Molecular ElectronicsFor Fun and Profit(?)

Prof. Geoffrey HutchisonDepartment of ChemistryUniversity of Pittsburghgeoffh@pitt.edu

July 22, 2009

http://hutchison.chem.pitt.edu

Moore’s “Law:” Transistor Count Doubles Every 18 Months

From Intel Corp. circa 2007

Gordon Moore: Co-founder of Intel

Similar trends observed in chip performance,price per peformance, etc.

Moore-San’s Law

Image courtesy M. Ratner

Don’t Underestimate Silicon...

Actually, these are from 1997(!)

Current Intel Roadmap

45nm 32nm

International Technology Roadmap for Semiconductors (ITRS)

http://www.itrs.net/reports.html

Moore’s Second Law: Exponential Economics

0

1,200,000,000

2,400,000,000

3,600,000,000

4,800,000,000

6,000,000,000

1966 1985 1992 1995 2002 2007

2007:~$6 billion!

Cost of a building a new fabrication plant doubles every 3-4 years:

Advances in conventional lithography are quickly becoming cost-prohibitive!

Cost

(U

S$)

Molecular Electronics: Just Nanoscale?

M2+

M2+

M2+

M2+

M2+

3D: Bulk Films

2D: Monolayer

1D: Chains

0D: Individual Molecules

Why Organic Electronics?

Relative to Inorganic Materials:

Pros

Greater “tailorability” through synthetic chemistry

“Wet” deposition

Wide-area

Inexpensive

Flexible, lightweight

Synthesis = 6.02x1023

Cons

Lower electron mobility

Slower switching speeds

Hard to obtain complementary p/n logic

Integration into industry?

Broad Applications for Molecular Electronic Materials

Light-Emitting Diodes

Hole transport layers

Flexible anodes

Photovoltaic Devices

Thin-Film Transistors

Flexible circuits

“Smart” Windows

Fast color changes

Anti-Static Films

Device fabrication

Photographic film

Batteries: Not Just Inorganics...

Energy Density for Secondary Batteries

0 50 100 150 200 250Gravimetric Energy Density / Whkg-1

Volu

met

ric

Ener

gyD

ensi

ty /

WhL-1

0

100

200

300

400Sm

alle

r

LighterLead

Ni-CdNi-MH

Li-IonLi-Metal

PolymerLi-Ion

Reaction Mechanism for LIBs

ElectrolyteOrganic Solvent

plusSupportingElectrolyte

(ex. LiBF4/PC)

AnodeGraphite

(IntercalationLayer)

CathodeLithium

Metal Oxide(ex. LiCoO2)

AnodeCurrent

Collector

CathodeCurrent

Collector

: Li+: Doped Li+

Reaction Mechanism for LIBs

ElectrolyteOrganic Solvent

plusSupportingElectrolyte

(ex. LiBF4/PC)

AnodeGraphite

(IntercalationLayer)

CathodeLithium

Metal Oxide(ex. LiCoO2)

AnodeCurrent

Collector

CathodeCurrent

Collector

e-e-

: Li+: Doped Li+

Ink-Jet Printing of Organic Electronics

Sekitani et al. PNAS 2008 p. 4976

Konarka: Ink Jet Printing for Solar CellsMar. 4, 2008Ink-Jet Printing

Transistors

GE: Ink Jet Printing for DisplaysMar. 11, 2008

Organic Field-Effect Transistors

Dielectric

Gate Electrode

Source DrainSemiconductor

Vsd

Isd ~ 0.0 A “Off”

Isd = 10x A “On”VgateDielectric

Gate Elecrode

Source DrainSemiconductor

Applied Gate Potential

Vsd

Dielectric

Semiconductor

Organic Field-Effect Transistors

Dielectric

Gate Elecrode

Source DrainSemiconductor

Applied Gate Potential

Operation of an OrganicField-Effect Transistor

S

S S

S

n

N

N

O O

OO

X

X

S

S

R

R

n

S

S

n

OO

O ONH

HN

n

Example OrganicSemiconductor Materials

Dielectric

Gate Electrode

Source DrainSemiconductor

Dielectric

Semiconductor

Back to the Nanoscale...

M2+

M2+

M2+

M2+

M2+

3D: Bulk Films

2D: Monolayer

1D: Chains

0D: Individual Molecules

Si Substrate

SiO2

PbSe PbSe

Enough Bulk Conductivity:How Do You Wire Up a Molecule?

STM tip

Many, Many Methods...

Crossbars

Break Junctions

Example: Device Fabrication

Sixty 15 mm x 6 mm chips are fabricated on a thin (200 µm) Si wafer.

Bonding pads are defined by photolithography.

Critical features are defined by e-beam lithography.

A timed etch is used to suspend the junctions.

Ralph Group, Cornell

Electromigration: Making a 1-3nm Gap

Current

Voltage

Ralph Group, Cornell

Single Molecule Transistors

V

Drain(Pt)

N

N N

NN

NN

N N

NN

NN

N N

NN

N

RuRu

Source(Pt)

Gate (Al)

Vg

Gate Oxide (Al2O3)

Ralph Group, Cornell

Note: No Guarantee of 2 Connections

Metal / Molecule Interface

WARNING: PowerPoint Science!

One-Site

“Bridging”

Three-Site

Coordination

“Coulomb Blockade”

Source Drain

“Coulomb Blockade”

Source Drain

e-

“Coulomb Blockade”

Source Drain

e-

Source Drain

e- e-

“Coulomb Blockade”

Source Drain

e-

Source Drain

e- e-

SourceDrain

e- e-

Coulomb Blockade in [Ru2(tppz)3]+4

0.00 0.02 0.04

7.5

0.0

-7.5

Gate (V)

Bias (mV)

0 1250 2500

Conductance (nS)

4K Temp. Single-MoleculeCoulomb Blockade Transistor

DrainSource Drain

e-

Source

e-

Charge passes when source, drain and molecule bridge electronic

states are aligned

Charge does not pass when molecule bridge electronic states

are above or below

Vsd

Is it Really Single-Molecule?

few and eventually a single atomic chain.10 A conductance

histogram constructed from 1000 consecutive traces shows

clear peaks around 1, 2, and 3 ! G0 (Figure 1). When the

single atom chain is broken in the absence of molecules,

the conductance either decreases exponentially with the

electrode displacement due to tunneling across the gap

between the two Au electrodes or drops from G0 to below

our experimental detection limit possibly due to the broken

ends snapping back as the contacts relax (Figure 2a, yellow

traces).

The conductance histograms of three differently substituted

aromatics, 1,4-benzenedithiol, 1,4-benzenediisonitrile, and

1,4-benzenediamine, are compared in Figure 2b. For each

of the molecules, a large fraction of the individual traces

show additional steps at conductance levels below G0, as

shown in Figure 2a. Conductance histograms are constructed

from more than 3000 consecutively measured conductance

traces without any data selection or processing so as to

include all possible molecule/electrode conformations in the

statistical analysis.11 In comparison to the data for the dithiol

or the diisonitrile, the conductance histogram for 1,4-

benzenediamine is strikingly well-defined. Indeed, within our

experimental accuracy, the histogram differs from that of

Au only for conductance between 0.003 and 0.015 G0. This

implies that the additional steps occur only within this

conductance range. From a Gaussian fit to this peak (inset

Figure 2b), we deduce the most prevalent conductance value

for the molecule to be 0.0064 ( 0.0004 G0. A Gaussian

cannot be fit to the diisonitrile data although a broad peak

is visible in Figure 2b between 0.001 and 0.1 G0. The

conductance for the dithiol occurs over a wide range. The

resulting histogram is diffused showing no well-defined trend

in the conductance.12 (See the Supporting Information for

other control experiments.)

The broadening of the conductance histograms of the

diisonitriles and dithiols is not unexpected. Thiols are known

to bind to gold in a variety of motifs, for example, a-top

and threefold hollow sites. The binding of sulfur over a-top

and threefold hollow sites has been calculated to yield a

difference of up to a factor of 3 in the conductance.13

Moreover, the dithiols may contain oligomeric molecules

formed through oxidative disulfide formation. Isonitriles14

suffer from similar complications. They are known to

oligomerize easily15 and also have a variation in their binding

to the surface gold atoms.16

We have measured the conductance of a series of alkanes

with amine end groups that progresses from 1,2-ethylene-

diamine to 1,8-octanediamine. The conductance histogram

for each alkane (Figure 3a) displays a prominent primary

peak.17 We determined the conductance of these molecules

from the center position of a Gaussian fit to the primary

peaks.18 Figure 3b shows these conductance values plotted

against the number of methylene groups on a semilog scale

along with an exponential fit to the data. The clear

exponential dependence provides additional evidence that the

peak in the histograms corresponds to the conductance of a

Figure 1. Conductance histogram constructed from 1000 tracesmeasured while breaking Au point contacts in the absence ofmolecules on a linear scale. The bias is 25 mV, and bin size is10-4 G0. The inset shows five traces at arbitrary x axis offsets.

Figure 2. (a) Sample conductance traces measured without molecules (yellow) and with 1,4-benzenediamine (blue), 1,4-benzenedithiol(red), and 1,4-benzenediisonitrile (green) shown on a semilog plot. All data were measured at 25 mV bias, although no bias dependencewas found up to 250 mV. (b) Conductance histograms constructed from over 3000 traces measured in the presence of 1,4-benzenediamine(blue), 1,4-benzenedithiol (red), and 1,4-benzenediisonitrile (green) shown on a log-log plot. The control histogram of Au without moleculesis also shown (yellow). Histograms are normalized by the number of traces used to construct the histograms. Inset: same data on a linearplot showing a Gaussian fit to the peak (black curve). Bin size is 10-4 G0.

Nano Lett., Vol. 6, No. 3, 2006 459

Do

wn

load

ed b

y U

NIV

OF

PIT

TS

BU

RG

H o

n J

uly

20

, 2

00

9P

ub

lish

ed o

n F

ebru

ary

1,

20

06

on

htt

p:/

/pu

bs.

acs.

org

| d

oi:

10

.10

21

/nl0

52

37

3+

Venkataraman, L. et. al. Nano Letters 2006 v. 6, 458-462

You must do statistics....

You must do statistics...

Careful controls(“no molecules”)

Dilution Experiments

Is it Really A Molecule?

Is it a nanoparticle?

What are molecular “signatures?”

What are control experiments?

What are our statistics?

WARNING: Be Skeptical!We can do theory...

But what’s the real experiment?

Prospectus

Statistics, Statistics, Statistics!

Charge Transport Mechanisms

More Charge Transport

Ensembles & Dynamics

Solar Energy: Where things get (more?) complicated