Escolar Documentos
Profissional Documentos
Cultura Documentos
and Bioelectronics
Institute for Biomedical Engineering
Diploma Thesis
Thomas Kathriner
1 Introduction 3
2 Theory 5
2.1 Physical Domain of the Problem . . . . . . . . . . . . . . . . 5
2.1.1 Sommerfeld Model of an Individual Colloid . . . . . . 5
2.1.2 Quantum Dots . . . . . . . . . . . . . . . . . . . . . . 6
2.1.3 Shuttle Mechanism for Charge Transfer . . . . . . . . 7
2.2 Coulomb Blockade . . . . . . . . . . . . . . . . . . . . . . . . 8
2.3 Potential Between Two Metal Spheres . . . . . . . . . . . . . 10
2.3.1 Bare Potential Between Two Colloids Using Image
Charges . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3.2 Image Potential due to the Tunneling Electron . . . . 13
2.3.3 Resulting Tunnel Barrier Height . . . . . . . . . . . . 15
2.4 Tunnel Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.4.1 WKB Approximation . . . . . . . . . . . . . . . . . . 16
2.4.2 Tunneling Current in the WKB Approximation . . . . 16
2.4.3 Approximation for Faster Computing . . . . . . . . . 19
2.5 Tunneling Current Between Two Colloids . . . . . . . . . . . 20
2.6 Capacitance of Two Colloids . . . . . . . . . . . . . . . . . . 22
2.7 One-Dimensional Distribution of Colloids . . . . . . . . . . . 23
3 Methods 24
3.1 Numerical Calculation of Tunnel Resistance . . . . . . . . . . 24
3.1.1 Potential at a Specific Point . . . . . . . . . . . . . . . 25
3.1.2 Finding Potential Maximum . . . . . . . . . . . . . . . 26
3.1.3 Finding Tunneling Boundaries s1 and s2 . . . . . . . . 27
3.1.4 Tunneling Current . . . . . . . . . . . . . . . . . . . . 27
3.2 Analyzation of Tunnel Resistance . . . . . . . . . . . . . . . . 27
3.3 One-Dimensional Distribution of Colloids . . . . . . . . . . . 28
4 Results 29
4.1 Range of Variables . . . . . . . . . . . . . . . . . . . . . . . . 29
4.2 Coulomb Blockade . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3 Effective Potential Between Two Colloids . . . . . . . . . . . 31
4.3.1 Number of Image Charges Needed . . . . . . . . . . . 31
4.3.2 Barrier Height without Image Potential . . . . . . . . 31
4.3.3 Barrier Height with Image Potential Included . . . . . 31
4.4 Resistivity of the Tunnel Junction . . . . . . . . . . . . . . . 35
4.4.1 Dependency on Gap Size . . . . . . . . . . . . . . . . 35
4.4.2 Dependency on Colloid Radius . . . . . . . . . . . . . 38
4.4.3 Permittivity Plots . . . . . . . . . . . . . . . . . . . . 41
4.4.4 Measurable Permittivity Changes . . . . . . . . . . . . 41
4.4.5 Rough Estimate of General Tunneling Resistance . . . 42
4.5 One-Dimensional Distribution of Colloids . . . . . . . . . . . 44
5 Conclusions 45
6 Outlook 46
References 49
Appendix
1 Introduction
Medical doctors and patients are in dire need of information on insulin con-
centration or the existence of HIV antibodies in a droplet of blood. The food
industrys survival depends on their ability to detect any E. Coli bacteria in
their products. Governments demand ways to detect toxic waste near rural
areas. Governments, the food and medical industry are but a few examples
of actors requiring biosensors a device that detects the presence of some
biological or chemical species.
Numerous techniques for biosensing exist. Today, most detection sys-
tems use some kind of label to tag the species of interest, e.g. a fluorescent
marker to identify bacteria. Possible future label-free biosensors, which are
currently under development, utilize nanogap (Dong et al. , 2007) or Local-
ized Surface Plasmon Resonances (LSPR) sensing (Xu & Kall, 2002).
Nanogap sensing seems a very promising area of research. Electrons
tunnel through the gap once a bias is applied. The tunneling current
can then be used to investigate the electrical properties of the gap and
changes thereof. For example if a macromolecule is caught between
electrodes, in- duced resistance or capacitance changes could be
monitored. Gaps can be fabricated using traditional lithographical
techniques. Gaps also typically occur in self-assembling systems of
nanoparticles.
Bioelectronics offers a bottom-up approach to biosensors whereas litho-
graphy is a top-down method. Nanoparticles are biologically functionalized
to enable them to attach to a substrate (MacKenzie et al. , 2008). It is
possible to form patterns on the carrier where nanoparticles can or cannot
stick to. Almost arbitrarily complex, self-assembling structures can thus be
built. The particles themselves may be conducting or semiconducting and
can, therefore, be utilized to construct self-assembling electrical circuits.
3
1 INTRODUCTION
4
1 INTRODUCTION
Many methods exist to fabricate these self-
assembling structures. Self-assembled nanowires
can be fabricated using the technique of
Stadler (2007): First, a Nb2 O5 coated substrate
is func- tionalized with PLL-g-PEG/PEGbiotin.
Neutra- vidin and bDNA complexes are then
adsorbed onto the surface by biotin/avidin
linkage. Finally, gold colloids tagged with the
complementary DNA are hybridized onto the
surface (see Figure 2). The gold colloids can be
selectively placed with a precision of about 10
nm using Extreme UV Interference Lithography
(EUV-IL) to form dot- and wire-structures
Figure 2: One way of (Stadler et al. , 2007). This method has
attaching gold colloids potential for mass production since the colloids
to a substrate can be dehybridized and transferred onto
(Stadler, another substrate.
2007). Self-assembled nanostructures produced using
this or a similar method could be used as biosen-
sors in the future. Gaps naturally occurring in
nanowires could be functionalized to adsorb certain specimen. Changes
in the current flowing through the structure are potentially usable to
detect or even quantify this adsorption. Yet before nanogap biosensing can
be ac- complished, charge carrier transport in self-assembled nanostructures
needs to be modeled and understood.
Gold colloids assembled in nanowires or nanowire-arrays seem a reason-
able starting point for this theoretical challenge (see Figure 1 and Figure 3).
Gold colloids are easily functionalized using Thiol and one-dimensional wires
form a relatively simple system. It is thus the goal of this thesis to find an
appropriate one-dimensional model that allows predictions on electron cur-
rent through such gold nanowires.
Figure 3: Gold colloids forming nanowires with small gaps below 1 nm (Blech
et al. , 2007).
5
2 THEORY
2 Theory
First it is shown that a single colloid may be described classically and not
quantum mechanically. On the other hand charge transport between indi-
vidual colloids will be modeled using the quantum mechanical tunnel effect.
Possible Coulomb blockade effects for nanowires are studied.
In order to calculate the electrical properties of a gold nanowire, one
needs to look at two colloids first. An expression for the potential between
two arbitrary metal colloids submerged in a homogeneous infinite medium
with a certain permittivity is derived.
The tunneling mechanism is looked at in detail. Simmons formulas for
a current per area between two electrodes are simply integrated using the
derived potential to yield an overall current between two arbitrary
colloids.
For the sake of completeness, the capacitance between two metal
colloids is revised. The problem of the distribution of colloids in a self-
assembled, one-dimensional nanowire is also addressed.
How does an applied bias change these velocities within the electrodes?
One can show that the drift velocity vD due to an electromagnetical field E
is negligible to the Fermi velocity vF . The electron drift velocity vD due
to an electric field E can be approximated using Drudes model. It states
that (Ashcroft & Mermin, 1976)
1
vD = E, (3)
en
where e > 0 is the elementary charge, n the charge carrier density and
the electrical resistivity. Taking EF = 5.5 eV (Ashcroft & Mermin,
1976), = 2 108 m (Kaye & Laby, 1966), n = 6 1028 m3 (Wyckoff,
1963) for gold and a strong field |E| = 100 kVm1 :
2.1 Physical Domain of the Problem 2 THEORY
|vD |
4 104 1. (4)
vF
The electrons movement due to the electromagnetic field in the colloids
is negligible.
Valence electrons are accelerated by an electromagnetical field until they
collide after a relaxation time . The value can be calculated
using
(Ashcroft & Mermin, 1976)
m
= . (5)
ne2
For gold, the mean free path of an electron between collisions is,
therefore, (Ashcroft & Mermin, 1976)
= vF 41 nm. (6)
200
Current [nA]
-200
-400
-20 -10 0 10 20
Bias [V]
h
Rgap RQ = 25.8 k. (7)
e2
Ignoring the rest of the nanowire, the electric potential (r) of the isolated
charged sphere at some point r in space can also be generated by a point
charge Q residing at the center of the colloid, where (Jackson, 1999)
2.2 Coulomb 2 THEORY
Blockade Q
(r) = (r) = . (8)
40 r
0 is the permittivity of free space, the permittivity of the surrounding
medium. Putting a small additional charge Q onto the colloid, the poten-
tial on its surface is changed by
Q
= . (9)
40 a
The isolated colloid thus has a self-capacitance of
dQ
C := = 40 a. (10)
d
Suppose an electron tunnels from a neighboring sphere through the gap
onto the isolated colloid. This will increase the potential of the metallic
island by
e e
= = . (11)
C 40 a
A second tunneling electron may have an insufficient total energy to
reach that higher potential of the island even if a bias V is applied over
the gap (see Figure 6). Both potential energy eV and thermal energy kB T
need to be considered. Since kB T is only a measure of the average
thermal energy of an electron, the electrons average total energy should be
much below the
Epot
Coulomb blockade
e2
2C
2 1
kB T e2
eV 2C
0
left colloids isolated colloid right colloids
e2
a (13)
80 (eV + kB T )
demonstrate a Coulomb blockade if a bias V is applied over the gap.
Islands consisting of two, three or more colloids may also exhibit
Coulomb blockade behavior if the temperature drops. Assuming that a
metallic island of m colloids has a self-capacitance of mC , blocks of
e2
m. (14)
C (eV + kB T )
2.3 Potential Between Two Metal 2 THEORY
Spheres
10
10
2.3 Potential Between Two 1Metal n n 2 THEORY
Spheres = = . (15)
Ctotal C 40 a
Suppose that all the metallic islands have approximately the same potential
difference to their nearest neighbors. The threshold bias VT to be applied to
the chain to observe Coulomb blockade effects is then given by
(Maheshwari
et al. , 2008)
e e n
VT = = . (16)
2Ctotal 80 a
The effective radius to observe a Coulomb blockade in a nanowire is
reduced by a factor 1/n.
Q = 40 aV.
larger
right sphere. To obtain = 0 on the second colloids surface, one can
put a charge Q1 a distance d1 away from the second spheres center (see
Figure 8). The combined potential of both charges Q and Q1 then put the
11
11
x
b
a z
s
a R + d1
Q2 = Q , d2 = a2 . (20)
|R + d1 | 1 |R + d1 |2
This process must be iterated infinitely many times. Even charges Q2n
come to lie within the first colloid, odd charges Q2n+1 within the second.
The magnitude of the image charges Qm decreases rapidly to zero. If
the radii are equal, a = b and the colloids touch each other, then
1 (1)m
Qm = Qm1 = m Q. (21)
2 2
If one defines c as the ratio b = ca (c 1 since b a) and the colloids
touch, then (
cn+1
(1+c) 2n+1 Q, m = 2n + 1
Qm = n (22)
+ (1+c)2n Q, m = 2n
c
x
Q Q2 Q1
z
d2 d1
Figure 8: A charge Q at the center of the left sphere induces a first image
charge Q1 in the right colloid. Q1 in turn causes a second image charge Q2
in the first colloid, add infinitum. R is the vector pointing from the origin
to the right colloids center.
cm/2
Qm (1)m Q (1)m cm/2 Q. (23)
(1 + c)m
In any case, one finds a rapid convergence
lim Qm = 0. (24)
m
Q0 = Q, d0 = 0. (29)
This is just the original charge Q at the origin. The overall, bare potential
(b) (r) of two metallic colloids is thus
X
(b)
(r) = m (r). (30)
m=0
An index has been introduced to indicate that these image charges result
from a first image charge on the left colloid. The recursion starts with
() () ()
Q0 = e and d0 = r R. The potentials m (r) at the tunneling electrons
()
location generated by each image charge Qm are
1 Q()
m = 2n + 1 : () (r) = ,
m
m ()
40 |r m |
d (34)
() 1 Q()
m
m = 2n : m (r) = .
40 |r R d() m|
The tunneling electron also directly causes an image charge on the right
colloid. This first image charge again induces a second image charge on the
first colloid and so forth. The formulas for the image charges beginning on
the right colloid are
(r) b (r)
m = 2n + 1 : Qm = (r) Qm1 ,
|R m1 |
d (r)
d(r) = b2 R dm1 .
m (r) 2
|R dm1 |
(35)
a (r)
m = 2n : Qm(r) = Qm1 ,
(r)
|R + dm1 |
(r)
d(r) R + dm1
m = a2 .
|R + d(r)
m1 |
2
(r) (r)
The recursion relations must start with Q0 = e and d0 = r. The
(r)
potentials m (r) amount to
m = 2n + 1 : (r) (r) = Q(r)
m
1 (r)
,
m
40 |r R dm |
(36)
(r) 1 Q(r)
m
m = 2n : m (r)
= .
40 |r m(r)
|
d
The tunneling electron is, therefore, subject to an additional image po-
tential (i) (r) of
X (r)
(i)
(r) =
h
i
() (r) + m (r) . (37)
m
m=1
Due to the linearity of Maxwells equations, this (i) (r) is simply added to
the bare potential (b) (r) of the last section. The resulting total electromag-
netical potential em (r) for the tunneling electron is
EF
Figure 9: Electrons tunnel from the left to the right electrode (plate or
colloid) through an insulating layer with permittivity due to an applied
voltage V . All energy bands up to the Fermi energy EF are filled. is
the metals work function to free the electrons. The tunneling electron
induces an image potential, thus effectively reducing the barrier height (z).
Tunneling takes place only between s1 and s2 .
The number N1 of electrons tunneling from the left electrode to the right
electrode is (Simmons, 1963)
Z vmax
Z Emax
1
N1 = vz n(vz )P (Ez )dvz = n(vz )P (Ez )dEz . (43)
0 m 0
Figure 9 depicts the situation. The variable n(vz ) is the electron density of
electrons in the velocity range [vz , vz + dvz ]. Emax = m can taken to be
v2 2 max
the Fermi energy EF .
It was shown in section 2.1.1 that the electron drift velocity vD due
to an applied bias can be neglected compared to the Fermi velocity vF .
The electrons velocity distribution is, therefore, isotropic.
Also, temperature effects need not to be taken into account. Fermi
energies are in the range of a couple of electron volts, whereas a room tem-
perature of T = 300 K contributes a small kB T = 25.8 meV.
Hence, the number of electrons N1 tunneling from the left electrode to the
right is (Simmons, 1963)
Z Z
4m2 EF
N1 = P (Ez )dEz f (E)dEr . (46)
h3 0 0
On the other hand, N2 electrons tunnel from the right to the left
metal plate. The tunneling probability P (Ez ) stays the same, yet the
electrons energy is lower by an amount of eV due to the applied bias:
Z Z
4m2 EF
N2 = P (Ez )dEz f (E eV )dEr . (47)
h3 0 0
The overall current per unit area J from the left to the right is, therefore,
(Simmons, 1963)
Z Z
4m2 e EF
J = e(N1 N2 ) = P (Ez )dEz [f (E eV ) f (E)] dEr .
h3 0 0
(48)
2.4.3 Approximation for Faster Computing
Simmons also derived an approximation of equation (48). Taking
Z s2
1
:= (z)dz (49)
s2 s1 s1
as the average height of the potential that an electron has to tunnel
through, the tunneling current per area is roughly given by (Simmons,
1963)
h p i
J = J0 ( eV ) exp A eV exp A (50)
where
e
J0 := ,
2h 2 (s2 s 1 )2
4 2m
A := (s2 s1 ), (51)
h Z s2
1
:= 1 ((z) )2 dz 1.
2
8 (s 2 s 1 ) s1
For very low voltages eV , one can use the Taylor series expansions
X
x xn
e = 1 + x, (52)
n!
n=0
X (1)n (2n)! 1
1+ x = xn 1 + x (53)
n=0
(1 2n)(n!) 4
2 n 2
to obtain
AeV
J J0 exp A ( eV ) . (54)
exp
2
After further expansion and because A 1, one has a tunneling current
per unit area of (Simmons, 1963)
2me V
exp (s2 s1 ) (55)
2 (s2 s1 )
J
with h2
4 2m 1
:= 1.02 eV1/2 . (56)
h
If eV is very small the bare potential (b) (r) is also very small since the
induced charge (18) is also negligible. Hence the mean barrier height
is approximately constant. The tunneling current shows, therefore, an
Ohmic behavior J V for a small bias V .
Assuming further that s2 s1 scales like the actual gap between the
electrodes, tunneling current per unit area is exponentially decaying with
increasing gap size s,
eC s
J , (57)
s
for some constant C .
2.5 Tunneling Current Between Two 2 THEORY
Colloids
w()
a b
z
s
20
20
2.5 Tunneling Current Between Two q 2 THEORY
Colloids w() = s + a(1 cos()) + 1 1
2
2
b2 sin () . (58)
a
b
21
21
in spherical coordinates by
ZZ
I= Jz (r) dA
half-sphere
Z 2 Z /2
= d d a2 sin() cos()J (w())
0 0
Z /2
2
= 2a d sin() cos()J (w()) (59)
0
This last equation integrates Simmons formula (50) to give an overall cur-
rent for two arbitrary colloids.
If one looks at two colloids of the same radius a and uses equation (57),
then
Z /2
e2aC cos()
2 C (s+2a) d sin() cos()
Ia e s + 2a(1 cos())
0
Z 1
te2C at
= a2 eC (s+2a) dt
s + 2a 2at
Z s+2a0
s + 2a eC z 1
= dz + eC (s+2a) eC s . (60)
4 s z 4C
where
c := a + s + b,
1 p
:= (c + a + b)(c a b)(c + a b)(c a + b),
2c
p
:= 1 + 2 /a2 (67)
,a
:= c a
b ,
:= .
C12 is the capacitance between the two spheres and can be used to
model an electrical circuit.
2.7 One-Dimensional Distribution of Colloids 2 THEORY
Reasonable results can be given if the length of the wire approaches infinity.
The expected density of colloids in this case is (Bankovi, 1962)
Z Z
E(Mx ) t
1 eu
C := lim = dt exp 2 du 0.748. (69)
x x 0 0 u
Also, the probability P(sx < h) that a gap picked at random of size sx
is smaller than 0 < h 1 can be calculated to (Bankovi, 1962)
Z Zt
2 1 eu
P(sx < h) = 2 dt exp ht 2 du . (70)
C 0 0 u
Two- or even three-dimensional self-assembled structures pose an ever-
greater mathematical challenge.
3 METHODS
3 Methods
MathWorks Matlab 7.6 was used for all numerical calculations. A program
was written to yield an overall resistance R = V /I for a tunnel
junction between two arbitrary colloids with an applied bias V . A
graphical user interface lets one plot and analyze the data. An algorithm
was developed to generate a random one-dimensional distribution of
colloids in a nanowire.
4.5
3.5
3
Potential (z) [eV]
2.5
1.5
0.5
-0.5
540 560 580 600 620 640 660
z
[A]
Figure 11: Form of a typical potential between two colloids of radius 600
A
at an angle = /8 with V = 1 V. Gap size is s = 10 A, barrier
height = 4.8 eV, permittivity = 1. The electron must tunnel only
where (z) > 0 between s1 and s2 .
a sin()
3.1 Numerical Calculation ofrTunnel
= Resistance 3 METHODS
(75)
z
with [0, /2] and z [zstart , zstop ], where
zstart = a cos(),
q (76)
zstop = a + s + b b2 a2 sin2 ().
and the colloids are separated by
0
R= . (77)
a+s+b
x
w()
r
a b
z
zstart s zstop R
Figure 12: The tunneling electron at r. In the model, all electrons tunnel
along straight lines parallel to the z-axis from zstart to zend . The gap at
angle has an actual width of w() = zstop zstart .
Let the number of image charges for the bare potential (b) (r) be N
(b)
and N (i) for the electrons image potential (i) (r). The function
colloidpotential approximates the potential (z) at r with
N (b) (i)
X NX
(z) + m (r) + m(l) (r) + m(r) (r) (78)
m=0
| { z } m=1
| { z }
(b ) (r)
(i ) (r)
(l) (r)
using the recursive relations for m (r), m (r) and m (r) given at the end
of sections 2.3.1 and 2.3.2 for a given and z.
This ensures that the vector grid with Nangle entries has more values
towards = 0 where the gap is smaller and more current flows. The
integration process thus puts more weight on values close to zero. 7 in
the exponent was randomly chosen and proved to provide sufficient
accuracy even for Nangle = 50 what was used for all numerical
calculations.
resistancemodel gives the tunnel junctions resistance R = V /I as a result.
y = AeBx , (80)
one takes the logarithm on both sides to fit the data to the first
degree polynomial
log(y) = Bx + log(A). (81)
The absolute change of resistance dR/d and the relative change R1 dR if
varying the permittivity can also be approximated for different d
independent variables such as the gap s or the smaller colloid radius a. A
polynomial is fit to the R plot for every value of the independent
variable. Matlabs
3.3 One-Dimensional Distribution of Colloids 3 METHODS
2. If there is no space left, start filling the next interval to the right: go
to step 1.
4 Results
The expected boundaries of all independent physical variables are given
first. After that, Coulomb blockade effects are investigated. Then the
resulting to- tal potential barrier through which an electron must tunnel is
studied. The current-voltage relationship of a tunnel junction is then
examined. The dis- tribution of colloids in a self-assembled one-dimensional
nanowire is studied finally.
Table 1: Range of variables in this thesis. Note that the work function of
gold, = 4.8 eV, has been used for all calculations.
VT
V
VT
over the gaps separating the colloid (see figure 5 on page 8), colloids of
radius
a 8 nm (82)
show single electron transfer if one applies low voltages to the whole
nanowire (equation (13)). This upper limit is even increased if the
permittivity is larger. A single specimen of the smallest colloids of 50 nm
radius could thus show Coulomb blockade effects.
In a self-assembled nanowire, more than just one single isolated colloid
are to be expected. The threshold bias VT to observe Coulomb blockade in
a nanowire of colloids of radius a is given by (Maheshwari et al. , 2008)
30
30
4.2 Coulomb 4 RESULTS
Blockade e n
VT = , (83)
80 a
where n is the number of these isolated colloids. If we arbitrarily
choose
VT = 1 V, = 3 and a = 5 nm then
n 21. (84)
31
31
4.3 Effective Potential Between Two 4 RESULTS
Colloids
Potential [eV]
Potential [eV]
Potential [eV]
4.6 4.6 4.6
4.4 4.4 4.4
4.2 4.2 4.2
a = b = 900 A, = /2
5 a = b = 900 A, = 5 a = b = 900 A, = 5
0 /4
4.8 4.8 4.8
Potential [eV]
Potential [eV]
Potential [eV]
4.6 4.6 4.6
4.4 4.4 4.4
4.2 4.2 4.2
4 4 4
32
Figure 14: Barrier height for a tunneling electron between two gold colloids without an image potential. Colloids have
the same radii a = b. See Figure 12 on page 26 for the definition of the coordinate system. Gap s = 10 A. Applied bias V
= 1 V, permittivity = 1.
4 RESULTS
a = b = 50 A, = a = b = 50 A, = a = b = 50 A, =
0 /4 /2
5 5 5
4 4 4
Potential [eV]
Potential [eV]
Potential [eV]
3 3 3
Potential [eV]
Potential [eV]
3 3 3
33
2 2 2
1 1 1
0 0 0
-1 -1 -1
900 902 904 906 908 910 630 810 990 1170 0 450 900 1350 1800
z z z
[A] [A] [A]
Figure 15: Total barrier height (r) with image potential (i) (r) due to the tunneling electron included (red curve). The
4 RESULTS
blue curve shows barrier height without the image potential. Gap s = 10 A, bias V = 1 V, permittivity = 1.
a = 50 A, b = 400 A, a = 50 A, b = 400 A, = a = 50 A, b = 400 A, =
=0 /4 /2
5 5 5
4 4 4
Potential [eV]
Potential [eV]
Potential [eV]
3 3 3
Potential [eV]
Potential [eV]
3 3 3
34
2 2 2
1 1 1
0 0 0
-1 -1 -1
50 52 54 56 58 60 35 40 45 50 55 60 0 16 32 48 64
z [ z [ z [A]
Figure 16: Total barrier height (r) with image potential for colloids of different radii (red curve). The blue curve
shows barrier height without the image potential. Gap s = 10 A, bias V = 1 V, permittivity = 1.
4 RESULTS
4.4 Resistivity of the Tunnel Junction 4 RESULTS
35
35
4.4 Resistivity of the Tunnel JunctionC s 4 RESULTS
Re (85)
for some constant C . The simulation confirmed this relationship even
for colloids of different radii, a = b. Figures 17 and 18 are two examples
showing the simulations output in blue and an exponential fit in red.
Figures 19 and 20 show fits for different permittivities and radii a
= b. Figure 21 shows the exponential dependence for different radii.
The exponential dependency holds nicely in all studied cases.
If permittivity is increased, the resistance increases as well. This is an
expected simple consequence of Maxwells equations. The exponential coef-
ficient C is only slightly modified for different as can be seen in Figure
19
and 20.
Larger colloids conduct better. A larger area leads to more electrons
tunneling and an overall increase in tunneling current.
The exponential coefficient C is constant for all investigated radii and
permittivities:
1
C = 2.20 0.08 . (86)
If one assumes that
= C = (equations (57) and (64)) then the
C
average barrier height colloid for an entire colloid is about
(87) This value is slightly lower than the work function = 4.8 eV
that was
used.
36
36
Radii a = b = 90 nm. Applied bias V = 1 mV. = 1
40
10
Resistance R [] 1030
1020
1010
100
5 10 15 20 25 30 35 40
Gap size s
[A]
Figure 17: Tunnel resistance R versus gap size s for colloids of the
same radius. Blue dots show numerical simulation. Red curve fits data
points to
R eC s.
1040
Resistance R []
1030
1020
1010
5 10 15 20 25 30 35 40
Gap size s
[A]
Figure 18: Same plot for colloids of different radii and higher permittivity .
R eC sstill holds.
a = b = 90 nm. V = 1 mV.
40
10
1035 = 100
= 10
= 2
1030
= 1
Resistance R []
1025
1020
1015
1010
105
100
5 10 15 20 25 30 35 40
Gap size s
[A]
Figure 19: Tunnel resistance R versus gap size s for colloids of the
same radius.
= 100
1040
= 10
= 2
1035 = 1
Resistance R []
1030
1025
1020
1015
1010
105
5 10 15 20 25 30 35 40
Gap size s
[A]
Figure 20: Tunnel resistance R versus gap size s for colloids of different
radii.
V = 1 mV, = 1.
1040
1035 a = b= 90 nm
a = b= 20 nm
30 a = b= 10 nm
10
a = b= 5 nm
Resistance R []
1025
1020
1015
1010
105
100
5 10 15 20 25 30 35 40
Gap size s
[A]
Figure 21: Tunnel resistance R versus gap size s for colloids of the
same size. Different radii are plotted.
Resistance R [] 6
0
0 100 200 300 400 500 600 700 800 900
Radius of colloids a = b
[A]
s = 10 A, = 1, V = 1 mV.
9
10
Resistance R []
108
102 103
Radius of colloids a = b
[A]
Figure 23: Double logarithmic plot of Figure 22. Colloids have the
same radius a = b. 1/a fit is plotted red.
Radius of larger colloid b = 90 nm. s = 10 A, V = 1.5 mV.
Resistance R []
108
40
40
102
Radius of smaller colloid a [A]
8 8
4 4
0 0
2 4 6 8 10 2 4 6 8 10
Permittivity Permittivity
Resistance [1012 ]
5 5
4 4
3 3
2 2
1 1
0 0
0 50 100 150 0 50 100 150
Permittivity Permittivity
41
41
= 80, s = 10 A, V = 1.5 mV, b = 900 A.
2.642
103
2.64
2.638
2.636
Sensitivity
2.634
2.632
2.63
2.628
2.626
2.624
3.15
Sensitivity
3.1
3.05
5 10 15 20 25 30 35 40
Gap size s
[A]
Figure 27: Sample plot of sensitivity for = 10. Again, is not signif-
icantly influenced by gap size s. A plot for a = b = 900 nm would look
almost exactly the same.
0 = 1.26 106
A,
1
C = 2.20 .
1
Probability [arbitrary units]
0
0 20 40 60 80 100
Gap size s
[A]
5 Conclusions
A simple model to describe tunneling current through a gap between
two metal colloids was presented. The colloids themselves are assumed to be
per- fectly conducting. Tunneling through the gap is estimated using a
quantum mechanical approximation made by Simmons (1963). His
assumption that electron velocity within the metal electrodes is
isotropically distributed still holds for the systems geometry. Simmons
equations are thus integrated to yield an overall tunneling current between
two arbitrary colloids.
Coulomb blockade effects limit charge carrier transport to
sequential tunneling in longer nanowires assembled of the smallest
colloids. Even at room temperature, these nanowires show Single Electron
Tunneling (SET).
Image charges induced by the tunneling electron itself are crucial to the
model. They lower the barrier height through which an electron has to
tunnel significantly.
Tunneling resistance between two gold colloids was studied. The
resis- tance R is exponentially dependent on gap size s for colloids of
arbitrary radii. R is also inversely proportional to the colloid radius a if
both colloids are of the same size. For permittivities around 80 (the
permittivity of water), one finds the approximation
eC s
R = 0 , 5 CONCLUSIONS
(93)
a
with
0 = 1.26 106 A,
1
C = 2.20 .
1
smax = log aN Rmax . (94)
C 0
6 Outlook
Current-voltage measurements of longer nanowires and nanowire-arrays
need yet to be made. Blech et al. (2007) could only investigate nanowires
made of 5-12 colloids.
The model at hand can be implemented to study a given nanowire. SEM
images could be analyzed to yield gap widths which in turn would be used
to calculate tunnel resistances. Current-voltage measurements could
then be used to test the models accuracy.
The model could also be extended to alternating currents. The gaps
capacitances (section 2.6) would then have to be implemented and a suitable
electrical circuit model would have to be found.
The self-assembling process itself should not produce gaps wider than
just a few nanometers. Otherwise the nanowires will be insulating. A way
of achieving this is further growth of the gold colloids after the colloids
have attached themselves by immersing them in a specific solution
(Stadler,
2007).
The electrical properties of nanogap sensing need to be further investi-
gated. The way a certain specimen caught in a gap between two
colloids changes the resistance and capacitance of the gap needs to be
studied.
Charge transfer through a shuttle mechanism might need to be included
in the current model if future experiments yield a characteristic hysteresis
current-voltage curve or discrete charge transfers.
Future nanogap biosensing using nanowires could be done with moni-
toring the electrical properties of the wire or wire-array. This information
could be combined with optical properties of the nanowire by investigating
Localized Surface Plasmon Resonances (LSPR).
This thesis presents a way to model one-dimensional, particle-based
nanowires. However, the basic concept of quantum mechanical tunneling
between two individual colloids could be expanded to two- or even three-
dimensional structures. The main challenge will be the modelling of the
distribution of the self-assembling colloids which will ultimately lead to
shortest-path problems. Once a shortest path with minimal gaps would be
found, the one-dimensional model presented in this thesis could be applied.
REFERENCES
References
Anderson, Paul A. 1959. Work Function of Gold. Physical Review, 115(3),
553554.
Ashcroft, Neil W., & Mermin, N. David. 1976. Solid State Physics. Thomson
Learning, Inc.
Aswal, Dinesh K., Lenfant, Stephane, Guerin, David, Yakhmi, Jatinder V.,
& Vuillaume, Dominique. 2005. A Tunnel Current in Self-Assembled
Monolayers of 3-Mercaptopropyltrimethoxysilane. Small, 1, 725729.
Braun, Erez, Eichen, Yoav, Sivan, Uri, & Ben-Yoseph, Gdalyahu. 1998.
DNA-Templated Assembly and Electrode Attachment of a Conducting
Silver Wire. Nature, 391, 775778.
Bronstein, Ilja N., Semendjajew, Konstantin A., & Musiol, Gerhard. 2005.
Taschenbuch der Mathematik. 6th edition. Wissenschaftlicher Verlag Harri
Deutsch GmbH.
Dong, Xiaodong, Xia, Yong, Zhu, Guoyi, & Zhang, Bailin. 2007. Molec-
ular Sensing with the Tunnel Junction of an Au Nanogap in Solution.
Nanotechnology, 18, 395502.
Shekhter, R. I.,
& Jonson, M. 1998. Shuttle Mechanism for Charge Transfer in Coulomb
Blockade Nanostructures. Physical Review Letters, 80, 4526.
Isacsson, A., Gorelik, L. Y., Voinova, M. V., Kasemo, B., Shekhter, R. I., &
Jonson, M. 1998. Shuttle Instability in Self-Assembled Coulomb Blockade
Nanostructures. Physica B: Condensed Matter, 255, 150163.
Kaye, George William Clarkson, & Laby, Thomas Howell. 1966. Table of
Physical and Chemical Constants. Longmans Green.
Lang, N. D., & Kohn, W. 1971. Theory of Metal Surfaces: Work Function.
Physical Review B, 3(4), 12151223.
Simmons, John G. 1963. Generalized Formula for the Electric Tunnel Effect
between Similar Electrodes Separated by a Thin Insulating Film. Journal
of Applied Physics, 34(6), 17931803.
Xu, Hongxing, & Kall, Mikael. 2002. Modeling the optical response
of
nanoparticle-based surface plasmon resonance sensors. Sensors and Ac-
tuators B: Chemical, 87, 244249.
A DESCRIPTION OF MATLAB PROGRAMS
A.1 Resistancemodel
The Matlab routine resistancemodel is a graphical user interface to
calculate the resistance R (Figure 29). The top part lets one choose the
simulations accuracy, the bottom part each variables range to be
investigated. The results are saved into a . mat file.
Number of image charges is the number of charges induced by the
tunneling electron (N (i) in equation (78)) and should at least be equal to 20.
Number of image charges for raw potential is the bare potentials accuracy
(N (b) in equation (78)) and should be around 50. Tunneling current
per area is integrated from zero to max , the Max integration angle.
Usually, max = /2. Mean potential integration steps and Angle theta
integration steps let one choose the precision with which equations (73)
and (59) are integrated.
The function resistancemodel will scan through all values of the variables
given in the bottom part of the graphical user interface. Barrier height is
the value of the work function in equation (39).
For instance if one chooses a minimal and maximal value of 1 and
100 for the permittivity and 100 steps, then the function will calculate R
for every
= 1, 2, 3, . . . , 100. (95)
This is done for every value of the other variables. Depending on the
computers speed and the variable grid chosen, calculations may take up to
a couple of years.
Hence, one should make a test run with only a few variables to estimate
the time needed for the full run.
50
50
A.1 Resistancemodel A DESCRIPTION OF MATLAB PROGRAMS
Figure 29: Graphical user interface to calculate the resistances R for different
values of the barrier height (approximated by the work function ) and
other variables. The top part of the window lets one choose different
degrees of accuracy. All results are saved into a . mat file.
51
51
A.2 Analyzeresistance A DESCRIPTION OF MATLAB PROGRAMS
A.2 Analyzeresistance
The . mat files generated by the Matlab function resistancemodel are
analyzed using another graphical user interface called analyzeresistance (see
Figure 30). Files must first be loaded. The range of the variables for which
a resistance was calculated can then be found in the six listboxes.
The resistance R can be plotted versus every variable using the first drop-
down list. The plots x-axis or both the x- and y-axis may be logarithmical.
Plot in one window only has an effect if a second variable is chosen
with the drop-down list Plot with different. The Matlab function will
then plot R versus the first variable for every single value of the second
variable. All plots are made using the specific values selected in the
listboxes.
Different fits may also be plotted if one changes the entry No Fit in the
very last drop-down list. The calculated fit parameters are saved into the
variable fitparameters in Matlabs workspace.
An analytical model of the resistance R can be plotted as well (like
the one given in equation (90)). If Plot approx is checked,
resistancemodel will call a Matlab . m file called modelresistance.m to get
approximations to the current plot. The values of the barrier height ,
permittivity , gap width s, colloid radius a and applied bias V are handed
to modelresistance.m in exactly this order. Currently, the colloids must be of
the same size to plot approximations.
Plot permittivity change may also be chosen in the first drop-down list.
A second variable to plot must then be chosen in the fourth drop-
down list. This will first plot R versus the permittivity for every fifth
value of the second variable together with a polynomial fit of the order
chosen in the field right next to Degree. One can thus check if the
polynomial fit
52
52
A.2 Analyzeresistance A DESCRIPTION OF MATLAB PROGRAMS
is acceptable. The absolute and relative change with permittivity, d and
dR
1 dR (the sensitivity ) are then plotted using the polynomial fits to get the
R d
slopes dR
d
for different values of the second variable.
One should not choose values of the permittivity close to the
boundaries of the investigated -interval since the polynomial
approximations usually fit only poorly.
The function analyzeresistance can also be used to extract the resistance R
for certain values of the other variables since the plotted R-values are
saved to the variable resistance in Matlabs workspace.
53
53
Figure 30: Graphical user interface to calculate the resistances R for different
values of the barrier height (approximated by the work function and other
variables) and other variables. The top part of the window lets one choose
different degrees of accuracy.
B MATLAB SOURCE FILES
B MATLAB SOURCE FILES
resistancemodel.m analyzeresistance.m
tunnelcurrent.m modelresistance.m
.mat file
colloidpotential.m peakdet.m
B.1 colloidpotential.m
f u n c t i o n p h i = c o l l o i d p o t e n t i a l ( s , V , b a r r i e r , K, a , b , z i n i t i
al ,theta , images , bare pot images)
% COLLOIDPOTENTIAL C a l c u l a t e s p o t e n t i a l btw two c o l l o
ids
%
% p h i = c o l l o i d p o t e n t i a l ( s , V , b a r r i e r , K, a , z , t h e t a , i m a
ges ,
5 % bare pot images)
%
% s: gap [
A ]
% V: bias [V
]
% b a r r i e r : b a r r i e r h e i g h t [ eV
]
10 % K: permittivity [1]
% a: colloid radius [
A ]
% z, theta : coordinates of point of int er
est
% i f r =( a s i n ( t h e t a ) ,
z)
% i m a g e s : number o f i m a g e c h a r g e s (
20)
15 % bare pot images : image charges for bare pot ent ial (50)
%
% phi : calculated pot ent ial at point (asin ( theta ) , z) i
n eV
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%
% Load p h y s i c a l c o n s t a n
ts
20 global e e 0
B MATLAB SOURCE FILES
physicalconstan
ts
B.1 colloidpotential.m B MATLAB SOURCE FILES
B.2 peakdet.m
f u n c t i o n maxtab = p e a k d e t ( v , d e l t a )
% PEAKDET D e t e c t s l o c a l maxima i n a v e c t o r
%
% MAXTAB = PEAKDET( V , DELTA)
5 %
% A p o i n t i s c o n s i d e r e d a maximum p e a k i f i t h a s t h e m axi m al
% v a l u e , and was p r e c e d e d ( t o t h e l e f t ) by a v a l u e l o w e r by
% DELTA. MAXTAB c o n s i s t s o f two c o l u m n s . Column 1 c o n t a i n s
% i n d i c e s i n V , and col um n 2 t h e f o u n d v a l u e s .
10 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%
maxtab = [ ] ;
mx = I n f ; mxpos = NaN ;
lookformax = 1;
f o r i =1: l e n g t h ( v )
15 t his = v( i) ;
i f t h i s > mx
mx = t h i s
; mxpos = i
;
end
20 if lookforma
x
i f t h i s < mx d e l t a
maxtab = [ maxtab ; mxpos mx ] ;
lookformax = 0;
end
25 end
end
B.3 tunnelcurrent.m B MATLAB SOURCE FILES
B.3 tunnelcurrent.m
f u n c t i o n J = t u n n e l c u r r e n t ( s , V , b a r r i e r , K, a , b , t h e t a i n i t
ial ,images , bare pot images , mean pot ent ial int egrat ion st eps
)
% TUNNELCURRENT C u r r e n t p e r a r e a a t s p e c i f i c a n g l e btw 2 c o l l o i d s
%
% J = t u n n e l c u r r e n t ( s , V , b a r r i e r , K, a , b , t h e t a , i m a g e s ,
5 % bare pot images , mean pot ent ial int egrat ion st eps)
%
% s: gap [ A ]
% V: bias [V]
% b a r r i e r : b a r r i e r h e i g h t [ eV ]
10 % K: permittivity [1]
% a: s m a l l e r c o l l o i d r a d i u s [ A ]
% b: b i g g e r c o l l o i d s r a d i u s [ A ]
% t h e t a : a n g l e , m e a s u r e d f r o m d i r e c t l i n e c o n n e c t i o n btw c o l l o i d
% centers , measured in radians
15 % i m a g e s : number o f i m a g e c h a r g e s ( 2 0 )
% bare pot images : images for bare pot ent ial (50)
% mean pot ent ial int egrat ion st eps : int egrat ion accuracy (100)
%
% J : c a l c u l a t e d c u r r e n t f o r a n g l e t h e t a [ A ] , NOT y e t i n t e g
rat edover phi
20 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%
% Load p h y s i c a l c o n s t a n
tsglobal e h m
physicalconstants
% beta is around 1
25 beta = 1;
% P relocation
t het a size = length ( t het a init ia
l ) ;J = zeros(1, t het a size) ;
% Calculate for every entry of vector t het a init ial
30 for o=1:1: t het a size
theta = t het a init ial (o) ;
% g e t a l l o w e d ( f r e e s p a c e ) z i n t e r v a l btw c o l l o i d s f o r a n g l e
theta z st art = acos( theta ) ;
z stop = a + s + b realsqrt (b2 a2sin ( theta )2) ;
35 gap = z s t o p z s t a r t ;
% t o l e r a n c e f o r f i n d i n g minima o f d e r i v a t i v e o f p o t e n t i a l .
% Essential since there are poles at int erval endpo
i n t s t o l e r a n c e = gap 1 0 6 ;
z g r i d = z s t a r t + t o l e r a n c e : ( gap 2 t o l e r a n c e ) / 2 5 : z s t o p
tolerance ;
40 p o t = c o l l o i d p o t e n t i a l ( s , V , b a r r i e r , K, a , b , z g r i d , t h e t a , 5 ,
5) ;
% F i n d maximum o f p o t e n t i a l u s i n g p e a k d e t m
f i l e maxtab = p e a k d e t ( p o t , 0 . 1 ) ;
maximum = z g r i d ( maxtab ( : , 1 ) ) ;
f u n c t i o n h a n d l e = @( z ) c o l l o i d p o t e n t i a l ( s , V , b a r r i e r , K, a , b
, z ,theta , images , bare pot images) ;
45 % C a l c u l a t e t u n n e l c u r r e n t i f f p o t e n t i a l b a r r i e r i s p o s i t i v and
% tunneling takes place
i f f u n c t i o n h a n d l e ( maximum) >0
% Z e r o s o f p o t e n t i a l btw c o l l o i d s
% Left zero
50 s 1 = f z e r o ( f u n c t i o n h a n d l e , [ z s t a r t +t o l e r a n c e , maximum ] ) ;
% Right zero
s 2 = f z e r o ( f u n c t i o n h a n d l e , [ maximum , z s t o p t o l e r a n c e ] ) ;
% E l e c t r o n t u n n e l s a d i s t a n c e d e l t a s btw s 1 and
s 2delt a s = s 2 s 1 ;
B.4 resistancemodel.m B MATLAB SOURCE FILES
B.4 resistancemodel.m
Only the most important program excerpts are given. Program parts asso-
ciated with Matlabs graphical user interface are omitted.
60
60
B.5 randomdistr.m B MATLAB SOURCE FILES
forP ermittivity , fors , fora , forb , forV)
2pi a2 1020;
% C a l c u l a t e r e s i s t a n c e f r o m c u r r e n t and b i a s
Data . r e s i s t a n c e ( f o r B a r r i e r , f o r P e r m i t t i v i t y ,
f o r s , f o r a , f o r b , f o r V ) = V/ J ( f o r B a r r i
er ,forP ermittivity , fors , fora , forb ,
end forV) ;
75 end
end
end
end
end
80 % Get t i m e and s a v e t o mat f i l e
Data . t i m e = t o c ;
s e t ( h a n d l e s . t e x t T i m e t o t a l , S t r i n g , n u m 2 s t r ( Data . t i m e / 6 0 ) ) ;
s a v e ( F i l e , s t r u c t , Data , . . .
numberofimages , number of raw images , thetamax , time , . . .
85 Barriermin , Barriermax , Barriersteps , ...
Permittivitymin , Permittivitymax , P ermittivityst eps , ...
mean pot ent ial int egrat ion st eps , ...
angle int egrat ion st eps , ...
s m i n , smax , s s t e p s , . . .
90 amin , amax , a s t e p s , . . .
bmin , bmax , b s t e p s , . . .
Vmin , Vmax , V s t e p s , . . .
resist ance ) ;
B.5 randomdistr.m
function [X, t ] = randomdistr (L, a)
% RANDOMDISTR G i v e s a random d i s t r i b u t i o n o f c o l l o i d s a l o n g 1 d w i r e
% T h i s i s Re n y i s P a r k i n g P r o b l e m ( o r Re n y i s P a c k i n g P r o b l e m )
% Theoretical density : 0.748...
5 %
% [X, t ] = randomdistr (L, a)
%
% L: length of wire
% a: radius of colloids
10 %
% X: vector containing coordinates of colloid midpoints
% t : calculat ion time
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% i n i t i a l i z i n g random num ber s
15 r a n d ( t w i s t e r , sum ( 1 0 0 c l o c k ) ) ;
t ic
% P relocation
size = round (L / (2a) ) +
1;X = zeros(1, size) ;
20 l = a ; r = L + a ;
m = 1;
X( 1 ) = L + a ;
% Start fillin g int erval
w h i l e ( r l >= 4 a ) | | ( r = L + a )
25 w h i l e r l >= 4
am = m + 1;
X(m) = l + 2 a + ( r l 4 a ) r a n d ;
r = X(m) ;
end
30 l = X(m) ;
r = min (X( f i n d ( X>l , s i z e ) ) ) ;
61
61
B.6 gapdistr.m B MATLAB SOURCE FILES
w h i l e ( r l < 4 a ) && ( r = L + a
)
l = r;
r = min (X( f i n d ( X>l , s i z e ) ) ) ;
35 end
end
% S o r t i n g and r e s i z i n g o f
X X = s o r t (X , a s c e n d ) ;
X = X(X>0) ;
40 X = X(X <= L ) ;
% Get t i m
e t=t o c ;
B.6 gapdistr.m
f u n c t i o n [ S ] = g a p d i s t r ( L , a , X)
%GAPDISTR g i v e s v e c t o r o f gap d i s t a n c e s
%
% [ S ] = gapdistr(L, a)
5 % OR
% [ S ] = g a p d i s t r ( L , a , X)
%
% L: length of wire
% a: radius of spheres
10 % X: vector containing coordinates of spheres
%
% S : v e c t o r w i t h gap d i s t a n c e s
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
i f e x i s t ( X , v a r )
15 X = randomdistr (L, a) ;
end
S i z e = l e n g t h (X) ;
S = zeros(1, Size + 1)
; S ( 1 ) = X ( 1 )a ;
20 f o r i =1 : S i z e 1
S ( i +1) = X( i +1) X( i ) 2 a ;
end
S ( S i z e + 1 ) = L X( S i z e ) ;