Chemical effects in surface-enhanced Raman scattering (SERS) Semion K. Saikin Department of Chemistry and Chemical Biology Harvard University

Ohio University

May 26, 2011

Outline • Introduction

• Canonical transformation approach to SERS

• Role of molecular adsorption

• Molecular profiling using nanostructures

Outline • Introduction

• Canonical transformation approach to SERS

• Role of molecular adsorption

• Molecular profiling using nanostructures

Fingerprinting molecules Raman scattering 101

wscatter= wlaser-Wvibration

wlaser

Wvibration

M. Ito, J. Chem. Phys. 42, 2841-2848 (1964)

• Molecules can be characterized by a specific pattern of lines in Raman spectra

Raman scattering in molecules Schematic picture

E

Energy

Virtual State

wfield

Excited electronic state Scattered field

Incoming field

wfield-Wvibration G

Ground electronic state

Normal coordinate, Q

Fingerprinting molecules

Raman spectra

Cocaine Glucose Caffeine

A. G. Ryder, G. M. O’Connor and Th. J. Glynn J. Raman Spectrosc. 31, 221–227 (2000)

• Can be used to separate good molecules from bad molecules

Fingerprinting molecules

H2O2??

• Can be used to explore unknown structures and phenomena

However, Raman scattering is very inefficient

Raman scattering in molecules How efficient it is? Rhodamine 6G Resonance excitation Raman cross-section @ 532nm:

  1022 cm2

To observe a single Raman scattered photon from the molecule with the laser intensity 50 mW/cm-2  1017 ph/s/cm2 one has to wait about 70 days!

Benzene

Lowest electron transition in UV Raman cross-section @ 785 nm: 

 1029 cm2

The waiting time extends to 2 million years

Surface-enhanced Raman scattering Molecules on nano-structured noble metal surface

Ag-on-”black silicon” substrate Mazur group, Harvard University

IntensitySERS ≈ 109 IntensityNeat per molecule

Ag nanoparticles Van Duyne group, Northwestern University

E.D. Dieblold et. al., J. Am. Chem. Soc. 131, 16357 (2009)

Single-molecule Raman spectroscopy is possible J.P. Camden et. al., J. Am. Chem. Soc. 130, 12616 (2008)

Surface-enhanced Raman scattering How it works Chemical: Charge transfer between the molecule and metal modifies the response

E

Electromagnetic: Collective oscillations of electrons in metal induce strong nearfields at the molecule

Ag

Total enhancement

=

Electromagnetic enhancement

×

Chemical enhancement

Can we develop a model with both mechanisms included?

Outline • Introduction

• Canonical transformation approach to SERS

• Role of molecular adsorption

• Molecular profiling using nanostructures

Partitioning the Hamiltonian M = molecule P = particle

+

Single-electron terms 2  1   pi  eA(ri )  H el     2me 4 0 i  

Molecular Hamiltonian: includes chemisorption

 kP

eqk 1  r i  Rk 4 0

Two-electron term



kM

eqk r i  Rk

Interaction: excitation and electron transfer

  1 1    2 4 0

 ij

e2 r i rj

Particle Hamiltonian: includes plasmons

Partitioning the Hamiltonian Single-electron terms a+

b+

b

a



*  H1   Em am am    nbnbn  VmnM am an  VmnP bmbn   Tmn am bn  Tnm bn am m

n

single-electron energies

m n

m n

scattering by external nuclear potential

m,n

tunneling



Plasmons Electron Hamiltonian Coulomb gauge: H P   Ei ai ai  1  vk vk k  k  1 i

v v 2

2 k 0

Single-electron Coulomb term energies

k  0, n

k k

an an

self-interaction

Canonical coordinates Pk and Qk are introduced

Hamiltonian gauge: scalar potential  = 0 H P   Ei ai ai  i





1 1 1 2 * * w Q Q  P P  v v    vk vk an an  V plel    P k k k k k k k k 2 k k c 2 k k c 2 k  0, n Plasmon modes

Subsidiary condition: Pk  ivk k  pl  0

screened Coulomb term

plasmon-electron scattering

D. Bohm & D. Pines, Phys. Rev. 92, 609 (1953)

How energy is transferred? Förster & Dexter mechanisms: P m k P

M m’

P

M

k’ M

m

m’

k

k’ M

P

particle

molecule

particle

Coulomb-assisted charge transfer: P m k

P

M m’

P m

k’

k

P

P m’ k’

P

M

B. N. J. Persson, Chem. Phys. Lett. 82, 561 (1981)

particle

molecule

molecule

Molecule-particle coupling F

The Förster term: H F 

 (k )P  dˆ   

k  k c ,i

i

k

i

k  k c ,i , j

ij

(k )Pk  qˆij  H FSE

Molecular dipole and quadrupole coupled to the plasmonic modes

Plasmon state

Molecular state

Molecular transitions Coupled to single-electron Excitations in the metal

CT

The charge transfer term: H CT 

 

m , k ,k k c

mk

(k ) Pk am bk  c.c.

Plasmon Surface Molecular state state state

Surface-enhanced Raman scattering 0, E

Near-field coupling Förster interaction

, G

P, G

Incoming field

Coulomb-assisted charge transfer

Scattered field

• Förster interaction couples the

0,G

metal

• Incoming optical fields excite plasmon modes in a metal particle

plasmon with molecular excited states

molecule

• The charge transfer term couples the plasmon with redox molecular states

Surface-enhanced Raman scattering Intensity borrowing picture ~

Schrieffer-Wolff transformation: H  e S HeS ~ E2

Metal borrows vibrational structure from the “molecule”

~ E1

~ P

The “molecule” borrows transition dipoles from metal

G

Surface-enhanced Raman scattering Metal-molecular complexes CT

Plasmon state

Surface Molecular state state

Locality

• Computational methods can be used to study the response

Outline • Introduction

• Canonical transformation approach to SERS

• Role of molecular adsorption

• Molecular profiling using nanostructures

Raman response of metal-molecular complexes Test system: benzenethiol on silver clusters

PhSAg6

PhSAg11

PhSAg

• Is atomic scale roughness important for SERS? PhSAg9 PhSAg7

• What are the processes contributing to chemical enhancement?

• How large is the chemical PhSAg8

enhancement factor for a specific system? PhSAg10

S. K. Saikin, R. Olivares-Amaya, D. Rappoport, M. Stopa and A. Aspuru-Guzik, PCCP 11, 9401–9411 (2009).

Raman spectra & Raman excitation profiles

Intensity (a.u.)

Isolated molecule

0

500

1000

1500

2000

2500

3000

3500

Cross section, 10-28 (cm2)

Wavenumber (cm-1)

10

w

w

w

w

Strongest SERS modes: w1 - ring breathing mode w2 - ring deformation mode w3 - C-S stretching mode w4 - ring stretching mode

PhSH

   4  ~ (w  wQ )  W Q

0.1 

All Four Modes Behave Similarly

1E-3 1.6

1.8

2.0

2.2

2.4

2.6

Excitation energy (eV)

2.8

3.0

Molecule, cluster, complex Electron excitation spectra 0.25 0.20

• Molecule: lowest transitions are in UV

0.15 0.10 0.05

Oscillator strength

0.00 2.5

1

2.0

0.1

1.5

0.01

1.0

1E-3

0.5

1E-4 1.0

• Cluster: Electronic excitations are in the visible part of the spectrum 1.5

2.0

2.5

3.0

0.0 2.5

Complex: • Strong transitions are quenched • All transitions have metal-molecular

1

2.0

0.1

1.5

0.01

1.0

1E-3

0.5

1E-4 1.0

1.5

2.0

2.5

3.0

0.0

1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 Excitation Energy (eV)

character • Excitation spectra are very sensitive to the structure

Metal-molecular excitations in SERS

Oscillator strength

10 times enhancement

PhS-Ag10

1000

Cross section, 10

Off-resonance shift

-28

cm

2

Raman excitation profile & absorption spectrum

10

Resonance effects

 

0.1

1000 times enhancement

1E-3 1 0.1

1.84 eV

2.17 eV

0.01 1E-3 1E-4 1E-5 1.6

1.8

2.0 2.2 2.4 2.6 Excitation energy (eV)

2.8

3.0

• SERS spectra are sensitive to the location of a specific excitation

Off-resonance effects Orientation Effect, PhS-Ag9

Scattering cross section, 10

-34

2

(cm )

Short-wave infrared excitation, 2000 nm 4

• Benzene ring is oriented

PhSAg9 (I)

2

towards cluster (non-bonding interactions of C and Ag)

0 12 PhSAg9 (II)

8

• Benzene ring is oriented away from the cluster

4 0 500

1000

1500

2000

-1

Wavenumber (cm )

• Off-resonance SERS spectra are sensitive to the molecular orientation

Can we identify chemical effects in experiments?

Raman spectra modifications SERS vs. Raman Benzenethiol on Au substrate neat

Benzenethiol molecule

w3=1002 cm-1

w4=1026 cm w1=415 cm-1 w6=1584 cm-1 -1 w2=700 cm w5=1094 cm

Intensity (a.u.)

-1

-1

SERS

• Lines are enhanced differently • Vibrational modes are shifted

400

800 1200 Wavenumber (cm-1)

1600

• Some lines appear/disappear

S. K. Saikin, Y. Chu, D. Rappoport, K. B. Crozier, and A. Aspuru-Guzik, J. Phys. Chem. Lett. 1, 2740 (2010)

Double resonance plasmonic structure Schematic picture

SEM image

Au disk array d

p

SiO2 spacer

Au film glass substrate

• The disk diameter d determines the localized plasmon resonance

frequency • The period of the disk array p determines the coupling between the

localized and propagating plasmons

Chu, Y.; Banaee, M. G.; Crozier, K. B. ACS Nano 4, 2804 (2010) Chu, Y.; Crozier, K. B. Opt. Lett. 34, 244 (2008)

Electromagnetic effects Extinction spectra & near fields

600

Measured

Computed (FDTD)

excitation 785nm

700

800 900 Wavelength (nm)

Intensity (a.u.)

Intensity (a.u.)

p = 350 nm p = 500 nm p = 770 nm p = 780 nm

1000

600

Solid line – extinction Dotted line – near field at a hotspot

700

800 900 Wavelength (nm)

1000

• Computed extinction spectra agree with the measured spectra

• Near-field profiles computed at hotspots are similar to the extinction spectra

Correction for plasmonic profile (PhSH/Au)

Intensity (a.u.)

Benzenethiol PhSH

Corrected SERS spectra

Intensity (a.u.)

SERS spectra

p = 350 nm p = 500 nm p = 770 nm p = 780 nm 200 400 600 800 1000 1200 1400 1600 -1

Wavenumber (cm )

200 400 600 800 1000 1200 1400 1600 -1 Wavenumber (cm )

• Spectra corrected for plasmon enhancement are reproducible for different substrates

Spectral modification due to adsorption Measured data Neat sample

SERS

w3 = 1002 cm-1

w4 = 1026 cm w1 = 415 cm-1 w2 = 700 cm-1

400

800

Cross section (a. u.)

Cross section (a. u.)

-20

-1

w5 = 1093 cm-1

1200 -1

Wavenumber (cm )

-9

-2 +6

w6 = 1585 cm-1

1600

-3

-6

400

800

1200

1600

Wavenumber, cm-1

• Lines are enhanced differently Strongest enhancement

• Vibrational modes are shifted

Spectral modification due to adsorption Model structures

PhS-Au1 PhSH

on-top binding motif

Isolated molecule

PhS-Au9 (I)

PhS-Au2 bridge binding motif small structure

PhS-Au9 (II)

bridge binding motif larger structures

(PhS)2-Au8 staple binding motif

Spectral modification due to adsorption Computed results Isolated molecule

PhS-Au9

w3

w5

w6

w4 w1

Cross section (a.u.)

Cross section, a.u.

-21

-7

+6 -1

w2

400

800

0

-4

1200

1600

Wavenumber (cm-1)

• Computed spectral modifications are consistent with the measured results

400

800

1200

1600

Wavenumber, cm-1

Strongest enhancement

Mode shifts: theory vs. experiment Modes 10

(a)

Measured

Wavenumber (cm-1)

0 -10 -20 -30 10

(b)

w1 = 415 cm-1

w2 = 700 cm-1

w3 = 1002 cm-1

w4 = 1026 cm-1

w5 = 1094 cm-1

w6 = 1584 cm-1

Computed

0 -10 -20 -30

w



w



w

w





w



w



Modes

Legend:

Experiment PhS-Au9(I)

PhS-Au1 PhS-Au9(II)

PhS-Au2 (PhS)2-Au8

• Molecule-cluster model describes mode shifts quantitatively

Mode enhancement: theory vs. experiment 15 Relative enhancement (a.u.)

10

Measured

5 0 15

Computed orientation-averaged

10 5 0 30

Computed orientation-fixed

20 10 0

w

w

w Modes

w

w

p = 350 nm p = 770 nm

p = 500 nm p = 780 nm

PhSH PhS-Au9 (I)

PhS-Au1 PhS-Au9 (II)

w

Legend: PhS-Au2 (PhS)2-Au8

• Molecule-cluster model describes relative mode enhancement qualitatively

Conclusions • It is possible to develop a unified approach for SERS that combines both electromagnetic and chemical effects.

• Chemical effects can be computed using a metalmolecular complex model • Chemical enhancement = static enhancement (10 times) + resonance enhancement (103-104 times) • Chemical effects can be identified in spectral modifications of analyte molecules • Chemical effects in SERS provide information about molecular adsorption

Outline • Introduction

• Canonical transformation approach to SERS

• Role of molecular adsorption

• Molecular profiling using nanostructures

Molecular spectroscopy and detection Raw sample Scattered field

In a complex environment Incoming field

Scattered field

Incoming field

collective excitation

• Molecules interact with the optical fields directly

• Interaction with fields is mediated by the environment

• Ensemble averaging over different molecules and orientations

• Signal is dominated by a small group of molecules, orientation may be important

• Optical fields are homogeneous

• Intermolecular coupling mediated by environmental modes may be sufficient

• Intermolecular coupling is small

Environment-assisted probing Probing field and signal enhancement Raman spectra

Plasmonic substrate neat

SiO2 Computed near field

Intensity (a.u.)

Au

SERS

400

800 1200 Wavenumber (cm-1)

1600

ISERS /Ineat ≈ 107-109 per molecule Dr. M. Banaee, Crozier’s group

• Allows for detection of small concentrations of analyte molecules

Environment-assisted probing Intensity borrowing Molecule

Environment + Molecule  EM2    E1 X env

EM2 Energy levels

E Forbidden transition

1 M



X

New states

 env

 EM1    E1 X env

Allowed transition

GM



 GM    G X env 

• Allows for probing intra-molecular transitions forbidden by selection rules • Provides information about system-environment interaction

Environment-assisted probing Spatial coherences Intramolecular coupling

Superradiant response Incoming field

Scattered field

• Environment excitations can induce long range coherences/correlations in single molecules and between molecules

Can we introduce a new paradigm for probing of molecular structures in engineered environments?

Plasmons

Molecule Excitons

Optical cavities

Acknowledgements

Theory and computation: Roberto Olivares-Amaya Dmitrij Rappoport Michael Stopa Alan Aspuru-Guzik

Experiment: Yizhuo Chu Eric Diebold Paul Peng Mohamad Banaee Eric Mazur Kenneth Crozier

Defense Threat Reduction Agency

Defense Advanced Research Project Agency

Thank you!

Supplementary slides

Testing TDDFT method Molecules, Time-dependent density functional theory Raman spectra of Benzenethiol

Absorption spectra Experiment, solution G. Di Lonardo, C. Zauli J.Chem.Soc.(A) 1306 (1969)

Cross section (a.u.)

Experiment: The Crozier group 532 nm excitation

0

500

1000

1500

2000

2500 Wavenumber (cm-1)

3000

Intensity (a.u.)

Computation: Turbomole basis: def2-QZVP functional: PBE0

3500

Computation: Turbomole basis: def2-QZVP functional: pbe0 Line width: 100 meV 32

34

36

38

40

Wavenumber, 103 (cm-1)

• Computed spectra agree with the available experimental data

42

44

Testing TDDFT method Metal clusters, Time-dependent density functional theory Absorption spectra Computation: Turbomole basis: def-SV(P) functional PBE0 Line width: 100 meV

Intensity (a.u.)

Ag7

Ag7 in Ar-matrix

Ag20

1.5

2.0

Ag20 in Ar-matrix

2.5 3.0 Energy (eV)

3.5

Experiment: J.Chem.Phys. 129, 194108 (2008)

Experiment: PRB 78, 075439 (2008)

4.0

Energy (eV)

• Computed spectra agree with the available experimental data

Qualitative analysis Au1

Au2

Isotropic components of polarizability derivatives are modified as compared to PhSH

S C1

C6

C2 C5 C3

C4

static charge transfer 0.3e

Qualitative analysis

Isotropic components of polarizability derivatives with respect to the C–S and C–C bond stretching modes for PhSH and PhS–Au2.

Au1

Au2

Au1–S

S C1

C6

C2 C5 C3

Mode

C4

PhS–Au2 0.06

PhSH 0

C1–S

-0.21

-0.04

C1–C2

0.21

0.04

C2–C3

-0.2

0

C3–C4

0

0.05

Chemical distortion Chemical modifications: - new Au-S bonds - ground state charge transfer

Chemical distortion is described reasonably well just static changes in polarizability derivatives. Why?

In experiment Reference mode

w4 = 1026 cm-1 w1 = 415 cm-1 w2 = 700 cm-1

400

800

w5 = 1093 cm-1

1200

Wavenumber, cm

w5 Cross section, a. u.

Cross section, a. u.

w3 = 1002 cm-1

1600

w3

w1

RE total (w n )  w4

I SERS (w n ) I R (w3 ) I SERS (w3 ) I R (w n )

w2

w6 = 1585 cm-1

-1

w6

400

800

1200

Wavenumber, cm

1600

-1

In theory (metal-molecular complexes) Orientation-averaged: RE

A chem

d C (w n ) d PhSH (w3 ) (w n )  d C (w3 ) d PhSH (w n )

No specific orientation of a metalmolecule complex with respect to the excited plasmon is assumed.

Orientation-fixed: 2

(w ex  w n ) 4 w3   zzC (w n )  d PhSH (w3 ) F   RE chem (w n )  (w ex  w3 ) 4 w n   zzC (w3 )  d PhSH (w n )

Metal-molecule complex is oriented in such a way to model a response from a “hot spot”