Columbia University in the City of New York

2012 Einstein Scholar Lecture for Chinese Academy of Sciences

Peking University May 21Peking University, May 21CAS National Center for Nanoscience and Technology, Beijing, May 17

CAS Suzhou Institute for Nanotechnology & Nano-bionics, May 11

New Columbia Interdisciplinary Science Building 2011

• This tutorial lecture describes unifying and basic principles of nanoscience, and also p p ,presents our most recent research on graphene highly doped by interfacial charge g p g y p y gtransfer.

Things ManmadeThings Manmade

The Scale of Things The Scale of Things –– Nanometers and MoreNanometers and More

Things NaturalThings Natural Things ManmadeThings Manmade

Head of a pin1-2 mm

Ant~ 5 mm

The Challenge

1 cm10 mm10-2 m

10-3 m e

1,000,000 nanometers= 1 millimeter (mm)

MicroElectroMechanical

gg

Fly ash~ 10-20 mHuman hair~ 60-120 m wide

mm

Dust mite

200 m

OP

O

O

owor

ld 0 01 mm

0.1 mm100 m10-4 m

Micr

owav

e MicroElectroMechanical (MEMS) devices10 -100 m wide

Red blood cells(~7-8 m)

O O OO

O OO O OO OO

O

S

O

S

O

S

O

S

O

S

O

S

O

S

O

S

Fabricate and combine l b ildi bl k

Mic

ro d 0.01 mm10 m10-5 m

10-6 m

Visib

le

1,000 nanometers = 1 micrometer (mm)

Infra

red

Zone plate x-ray “lens”Outer ring spacing ~35 nm

Red blood cells

Pollen grain

~10 nm diameter Nanotube electrode

nanoscale building blocks to make useful devices, e.g., a photosynthetic reaction center with integral semiconductor storage.

0.01 m10 nm

0.1 m100 nm10-7 m

10-8 m

Nan

owor

ld

Ultra

violet

Self-assembled,Nature-inspired structureMany 10s of nm

Quantum corral of 48 iron atoms on copper surfaceiti d t ti ith STM ti

ATP synthaseNanotube electrode

Carbon nanotube~1.3 nm diameter

0.1 nm

1 nanometer (nm)10-9 m

10-10 m

N

Soft

x-ra

y

Carbon buckyball

~1 nm diameter

Atoms of silicon

DNA~2-1/2 nm diameter

carbon nanotubessemiconductor nanocrystals

positioned one at a time with an STM tipCorral diameter 14 nm

Office of Basic Energy SciencesOffice of Science, U.S. DOE

Version 05-26-06, pmd

siliconspacing 0.078

nm

~2-1/2 nm diameter

Nanoscience: Graphene, Carbon Nanotubes andNanoscience: Graphene, Carbon Nanotubes and Semiconductor Nanocrystals

Louis BrusColumbia University

New York, NY

CdSeNanocrystal

Graphene Electron Correlation and Charge TransferGraphene Electron Correlation and Charge Transfer

1. Single electron electronic structure, optical Properties and Electron Correlation inProperties and Electron Correlation in Graphene: comparison with Nanocrystals and SWNT.

2. Charge Transfer Doping and Fermi Level Shift by Adsorbed and Intercalated Species. Asymmetric doping on one side only. Extreme Electron Doping by Alkali atoms.

731 million transistors on one Intel Core i7 chip!!

M l l Di iMolecular Dimensions

Intel Corp.Intel Corp.

Nanoscience: The evolution from l l t b lk t lmolecules to bulk crystals

Intellectual challenge: how to quantitatively describe electronic properties as a function of size? Physics concepts? Chemistry p p y p yconcepts?

Synthetic challenge: can we actually make well defined nanocrystals and nanotubes ? Can we vary the surface chemistry?and nanotubes ? Can we vary the surface chemistry?

Characterization challenge: can we measure physical properties one nanocrystal or nanotube at a time?

Technology Challenge: Nanocrystals and Nanotubes are new families of large molecules. Can we use them in technology?

Electrons: Evolution from Molecular Orbitals to Continuous Bands

Local chemical bonding around each Si atom is the same in the

l l d l h d lmolecule and crystal – tetrahedral covalent sp3 hybridization

Nanometer crystallite should have larger band gap than bulkhave larger band gap than bulk

crystal

Bulk Crystalline SiliconBulk Crystalline Silicon

Disilane molecule

Silicon Nanocrystal

Quantum Size Effect – “Particle in a box”

E |c>

Empty conduction band

E k

.band

g k

|v>Fullvalenceband

De Broglie Electron Momentum mv=hk=h/λ

band

electron wavelength λ quantized in particle:

L. Brus, J. Chem. Phys. 79, 5566 (1983); 80, 4403 (1984)

Synthesis 1986 Steigerwald etal, JACS 110, 3046 (1988)

Synthesis 1988 Bawendi etal J. Chem. Phys. 91, 7282 (1989)

Carbon Nanoscience

Graphene and Single Wall Carbon Nanotubes

Hückel molecular orbitals for P electrons

Butadiene

π*

π

From Wikipedia

k l d * ( d ) hHuckel π and π* MOs (Band structure) of grapheneMolecular Orbital Energies are a function of electron momentum k(x,y) in the plane of g ( y) p

graphene

Electron momentum k is continuous for infinite plane of graphene

beta = 3.033 eVResonance integralResonance integral

π*

Optical transitions: π to π* just as in

π aromatic molecules

Saito and Kataura, Topics Appl. Phys. 2001, 80, 213

Graphene: atomically thin SP2 carbon network

Graphene Band Structure(Wallace, 1947)

Density of StateDensity of State

EDensity of StateDensity of State

• Atomically thin 2D system

• Zero carrier mass: high mobility

N2D(E)• Graphene is zero-gap semiconductor

Zero carrier mass: high mobility

• Density of State is zero at E = 0.

Philip Kim

Graphene electronic absorption π interband transitionsπ interband transitions

No band gap continuousEEE

No band gap – continuous absorption in IR, visible, and UV

kxky

kxky

kxky

Optical (sheet) conductivity:

2

2

(1)

2eh

= (π/4) G0

2

2 3%e

Optical absorption:

A( ) 2.3%137c

V. P. Gusynin et al., PRL 96, 256802 (2006)S. Ryu et al., PRB 75, 205344 (2007)D. S. L. Abergel et al., PRB 75, 155430 (2007)

A(ω) =Mak and Heinz

Continuous electronic absorption across the IR, Visible and UltravioletNo band gap

Absorption increases in UV at M saddle point

I d d l d l l iIndependent electron model: no correlation

E0

2 ln 1DDOS

0

Symmetric line shape near the saddle point, assuming

0

ln 1SPDOS E no electron correlation

Mak and Heinz

An Exciton ( electron-hole bound state) with 0.25eV Binding Energy forms at the M saddle point, even though the system is metallic

5/2h) Experiment

Ab initio with excitonic

3

4

ts o

f e2

effect (50 meV) Ab initio with excitonic

effect (250 meV)The universal optical

1

2

(in

uni

t The universal optical conductivity

0 1 2 3 4 5 6 70

Photon Energy (eV)

Existence of strong e-h interactions at the graphene saddle pointQuasiparticles’ lifetime near the M-point ~ 2.6 fsAlso Chae etal, (von Klintzing group) Nano Lett. (2011) 11, 1379.Fano lineshape at saddle point

Yang et al. PRL 103, 186802 (2009)Mak and Heinz

F ti l Q t H ll Eff tAndre Geim & Philip Kim

Strong Correlation: Graphene Fractional quantum Hall effect at low temperature

Fractional Quantum Hall Effect____ he2

12

15

h1

5

10

(k

)

____ he2

16

____ he2

110

h1

Classic signature of Correlated Electron Motion at low temperature

0

5

esis

tanc

e ____ he2114

h1

Robust effect implies that electron-electron interactions are essentially unscreened.

-5

Hal

l Re h1 ____

e2-14

e2____ h1

-10

h1

Bolton et al, Nature 2009, 462, 192Du et al, Nature 2009, 462, 192.

-15

-10____ he2

1-6

____ he2

1-2

Remarkable properties all res lt from 15 -50 0 50Vg (V)

T= 1.5K, B= 9TPhilip Kim

Remarkable properties all result from strong aromatic pi chemical bonding

If electron correlation is strong in graphene, it should be stronger in semiconducting SWNT:

Lower dimensionality

Existence of a Band Gap; less screening

(7,12) Chiral Semiconducting Tube

Semiconductor Carbon NanotubesIndependent Electron Model: no correlation

Metallic:

Si Q ti ti i N t bSize Quantization in Nanotubes:

Electron momentum k

Semiconducting:

quantized around circumference.

Electron momentum remains

empty

fill dElectron momentum remains continuous along length

filled

Correlation: Neutral Bound Excitons Due to electron-hole attraction

h

h

Exciton envelope wavefunction:

Neutral excited state moves as a unit along the SWNT

Exciton Bound states belowthe van Hove Band Edge

Two photon luminescence excitation spectra shows strongly bound exciton

F W l S i 308 838(2005)

band gap 1.7 eV

F. Wang etal, Science 308, 838(2005)

1g 2u

(6 ) bContinuum states

E2p – E1s0 31 eV

(6,5) nanotube

dt = 0.76 nm

E

0.31 eV

Ebinding0.43 eV

F i Poly(phenylene vinylene) ~ 0.35 eVSemiconductor nanowires ~ tens of meV

For comparison:

Why? Increased electron-electron interaction in Nanostructures

Two independent factors: dimensionality and screening

1. Dimensionality with no change in screening:y g g

3D bulk semiconductor: weak Coulomb, excitons not important2D confinement in plane: Coulomb interaction up by 4x1D confinement on line: Coulomb interaction diverges!Low Dimensionality implies increased electron-electron correlation

0D t d t i diff t di i ti l l t d ti0D quantum dot is a different case: no dissociation, less correlated motion, Finite Coulomb interaction, kinetic energies larger

2 Screening of Coulomb interaction:2. Screening of Coulomb interaction:

reduced screening in 1D, almost full screening in 0DgReduced screening implies increased electron-electron correlation

Brus, Nano Letters 10, 363 (2010)

Quantum Dots compared to Carbon Nanotubes

4 CdS l 1 C b N b4nm CdSe nanocrytal

Particle in box orbitalsSurface states and ligands

1nm Carbon Nanotube

Plane wave basis orbitalsNo surface states or ligandsf g

Rigid structureHard to collect photocurrent

B d 2 2 V

f gRigid structure: minor Franck-condon shiftEasy to collect photocurrent

B d G 1 7 VBand gap: 2. 2 eVKinetic energies 0.4 eVCoulomb energy 0.1 eV

Band Gap: 1.7 eVKinetic Energies 0.4 evCoulomb Energy: 0.8 eV

Carbon Nanotubes apparently have the strongest electron-electron interaction of any Nanosystem

Enhanced impact ionization

Decay of accelerated “hot” electrons by exciton creationrather than phonon creation

Hot Carrier Impact Exciton generation rate 10+15 s-1

1. High MEG: Either increase electron-electron interaction or slow down vibrational relaxation of “hot” Electrondown vibrational relaxation of “hot” Electron

Direct electrical current collection from strong MEG in Carbon Nanotubes

Nobel Prize in Physics 2010Nobel Prize in Physics 2010Andre Geim and Konstantin Novoselov

For GrapheneFive years from discovery to Nobel Prize!!Five years from discovery to Nobel Prize!!

A simple Nobel Prize experiment

Micro-mechanical exfoliationusing “Scotch tape”

1L2L

g p

3L

5 m

Ki h hit 300 SiO /SiKish graphite on 300 nm SiO2/Si

32

-Scientific American@ Columbia University

Novoselov, Geim et al. PNAS 102, 10452 (2005)

Large Scale CVD Grown Graphene

Jing Kong Group (MIT)A. Reina et al., Nano Letters (2009)

ne/ N

i

Byung Hee Hong Group (SKKU)Gra

phen

Byung Hee Hong Group (SKKU)

K. Kim et al., Nature (2009)

Rod Ruoff Group (Austin)Cu

Rod Ruoff Group (Austin)

X. Li et al., Science (2009)

raph

ene/

CG

r

Molecules adsorbed on Graphene:

Strong charge transfer electrical dopingStrong charge transfer electrical doping

Strongly doped graphene is very different than neutralStrongly doped graphene is very different than neutral graphene:

Increased in plane DC conductivity, and decreased optical absorption in IR and Visible

Strong G band Resonance Raman scattering,coupled to interband electronic transitions

1000

800

1000

unit)

n-layered 1 2 3

G D* Micro-Raman setup

400

600

ensi

ty (a

rb. u

0

200Inte

D

• G mode: C-C stretching (~1580 cm-1)

1300 1400 1500 1600 2600 2700 2800

Raman shift (cm-1)

- A. Ferrari et al. PRL (2006)- J. Yan et al. PRL (2007)…

Resonant and off-resonant intensity enhancement:D. Basko, N. J. Phys. 11, 095011 (2009)

35

• D mode: ring-breathing (~1350 cm-1) activated by structural defects – sp2 carbon to sp3 carbon • D* mode: overtone of D mode (~2700 cm-1) indicates thickness

Doping bleaches the electronic spectrum, and shifts the Raman G band e G b d

Uniqueness of graphene:

• All electrons are confined in a single atomic layer:

large electron density change for given gate voltagevoltage.

• Small density of states close to Dirac point:

large fermi energy shift for given electron density changechange.

EF shift of hundreds of meV:

Bleach Optical transitions into near pinfrared and visible

Shift position of G Raman lineFeng Wang

Failure of Born-Oppenheimer Separation of Vibrational and Electronic Degrees of Freedom

- Phonon energy renormalization: G when |EF| - Narrowing of the the G band width

g

J. Yan et al. PRL 98, 166802 (2007)Yan et al, PRL 102, 136804(2008)

Pinczuk groupCharge transfer shifts much larger

NO2 Adsorption: Surface Hole Doping

pristine graphene

EF

Crowther et al, ACS Nano 6, 1865 (2012)

2L shows only one G line: Gas adsorption on both top and bottom layers

Surface Hole doping extends inside only 1-2 layers

One sided doping of Bilayer Graphene on h-BN

Th S h BN i i bl k diff i f NO• The Strong graphene-BN interaction blocks diffusion of NO2between graphene and BN, introducing one-side doping to bilayer (2L) graphene.

1200

G+

BN2L Graphene

NO2 Raman Spectray ( ) g p

Electric Field

80030 torr NO2si

ty

G-90 torr NO2

BN

Si/SiO2

Optical Image

400

2

2 torr NO2

0.1 torr NO2

Inte

ns

BN

Optical Image

1560 1580 1600 16200

2

Bilayer Graphene

Raman Shift (cm-1) Bilayer Graphene

Direct observation of Optical Bleach Due to Hole Doping

Bleaching of electronic absorption due to surface hole doping

In 1L graphene, Fermi level shift Of b 0 75 V f b h h ROf about 0.75 eV from both the Raman Shift and the electronic bleach

Bl h d t h l d iBleach due to hole doping

Molecules between the layers:Graphite Intercalation Compounds

FeCl3, Br2, I2, NO2 K, Rb, Li, Ca

Donor: Alkali Metal, Alkaline earth metals lanthanidesearth metals, lanthanides.

Acceptor: Lewis acid Intercalants(Halogens, Acidic Oxides, Strong Bronsted acids)

• More than 100 reagents can be intercalated into graphite.

44Dresselhaus. Adv. Phys. 51, 1 (2002)

Bromine forms a stage 2 Bulk Intercalation CompoundBromine

Graphene

• Stage # : The number of graphene layers per intercalate layer

pIntercalate Layer

Stage # : The number of graphene layers per intercalate layer.• Doping level varies layer to layer

Dresselhaus. Adv. Phys. 51, 1 (2002)45

Br2  Vapor Exposed Graphene: both intercalation and adsorption 

G

Bulk

4L

y(a.

u) 1612

1612

3L

2LInte

nsity

1624

1612

16201613

1590 1620 1650

1L

Wavenumber

1624

I t i i 1580

• G peak up‐shift as charge density of graphene increases.• Stage 2 Intercalation observed similar to Bulk HOPG.

Intrinsic 1580

g• Bromine intercalates 3L and higher nL• 1L G peak shifts the most due to two layers of adsorption, on the top and bottom of graphenebottom of graphene.• 2L, 4L, and Graphite show same charge transfer amount.

N. Jung et al Nano Letters 9, 4133 (2009)

Very high electron doping: stage 1 alkali intercalationGruneis PRB 79 205106 (2009)( )

Fermi level shifts for C8Cs, C8K and C8Rb are respectively 1.0, 1.4 and 1.8eV. Graphene Layers are decoupled AA stacking.  No band gap at K point. 

Fermi level at M point, which is folded back on Gamma point by crystalline alkali intercalation layers

Interband optical absorption completely gone

Intraband Drude free electron optical absorption strong

Extreme Electron Doping in bulk  Alkali Metal GICs

• Shifts in Fermi level in intercalation compounds can be quite large: Fermi level shifts for Stage 1 C8Cs, C8K and C8Rb are respectively 1.0, 1.4 and 1.8eV. Layers are also decoupled. Expect complete electronic1.4 and 1.8eV. Layers are also decoupled.  Expect complete electronic bleaching in the visible spectrum. 

• Free electrons on metal layers and graphene create a 2D Drudeelectron gas with a plasma edge in the visible spectrum

)/()(

2

D

p

i

)( D

GIC dielectric constant in the visible d IR i ll fit b D d fand IR is well fit by a Drude free

carrier model with , effective plasma frequency and mean carrier lifetime.

Extreme Bleaching of 2L graphene visible interbandoptical absorption, at very high electron charge transfer

0.4

Pristine 2LK doped 2Lp

Stage 1 Simulation Stage 1 - 15% wp decreaseSt 2 Si l ti

0.2

Stage 2 Simulation

Con

trast

0.0400 600 800

Wavelength

Brus group

2

K doped Graphene 161L

Graphene: Drude intraband optical absorption of free electrons

1

K doped Graphene

Con

trast

4L6L

8L

400 500 600 700 8000

2

2L

1 131Ltra

st

Rb doped Graphene

1 131L

8L

Con

400 500 600 700 8000

Wavelength(nm)

5L

Bulk alkali intercalated graphite:  G band Raman Downshift and Fano Lineshape

Continuumelectronic band2 K-doped GraphiteRb-doped Graphite

E2g(G) phonon

Rb doped Graphite

ized

y(

a.u.

)

1

Nor

mal

iIn

tens

ity

900 1200 1500 1800

Wavenumber(cm-1)

• The BWF (Breit‐Wigner‐Fano) Raman shape of G mode in bulk graphite is due to coupling of discrete G phonon to the continuous interlayer electronic p g p yband along the kz axis.

51

New Raman Spectra for alkali doped Few layer GraphenesJung etal, Nano Letters (2011), 11, 5708

ty(a

.u.)

K doped 2L

lized

Inte

nsit

Rb doped 2L

Nor

mal

1000 1500

Pristine 2L

Wavelength(nm)Wavelength(nm)Bulk Fano Raman Lineshape Develops only slowly with increasing thickness

New resonance Raman effect, not from Basko Analyzed interband optical transition

Less electron transfer for few layer graphenes

Intrinsic silent graphene modes activated in Raman by p(2x2) alkali metal zone folding

K Doped 4L

nsity

(a.u

.)

11301260

Inte

n

Rb Doped 4Lp(2x2) Superlattice

M K1000 1500Wavenumber(cm-1)

p(2x2) Superlattice

Г• Potassium and Rubidium intercalated  graphene develops the identical Raman peaks at 1130 and 1260 which is due to folded M point Raman.

53

K and Rb intercalated Few layer GrapheneK-doped Rb-dopeddoped b doped

• While adsorbed potassium fully doped graphene as intercalated layer, adsorbed rubidium does not dope graphene much as potassium does. 54

Conclusions:

Both CdSe Nanocrystals and SWNT show simple quantum y p qconfinement. Electron Correlation is high in graphene and SWNT, but low in CdSe nanocrystals.

High graphene charge doping from molecular absorption, without application of voltage

B d b di f f d i d hBand bending from surface doping extends one or two graphene layers.

One sided chemical doping can be achieved on BN substrateOne sided chemical doping can be achieved on BN substrate

With K and Rb stage 1 intercalation, complete bleaching of interband optical absorption occurs in few layer graphenesinterband optical absorption occurs in few layer graphenes. Drude optical plasma edge of doped electrons observed in visible.

High charge transfer of stage 1 K intercalated graphite develops only slowly as a function of the number of graphene layers.

Thanks to NSF, DOE, Keck Foundation, AFOSR, ll b d i ll d !collaborators and especially students!

Graphene students and postdocs:

Sunmin Ryu, Stephane Berciaud, Haitao Liu, Andrew Crowther, Naeyoung Jung, Li Liu, Zheyuan Chen, Yinsheng Guo, Elizabeth Thrall