.

We thank A.G. Rinzler, R.E. Smalley, and H. Dai for providing us with the single- and multi-wall nanotubes.

REFERENCES

* e-mail address:[email protected] M.S. Dresselhaus, G. Dresselhaus, and P.C.

Eklund, Science o f Fullerenes and Carbon Nanotubes, Academic Press, San Diego, 1996

2. T. Hertel, R. Martel, and Ph. Avouris, J. Phys. Chem. B 102 (1998) 910

3. S. J. Tans, A.R.M. Verschueren, and C. Dekker, Nature 393 (1998) 49

4. R. Martel, T. Schmidt, H. R. Shea, T. Hertel, and Ph. Avouris, in press, Appl. Phys. Lett., Oct.1998

5. H. Grabert and M.H. Devoret, eds., Single Charge Tunneling: Coulomb Blockade Pheno- mena in Nano-structures, N A T O ASI Series, vol. 294, Plenum Press, N Y 1992

6. A. Bezryadin, A.R.M. Verschueren, S.J. Tans, C. Dekker, Phys. Rev. Lett. 80, (1998) 4036

7. A. Rochefort, D.R. Salahub, and Ph. Avouris, Chem. Phys. Lett., in press