Molecular Electronics

Università degli Studi di Catania
Facoltà di Scienze Matematiche, Fisiche e Naturali
Corso di Laurea Magistrale in Fisica
Dipartimento di Fisica e Astrofisica
Valentina Ferro
Molecular Electronics
Esame di fisica delle nanostrutture
Prof.ssa Maria Grazia Grimaldi
Dott. Francesco Ruffino
Academic Year 2012-2013

Contents
Introduction III
1: Physics in Molecular Devices 1
1.1: Energy Levels in a Molecule . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2: Transport Regimes in Mesoscopic Systems . . . . . . . . . . . . . . . . . . . 3
1.2.1: Landauer’s approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2: Non-Equilibrium Green’s Function . . . . . . . . . . . . . . . . . . . 5
1.2.3: Mechanics of Molecular Transport . . . . . . . . . . . . . . . . . . . 5
1.3: Other relevant properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1: Timescale Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.2: Heating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.3: Spin Eﬀects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.4: Geometric Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2: Molecular Devices:
Fabrication and Application 9
2.1: Molecular Junctions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2: The “ability” of Self-Assembly . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3: Molecular Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.1: Indications in Fabrication Procedures . . . . . . . . . . . . . . . . . 14
2.3.2: Molecular Rectiﬁers . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.3: Molecular Switches . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3.4: Molecular Transistors . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.5: Connecting Molecular Devices . . . . . . . . . . . . . . . . . . . . . 20
I

2.4: Molecular Electronics Circuits . . . . . . . . . . . . . . . . . . . . . . . . . 21
Conclusions i
Bibliografy ii
II

Introduction
The Moore’s Law, that states that transistor density on a chip doubles every 2 years,
has been dominating the industry of semiconductors since now. Nevertheless, it is now
facing its first problems. Down-scaling the technology, computer architects have to work
with new challenges and new physics features to study, to avoid or to use [1].
The main challenges are [1]:
• Power Consumption: it derives from the breakdown of Dennard’s scaling rules.
Dennard’s rules stay that smaller transistors give more power-efficient circuits and
keep the power density constant. Nevertheless, static power consumption was com-
pletely ignored. But nowadays we know that transistors are never truly off, allowing
leakage currents to flow.
• Memory Wall: it arises from the discreancy between logic speed (related to cal-
culation) and memory. In fact if the speed of logic circuits doubles every 2 years,
memory technology increases speed much more slowly.
• Heating: the increase in density due to scaling also leads to an increase in generated
heat.
• Limits of lythography: in fact lythography, and more in general all the technologies
used in transistor and CMOS production, are now approaching their physical limits
or too elevated costs. The resolution of the modern lithographic equipment is about
5nm [2].
• Secondary effects: for transistor, we can think about DIBL (Drain-Induced Barrier
Lowering) and Surface Scattering in the channel.
Due to all these limitations in the scaling process, the 2010 update of the International
Technology Roadmap for Semiconductors predicts a decline in development and a relevant
III

changing slope in Moore’s law, with transistor count doubling only every 3 years starting
2013 [1].
Figure 1: Taxonomy for emerging research information processing devices (from International
Technology Roadmap for Semiconductor).
New concept of electronics devices is needed in order to supply or integrate the semi-
condactor industry (fig.1). In this scenario, molecular electronics offers an alternative,
but also a complement, to existing semiconductor technologies. Advantages of the use
of molecules are numerous (fig.2). Nevertheless, the approach in processing molecular
devices has a completely different know-how than the silicon industry and needs to be
improved both in performance and in cost.
Advantages of Molecular devices
Molecular electronics emerging applications are exciting: molecules can function as
transistors, switches, rectifiers, interconnects, photovoltaics, memories, and sensors [3].
Using molecules as electronic components has several potential advantages [3,4]:
• The size of molecules is in the length scale of approximately 1-100nm range. This
leads to advantages in cost, efficiency and power dissipation.
• The molecular recognition can be employed in changing electronic properties of
IV

molecules, which can provide sensing or switching capabilities. Moreover speciﬁc
intermolecular interaction can help to form structures by self-assembly techniques.
• With their variability of compositions and geometries, one can tune the transport
properties of molecules through chemical synthesis.
• A large number of identical molecules can be cheaply produced.
However, molecules have obvious disadvantages, like stability and robustness at ele-
vated temperature, problems to connect them to electrodes, weak ability to interface with
other components in extended systems (2).
Figure 2: An indicator of increasing interest in molecular electronics is provided by the number
of citation per year.
Outline
This paper want to be a quick overview about molecular electronics current status
and has the aim to give the reader all the tools could be necessary to understand the
main implications of this technology. In the ﬁrst part, basics concepts behind molecular
V

electronics will be discussed, citing the theoretical approach in the study of energetic
levels and electronic transport in a molecule but avoiding a deep and comprehensive
analysis (see chap. 1). The second part instead will have a more experimental approach,
investigating which type of devices can be produced with molecules and citing, when it
will be necessary, some details about implementations (see chap. 2). In sake of simplicity,
the molecular electronics related to the use of carbon allotropic forms won’t be a subject
of this paper.
VI

Chapter 1
Physics in Molecular Devices
It is well known that in mesoscopic devices classical law of physics cannot be always
applied. Ohm’s law fails, i.e. resistance of quantum wire does not depend on its length
since the electron transport is not a diffusion process as described by this simple law.
To describe small devices, one needs a quantum-mechanic treatment of the electrons, in
which the quantum effects, such as the discreteness of charge, the energy quantization,
and the quantum coherence have to be considered [3].
Experimental progress in molecular electronics has given rise new challenges to theory
in developing theoretical tools to describe the electron transport. The density functional
theory (DFT) is now a well-established method for electronic structure calculations of
the ground-state properties of metals, semiconductors and insulators. However, when a
bias voltage is applied to the system, the DFT method cannot rigorously handle an open
system under the nonperiodic and out of equilibrium conditions. Therefore, molecular
electronics devices need to be taken into account by the Green’s function formalism. Up
to now, the combined DFT and non-equilibrium Green’s function (NEGF) methods have
been widely used to study the quantum transport through nanoscale devices.
1.1 Energy Levels in a Molecule
For electronic transport studies, it is extremely important to know the energy distribu-
tion in the molecule and the right terminology. In a molecular system, we will refer often
to the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecu-
lar orbital (LUMO), similar to the valence band and the conduction band in solid-state
materials, respectively, but with discrete energy levels caused by quantization effects in
1

CHAPTER 1. PHYSICS IN MOLECULAR DEVICES
constricted dimension of a molecule. The difference in HOMO and LUMO is referred to
as the energy bandgap Eg [5].
By adjusting the chemical functionalization of the molecule with electron-donating or
electron-withdrawing substituents, thus altering the energy and geometry of the molecular
orbitals, one can “dope” the molecule [6].
To describe the transport through a molecular system having HOMO and LUMO
energy levels, one of the applicable models is the Franz two-band model, that we will just
cite. This model provides a non-parabolic energy-momentum E(k) dispersion relationship
by considering the contributions of both the HOMO and LUMO energy levels [7]:
k2
=
2m∗
2
E 1 +
E
Eg
(1.1)
This model could be useful in interpretation of some result, and can give information
about energy level in the studied system.
The mobility of electron in the molecular system can be seen from the point of view
of delocalization of the electrons accross molecular states. If in inorganic semiconductors,
the delocalization of electronic states is due to the overlapping of atomic orbitals of each
atoms, likewise in a molecule a set of overlapping electronic states accross the entire
molecule is necessary for efficient electronic condunction [5]. We can provide some useful
examples of the different kinds of bonds that can occur in molecular system:
• Delocalized Bonds in Aromatic Molecules. More than two adjacent π orbitals can
combine to form a set of molecular orbitals with electron pairs shared by the atoms
in the bond. Moelcules with this kind of orbital are called aromatic; an example
could be benzene (C6H6) (fig.1.1).
Figure 1.1: Delocalized π bonds of benzene.
• Aliphatic Molecules. They act as insulators because no π conjugated orbitals are
present.
2

• Conjugated Oligomers. The term conjugated means an alternation of multiple and
single bonds linking a sequence of bonded atoms; also in this case π electrons are
involved. The principle that lies behind the conduction mechanism in this kind
of molecules is the analog of the symmetry in cristalline semiconductors. In fact
oligomers are synthesized by joining one molecular unit (monomer) to itself a few
times, as the repetition of the unit cell in solid state. Good examples are polypheny-
lene and polyphenylene-based molecules [5].
1.2 Transport Regimes in Mesoscopic Systems
Down-scaling to 1-10nm means working with structures with dimensions comparable
to some characteristic physical length. In order to distinguish the transport regime in these
systems, we can mention at least two characteristic lengths, the momentum relaxation
length, Lm , and the phase relaxation length, Lφ. The ﬁrst one, also known as electron
mean free path, is the average distance which an electron can travel before its momentum
is lost, while the latter is the average distance which an electron can travel before its phase
is destroyed. The simple Ohm’s law is applicable just when the length of the device, L, is
longer than Lm and Lφ. When this is not the case, the quantum character of the electron
need to be considered, and we have to keep in mind that it can behave like waves and
show interference eﬀect. We can distinguish three transport regime:
1. Ballistic transport regime, L Lm, Lφ: electrons can propagate freely from one
contact to the other without scattering. The resistance is due to the scattering
phenomena with the contacts and the conductance does not depend on the length
of the device.
2. Elastic and coherent transport regime, L < Lφ: electrons go under elastic scattering,
without lowering their energy and then their phase. It is typical of short molecules
in which electrons move too fast to interact with the molecular vibration.
3. Inelastic and incoherent transport regime, L > Lφ: electrons can interact with other
electrons or phonons loosing energy and coherence in phase.
There are two approaches that have been widely used to study the transport prob-
lem: the Landauer method and the Non-Equilibrium Green’s Function (NEGF) method.
3

The Landauer approach allows us to describe the non-interacting electron transport cor-
responding to the ballistic or coherent transport regimes, while the NEGF approach is a
more sophisticated method that can be used in all three transport regimes.
1.2.1 Landauer’s approach
Figure 1.2: Landauer’s Scheme of a Device. Fermi’s functions of electrodes are also repre-
sented.
We can imagine a simplified system consisting of a central region (C) sandwiched
between left (L) and right (R) leads, of electrochemical µL and µR (fig. 1.2). The
electrochemical potential difference, adjusted by applied bias voltage, between the two
leads causes the current to flow. The problem is then faced as a scattering problem. An
incident wave function, propagating along the central region, is scattered by a potential
connecting to the two leads and then transmitted to the other lead; in this scenario the
current flow is the probability of the electron to be trasmitted from one lead to the other.
Mathematically, this leads to the Landauer formula for the conductance [4]
g =
2e2
h ij
Tij (1.2)
where Tij is the scattering probability from incoming channel i to outgoing channel j. If
there is just one open channel, without scattering ( ij Tij = 1), we obtain the quantum
4

of conductance.
g =
2e2
h
= (12.8kΩ)−1
(1.3)
1.2.2 Non-Equilibrium Green’s Function
The NEGF method was established in the 1960’s through the classic work of Schwinger,
Kadanoff, Baym, Keldysh and others. The development of this method is made by the
use of many-body perturbation theory (MBPT).
A more general approach in the study of transport, makes use of this theory to include
not only elastic scattering, but also such issues as interaction between the molecule and
the electrodes and the coupling to vibrations and to external fields such as light or thermal
gradients (Keldish-Kadanoff-Baym approach) [8]. In the limit of small applied voltage V,
the coherent conductance can be written as
g =
2e2
h
TrM ΓR
GM ΓL
G∗
M (1.4)
where GM describes the propagation through the molecule, and ΓR
and ΓL
are respec-
tively the spectral densities coupling the molecule to the electrodes on the left and the
right. These spectral densities describe the effective mixing strength between molecule
and electrodes [4,8]. The dependence of the conductance on the molecular species arises
mostly from the Green’s function G. In a very simple single-determinant description, if
we use i, j to denote atomic orbitals and µ for molecular orbitals, the matrix elements of
G are
Gi,j =
µ
i | µ µ | i
E − Eµ − Σµ
(1.5)
with Eµ and Σµ the molecular orbital energy and the self-energy [8].
1.2.3 Mechanics of Molecular Transport
The theoretical approaches to understand these mechanisms can be really complicate
and they are not the main topic of this digression, but can be summarized into five
different regimes:
1. Coherent electron motion: nonresonant.
It takes place in absence of strong dissipation or trapping effects [5], when the Fermi
level of the contacts is in the middle of HOMO/LUMO gap. The conductivity
5

exponentially decreases with the length of the molecular bridge; it can be expressed
as
g = Ae−βN
(1.6)
with N the number of sites in the wire (it is proportional to wire length), β the decay
parameter (it depends on the electronic structure of the wire) and A related to the
contact conductance [9]. This transport mechanism suits very well short wires with
large HOMO/LUMO gaps, as oligoalkanes [5].
2. Coherent electron motion: resonant.
It takes place when the energy of the tunneling electrons is resonant with the conduc-
tion band of the wires [5]. The conductance is dominated by the contact scattering
and it is indipendent of length; it linear increases with the number of allowed modes
in the molecular wire. The rate of electron transport obeys, in fact, the Landauer
formula [5].
3. Incoherent transfer: Ohmic behavior.
It takes place when the wire is strongly coupled with the enviroment, allowing more
available modes and giving rise to an effective intramolecular lifetime. In this regime,
the wire behaves like a regular electrical resistive wire, with conductivity inversely
proportional to the length. It is usually more efficient than exponential nonresonant
transport in the case of longe chains [9].
4. Quasiparticle formation and diffusion.
The carriers can be charged solitons (a structural defect with an extra electron
sitting on a free radical site - polaron); it usually takes place in the case of degenerate
electronic ground states, as polyacetylene. In the case of non-degenerate conductors,
as poly(p-phenylene), it is possible to have bipolarons [9].
5. Gated electron transfer.
In molecules it is really common to have different transfer rates dependent on ge-
ometrical modification. A dynamical change in a molecule can be then used as an
electronic gating mechanism [9].
Conduction mechanisms can be listed also in order of the interaction between molecule
and metal electrodes, as summerized in table (fig.1.3) with their characteristic behavior,
6

temperature dependence, voltage dependence, and schematic band diagram [5].
Figure 1.3: Possible conduction mechanisms
Hopping Conduction
It is worth mentioning Hopping Conduction. It refers to the process in which thermally
excited electrons “hop” from one isolated state to the next and it can really dominate the
conduction in some condition, i.e. with high defect density.
1.3 Other relevant properties
1.3.1 Timescale Issues
Many times in application, it is useful to know the scale of the time spent by an
electron in its trasmission, in order also to evaluate the probability that it has to interact
with the matter. If the trasmission is fast on the timescale of nuclear motion, like usually
happens, the latter can be taken static and the resulting trasmission is just the average
of the trasmission probability over the relevant ensemble of nuclear conﬁguration. For
resonance electron transfer, the timescale is related to the electronic coupling with the
electrodes. In the case of oﬀ-resonance tunneling, we need to consider the traversal time,
which is the time available for a tunneling electron to interact with degrees of freedom
7

localized in the barrier region (this peculiar time can be easily estimated, and it lie in the
range 0.1 − 1fs) [8].
1.3.2 Heating
As we wrote, one of the problem of nanoscale transistors is the heat dissipation. It
worths then asking how the heat is dissipated in a molecule and how it effects the conduc-
tion mechanism. The rate of electronic energy dissipation (comprising heat generation) is
κIΦ, with I the carried current by the molecule, Φ the applied bias, and 0 ≥ κ ≥ 1. If we
want to evaluate the heat conduction, the classical representation fails because phonon
motion in a small molecule is essential a ballistic motion, making the classical conduc-
tion concept invalid. According to this osservation, the heat transfer should be greater
than what expected in a macroscopic conduction. Nevertheless, a molecular system is a
restricted environments, limited in the number and spectrum of available phonon modes.
This actually reduces the efficiency of heat dissipation. Experimentally, it is confirmed
that in a molecular bridge the junction stability is compromised because of heating [8].
1.3.3 Spin Effects
Because of the Pauli’s Exclusion Principle, interactions between same spin and oppo-
site spin electrons are different. Spin effects are strongly marked in situations involving
open shell transition metal ions, i.e. Cobalt [8]. In addition peculiar devices that make
use of spintronic effects can be though [10].
1.3.4 Geometric Behavior
Modifications in structure lead to modifications in molecular properties, and vicev-
ersa. An increasing number of experiments have demonstrated that electronic currents
in junctions can actually break chemical bonds, excite vibrations, and otherwise alter the
structural chemistry of the molecules.
8

Chapter 2
Molecular Devices:
Fabrication and Application
In this section we will focus on the possible application of molecules in electronic
devices. The chemical production of the molecules themselves will not be matter of this
discussion, in favour of measurement results.
2.1 Molecular Junctions
Molecular Junctions are not themselves molecular devices, but they provide the sim-
plest way to analyze and to apply transport mechanisms we described before. In ﬁg.2.1
four types of junctions are represented [4]:
Figure 2.1: Examples of molecular transport jucntions.
(a) A linear chain (an alkane) as a bridge between two electrodes, one that functions as
9

CHAPTER 2. MOLECULAR DEVICES:
FABRICATION AND APPLICATION
donor, the other as acceptor. The transport mechanism is, for alkanes up to a certain
length and for small applied voltages, the coherent electron motion (see 1.2.3). The
alkane behaves like a simple energy barrier separating the two electrodes.
(b) DBA: Donor-Bridge-Acceptor molecular junction. In this case the donor and ac-
ceptor sites are part of the molecule, separated by a bridging component that has
molecular orbitals of different energy respect to their lowest unoccupied levels. In
this kind of structure two transport mechanism can occur: the direct D-A coupling
or the more interesting superexchange process; in the latter, an electron tunnel
from an electrode to the acceptor, then, under applied bias, it may transfer to the
donor and tunnel another time from the donor to the other electrode. An equivalent
process can occur for holes [4]. It is known that both direct and superexchange in-
teractions are expected to have a competive exponantial dependance with distance,
as we can expect if we have a look to the effective electronic coupling V [11]:
V ∝ exp(−RDA) (2.1)
with RDA the distance between Donor and Acceptor. For short bridges (5 − 10 ˙A),
the superexchange mechanism will almost always dominate. For longer bridges, the
superexchange tunneling process is very slow and actual population of the bridge by
the charge becomes a competing process. In such cases, the charge can transfer by
a multistep hopping mechanism, and the dependence will be no more exponantial.
If the donor and acceptor are designed to differ energetically, this device behaves
like a rectifier: the asymmetric energy landscape lead the current to flow just in
one direction. Thanks to the ability of isomerize of some molecule, with subsequent
change in transport mechanism, some DBA could be used as switch devices.
(c) A molecule bridge could be formed also by a molecular quantum dot. The common
use of this king of brigde is to create an electrical break junction. In addition a
gate electrode can be added, in order to tune the energy levels of the molecule with
respect to the Fermi levek of electrodes. When a gated measurement is carried out
at low temperature, it can resolve energy levels to a few meV, allowing to study the
signatures of electron-vibration coupling.
(d) The last one is the more general case in which several functional groups are in the
10

molecular bridge. For very short-distance motion, coherent tunneling can occur. For
transfer over more than six or seven base functional groups, inelastic hopping has
been strongly suggested (fig.2.3). In general, a model developed by Marcus, Hush
and Jortner, can be here applied. The model assumes that for a charge tunneling
from a donor to an acceptor site, the height and the width of the tunnel barrier
are modulated by interaction with a bath of harmonic oscillators that account for
the chemical environment. This leads to several complication in the interpretation
of experimantal results, which are often the results of a big mixture of different
mechanism of conduction. One interesting example for molecules in which this
Figure 2.2: Competion between cohenrent tunneling and hopping conduction in DNA: the
theoretical model by Marcus et al (solid line) and experimental data (triangles and circles) are
plotted.
behavior can be observed is DNA (fig.2.3). As a confirmation of the complexity
of electron transport in this kind of devices, the results about DNA show that it
can behave as semicondactor, as metal and as superconductor. Most probably it
is a wide-gap semiconductor, characterized by localized hole hopping between low-
energy pair (guanine-cytosine). Nevertheless, significant effects should arise from
other processes, such as polaron-type hopping or Anderson-type localization [4].
The DNA is for this reason one of the most studied complex molecule, so also a lot
of recent simulations, made by the use NEGF-DTF, have also been done [3,12].
11

In a molecular junction, not only the molecule itself, but also the molecule-electrode
interface plays an important role in determining the characteristichs of the final device.
Poor covalent bonding usually exists between the molecule and the electrode. Then, at
zero applied bias some charge must flow to equilibrate the chemical potential accross
the junction, causing possibly partial charging of the molecule and local charge buildup
could give Schottky-like barriers to charge flow though the interface. Mechanism like
that, can mask the molecule’s signature. In addition, it is not known so much about the
alignment of molecular orbitals’ energy levels with the Fermi levels of electrodes. Finally,
this interface can modify the number of modes in the molecule itself with a change in
the conductance (remember that the density of modes or channels in the molecules is a
parameter in the Landauer approach). The clear implication is that the molecule and
interface junctions are inseparable and thus must be considered as a single system.
Transport measurement
Transport measurement can be done with basically two different techniques. In the
first one, the molecules are positionated between electrodes in some way. Often their
properties are studied in the so-called self-assembled monolayers (SAM), in which the
substrate functions as one electrode while a scanning tunneling microscopy tip is the other
one. The second approach makes use of photo-induced electron transfer, a time-resolved
spectroscopy of the charge transfer between donor and acceptor upon excitation [11]. In
both these techniques, the rate charge transfer is determined to a large extent by the
properties of the donor and acceptor or the molecule-electrode coupling.
A more tricky method allows instead to “follow” the charge during its motion. This is
possible by generating the charges initially on the conjugated chains, this kind of charge
can move along polymer chains. Then the motion of the charge can be probed by optically
spectroscopy, i.e. detecting the motion of charges toward appended traps at the chain
ends. Being the charge localized on the chain, the motion is truly probed [11].
2.2 The “ability” of Self-Assembly
As we already noticed, to probe individual molecules electronically, one of the first
issues that arises is how to attach the probe electrodes to either side of a molecule. One
12

of the advantage to use molecule, allow the use of self-assembly process to adsorb the
molecule on the electrode. Self-assembly offers a selective chemistry where the covalent
attachment of the molecule’s linking group (i.e., S, O, N and today known as “alligator
clip”) can be used to bind the molecule selectively to specific transition metals. There are
many potential alligator clip groups, one of the most used is sulfur headgroup of thiolates
R-S (e.g., -alkanethiols) adsorbed to Au to form self-assembled monolayers (SAMs), which
is nominally complete one molecule thick layer of adsorbates where a nonreactive “tail”
functionality prevents further growth normal to the surface. When an Au surface is
exposed to an organothiol, a strong covalent bond (1.9eV ) forms between sulfur and gold.
The SAM is formed by two driving forces: the formation of a strong covalent bond between
the metal substrate (in this case Au) and the linking group (S) that binds the molecule to
the surface, and the close packing of the hydrocarbon chains as a result of van der Waals
interactions between adjacent hydrocarbon tails. These interactions form a well-ordered,
energetically favorable, though kinetically overlayer structure [5].
Figure 2.3: Cross-sectional schematic of SAM on a substrate
Because alkanethiols are easily assembled in SAMs and chain length and chemical
functionality can be selected or modified, they have been used as model systems for
measurements of electron transport through molecules.
The alkanethiol SAM Metal-Insulator-Metal system has been widely electronically
characterized. Making use of the Franz two-band model, the effective mass of the electron
tunneling through the molecular wires can be deduced by knowing the barrier height of
the metal-SAM-metal junction [5] or all the band diagram can be obtained.
13

2.3 Molecular Devices
In order to produce a device, candidate molecules have to satisfy some criteria, i.e., the
presence of extended π-conjugation (overlap and delocalization of electron orbitals). Good
molecules could be polyporphyrins, polythiophenes, polyphenylenes, and oligo(phenylene-
–ethynylene)s or (OPEs). They have been identified to have low HOMO–LUMO gaps,
they are fully conjugated ad furnished of terminal groups theat easily allow covalent bond-
ing to metal surfaces. In addition, substituents can be added to the molecule that modify
its electronic properties or its solubility (ease of processing). By manipulating these
components of the molecule, properties ranging from high-conductivity wires, rectifica-
tion, Negative Differential Resistance (NDR) 1
, bistable conductance have been demon-
strated [6].
Anyway not only the thought devices have the same probability of success. In fact,
while silicon microelectronics is based largely on three-terminal devices such as transistors,
molecular electronics will most likely remain limited to two-terminal devices because
the alignment and contact of two points of a molecule is already difficult to obtain.
Nevertheless, full logic systems can be developed from diode-based logic [6].
2.3.1 Indications in Fabrication Procedures
In molecular device fabrication, the main difference compered to CMOS industry is
the possibility to use a “bottom-up” approach. The first step in bottom-up manufacturing
is to create the devices and wires. Bottom-up, scalable integration techniques, such as
Langmuir–Blodgett films, random assembly, biologically assisted assembly, self-assembled
monolayers, catalyzed growth, can form random or regular structures but it is not possible
to predetermine exactly where a particular element will be located in the structure in a
deterministic way. A good example of bottom-up technique is the one used to create two-
dimensional meshes of wires. In one example of production, nanowires are suspended in a
fluid which flows down a channel. The nanowires align with the fluid flow and occasionally
stick to the surface they flow over. The average spacing between wires and the wire density
can be controlled by varying the rate and duration of the flow. By performing the flow
1
NDR is characterized by a decrease in current in respond to an increase in voltage, for some voltage
range. A famous example of device that shows NDR is Esaki diode. About NDR, typical molecular
performance exceeds what observed in solid-state quantum well resonant tunneling heterostructures
14

operation twice, first in one direction and then in the orthogonal direction, arrays of wires
at right angles can be formed. Another approach for creating crossbars is to use nanoim-
printing. Researchers at Hewlett-Packard (HP) have recently used this technique to make
a 64-b molecular memory [13].
The use of bottum-up approach forces the device architechs to accept some comprise.
I.e., it will be difficult to create precise alignment between components or deterministic
aperiodic structures. Two-terminal devices are preferred as well as connections made by
overlapping wires.
2.3.2 Molecular Rectifiers
The molecular rectifier represents the first idea for an electronic component consisting
of a single molecule, as described by Aviram and Ratner in 1974 and then realized by
Metzger in 1997 (fig. 2.4). The model consisted of D-σ-A system: an electronic donating
moiety (D) with low ionization potential, tetrathiafulvalene, connected to an accepting
group (A) with high electron affinity, tetracyanoquinodimethane, by an “insulating” σ
bonded spacer. The excited state D+
-σ-A−
would be relatively accesible from the ground
neutral state D-σ-A, while the opposite D−
-σ-A+
would lie several eV higher and then
inaccessible: this is the way to have an asymmetric resonant tunneling structure [9,11].
Figure 2.4: (a) Aviram-Ratner proposal for a single-molecule rectifier. (b) Molecular rectifier
realized by Metzger.
2.3.3 Molecular Switches
One example of a molecular switch is the photochromic switch consisting of a dithienylethene
molecule. The switch behavior is actived by photons in this case: if the molecule is illumi-
nated by UV light, thienyl rings assume a closed shape, closing the bridge, the molecule
can then be switched back to its open form by irradiation with visible light. A possible
15

use of this kind of switch is as a memory element, using the open and closed form as “on”
and “off” bits [11].
In literature, we can find more than one example of molecular switches. We will
describe the one reported in [9]. The molecule used was 2’-amino-4-ethynylphenyl-4’-
ethynylphenyl-5’-nitro-1-benzenethiolate (fig. 2.5(a), compound 1c). To contact them, a
well know chemical reaction that form the thiolate group upon exposure to Au was used.
A series of control experiments were performed with alkanethiol-derived SAMs. Typical
I(V) characteristics of a Au-chain-Au device at 60 K are shown in fig.2.5.
Figure 2.5: (a) The active molecular compound 1c and its precursors. (b) IV characteristic
with the NDR peak.
Positive bias corresponds to hole injection from the chemisorbed thiol-Au contact and
electron injection from the evaporated contact. This device exhibits a robust and large
NDR. The performance exceeds that observed in typical solid state quantum well resonant
tunneling heterostructures.
Molecular Memories
A memory device operates thanks to the storage of a high or low conductivity state.
In [5], memories molecular devices with several molecules are suggested and compered.
The conductivity state has to be easy to write, fast to be read and to be erased, stable to
16

be stored in time. For the writing process, an high conductivity state is obtained upon the
application of a voltage pulse. Then if the applied bias is inverted, an erasing operation
is obtained (fig. 2.6).
Figure 2.6: An initial low conductivity σ state is changed into a high σ state upon application
of a voltage. The high σ state persist as a stored “bit” and can be erased inverting the sweep
of the voltage.
A characteristic bit retention time was measured, by monitoring one of the device at
various intervals. The state decay in time with an exponential constant of approximately
800 sec. Analyzing this decay in funtion of the temperature, it is shown that it follows a
1/T dependence, suggesting the existence of an activation energy for the memory effect
and then the presence of some traps. The observation lead to a simple model of the phe-
nomen: the different conductivity states could be caused by different trap levels occupied
by electrons (fig.2.7); different traps could be related either to different electronic states,
or to different geometric conformation in the molecule, or both.
Figure 2.7: Proposed model for the memory effect.
17

A complete implementation of this device in a memory circuit shows satisfactory
results, confirming the memory behavior of the molecular device.
2.3.4 Molecular Transistors
The switches mentionated above make use of conformational changes. This limits the
possible switching speed to a few kHz. An approach that should in principle allow much
faster switching speeds is a switch (or transistor) that relies on a single electron transfer.
Aviram was a precursor also in this case, proposing first in 1988 a field-effect transistor
consisting of a single molecule (fig.2.8(a)). This transistor consists of a semiconduct-
ing piece of polythiophene, connected to a doped (oxidized) piece of polythiophene that
charge transfer between these two parts of the molecule is inefficient. The oxidized poly-
thiophene is conducting and the nondoped polythiophene will be nonconducting up to a
certain threshold voltage, but the application of an electric field can result in tunneling
of an electron between the two parts. In this way, the conduction of both polythiophene
channels can be switched by application of an electric field [11].
Figure 2.8: (a) Aviram proposal for a single-molecule transistor consisting of a photoconductor
coupled to a conductor. (b) A central quantum dot unit connected to three electrodes by
conjugated chians.
Another example could be constituted by a quantum dot unit (i.e. a single thiopene
ring) connected to three conjugated arms (fig.2.8(b)). When two arms are connected to
electrodes, the dot acts as a tunneling barrier. However the barrier can be modulated
by applying a potential to the third terminal, which behaves as a “gate”. In this case a
18

switching frequency up to 10THz can be reached [11].
SET Transistor
The Single-Electron Tunneling (SET) transistor represents on of the best answer to
the problems of CMOS down-scaling. It has small size and small power consumption.
One big disadvantage is that is not so straightforward to work at room temperature.
Implementation of SET transistor by use of cluster molecules has been suggested and
developed [2,14]. The possibility to use cluster molecules can be garanteed by the presence
of a great quantity of close-together molecular orbitals, that provides sufficient stability of
the cluster molecule skeleton when adding or removing an electron (electronic reservoir).
This can let us think to the molecule as a metallic quantum dot.
An implementation of this kind of device can be found in [2,14]. The more effective
technique was demostrated to be the one that made use of Langmuir-Blodgett technique
for cluster monolayer formation and cluster fixation to the surface. The topographic study
has been done by STM. The same instrument was used also for electrical characterization,
which showed the main single-electronic features, like staircase current-voltage curve and
Coulomb blockade.
Figure 2.9: (a) STM topography of the Tl-substituted carborane cluster molecule and (b)
corresponding projection on the substrate.
To make the SET transistor an electrode was added by electronic nanolithography
in order to control the tunnel current. Two completely different molecules were used as
the basis of a molecular tunnel nanosystem:the first from the metallorganic cluster family
(Tl-substituted carborane - fig. 2.9), the second a Pt5(CO)6 [P(C2H5)3]4 cluster. Such
19

cluster molecules represent ready ‘elementary cells’ for tunnel nanostructures (metal core
- electronic reservoir + the tunnel barrier - a ligand shell).
The control current through this molecular transistor change periodically for linear
increase of the control electrode voltage. According to the theory, this behavior is typical
for the SET transistor and the period corrispond to the change of an effective cluster
charge by one single electron charge value (fig.2.10).
Figure 2.10: IV curve in the case of tunneling via the cluster molecule.
An interesting feature of the reported result is that almost the same curve are observed
for several kinds of molecular cluster, despite their difference in the nature of their struc-
ture, in the size and above all in the value of HOMO/LUMO separation. This suggests
that the area of energy levels near HOMO–LUMO separation is not a governing factor for
the electron transport through the cluster molecule, and it determines only the common
shape of the IV curves [2]. Also really interesting is the fact that all the measurements
where carried out at room temperature.
2.3.5 Connecting Molecular Devices
It worths to stress that, in order to take advantages from the small dimension of the
molecule devices, also the connection among them should be of the same dimension.
It is convenient to have a piece of a conjugated polymer anologous to the polymers
used for the device itself (fig.2.11). The advantages of using molecular wires is related
20

Figure 2.11: (a-c) Examples of simple conjugated wires.
also to the possibility to characterize completely the rigidity and the torsional disorder
of the wire [11], using in the meanwhile the big pool of transport mechanism mentioned
above.
2.4 Molecular Electronics Circuits
The debate about the possibility of create a molecular circuit is open and rich. Such a
circuit has ﬁrst to satisfy some key issues: scalability to molecular dimensions; tolerance in
manufacting defects; integration of non-traditional fabrication methods; bridging achiev-
able; fabrication semplicity [4]; those are in addition to the traditional requirements, like
noise tolerance, ability to support fan-out (which means high input impedance and low
output impedance), and to requirements the device has to satisfy in order to be competive,
like high switching speed and low power [13]
The dominant circuit structure that has arisen from those considerations is the cross-
bar, an expanded ticktacktoe board formed from wires and having individual molecular or
molecular-scale devices sandwiched within the junctions. Both memory and logic circuits
have been demonstrated [4].
We can distinguish several kinds of circuit [13]:
• Resonant Tunnelin Diode Circuits
An NDR device used as a memory can be a signiﬁcant example of the basic unit of
this kind of circuit.
The circuit can only support a fan-out of one and has a low noise margin. NAND
and NOR gate have been demostrated by simulations. Its biggest limitation is that
it is almost impossible to integrate given current nanofabrication techniques.
21

• Crossbar Circuits
Regular crossbar architecture is nowaday the most promising one (fig.2.12).
Figure 2.12: Schematic of a crossbar.
One possible realization is an electrically configurable monolayer of bistable molecules,
i.e. rotaxane and catenane. The idea proposed by HP and UCLA, is an array
of crossed nanoscale wires with molecules present at each junction, forming two-
terminal devices that can be electrically configured to behave as low or high resis-
tance diodes. These molecules create a programmable computing fabric that can
be used for memories, logic arrays, etc A big advantage is the possibility of high
computational densities, that has to face the inability to achive signal gain and to
implement an inverter (due to the fact that most of the devices are designed to be
diode-like).
The crosspoint circuit alone is not sufficient for memory or logic, but it needs the
presence of a decoder that can provide the framework for addressing memory for
data storage [4,13].
It is interesting to think about CMOS/Nano Mixed Architectures. In fact, despite of
all the advantages of molecular electronics it is difficult to imagine a complete replacement
of Si indutry. A new technology cannot easy compete with silicon’s large-scale fabrication
infrastructure, proven design methodologies, and economic predictability [13].
An alternative approach is the integration of silicon with nanoelectronics. An extreme
approach see the CMOS used as the primary computation medium while the nano on top
is used as a supplement to better achieve integration goals. At the other extreme, the
nano portion would be the primary computation medium while the underlying CMOS
would be used simply to provide signal gain and latching capabilities.
22

Conclusions
With this quick overview on molecular electronics, we demostrated how much this
field could be an emerging solution for the bottleneck of the modern semiconductor in-
dustry. Highlighting on the implementation that has been done in the past decade, a
really versatile technology arose, demostrating both the possibility to produce and to de-
sign competive molecular devices making use of quantum physical concepts that in the
traditional device represent parassite process.
Nevertheless, still many disadvantages are present, suggesting that a first introduction
of molecular electronics in the productive process could see it integrated in actual CMOS
technology. The amount of advantages and the possibility of this first integration, lead
to a renovate interest in future studies on this topic.
i

Bibliography
[1] Lorente N. and Joachim C. “Architecture and Design of Molecule Logic Gate and
Atom Circuit”, 2013.
[2] Gubin S. P. et al. “Molecular clusters as building blocks for nanoelectronics: the ﬁrst
demonstration of a cluster single-electron tunnelling transistor at room temperature”.
Nanotechnology, 13:185–194, 2002.
[3] Prasongkit Jariyanne. “Molecular Electronics - Insight from An-Initio Transport
Simulation”, 2011.
[4] Heat James R. and Ratner Mark A. “Molecular Electronics”. Physics Today, pages
43–49, 2003.
[5] Chen J., Lee T., Su J., Wang W., and Reed M.A. “Molecular Electronic Devices”.
Encyclopedia of Nanoscience and Nanotechnology, X:1–30, 2004.
[6] Mantooth B.A. and Weiss P.S. “Fabrication, Assembly, and Characterization of
Molecular Electronic Components”. In Proceedings of the IEEE, volume 91, 2003.
[7] Wang W., Lee T., and Reed M.A.R. “Electronic Transport in Molecular Self-
Assembled Monolayer Devices”. In Proceedings of the IEEE, volume 93, 2005.
[8] Cuniberti G. and Fagas G. Richter K. Introducing Molecular Electronics, volume 680
of Lect. Notes Phys. Springer, 2005.
[9] Reed M. A. “Molecular Electronics - Current Status and Future Prospects”. FED
Journal, 2000.
[10] Winpenny R.E.P. “Strecht for a moment”. Nature nanotechnology, 8:159–160, 2013.
ii

[11] Grozema F. C. and Siebbeles L. D. A. Charge and Exciton Transport through Molec-
ular Wires, chapter “Introduction: Molecular Electronics and Molecular Wires”.
Wiley-VCH, 2011.
[12] Sharma I., Kamal E., Randhawa K., and Singh L.K. “Quantum Simulation Study of
DNA nucleotide Thymine for use in Molecular Devices”. IOSR Journal of Electronics
and Communication Engineering, 2012.
[13] Stan M.R. et al. “Molecular Electronics: From Devices and Interconnect to Circuits
and Architecture”. In Proceedings of the IEEE, volume 91, 2003.
[14] Soldatov E.S. and Gubin S. P. et al. “Room Temperature Molecular Single-Electron
Transistor”. 2001.
iii

Molecular Electronics

Recomendados

Recomendados

Más contenido relacionado

La actualidad más candente

La actualidad más candente (20)

Destacado

Destacado (8)

Similar a Molecular Electronics

Similar a Molecular Electronics (20)

Más de Valentina Ferro

Más de Valentina Ferro (8)

Último

Último (20)

Molecular Electronics