Carbon nanotube electronics

Properties of CNT
Quasi 1-D structure
Ballistic transport

All chemical bond of the C atoms

No need for chemical passivation

Strong covalent bond

High mechanical and thermal stability
Resistance to electromigration 109A/cm2

Controlled by chemistry
Ideally, active device and interconnect from CNTs is possible

Chirality dependent

2.46 n1 + n1n2 + n2
d=
2
2

Metallic: 2n1+n2=3q
Armchair structure always metallic

Band Structure
a) Graphite
Valence() and Conduction(*) states touch
at six points(Fermi points)
Carbon Nanotube: Quantization from the confinement of
electrons in the circumferential direction
b) (3,3) CNT
; allowed energy states of CNT cuts pass through Fermi
point metallic
c) (4,2) CNT
; no cut passes through a K point semiconducting
Semiconduction CNT band gap

4hvF
Egap =
3d CNT

Resistance of CNT
Landauers Equation

2e 2
N
G=
i Ti
h
2e2/h : Quantum of conductance
Ti : Transmission of a conducting conduction channel(subband)
The resistance of metallic CNT
N = 2, Ti =1

R=

1
h
= 2 6 .5 K
G 4e

Contact resistance due to the mismatch of the number of conduction

channels in the CNT and the macroscopic metal leads

CNT TEF 1(1998)

Single wall CNT
Diameter ~ 1.6nm
P-type FET
I(on)/I(off)? 105
High parastic resistance 1M
Low drive current few nano ampere
Low transconductance ~ 1nS
High inverse subthreshold slope S ~ 1-2V/dec
BAD CONTACT :
weak van der walls force(SWCNT-Au)

How to reduce contact resistance?

CNT first then electrode deposition, annealing
TiC : stronger coupling

CNT FET 1
Vg = 0V: linear; 2.9M
Vg < 0V: linear
Vg >>0V : non linear
Current decreases with increasing gate voltage
1. shows field effect transistor
2. majority carrier : holes

RT = RNT + 2 RC

VG< 0V: current saturation

contact resistance dominating(~ 1.1M)

CNT FET 2(2001)

Co electrode and annealed
P-type
High on-off ratio ~ 10 6
Low contact resistance ~ 30k
Higher current range ~ A
Transconductance ~ 0.34 S
higher than previous CNTFET

To improve performance gate cap. should be increased

High k or thin oxide even individual gating

CNT FET 3(top gate)(2002)

Ti electrode: TiC formation
Stronger coupling
Tox ~ 15-20nm
Vth ~ -0.5V(back gated -12V)
Transconductance = 3.3 S

Metal-CNT contact

Interchanging of source and drain

: non symmetric I-V
Saturation current should be same if operation
is dictated by bulk of CNT
Dominated by barrier (Schottky barrier)

Ohmic vs. Schottky

Evac

Evac
EF

Evac
EF

EF
M < +Ec-EF = S
For n-type
semiconductor.

Work Function(CNT)
MWCNT ~4.9eV
SWCNT ~3.7eV

M > +Ec-EF = S
Work Function(Metal)
Ni 5.2eV, Ti 4.3eV,
Ag 4.3eV, Au 5.2eV

For n-type semiconductor

Very high density of surface states
pinning the Fermi level at the surface
w.r.t. the conduction band (Example:
GaAs)

Conduction Mechanism
Thermionic emission
Electrons emit over the barrier
Low probability of direct tunneling
Valid for low doping (ND < ~ 1017 cm-3)

Thermionic-field emission
Electrons use thermal energy to tunnel trough the
thin barrier in the upper end of the conduction
band
Valid for intermediate doping (~ 1017 cm-3 < ND <
~ 1018 cm-3)

Field emission
Direct tunneling, as depletion region is very narrow
Valid for heavy doping (ND > ~ 1018 cm-3); almost
ohmic

SB-FET
Standard FET, conductance is controlled by the
electrostatic potential in the channel
increase capacitance
SB-FET, performance is controlled by
electric field at the contact

Doping

Oxygen doping
Ambipolar property

Potassium doping
Non Ambipolar property

But two cases show contrast of dopants

Transport and switching is controlled by the Schottky
barrier at the contacts

SB-FET

a) Work function change of metal electrode

b) The charge on tube plays main role
The principle effect of oxygen exposure is not to dope CNT body
but change the work function of exposed portion of the metal electrode

Schottky barrier
S = ln 10

dVG
k T
C
B ln 10(1 + D )
d (ln I D )
q
CG

Fully depleted device, Cd = 0

CNT has no depletion capacitance
because of no charge variation
across tube circumference
Inverse subthreshold slope only
depends on temp ~60mV/dec at
300k
Below 200k S is almost constant not
temp. dependent

Bundle destruction

CNTs grown from arc discharge or Hipco form

rope or bundle
: How do we select semiconducting or
metallic?
Separation
CONSTRUCTIVE DESTRUCTION

Shell Breakdown
Vds increase: increase the avg. energy of carrier
coupling between shells
Higher electron energy
energy dissipation
breakdown
Less breakdown voltage in air than vacuum
oxidation
Band gap of CNT can be controlled by
electrical breakdown
Egap ~ 1/dcnt

CNT Logic Gate (inverter)

Initially both CNTFETs were p-type

FET
Annealing in vacuum make n-type
Expose O2
O2 exposure makes p-type FET

From single CNT

PMMA coating and e-beam litho.
Potassium doping n-type FET

Selective Local growth

Heavily doped Si layer and Si3N4
Electrolytical etching Si to form porous Si on sidewall
PR protecting and catalyst deposition
CNT growth

Conclusion
Pros
High current density and stability of metallic CNT for intercon
nect in nano devices
Semiconducting CNT shows better performance than state of
the art Si FET
Cons
Fabricating single type of CNT
Seeded growth(?), separation
Self assembly

Environmental sensitivity
Good candidate for sensor