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:
1022 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:
1029 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
kP
eqk 1 r i Rk 4 0
Two-electron term
kM
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 nbnbn VmnM am an VmnP bmbn 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 vk 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 vk an an V plel 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
PhSAg6
PhSAg11
PhSAg
• Is atomic scale roughness important for SERS? PhSAg9 PhSAg7
• What are the processes contributing to chemical enhancement?
• How large is the chemical PhSAg8
enhancement factor for a specific system? PhSAg10
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
PhSAg9 (I)
2
towards cluster (non-bonding interactions of C and Ag)
0 12 PhSAg9 (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 E1 X env
EM2 Energy levels
E Forbidden transition
1 M
X
New states
env
EM1 E1 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”