py4p04 1-d 2013
DESCRIPTION
dank nanoscience memesTRANSCRIPT
• One Dimensional Materials
• Electronic structure in 1D
• Electrical transport in 1D
• Most of this stuff is elementary quantum mechanics:
• Variations on infinite potential well
• Quantum mechanics, SM McMurry
• Quantum transport, Supriyo Datta
• Intro to Solid State Physics (8th Edition) C Kittel, (with PMcEuen)
Ch 18
PY4P04 Nanoscience: One Dimensional
Nanostructures
• What is a 1D Material?
• Strictly speaking a 1D material is one where electrons have
freedom to move in 1D ONLY.
• Ie a perfectly linear array of atoms
• However such structures are virtually non-existent
• In reality we have quasi 1D materials
• These are materials that are small but finite in 2 dimensions
• Take a 3D structure and shrink its size in the x, and then the y
direction
x z
y
Lx
Ly
Lx, Ly< 100 nm
• This can be done in 2 ways
• Top down by etching bulk semiconductors
• OR Bottom up
• This means building up one dimensional structures atom
by atom
We can do this directly by moving
atoms with a scanning tunnelling
microscope, but only at small scales
because it is just not practical!
We need to convince atoms to
arrange themselves into useful
structures!!!!!!Self assembly
Types of quasi 1D material
• There are a
wide range of
these materials
available today
ZnO nanowires
grown through
evaporation of zinc
powders
Gold nanowires
grown by self
assembly of Au
clusters templated
by surface terraces
ZnO nano-swords
Ni nanowires
templated by Alumina
membranes
CdSe nanowires grown in solution
Also 1D versions or “normal semiconductors”
Si nanowires grown by chemical
vapour deposition
Si nanowires grown by molecular
beam epitaxy
In 1991 carbon tubes were found in a machine built to make C60: Carbon Nanotubes
Most common 1D material has to be the
carbon nanotube
Roll up graphene sheets
Graphite consists of Graphene sheets that
are held together by van der Waals forces.
Each graphene sheet is formed from sp2
bonded carbon atoms in a hexagonal array
A graphene sheet can be rolled up to form a
single walled nanotube
How are they formed?
http://www.nanoonline.net/articles/ swntproperties.htm
A graphene sheet can be rolled up into a cylinder, diameter D in an
infinite number of ways.
To make the nanotube, take the strip defined by the dotted lines and
roll around the direction defined by T, such that DB and AO
The ends of the nanotube are then capped by hemispherical carbon
cages
A nanotube is rolled around an axis ie [OB]
D
A given nanotube can be
identified uniquely by its
circumference vector (red), C.
This vector is at an angle to the
a1 unit vector. is the chiral angle
C is specified by how many unit
vectors in the a1 and a2 directions
it equals:
Here
This is represented as (4,2)
The vector T represents the
length of the nanotubes unit cell.
21ˆ2ˆ4ˆ aaC
21ˆˆˆ anamC
Roll-up vectors
Nanotube type
Rolling up a nanotube like this gives a specific nanotube specified by
(n,m)
Nanotubes are characterised according to 3 basic types:
Armchair tubes: n=m, =30o
Zigzag tubes: m or n =0, =0o
Chiral tubes: All others, 0o < < 30o
Armchair: (m,n)=(8,8)
Zigzag: (m,n)=(10,0)
a1
a2
(5,0)
(5,5)
(4,3)
(2,2)
120o
223mnmnad cc
length C-C
nm14.0 cca
mn
m
2
3tan 1
NB does not uniquely
specify the nanotube
Diameters: 0.7 < D < 1.5 nm Lengths: 0.2 < L < 10,000 m
Tube types are often specified by a
diagram like this one.
Each tube type is represented by a
(non-green) dot
The circumference vector goes from
the green dot to the dot representing
the tube type.
m and n can be counted off
We can easily calculate the nanotube diameter and chiral angle
for a given (n,m) [cosine rule]
1 2 3 CCa a a
• Can also get Multi-walled
nanotubes (MWNT)
• Concentric cylinders of graphene
separated by 0.35 nm
• Each separate shell tends to have a
different (n,m)
Nanotube Synthesis
carbon + catalyst + energy => nanotubes)
I. Arc vapourisation of a catalyst-containing electrode
II. Laser ablation of a catalyst-containing target
III. Decomposition of vapour phase catalyst/carbon
mixtures at high temperature (a.k.a. CVD - chemical
vapour deposition)
Most commom type of CVD process:
Recently developed commercial process - ‘HiPco’ (High
Pressure Carbon monOxide (CO)).
Proceeds via decomposition of acetylene and Fe(CO)5 under
conditions of high pressure and temperature in the presence
of excess CO.
I. Apparatus for arc vapourisation
synthesis of SWNTs showing
enlarged hollow catalyst-containing
anode.
II. Schematic depiction of laser ablation
apparatus.
III. CVD apparatus for forming
nanotubes via decomposition of
acetylene and Fe(CO)5 using argon as
a carrier gas.
Synthesis apparatus
(A)SWNT imaged by transmission electron microscopy (TEM).
(B)TEM image of a MWNT made up of 10 concentric SWNT cylinders
(C) and (D) TEM and scanning electron microscope (SEM) images of SWNT raw
material.
(E) and (F) TEM images of SWNT bundles. Note the ordered 2D lattice in (F).
(G) and (I) STM images of discrete and bundled SWNTs.
Microscopy
of carbon
nanotubes
SWNT
MWNT
Electronic structure
Nanotubes can be metallic
or semiconducting
depending on (n,m)!!!
If
Where p is an integer, then
the nanotube is metallic: 1/3
tubes metallic
Otherwise the NT is
semiconducting with
bandgap Eg.
Eg depends on diameter, D.
pnm
3
D
aE cc
g
2 2.7 eV
Ouyang M et al. Science 292 (2001) 702.
E-Ef (eV)
• When you apply a force to a material it stretches.
• We usually write the stress, and the extension, as
Mechanical Properties 101
L0
A
F
L0+L
A
A
F
0
L
L
The stiffness, or resistance to stretching, is
given by the Youngs Modulus, Y or E.
The strength, F, is the stress at fracture.
The ductility is the strain at fracture, F
ddY 0.00 0.02 0.04 0.060
2x1010
4x1010
6x1010
F
(P
a)
Y
F
F
FY
For a brittle material:
Y
• How strong are nanotubes?
• We need to work out the force per area required to break
all the bonds around the circumference of a nanotube.
acc
2accSin60
There are N bonds around a circumference
602 Sina
DN
cc
The total force to break the NT is FT:
FCC is the force to break 1 C-C bond CCT NFF
The total stress to break the NT
(strength) is A
NFCCF
A is the area of
the bonds to be
broken
Where d is the
thickness of the
NT wall
DdA
3 3
CC CCF
cc cc
F FD
Dda a d
• What is the force required to break the C-C bond?
When a Force, FCC, stretches a bond by lF just before breaking,
The work done is the bond energy, ECC: Fcccc lFE
/
3
CC FF
cc
E l
a d
F F ccl a
• For a C-C bond, Ecc ~3105J/mol or
510-19J per bond, F~20%
• acc=0.14 nm and d=0.35 nm
This gives our estimate as
F~200 GPa
Cf High Strength Steel Wire: F=2.4 GPa, Y=210 GPa
23
CCF
F cc
E
a d
1000F
F
Y GPa
,
0,
NT FFF
cc NT
Ll
a L
We assume the strain in
one bond is the same as
the overall strain at failure
Yu et al Science, 287, 637 (2000)
Y F
Nanotubes ~ 1000 GPa ~ 100 GPa Carbon fibre 200-700 GPa 1.5-5 GPa
High Strength Steel Wire 210 GPa 2.4 GPa
The mechanical properties can be measured in a modified electron microscope
They are the strongest materials known to man!!!!!!!
The stress strain curves are linear up to fracture indicating a brittle
material. We can measure Y and F. They are close to our estimate.
Current carrying ability
712 10 /NT
lS m
R A
76 10 /Ag S m
However Jmax=21010 A/cm2!!
This is about 100 times better than
copper, silver or gold
This is because electrons in
nanotubes don’t scatter Joule
heating is very low
MWNT
Wei BJ, APL, 2001, 350, 6-14
Optical properties
of nanotubes
The optical properties of SWNTs are
dominated by the spikes in the DOS.
Photons with energy E>0.8 eV (< 1500nm)
can excite electrons from the valence band
to the conduction band
For nanotubes, this means VB and CB for
the same subband
Fermi’s golden rule tells us that the
absorption co-efficient is proportional to the
DOS at the energy the photon brings the
electron to: (E) g(E)
This means the absorption is large for
transitions between spikes
Called S11, S22 etc transitions
S11 S22
Thus the absorption spectrum for one tube would be a
set of spikes.
The lowest energy spike would be at an energy equal
to the energy difference between the first subbands of
the VB and CB
The second lowest energy spike would be at an
energy equal to the energy difference between the
second subbands of the VB and CB etc
S11 S22
HOWEVER, no sample ever has one
tube type.
They always have a range of (n,m)
Thus you get an array of sets of spikes
0.5 1.0 1.5 2.0 2.5 3.0 3.50.00
0.05
0.10
0.15
0.20
Absorb
ance (
au)
Photon energy (eV)
Nanotubes
S11 S22
M11
Red:S11 Green, S22 Blue, M11
What happens to the excited electronic state after an electron is
promoted to the conduction band?
S11 S22
Semiconducting nanotubes can photoluminesce!
For about 1% of the photons absorbed, the electron
and hole recombine by emitting a photon
Before emission of a photon, the electron (and
hole) decay to the bottom (top) of the lowest
subband by emission of phonons.
Thus for any semiconducting SWNT, the emission
energy is the same as the S11 absorption energy
For a set of nanotubes, the emission peaks occur
at the same energy as the S11 absorption peaks,
no matter what energy the sample is excited at.
Photoluminescence Excitation
The Pl intensity is proportional to the absorption co-efficient, , of the emitting
tube type. (more photons absorbed: more photons emitted)
We can measure the Pl intensity at one energy, say 1 eV while scanning the
excitation wavelength from 1-3 eV, for example.
The Pl will scale in proportion to the absorbance.
This will give a PLE, or excitation spectrum. For a simple material:
However, if a material has 2 or more emitting components, we can detect at
the emission energy for one component only. Scanning the excitation
wavelength gives us the absorbance spectrum of that component alone.
We can measure the absorption spectrum of an individual tube type in a
mixture!
PLEI
500 1000 15000.0
0.2
0.4
Ab
so
rba
nce
Wavelength [nm]
PLE spectrum of a nanotube sample gives the absorption coefficient of 1
nanotube only
In fact there is an even better way to pick out both the absorption and
emission of individual nanotubes.
Make photoluminescence measurements on a mixture of nanotubes, usually
dispersed using surfactants.
The difference is that instead of measuring one PL spectrum with a fixed
excitation energy, say 2 eV, many spectra are measured with many
excitation energies ie 1.05 eV, 1.1 eV, 1.15 eV……..2.9 eV, 2.95 eV, 3.0 eV
Absorption
PLE S22
S33
The beauty of this is that we have data on:
1 The emission intensity at all emission energies for a given excitation
energy (PL)
2 The emission intensity for one emission energy for all excitation energy
(PLE)
We have this data for all tubes in the sample. This data is generally plotted
as a Pl contour map. Each peak represents PL from a different nanotube
type - Tells you what semiconducting tubes you have in a sample
Can also plot as a 3D intensity profile
Excitation
Wavelength
(nm) Emission energy (eV)
Electronic Properties of 1D structures
Particle in a 1-D Box (infinite square well)
Nanoscience is about systems where one or more dimensions are confined to the nanoscale
The electrons in a material are confined by the strong attraction to the atomic nuclei.
In some materials electron shells are unfilled you can have (nearly) free electrons metals
However the electrons are still trapped within the material
Model this by a potential energy of zero in the material and infinity outside Infinite square well
This confinement results in the formation of quantised energy levels.
Solve Schroedinger equation for 1-D
infinite square well, length L.
Inside the well the potential energy: V(x)=0
x L 0
V
2 2
22V E
m x
2 2
22E
m x
Trial Solution:
What is k? Use Boundary Conditions
ASin kx
2 2 2 2 2 2 2
2 22 2 8n
k n n hE
m m L mL
2n
n xSin
L L
kL n
• What about real systems?
• How about Hexene (C6H8)
• Each C has 3 valance electrons ( electrons) involved in bonding leaving 1 free electron ( electron)
• Treat the electrons (1 from each C) as completely free ie a free electron gas.
• These then fill up the
energy levels (2 per level, )
Alkenes:
0 0.2 0.4 0.6 0.8 1.0
9
4
1
x /L
E (
h2/8
mL
2)
, Ef
C
H
C
H
C
H
C
H
C
H
C
H
H H
a
C
H
C
H
C
H
C
H
C
H
C
H
H H
a
Alkenes in general
• What about alkenes in general? H[CH]NH
There are N free electrons
The length is L=(N-1)a
The Fermi Energy is the energy of the top most filled level
That is: En when n=N/2 (1 e- per C, 2 e- per level)
If N big then: independent of N
C
H
C
H
C
H
C
H
C
H
C
H
H H
a
N=6
2 2 2 2
228 8 1n
n h n hE
mL m N a
2 2 2 2
2 22
( / 2)
32 18 1F
N h N hE
ma Nm N a
2
232F
hE
ma
We can work out the Fermi energy for a long alkene (polyacetylene)
a0.15 nm Ef 4.18 eV above the bottom
of the well
If all the electrons are below Ef 4.18 eV for any chain length, then the
energy levels must get closer together for larger and larger N.
Lets work out the spacing between 2 adjacent levels
2
232F
hE
ma
2 2 2 2
228 8 1n
n h n hE
mL m N a
2 2 22 2
1 2 2 2
1 2 1
8 1 8 1 8 1n n
n h n hn hE E E
m N a m N a m N a
• The spacing increases with n
• Lets work out a representative spacing
ie the energy difference between Ef and the level below.
The Fermi energy is when n=N/2 or N=2n
This gets smaller as N gets bigger!!!!!
The levels get closer together as N !!!!!!!!
2
2
2 1
8 1
n hE
m N a
2 2
2 2
1
8 18 1
N h hE
ma Nm N a
16.7
1
eVE
N
0
5
10
15
0 10 20 30 40 50
N
E
(eV
)
• At some point the levels are so close together it is hard to tell them apart
• They have become a band.
• When does this happen?
• We could imagine that this is when the spacing between the top 2 levels is kT (cf Kubo criterion)
• Then we have or
• At room temp this is N670 or L 670a 100 nm
This suggests that we see energy bands (like in bulk materials) for 1D objects bigger than 100nm
BUT
Quantised energy levels for objects smaller than 100nm
2
28 1
hE kT
ma N
2
21
8
hN N
ma kT
Density of States
• Thus: when we have 1D objects longer than 100 nm we need to
consider energy bands and so think in terms of Density of States.
Total number of states per
unit energy per unit
length including spin
degeneracy
• Where is the Fermi Energy?
We calculate it in exactly the
same way as we would do in 3D
0 1 2 3 4 5
0
5
10
15
20
25
giD
(e
V-1n
m-1
)
E (eV)
E
mEg D
12)(1
• The available states are filled until all electrons are
allocated.
• The energy of the highest state is then the Fermi
Energy.
• The available states that get filled are those with k
values between -kf and +kf.
• We know that
• Where each state is indexed by n
• This means, the amount of k space per state is
2xk L n
2xk
L
2x
nk
L
• So the total number of electrons, N, is the total amount of k space
divided by the amount of k space per electron (times 2 for spin)
Or
• kf is the maximum value of k for an (ground state) electron in our
material (at 0K)
• But remember, for a wave
• Thus is the smallest occupied (at 0K)
electron wavelength
• From above
• If each atom contributes 1 electron then
• And we have
2 2
2 /
fkN
L
2f
Nk
L
2 /k 2 /f fk
4 /f L N
/ 1/N L a
4f a
4a
Back to the Fermi Energy
• As we saw, when the potential energy, V=0
Thus
• As before
2 2
2
kE
m
2 2
2
f
f
kE
m
22
2 2f
NE
m L
22
32f
h NE
m L
2
232f
hE
ma
0 1 2 3 4 5
0
5
10
15
20
25
giD
(e
V-1n
m-1
)
E (eV)
Ef
N/L is the linear free electron density
If each atom contributes 1 free
electron, then N/L=1/a, giving
• Another way to find Ef is to remember that
• This follows from the definition of g1D(E) as the number
of states per unit energy per unit length
• Integrating gives
dEEgL
NfE
D )(0
1
dEEm
E
mdE
L
Nff EE
0
2/1
0
212
fEm
L
N2
2
Rearranging and remembering that
N/L=1/a gives 2
232f
hE
ma
• What about finite temperatures?
• At T>0 it is no longer the case
that all electrons are in states
below Ef and all states above Ef
are empty.
• Thermal occupation
• Then the number of electrons per
unit length in states between E
and E+dE is
• Where
Fermi-Dirac function
1( ) ( )e DN f E g E dE
0 1 2 3 4 50.01
0.1
1
10
f(E
)giD
(E)
(eV
-1n
m-1
)
E (eV)
Ef
1/exp
1)(
kTEEEf
f
4.1 4.2 4.3 0.00
0.05
0.10
0.15
0.20
0.25
f(E
)g iD
(E)
( eV
-1 nm
-1 )
E (eV)
E f
g iD (E)
Peierls Distortion!!!! • Beware!!!!!!
• The example of a long alkene
(polyacetylene) being a metal is WRONG!!
• This is because 1D metals tend to be
unstable
• They can lower their energy by a process
known as bond alternation.
• The array of single bonds distorts into and
array of single and double bonds, doubling
its period.
• There is no longer a free electron per carbon
as they have gone into double bonds
resulting in the formation of a bandgap at Ef
• The extra strain energy is more than
compensated by the energy reduction due to
the formation of the gap
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
C
H
k
E
kf
Ef
k
E
kf
Ef Eg
Everything has some size in the lateral dimensions !!!!!!!!
Even a carbon nanotube has a width (diameter) of at least 0.5nm.
These are known as Quasi 1D materials
How do we model the electronic structure?
We can model a quasi 1D material as a very long rod with a rectangular
cross section
There are virtually no ID materials!!!!!
x
z y
Ly Lx
Lz
V(x)=0 for 0xLx, 0yLy, 0zLz
V = everywhere else set up Schroedinger Eqn
Solving the Schroedinger equation gives the energies of any free
electrons (ie not tightly bound to atoms)
2 2 2 2
2 2 22E
m x y z
( , , ) ( ) ( ) ( )x y zx y z x y z
8
x y z x y z
x y z
Sin k x Sin k y Sin k zL L L
2 2 2 2 2 2
2 2 28 8 8
x y z
x y z
n h m h p hE E E E
mL mL mL
We generally solve this by separation of variables where we assume
the wavefunction is the product of three wavefunctions each
dependent only on x,y or z.
Then the Schroedinger eqn becomes
In 3D the Schroedinger
Eqn, when V is zero, is
Electron
energies:
Remember, in our
quasi 1D material. Lx
is VERY big so the
spacings between the
“x” energy levels are
very small.
2 2 2 2 2 2
2 2 28 8 8
x y z
x y z
n h m h p hE E E E
mL mL mL
As we will see, this
allows us to explain
the presence of
quantum subbands
Model a nanotube as a hollow cylinder with infinitesimally thin
walls
In cylindrical polar co-ords (z=z, x=rCos and y=rSin)
2 2 2 2
2 2 22E
m x y z
2 2 2 2
2 2 2 2
1 1
2E
m r r r r z
2 2 2
2 2 2
1
2E
m R z
r
z R
Lz
• But here r=R = constant so
derivatives w.r.t. r disappear
• As before ( , ) ( ) ( )zz z
2 2
2 2 2 2
1 2( ) ( ) ( ) ( )z z
mEz z
R z
2 2
2 2 2 2
1 1 1 2z
z
mE
R z
As before we separate the variables
2 2
2
2 2
22
2 2
2
2
l
z zz z z
mR Em
mEk
z
NB
We have seen this equation before:
The solution is
zSinkL
z
z
z
2
EEE z
Boundary
conditions give
2
22
8
/
z
z
zz
mL
hnE
Lnk
• What about
• This is slightly different
• Unlike z there is no special value of where is zero
• One solution to the equation above is
(1 e- moving clockwise +
1 e- moving anti-clockwise)
• To find ml we use “periodic bound conditions”. Ie the wavefunction must
be the same if you around the circumference by 1 revolution: ie
2
2
2
2
2 2lm
EmR
ll imim
BeAe
)2()(
22 llll imimimim
BeAeBeAe
22 llllll imimimimimimeBeeAeBeAe
This is only true if ml=0, 1, 2, 3….. 12
lime
• What about the energy?
• We can understand this another way
• The circumference of the tube must be equal to an integral number
of wavelengths of the wavefunction or there will be destructive
interference effects.
2
2
22
EmR
ml 22
22
2
22
82 Rm
mh
mR
mE ll
lmR 2 But for a wave
2k
lmRk
2
2222
22 mR
m
m
kE l
Same as ABOVE
2
22
2
222
,22 mR
m
mL
nEEE l
z
zmn l
• The total energy is
• Remember for a nanotube, Lz is very long
• This means that the energy levels are very close together
2
222
2
22
2
222
2
22
,2222 z
l
z
lmn
L
n
mmR
m
mL
n
mR
mE
l
kz
2
m
kkE z
mzm ll 2)(
22
This means that for every ml we
have a band of energy levels
associated with different kzs
The minimum energy of each band
is then when kz = 0 ie ml
•As Lz gets very big, n/Lz starts to look like a
continuous function
•Ie kz becomes continuous Band
•However R is small so ml2/R2 is big
quantized levels
• This means that there is an energy band associated with every one
of the quantised energy levels for the direction
• The lowest level of each band is ml
• These are known as Quantum subbands.
• Before we said that
• Now, for quasi 1D materials we need to say
• Ie the DOS of the subband scales with the energy above the
associated quantised level.
1( ) ( , )
l
T D l
m
g E g m E•The total density is states is just the sum
over all the subbands (each with different
ml.
For E>ml
For E< ml, g1D=0, for that subband
E
mEg D
12)(1
lm
DE
mEg
12)(1
m
kkE z
mzm ll 2)(
22
2 2
22l
lm
m
mR
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
0
100
200
300
400
500
ml=4ml=3
ml=2
gT(e
V-1nm
-1)
E (eV)
ml=1Calculated low
energy DOS for
a R=1 nm
nanotube
2
22
2mR
mlml
2
22
42
4
mR
2
22
32
3
mR
2
22
22
2
mR
2
22
12
1
mR
1( ) ( , )l
T D l
m
g E g m E
The Fermi energy can be found from dEEgN
fE
T )(0
Spikes are when E=ml
and are called van
Hove Singularities
lm
DE
mEg
12)(1
• How accurate is this?
•Our model suggests all nanotubes
are metallic.
•We have used a very simple particle
in a box model.
•We have not even considered atomic
structure
It is possible to do more sophisticated
calculations, taking into account the
way the sheet of graphene is rolled
up.
At low energy the DOS looks very
similar to ours.
However there are big differences
close to Ef.
NT can be metals or
semiconductors!!!!!
Ouyang M et al. Science 292 (2001) 702.
• Does this match experiment?
• We can do Scanning Tunnelling Microscopy measurements to image the nanotube.
• Here an atomically sharp tip is brought close to the sample, A voltage is applied. Electrons tunnel between sample and tip.
• As the tip is moved across the sample more electrons tunnel from regions of high electron density (orbitals)
• Generate image
• From the image we can tell that this is an (8,8) nanotube
• We can measure tunnelling current for different volatages.
• dI/dVg(E)
• We see spikes!!!!
• Experiment matches Theory!!!!!
Theory
Exp
Ouyang M et al. Science
292 (2001) 702.
Optical absorption spectra also show
sharp peaks
• Optical transitions (light absorption) can
occur between conduction and valence
band states of semiconducting
nanotubes.
• Fermi’s golden rule says optical
absorbance is proportional to the DOS
In an experimental
spectrum we see many
many sharp peaks,
each due to transitions
between pairs of DOS
peaks in a wide range
of nanotube types.
• Shine in a high energy (low
wavelength) photon and an
electron is promoted to the
conduction band.
• This then gives off phonons to
decay to the bottom of the CB
before decaying back to the
valence band by emitting a
photon.
• We can measure the
photoluminescence (PL)
spectrum which consists of
sharp spikes because of the
spike in the DOS at the
bandedge.
• Measuring the PL spectrum for
many incident wavelengths
and you build up a contour
map.
• Spikes in this map appear for
individual (n,m) types.
Excitation
Emission
800 900 1000 1100 1200 1300 1400
0
50000
100000
150000
200000
250000
300000
Ph
oto
lum
inesce
nce
in
ten
sity
Emission wavelength (nm)
Excite=655 nm
Photoluminescence also shows
spikes in the DOS
Nanoscale Electrical Transport
• Imagine a 1D object, of length L (L<100 nm).
• It has a set of energy levels given by
• Imagine two large metal contacts are brought close to each end. Each
contains a band of states filled up to Ef
• If they are brought into intimate (atomic) contact, the barriers between
them disappear and electrons can flow from material to material
• This happens until all the Fermi energies coincide
2
222
2mL
nEn
Ef1 Ef2
n=1
n=2
n=3
The case for
L=5a
(N=6)
What happens if a voltage, V, is applied between the metal
contacts?
• The Fermi energies of the contacts separate by an amount
• What happens then?
• This means that the 1D channel is in contact with 2 materials with 2
different Fermi energies.
• The Fermi energies want to equalise to lower the total energy.
• They generally do this by electrons flowing towards lower energy till
the Fermi energies are equalised
eVEE ff 21
Ef1
Ef2
n=1
n=2
n=3 eV
•This means electrons move from the top energy level in the channel to
material 2 to reduce their energy
•But electrons can also reduce their energy by moving from material 1 to
the channel
•However, this does not equalise the energies because the Fermi energies
are held apart by the applied voltage
This results in a net electron flow A CURRENT
This will happen so long as there is a channel energy level between Ef1and
Ef2
Ef1
Ef2
n=1
n=2
n=3 eV
•How do we calculate current flow?
•Contact 1 wants to give the channel electrons until the probability of
occupancy in the channel is equal to the probability of occupancy for an
energy level in the metal at the same energy as the channel ().
•Contact 2 wants to take electrons from the channel until the probability of
occupancy in the channel is equal to the probability of occupancy for an
energy level in contact 2 at the same energy as the channel ().
•Remember the probability of occupancy is given by the Fermi Function
•The thing is, the two metal contacts have different Efs and so different Fermi
functions and so different occupancies at a given energy!!!!!!!
Each metal contact is trying to set the channel occupancy at a different level!!
1/exp
1)(
1
1
kTE
ff
1/exp
1)(
2
2
kTE
ff
1/exp
1)(
kTEEEf
f
Channel
energy =
• This means that contact 1 wants to pump electrons into the channel
• Channel 2 wants to take them out
• Contact 1 wants the probability of occupancy of the upper level of
the channel to be f1() and so the population to be 2f1() (spin up
and spin down)
• Contact 2 wants the probability of occupancy of the upper level of
the channel to be f2() and so the population to be 2f2() (spin up
and spin down)
• Say the average population is 2fCh
• We might imagine the electron flow into the channel across the left
junction to be
• And the electron flow into the channel across the right junction to be
1 12 ( ) 2 ChI f f
2 22 ( ) 2 ChI f f
• Putting in proportionality constants
• At steady state, the net inflow into the channel will be zero, and so
• Solving for fCh, gives
• The total current is given by
11 1
22 2
2
2
Ch
Ch
eI f f
eI f f
1 2 0I I 1 1 2 2
1 2
Ch
f ff
1 2I I I
1 1 1 2 2 1 2
1 1 2
1 2 1 2
2 2( ) ( )
e f f eI f f f
It will become clear later why we write the
proportionality constants as and 1 /e 2 /e
Rem: depends on the voltage 1 2( ) ( )f f
• This equation is interesting
• It means that current will only flow IF f1() f2()
• This is important. If the energy level in the channel closest to Ef1 and Ef2
is very high in energy (>> Ef1, Ef2), then
1 2
1 2
1 2
2( ) ( )
eI f f
1 2( ) ( ) 0f f
However, if it is very low in energy (<<
Ef1, Ef2), then
1 2( ) ( ) 1f f
In both these cases, I0
This means that I is only appreciable
for energy levels between Ef1 and Ef2
NB the state can be either filled or empty and
still pass current
• What are 1 and 2?
• The currents flowing into the channel from each
contact are
• This means the number of electrons per second
entering from the left is
• If the probability of occupancy of the level in metal 1
at energy is high (f1()1) and the channel level at
is initially empty (fCh=0)
Then the maximum initial rate of electron flow into the
channel will be
• The minimum time per electron to enter the channel
will be the inverse of the rate
11 1
22 2
2
2
Ch
Ch
eI f f
eI f f
1 11
2Ch
If f
e
1 12I
e
1
12
• Similarly we can show that the rate of escape of an electron to
the right hand contact is
• The min time taken to escape is then approximately
• Then 1 or 2 can be considered the lifetime of the electron in
the channel.
• If 1 and 2 are big, then electrons can get in and out easily:
They don’t stay long in the channel
• But the Uncertainty principle says that if the lifetime is finite
then there must be a spread in the energy of the level
• But this means that
• That means connection of the channel to contact 2 broadens
the energy level to a width of (at least) 2
• Connection to both electrodes broadens the level to a width of
2 22I
e
2
22
22
E
2
2 2
1 1
2 2 / 2E
1 2
• Remember
• If eV=Ef1-Ef2 is greater than a few kT then
f1()1 and f2()0 In this case
1 2
1 2
1 2
2( ) ( )
eI f f
1 2( ) ( ) 1f f
If the contacts are identical then
Under these circumstances, the current is:
This suggests that the better the
connection to the electrodes (higher 1),
the higher the current, independent of
voltage
1 2
1eI
• This couldn’t be true because we know IV !!
• What have we missed?
• Remember we said earlier that I is only appreciable for energy levels
between Ef1 and Ef2.
• This is the key
Ef1
Ef2
If we increase 1 then we also increase the
width of the energy level.
This means more of the level will lie outside
the bounds set by Ef1 and Ef2 and so will not
contribute to the current.
This means we need to calculate the current,
realising it only flows through the part of the
broadened energy level that lies between Ef1
and Ef2.
• To do this we need to express the broadening of the line
in terms of Density of states
• The exact expression for the current is
• However it is easier to use an approximation
• Previously we expressed the current as
• The correct current is approximately this current times the fraction of
DOS between Ef1 and Ef2
• We can assume the line broadens such that we can express the
density of states as a Lorentzian function
Where is the FWHM and is the mean energy
2 2
2( )
( ) ( / 2)D E
E
1 21 2
1 2
2( ) ( ) ( )
eI f E f E D E dE
1 2
1 2
1 2
2( ) ( )
eI f f
• Then
• Remember if eV>>kT then
• Taking in which case
• Then
• If the channel is well connected to the electrodes then 1 and 2 are big and the level is broad
1
21 21 2
1 2
( )2
( ) ( )
( )
f
f
E
E
D E dEe
I f f
D E dE
1 2( ) ( ) 1f f
2 2
2( )
( ) ( / 2)D E
E
( ) 1D E dE
1
2
1 2
2 2
1 2
2 2
( ) ( / 2)
f
f
E
E
eI dE
E
• If the channel is well connected to the electrodes then 1
and 2 are big and the level is broad
1 and 2 small 1 and 2 big
• When 1 and 2 are small we can approximate the DOS
as constant between Ef1 and Ef2.
Ef1
Ef2
Ef1
Ef2
• We will use the value of D(E) at E= while realising that this is not
the best approximation in the world
• Then
• And so
Where we note that
If the contacts are identical such that and we remember
that
2( )D
1 2
1 2
1 2 1 2
2 2
( )f f
eI E E
1 2
1 2
1 2f fE E eV
22eI V
h
Ohms law says
Where G is the conductance
1I V GV
R
Thus nanosized objects have
quantised conductance
2
0
2eG
h
NB: This is exact as all
the approximations
cancel!!!
• This means that a nanosized object has a limit to the
amount of current it can carry even in the absence of
scattering!!!
• Also the conductance is independent of sample length!!
(unlike Ohms law)
• Is this true? Experiment says so
• This is known as ballistic transport
• What about a long 1D object where the energy levels are
so close together that there is effectively a band?
• We can think as the contacts as reservoirs (1) and acceptors (2) of
charge
• Of those electrons in contact 1 travelling right-wards as many as can
be accommodated will enter the nanotube
• By accommodated we mean they can only go where there are
states to take them
• In addition electrode 2 takes as many electrons as possible from the
nanotube
1 2 - +
• In contact 1 (at T=0) there are only electrons below Ef1.
• In the nanotube there are only available states above the
nanotube Fermi energy, Ef,NT
• Only half of these states are associated with electrons moving
to the right, (ie states with k>0)
• Thus half the states in the nanotube between Ef1 and Ef,NT get
filled from contact 1 with right moving electrons
• Contact 2 has only available states above Ef2.
• Thus contact 2 can take electrons from all states in the
nanotube above Ef2 filled by electrons moving to the right.
• That is half the states between Ef1 and Ef2.
• Only these right moving electrons can enter contact 2
E
x
Electrode 1
Filled states
Electrode 2
Filled states
Electrode 1
Empty states
Electrode 2
Empty states
Nanotube
Filled states
Nanotube
Empty states
Electrons with k>0
Electrons with k>0
• In 3D the current density is given by
Where n is the carrier density (m-3) and v is the velocity
• In 1D the equivalent equation is for the current
Where n is the number of carriers per unit length.
Here n is the number of states per unit length, from which
contact 2 can take electrons.
That is half (right travelling) those states between Ef1 and
Ef2.
Remember
That means n is just half the number states in the energy
interval eV above the energy Ef2.
J nev
I nev
1 2f fE E eV
• Imagine a very small voltage, dV is applied
• Remember the density of states is the number of states per unit energy per
unit length.
• Then, the number of carriers, dn is
• Which means the current is
• Remember that
• An electron with kinetic energy E has a velocity given by
• Therefore the DOS can be written as a function of the electron velocity
1 ( )
2
D fg Edn edV
E
mEg D
12)(1
1 ( )
2
Df f
g EdI ev dn ev edV
2
2
f
f
mvE
1
4( )D f
f
g vhv
The vf’s cancel giving as before 22dI eG
dV h
• This means that a perfectly conducting one-dimensional channel has a finite conductance that depends only on fundamental constants, e and h.
• If more than 1 energy level (or more than 1 subband for a very long object) lay between Ef1 and Ef2 than we would get a contribution of 2e2/h from each channel (ie each level or subband).
• These act like conductors in parallel and so the conductance's add.
• Thus if n channels have energy between Ef1 and Ef2 then the total conductance is
• The conductance is quantised
22ne /hTG
What if the 1D object is not perfectly conducting??
• If the object is not perfectly conducting the conductance can be lower than 2e2/h
• What happens if there is a defect in the channel?
• Lets go back to our model 1D system, our chain of C atoms
• What if we replace one of them by a Si atom?
• An electron in the vicinity if this Si atom will experience a different local potential energy
• This will scatter electrons (see McMurry p68 or any other QM text)
• These scattered electrons are scattered backwards and so are not transmitted through the channel
• The transmission probability, T <1
• This allows us to write the conductance as
Where T is the probability of transmission
(transmission co-efficient) of the electron
through the channel
If there are n channels between Ef1 and Ef2 then
This is the famous Landauer formula.
Because the resistance is the inverse of the
conductance we can write the channel
resistance as (considering only 1 level, n=1)
This can be rewritten as
22eG T
h
22n
n
eG T
h
2
1
2
hR
e T
2
(1 )
2
h T TR
e T
• The Resistance is
• Remember, (1-T)=, the reflection co-efficient.
• This means
• So R has two components, one constant and one
controlled by properties of the channel (T, )
• These components are interpreted as a contact resistance,
RC and the intrinsic resistance of the channel itself, RI.
• The latter can be considered as a resistance, intrinsic
to the material
2
(1 )
2
h T TR
e T
2 2 22 2 2C I
h T h hR R R
e T e e T
• The contact resistance is the resistance associated with
electrons getting in and out of the material and is fixed
• As we have just seen the channel resistance is controlled by
the properties of the channel, ie the probability of
transmission
• Transmission can be impeded if electrons can be scattered.
• Electrons can be scattered elastically by the presence of
potential wells or barriers in the channel
• For a single potential barrier (or well) the transmission can
easily be calculated (seeMcMurray p68, Scattering by a 1D
square well)
• Things are more interesting when there are 2 potential
barriers.
• Imagine if our chain of C atoms has 2 silicon atoms and hence 2 potential barriers
• The presence of these barriers can cause scattering
• We can model these as square barriers
• Now we need to work out the transmission probability for this system
Resonant Tunnelling
V(x)
x
0
1 2
0 L
• Imagine an electron incident from the left.
• It is described by a wavefunction
• When it encounters barrier 1 part of its amplitude is
transmitted and part is reflected
• The fraction transmitted is given by
• The fraction reflected is given by
• These quantities are complex as there may be a phase
change on reflection/transmission. In addition they
depend on the physical properties of the barrier.
ikxAe
1
1 1
it t e
1
1 1
ir r e
V(x)
x 0
1 2 ikxAe
ikx
t Be1ikx
r Ce1
0 L
• The transmitted wave encounters the second barrier
• When it encounters barrier 2 part of its amplitude is
transmitted and part is reflected
• The fraction transmitted is given by
• The fraction reflected is given by
• However, the reflected bit can be re-reflected back off
barrier 1 (see red arrow)!!! This adds a further
contribution to t1.
2
2 2
it t e
2
2 2
ir r e
V(x)
x 0
1 2
ikx
t Be1 ikx
t De2
ikx
r Ce2
0 L
3
ikx
r Ee
• Remember, what we want is the transmission probability
• This is the probability that an electron starting on the left gets to the
right
• Now we use the expressions for reflected and transmitted fractions of
the wavefunction to work out D
• D is the fraction of B transmitted
• C is the fraction of B reflected
• B is the fraction of A transmitted + the fraction of C re-reflected (ie E).
• We also need to take into account the extra phase accumulated by
r2 compared to t1 as (at a given position x) it has travelled an extra
distance of 2(L-x) and hence has phase greater by 2k(L-x)
2
2
A
DT
kLiikxkLiikxxikLikikxxLxik eeeee 2222)(2
This means
CrAtB
BerC
BtD
kLi
11
2
2
2
Aerr
ttD
kLi2
21
21
1
Putting these together gives
V(x)
x 0
1 2
ikx
t Be1 ikx
t De2
ikx
r Ce2
0 L
3
ikx
r Ee
Aerr
ttD
kLi2
21
21
1 But remember t1, t2, r1, r2
are complex
kLikLi errerrrr
tt
A
DT
2
21
2*
2
*
1
2
2
2
1
2
2
2
1
2
2
1
1
1 1
it t e
1
1 1
ir r e
2
2 2
it t e
2
2 2
ir r e
Cosrrrr
ttT
21
2
2
2
1
2
2
2
1
21212 kL
• This means the transmission is a maximum when Cos=1
• That means when
• Imagine for a moment there is no phase change on reflection ie 1 and 2 are zero
• Then the transmission is a max when
• This is when
• This is when we have constructive interference of all the reflected waves
• When the transmission is maximised the electrons are said to tunnel resonantly through the barrier
Cosrrrr
ttT
21
2
2
2
1
2
2
2
1
21
2122 kLn
kLn
Ln 2
/2k
• When the barriers are identical then t1= t2= t and r1= r2= r
• In this case, at resonance
• But remember the probability is the amplitude squared and
the electron must be reflected or transmitted
Therefore and hence, T=1
When does resonance occur? When k=n/L
But remember Therefore resonance occurs for
electron energies given by:
2241 rtT
122 rt
m
kE
2
22
2
222
Re2mL
nE s
These are the same as the energy levels of
the potential well defined by the barriers!!
This is a general situation: Tunnelling through a confined electron
system is always enhanced for energies matching the bound electron
energies
Incoherent addition of waves: Ohms Law
• The previous analysis relied upon us keeping track of the phase of the scattered electron. This added an ei2kL term
• However, in real materials the electrons can be scattered inelastically by phonons.
• This results in a change in their energy and hence in their momentum
• Thus, every time an electron is scattered by a phonon, k changes, which means the phase changes.
• If the phase changes unpredictably we need to treat the electrons as incoherent.
• This happens even in the absence of defects as long as T>0
• As electrons travel through a 1D material there is more and
more chance they will be backscattered by interaction with a
phonon, defect etc
• Thus the longer the channel is, the lower the transmission!
• To see this, imagine a long channel, which we consider as
consisting of 2 sections, 1 and 2
• We give each section a probability of transmission, T and a
probability of reflection, R.
• The probability that the electron is transmitted though the
first and then the second section at the first attempt is T1T2
T2=1-2 T1=1-1
T1T2
T12 1T2
T1 2 1 2 1T2
• However an electron that is transmitted by section 1 but
reflected by section 2 can the be reflected back towards the
right by section 1 and then transmitted by section 2. (see red
arrows)
• The probability of this is T1R2R1T2
• The same thing can happen with 4 reflections (blue arrows)
with probability T1R2R1R2R1T2
• In fact the electron can be transmitted with any even number
of reflections
• The total transmission is
• This is just a well known series
• NB we don’t have to account for phase as we are considering
incoherent electrons
.....12
212121 TTT
21
21
1
TTT
• Remember that
• Rearranging gives
Remember that the transmission must be a function of length
ie T=T(L)
If the length of section 1 is L1, of section 2 is L2 and the total
length is L1+L2.
Then
The following function fits perfectly
T 1
1111
21
TTT
2121
21
TTTT
TTT
1)(
1
)(
1
)(
1
221121
LTLTLLT
( )T LL
Is a constant of order to the mean free path
and represents the length where the
probability of transmission is 0.5
• Remember the conductance per channel is
• We can now write the conductance as a function of length
• The resistance is then
• Again this is the sum of contact and intrinsic resistances
• But can be rearranged to give
• This gives the resistance as
22eG T
h
Lh
eLG
22)(
L
e
h
e
hL
e
hLR
222 222)(
LT
TT
TL
1
Te
h
e
hLR
22 22)(
Which is what we had before.
C IR R R
What about Ohms Law?
• This says, for a macroscale 3d conductor
• So far we have an expression for the
intrinsic resistance of a 1D material (per
channel)
A
LR
L
e
hLRI 22
)(
L
e
h
nLR TI 2,
2
1)(
However, remember we said that if you has a quasi 1D material, each
subband near the Fermi energy would contribute to the conductance.
The bigger the material, the more subbands contribute..
•If there are n subbands and they all contribute equally, then we
can think of them as resistors in parallel, giving a total intrinsic
resistance, RI,T
,
1 1 1 1......
I T I I I I
n
R R R R R
• How many subbands are there?
• Only the subbands or modes in the energy range E where
contribute to the current.
• For a rectangular cross section material of width, a, and height, a, the
subbands are at energy
2n a cA
A
L
ce
hLR TI
2,2
)(
Where c is a
constant, A is
area. Therefore
nece
h 1
2 2
This is Ohms law,
where is the
resistivity
2 22 2
2( )
8x z
hE n n
ma
2
2 2
1
1 8
2 ( 1)f ff f x z
N ma
E E E h n n
2
2 2
8
2 ( 1)f fx z
N man E pkT
E h n n
Close to Ef, the number of bands per unit energy is N/E
The number of subbands in the range E is then
nxf and nxf represent the bands near Ef
1 2( ) ( ) 0f f
P~10
Applications
1 Nano electronics: Beating Moores Law
It will be difficult to keep making Si transistors smaller and
smaller.
Need new technology
Transistor feature size
Carbon Nanotube Transistors
•Individual NTs can be laid directly
onto contacts via nano-
manipulation and their transport
characteristics studied
(A) Output and (B) transfer characteristics of a
CNFET with cobalt electrodes See Ouyang, Huang and Lieber,
Accts. Chem. Res. 35, 1018-1024 (2002)
Drain
VDS
VGS
• Between 2 metals, the
energy profile looks
approximately like this
Nanotube applications:
Field Emission
Vacuum level
Ef Ef
• However apply a
voltage, V, and the
picture changes
• In the field between the
metals the electron
potential energy, VPE(x)
is
(blue line)
• This allows electrons
to tunnel out of the
left hand metal
• Field emission
( )PEV x eEx
: Workfunction
Ef
Ef
V
- +
Vacuum
level
Vacuum level
eV
e-
x
VPE
Remember, the potential energy of an e- between nanotube and metal
is
Therefore
What if you replace the left metal with a nanotube?
2
0
1( )
4
qE r
rWhere q is the charge on the sphere and r is
the distance from the centre of the sphere
( )PEV x eEx
Gauss’ Law says
that the Electric field
in the vicinity of the
surface of a charged
sphere (radius R) is
given by
( )PEdV xeE
dx
Far from the nanotube, E=V/L
Ef
Ef
V
- +
Vacuum
level
Vacuum level
eV
e-
x
VPE
L
However, close to the
nanotube,
is very BIG, because r is
very small
2
0
( ) 1
4
PEdV x qe
dx r
( )PEdV xeE
dx
( )PEdV x Ve
dx L
[Here q=CV]
As for a
normal metal
Makes it very easy
for electrons to
tunnel out
Transparent Electrode
So how do the CNT FEDs work ?
Samsung CNT display
Electrons are
emitted by
nanotubes grown on
a metal base
Current controlled
by a gate
Emitted electrons hit
phosphor-coated
transparent
electrode (ITO)
Can make very
bright displays
Probably most
realistic short term
application of NTs
Composites
• Nanotubes are very strong and stiff (see page 18)
• Add them to a plastic to form a composite. It should be stronger
and stiffer than the base plastic
• Nothing new, composites have been around for 4000 years
• Never before with such strong fillers however
F-111 Fighter Bomber Mud Brick Wall in N’gev Desert
Straw Carbon Fibres
• How strong and how stiff?
• The strength and stiffness of the composite are found from
the weighted average of the plastic and nanotube properties
• Say nanotubes are added to a plastic such that the nanotube
volume fraction is Vf.
• The plastic volume fraction is then 1-Vf.
• The weighted averages for strength and stiffness are:
1
1
Comp NT f Poly f
Comp NT f Poly f
V V
Y Y V Y V
Typically:
YNT=1000GPa, YPoly=1GPa
NT=50GPa, Poly=0.05GPa
This gives scope for large reinforcement at low nanotube
volume fraction
0 1 2 3 4 5 6 70
100
200
300
400
Str
ess, ,
(MP
a)
Strain, , (%)
0.6%
0.3%
0.15%
0.08%
Plastic
Vf
Mechanical properties of composites for different Vf
0.000 0.002 0.004 0.0060.0
0.1
0.2
0.3
0.4
0.5
C
omp
(GPa
)
Fractional volume Vf
0.000 0.002 0.004 0.0060
2
4
6
8
10
YC
om
p (
GP
a)
Volume Fraction, Vf
4 increase in both strength
and stiffness with less than
1% nanotubes
Coleman et al, Adv. Func. Mat. 14, 8, 791-798 (2004)