Mesoscopic Physics

Victor Moshchalkov (coordinator) Andr Stesmans - Michel Houssa

OUTLINE

I. Introduction to the mesoscopic scale II. Electronic transport in mesoscopic systems III. Graphene and carbon nanotubes IV. Magnetism and superconductivity at mesoscopic scale

I. Introduction to the mesoscopic scale

Meso (greek) means in-between Mesoscopic scale lies in-between microscopic and macroscopic world

Microscopic
100 102 104 106 108

Macroscopic
1010 1012

Number of atoms

Mesoscopic
Atom pendulum

Quantum mechanics

Classical mechanics

A typical example : electrical resistance

The electrical resistance R of a (macroscopic) 3D metallic bar of length L and cross section A=W*d is given by

L R= Wd

d I L
+ -

is the electrical resistivity of the material (a materials property independent of its dimensions). In the 2D-limit (e.g. for d0), the resistance is given by

L R= W

The resistance is thus expected to scale with L and W-1. In this classical picture, the resistance can take any value, depending on the materials resistivity In mesoscopic systems, the resistance is quantized,

h R=n 2 2e
Quantization of R is a consequence of the quantization of the electron energy levels in the system
G (e2/h)

-2.2

-2.0

-1.8

-1.6

-1.4

Vg

Mesoscopic scale is thus related to the size of a system where quantum mechanical effects can be observed In the case of electrical transport, the classical expression of the resistance breaks-down when the size of the system becomes small compared to 3 characteristics lengths - the electron wavelength - the electron mean free path - the phase coherence length

Electron wavelength In quantum mechanics, electrons are described by their wave function (r,t)
(r , t ) = Aeik .r eit

k: electron wave-vector : pulsation

k = 2 /

= E/

is the electron wavelength (also called de Broglie wavelength) Typical values of : 1-5 in metals 10-100 nm in semiconductors

Electron mean free path The electron mean-free path l corresponds to the typical distance traveled by an electron, subjected to an electric field, before encountering a collision or scattering event, leading to the change in its momentum p (or velocity v)

defect

p = mev
e

p = me v
e

The electron mean free path is strongly dependent on temperature. In the Drude model, l is related to the electron mobility through the expression

2 n l= e
l~100 m at 10 K

n: electron density

Electron mobility and density in a 2DEG

l~1 m at 300 K

GaAs/AlGaAs heterostructure

Phase-coherence length
The phase-coherence length L is related to the wave nature of the electrons. It represents the characteristic length the electrons can travel before their phase is changed or destroyed, through a collision. Consider an electron beam which is splitted into 2 paths, and then recombined. In a perfect crystal, the 2 paths would be identical, and interference is constructive.
Path 1

: average time by which the amplitude of the electron wave is reduced by exp(-t/) after time t

Path 2

If impurities are present in the 2 arms of the loop, or if electrons can be scattered by phonons, or other electrons, the interference of the 2 outgoing electron waves may be (partly) destructive, leading to the loss of the initial phase of the electrons (phase randomizing event).

Such phase relaxation effects have been observed e.g. in ring-shaped resistors. The temperature dependence of the inverse of the phase relaxation time is shown below, for the case of a GaAs sample.

Applying an AC magnetic field to the structure allows one to obtain the phase of the electrons

The phase coherence length can be calculated as

L = VF
This equation is valid if the phase relaxation time is of the same order or shorter than the momentum relaxation time, m. In this case, the motion of electrons over a phase-relaxation time is nearly ballistic. This is usually the case in high-mobility semiconductors. Typical values for GaAs are few hundreds of nanometers. In low mobility semiconductors (and also polycrystalline metal layers), momentum can relax much faster than the electron phase. In this case, electrons experience a diffusive motion during , and the phase relaxation length is then related to the diffusion coefficient D of the material

L = D

where

k BT D= e

(Einstein relation)

Examples of mesoscopic systems

III-V HEMT Graphene-baseddevices

Atomic clusters

Ag cluster

Example: atomic clusters

Aggregate of few to 105 atoms, also called nanoparticles Number of atoms 1-10 Molecule

Ag cluster

Cluster

TEM picture of a ~40 nm diameter silver cluster

106

Bulk material

Typical cluster sizes: a few nm to few hundreds of nm

Metal-insulator transition in metallic clusters A "macroscopic" metal is a very good conductor of electricity, but what about a cluster with a finite number of atoms?

BULK METAL

CLUSTERS

ATOMS & MOLECULES

kT defines an energy to be compared with the energy gap of a particle (Kubo gap) Approximate energy spacing :

EF n

Hg:(Xe)4f145d106s2

Electronic density of states of Hg clusters

p-states s-states

Work function/electron affinity of Hg clusters

High workfunction, typical of an insulator

Work function/electron affinity (eV)

Vacuum level WF EF Metal WF VB Insulator CB

Low workfunction, typical of a metal

Cluster size-induced metal/insulator transition

Experimental realization of nano-scaled structures a. Top-down approach Photolithography and etching Photolithography consists in transfering a pattern, drawn on a mask, on a layer or a substrate. A typical lithography process used for patterning a metal gate on an oxide layer is illustrated below
1) Photo-resist
Oxide

2)

Light (UV) exposure

Mask
Metal

Semiconductor Deposition of a photoresistive layer Exposure of the photoresist through a mask

3)

4)

Etching of the resist exposed to light

Etching of the (unprotected) Al layer

5) Patterning of MOS structures

Resist removal

The photo-resist is a material (polymer) which undergoes a chemical transformation when exposed to light. The exposed part of the resist can then be dissolved in an appropriate solvent (developer).

In order to pattern nanostructures (with typical sizes of few tens of nanometers), the projection lithographic process is used. In this case, a system of lenses is used to project the image of the mask on the layer to pattern. The minimal size lm of the pattern that can be transferred on the layer is related to the numerical aperture of the optical system and the wavelength of the light source
Reticle Optical system Focal plane

lm

NA

For a lens, NA=sin

Layer

Decreasing wavelength allows to pattern smaller structures

X-rays
600

Minimum feature size (nm)

500 400 300 200 100 0 0 100

UV light

200 300 Wavelength (nm)

400

500

Potential of X-rays for patterning nm sized structures

During the fabrication of nanostructures, a photolithography step is usually followed by an etching step. Wet etching is based on the chemical attack (dissolution) of the layer to pattern in a solution.

Material to etch

Chemicals

Etching rate (m/min)

Si

1 ml HF 5 ml HNO3 2 ml CH3COOH 28 ml HF 170 ml H2O 113 g NH4F H3PO4 1 ml HNO3 4 ml CH3COOH 4ml H3PO4 1 ml H2O

7.4

SiO2

0.1

Si3N4 Al

0.01 0.035

A limitation of the wet etching process lies in the fact that it is an isotropic process, i.e. the material is etched in each directions with the same etching rate, i.e.
vl Af = 1 0 vv

where vl and vv represents the etching rate along the lateral and vertical directions, respectively. If Af1 (vv>>vl), etching is anisotropic

Isotropic etching

Anisotropic etching

Isotropic etching of a layer leads to the partial loss of the resolution of the structure to be patterned on the layer, called over-etching. This effect has to be avoided when structures of few tens of nm have to be fabricated.

Resist

Resist

over-etching

Over-etching can be avoided by using a dry etching process, which consists in exposing the layer to be patterned to a plasma, containing chemically active ions. The plasma is generated from a gas, which is ionized by a high electric field (usually an AC electric field pulsed at radio frequencies). Dry etching of the layer then proceeds from a chemical component (due to the active ions like F- or Cl-) and a physical component, due to the ablation of the material induced by its bombardement with the high energy ions.
Layer to be patterned

Substrate holder under bias

The chemical component is isotropic and selective towards different materials, while the physical component is anisotropic and non or weakly selective. Proper adjustment of the properties of the plasma (chemical content, pressure, energy of the ions impacting on the surface) allows to maximize selectivity and anisotropy of the etching process. As an example, using CF4 and O2 enables the selective etching of Si with respect to SiO2 with a selectivity ratio of 12:1.

Material to etch Si SiO2 Si3N4 Al, Ti, W Resist

Chemicals CF4+O2, CCl4, Cl2 CF4+H2, CHF3+O2 CF4+O2 CCl4, SiCl4, BCl3+Cl2 O2

b. Bottom-up approach Self-assembly Self-assembly corresponds to the spontaneous organization of pre-existing components (atoms or molecules) by means of electrostatic or hydrophobic forces, this self-organization process allowing to reduce the energy of the system. Starting from atoms/molecules to build up a functional structure or device, so called bottom-up approach. Inspired from biology: DNA
Adenine Guanine Bonding through hydrogen bonds

DNA molecule

Cytosine

Thymine

A typical example of self-assembly: alkanethiols on a gold surface Alkanethiol: an organic chain, terminated by an S-H (functional) group
R (C,H)

S H

The self-assembly of alkanethiols on Au substrates originates from the strong Au-S bonding (~1.7 eV), which follows from the dissociation of the S-H group on the surface:

1 R S H + Au R S Au + H 2 2

The process then leads to the formation of an ordered monolayer of R-S on the Au surface, due to the tetrahedral arrangement of the S-H bond on the surface, as well as the (weak) Van der Waals interactions between the alkanethiols molecules.

30

Growth of carbon nanotubes (CNT) from catalysts Chemicalvapordeposition(CVD)ofCNT,usingFe,CoorNi nanoparticlesascatalysts.TypicalcarbonsourceisC2H2andthe growthprocesstakesplaceatabout1000C.

Fe, Co, Ni nanoparticles CNT grown on nanoparticles

1012 1013 CNTs/cm2

CNT on Fe nanoparticles

Combining top-down and bottom-up approaches: CNT interconnects

Top-down: definition of vias and trenches on dielectric layers Bottom-up: growth of CNT

Via level

Trench level

catalysts at bottom of via

catalysts at side wall

Replacement of Cu by CNT in nanoelectronic devices?

References
Y.Imry Introductiontomesoscopicphysics (OxfordUniversityPress,2002). C.P.PooleandF.J.Owens, IntroductiontoNanotechnology (Wiley,2003). M.Wilson,K.Kannangara,G.Smith,M.Simmons,andB.Raguse, NanotechnologyBasicscienceandemergingtechnologies (CRCPress,2002).