Supporting Information: Plasmonic Substrates Do Not Promote Vibrational Energy Transfer at Solid-Liquid Interfaces

Jan Philip Kraack*, Laurent Sévery, S. David Tilley, and Peter Hamm‡ Department of Chemistry, University of Zurich, Winterthurerstrasse 190, CH-8057

Corresponding Authors *phili[email protected] Phone: +41 44 63 544 77 Fax: +41 44 63 568 38

‡

[email protected]

Phone: +41 44 63 544 31 Fax: +41 44 63 568 38

SI 1

I. Determination of Enhancement Factor on Au, ITO and TiO2 The determination of the enhancement factors is analogous to the procedures described in refs.1–4 First, the stationary ATR IR absorption spectra are measured in situ in the ATR sample cell (Fig. SI 1 (a), (c), and (e)) for monolayer samples (black) as well as for a bulk solution reference samples (red). Second, nonlinear pump-probe spectra at about 1 ps population delay are measured from monolayer as well as bulk solution samples (Fig. SI 1 (b), (d), and (e)). Calculating the ratios between the pumpprobe signals and the ATR IR absorbances yields information about the “nonlinear signal per mOD stationary absorption” for each sample.1,2 Finally, the ratio between these values from the monolayer sample to the reference sample yields the enhancement factor, which is independent of the surface coverage. For the rhenium carbonyl complex on 0.5 nm thick Au layers we obtain an enhancement

in-situ ATR absorption / OD

0,04

0,5

(a)

(b) 0,0

0,02

Au 0,00 1960 0,04

-0,5

ML bulk 1980

Au 2000

2020

2040

(c) 0,02

ITO ML bulk

0,00

2000 0,04

ML bulk x100

-1,0

norm. pump-probe signal

in-situ ATR absorption / OD

in-situ ATR absorption / OD

factor of 31. For ITO and TiO2 layers we obtain enhancement factors of about 2 and 1, respectively.

2020

1960 1

1980

2020

2040

(d) 0

ITO

ML bulk

-1

2000

2040 0,5

(e)

2000

2020

2040

(f)

0,0 0,02

TiO2

0,00 1960

-0,5

ML bulk 1980

TiO2 -1,0 2000

2020

2040

1960

detection frequency / cm-1

ML bulk 1980

2000

2020

2040

probe frequency / cm-1

Figure SI 1. Determination of the enhancement factors for the rhenium carbonyl complexes on (a) and (b) Au, (c) and (d) ITO, as well as (e) and (f) TiO2 layers. (a), (c), and (e) are in-situ measured stationary ATR IR absorbances of the monolayer (black) and a bulk solution reference samples (red), whereas (b), (d), and (e) are normalized nonlinear pump-probe signals at about 1 ps population delay. Note that in (b) the bulk solution sample is multiplied by a factor of 100. SI 2

II. Surface Characterization of ITO and Au layers by use of Scanning Electron Microscopy Characterization of the conducting sputter-coated Au and ITO layers was performed by scanning electron microscopy imaging on a Zeiss Auriga 40 Scanning Electron Microscope (SEM, Carl Zeiss, Oberkochen, Germany) using an In-lens secondary electron detector and an acceleration voltage of 5 kV. Sputter-coating results in the generation of nanoparticle layers (light regions) on top of the CaF2 substrates. The surfaces are equally covered with nanoparticles of sizes between 3 – 10 nm for both samples. Similarly sized gaps (dark regions) exist between the nanoparticle for both samples.

(a)

(b)

50nm

50nm

Figure SI 2. Scanning electron microscopy analysis of the sputter-coated (a) Au and (b) ITO layers. Note that the Au substrate has been coated with 2 nm of Carbon on top of the Au layer to obtain high enough conductivity that allows a high-resolution SEM analysis. All images have been acquired with an acceleration voltage of 5 kV. The magnification is 300 k for (a) and 306 k for (b) resulting in image pixel sizes of 372.2 pm and 363.8 pm, respectively.

SI 3

III. Spectral Characterization of the Plasmon Resonance in Au Layers As demonstrated in Figure SI 2, the sputter-coating process results in the generation of Au nanoparticles on the CaF2 substrates. These arrays of nanoparticles are plasmonically active and result in tuneable surface enhancement also in the IR spectral range.2 The maximum of the plasmon resonance of the nanoparticles is, however, located in the visible spectral range at about 625 nm (Figure SI 3).

wavenumbers / cm-1

24000 20000 0,2

16000

OD

(a)

0,1

0,0 400 450 500 550 600 650 700 750 800

wavelength / nm

Figure SI 3. UV-VIS Spectral characterization of the plasmon resonance of sputter-coated Au layers as used in this study. A broad and featureless plasmon resonance exists with a maximum at about 625 nm.

SI 4

IV. Coupling of Two Dipoles on a Metal Surface In order to calculate the effective coupling of two transition dipoles sitting atop of a metal surface, we start with the electrostatic energy5 𝑉

=

∫

|𝐸⃗ | 𝑑 𝑟

(S1)

where 𝐸⃗ is the electric field generated by the contributing dipoles. In our concrete case, the integral goes over the upper half-space (𝑧 > 0) only, since the electric field vanishes inside the metal. For simplicity we furthermore assume that the dielectric constant outside the metal is =1. The overall electric field is the sum of four contributions, i.e., the fields from the two dipoles 𝜇⃗ and 𝜇⃗ and their corresponding image dipoles 𝜇⃗ and 𝜇⃗ , respectively (Figure 4a): 𝐸⃗ = 𝐸⃗ + 𝐸⃗ + 𝐸⃗ + 𝐸⃗

(S2)

Expanding Eq. S1, we obtain two terms: 𝑉 𝑉

=

∫

=𝜀 ∫

|𝐸⃗ | + |𝐸⃗ | + |𝐸⃗ | + |𝐸⃗ | 𝑑 𝑟 =

∫ |𝐸⃗ | + |𝐸⃗ | 𝑑 𝑟

(S3)

𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ 𝑑 𝑟

(S4)

In Eq. S3, the contributions of the image dipoles in the upper half-space (𝑧 > 0) are the same as those of the real dipoles would be in the lower half space (𝑧 < 0), if the metal would not be present. The lhs. of Eq. S3 therefore is the same as that of just the two dipoles without metal, which is what the rhs. of Eq. S3 describes with the integral going over the full space. 𝑉

thus represents the infinite (in the

case of point dipoles) but distance-independent self-energy of the two dipoles, which can be subtracted out. The six terms of 𝑉 , on the other hand, describe the pairwise interactions between the four dipoles, as indicated by the six arrows in Figure 4a. We group these terms as follows: 𝑉, +𝑉

,

≡𝜀 ∫

(

𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ 𝑑 𝑟 = 𝜀 ∫ 𝐸⃗ 𝐸⃗ 𝑑 𝑟 =

≡ 𝑉,

)

(S5)

where the integral in the rhs. goes over the full space for the same reason as in Eq. S3, and describes the normal dipole-dipole coupling without the presence of any metal surface. Here, r is the distance between the two dipoles on the surface (see Figure 4 (b)). The coupling of each dipole with its own image dipole is given by: 𝑉, +𝑉 , ≡𝜀 ∫ where the two terms 𝑉 ,

𝐸⃗ 𝐸⃗ +𝐸⃗ 𝐸⃗ 𝑑 𝑟 = 𝜀 ∫ 𝐸⃗ 𝐸⃗ 𝑑 𝑟 = −2 and 𝑉 ,

(

(

)

≡ 𝑉,

)

(S6)

are equal due to the symmetry of the arrangement, and at the

same time only half of the size of the usual dipole-dipole coupling, since the integrals go over the SI 5

upper half-space only. The two terms thus can be lumped together into one term with the integral going over the full space (rhs.), i.e., as if the metal surface was not present. Finally, the coupling to the corresponding other image dipole is given by: 𝑉, +𝑉 , ≡𝜀 ∫

𝐸⃗ 𝐸⃗ +𝐸⃗ 𝐸⃗ 𝑑 𝑟 = 𝜀 ∫ 𝐸⃗ 𝐸⃗ 𝑑 𝑟 =

) /

(

(1 − 2cos 𝜙) ≡ 𝑉 (,

)

(S7) with tan 𝜙 = 𝑟/2𝑅. As in Eq. S6, both terms are identical and can be lumped together. We see that the configuration in Figure 4 (a) can be reduced to that of Figure 4 (b), removing the (

metal and having only three effective interaction terms 𝑉 , (

and 𝑉 ,

)

)

(

, 𝑉,

)

(

, and 𝑉 , (

contribute to the transition dipole coupling of Eq. 1, while 𝑉 ,

)

)

(

. The terms 𝑉 ,

)

does not. However, since

the distance between the two dipoles r (4-5 Å) is much closer than that to the image dipoles (

√4𝑅 + 𝑟 (20 Å) in our concrete experiment, 𝑉 ,

)

dominates and the presence of the metal

surface hardly affects the vibrational energy transfer rate. It is important to keep in mind that the image dipoles are only a construct that enforce that the electric field lines enter the metal surface perpendicularly. What is happening in reality is surface charges that are induced at the metal/liquid interface. They generate an electric field that equals that of the image dipoles in the upper half-space and at the same time cancel out the field of the real dipoles in the lower half-space. The derivation given above is therefore not affected by that change of perspective.

SI 6

References (1)

Donaldson, P. M.; Hamm, P. Gold Nanoparticle Capping Layers: Structure, Dynamics, and Surface Enhancement Measured Using 2D-IR Spectroscopy. Angew. Chemie-International Ed. 2013, 52, 634–638.

(2)

Kraack, J. P.; Kaech, A.; Hamm, P. Surface Enhancement in Ultrafast 2D ATR IR Spectroscopy at the Metal-Liquid Interface. J. Phys. Chem. C 2016, 120, 3350–3359.

(3)

Kraack, J. P.; Hamm, P. Vibrational Ladder-Climbing in Surface-Enhanced, Ultrafast Infrared Spectroscopy. Phys. Chem. Chem. Phys. 2016, 18, 16088–16093.

(4)

Kraack, J. P.; Hamm, P. Surface-Sensitive and Surface-Specific Ultrafast Two-Dimensional Vibrational Spectroscopy. Chem. Rev. 2017, 117, 10623–10664.

(5)

Jackson, J. D. Classical Electrodynamics; John Wiley & Sons Inc.: New York, 2007.

SI 7

Jan Philip Kraack*, Laurent Sévery, S. David Tilley, and Peter Hamm‡ Department of Chemistry, University of Zurich, Winterthurerstrasse 190, CH-8057

Corresponding Authors *phili[email protected] Phone: +41 44 63 544 77 Fax: +41 44 63 568 38

‡

[email protected]

Phone: +41 44 63 544 31 Fax: +41 44 63 568 38

SI 1

I. Determination of Enhancement Factor on Au, ITO and TiO2 The determination of the enhancement factors is analogous to the procedures described in refs.1–4 First, the stationary ATR IR absorption spectra are measured in situ in the ATR sample cell (Fig. SI 1 (a), (c), and (e)) for monolayer samples (black) as well as for a bulk solution reference samples (red). Second, nonlinear pump-probe spectra at about 1 ps population delay are measured from monolayer as well as bulk solution samples (Fig. SI 1 (b), (d), and (e)). Calculating the ratios between the pumpprobe signals and the ATR IR absorbances yields information about the “nonlinear signal per mOD stationary absorption” for each sample.1,2 Finally, the ratio between these values from the monolayer sample to the reference sample yields the enhancement factor, which is independent of the surface coverage. For the rhenium carbonyl complex on 0.5 nm thick Au layers we obtain an enhancement

in-situ ATR absorption / OD

0,04

0,5

(a)

(b) 0,0

0,02

Au 0,00 1960 0,04

-0,5

ML bulk 1980

Au 2000

2020

2040

(c) 0,02

ITO ML bulk

0,00

2000 0,04

ML bulk x100

-1,0

norm. pump-probe signal

in-situ ATR absorption / OD

in-situ ATR absorption / OD

factor of 31. For ITO and TiO2 layers we obtain enhancement factors of about 2 and 1, respectively.

2020

1960 1

1980

2020

2040

(d) 0

ITO

ML bulk

-1

2000

2040 0,5

(e)

2000

2020

2040

(f)

0,0 0,02

TiO2

0,00 1960

-0,5

ML bulk 1980

TiO2 -1,0 2000

2020

2040

1960

detection frequency / cm-1

ML bulk 1980

2000

2020

2040

probe frequency / cm-1

Figure SI 1. Determination of the enhancement factors for the rhenium carbonyl complexes on (a) and (b) Au, (c) and (d) ITO, as well as (e) and (f) TiO2 layers. (a), (c), and (e) are in-situ measured stationary ATR IR absorbances of the monolayer (black) and a bulk solution reference samples (red), whereas (b), (d), and (e) are normalized nonlinear pump-probe signals at about 1 ps population delay. Note that in (b) the bulk solution sample is multiplied by a factor of 100. SI 2

II. Surface Characterization of ITO and Au layers by use of Scanning Electron Microscopy Characterization of the conducting sputter-coated Au and ITO layers was performed by scanning electron microscopy imaging on a Zeiss Auriga 40 Scanning Electron Microscope (SEM, Carl Zeiss, Oberkochen, Germany) using an In-lens secondary electron detector and an acceleration voltage of 5 kV. Sputter-coating results in the generation of nanoparticle layers (light regions) on top of the CaF2 substrates. The surfaces are equally covered with nanoparticles of sizes between 3 – 10 nm for both samples. Similarly sized gaps (dark regions) exist between the nanoparticle for both samples.

(a)

(b)

50nm

50nm

Figure SI 2. Scanning electron microscopy analysis of the sputter-coated (a) Au and (b) ITO layers. Note that the Au substrate has been coated with 2 nm of Carbon on top of the Au layer to obtain high enough conductivity that allows a high-resolution SEM analysis. All images have been acquired with an acceleration voltage of 5 kV. The magnification is 300 k for (a) and 306 k for (b) resulting in image pixel sizes of 372.2 pm and 363.8 pm, respectively.

SI 3

III. Spectral Characterization of the Plasmon Resonance in Au Layers As demonstrated in Figure SI 2, the sputter-coating process results in the generation of Au nanoparticles on the CaF2 substrates. These arrays of nanoparticles are plasmonically active and result in tuneable surface enhancement also in the IR spectral range.2 The maximum of the plasmon resonance of the nanoparticles is, however, located in the visible spectral range at about 625 nm (Figure SI 3).

wavenumbers / cm-1

24000 20000 0,2

16000

OD

(a)

0,1

0,0 400 450 500 550 600 650 700 750 800

wavelength / nm

Figure SI 3. UV-VIS Spectral characterization of the plasmon resonance of sputter-coated Au layers as used in this study. A broad and featureless plasmon resonance exists with a maximum at about 625 nm.

SI 4

IV. Coupling of Two Dipoles on a Metal Surface In order to calculate the effective coupling of two transition dipoles sitting atop of a metal surface, we start with the electrostatic energy5 𝑉

=

∫

|𝐸⃗ | 𝑑 𝑟

(S1)

where 𝐸⃗ is the electric field generated by the contributing dipoles. In our concrete case, the integral goes over the upper half-space (𝑧 > 0) only, since the electric field vanishes inside the metal. For simplicity we furthermore assume that the dielectric constant outside the metal is =1. The overall electric field is the sum of four contributions, i.e., the fields from the two dipoles 𝜇⃗ and 𝜇⃗ and their corresponding image dipoles 𝜇⃗ and 𝜇⃗ , respectively (Figure 4a): 𝐸⃗ = 𝐸⃗ + 𝐸⃗ + 𝐸⃗ + 𝐸⃗

(S2)

Expanding Eq. S1, we obtain two terms: 𝑉 𝑉

=

∫

=𝜀 ∫

|𝐸⃗ | + |𝐸⃗ | + |𝐸⃗ | + |𝐸⃗ | 𝑑 𝑟 =

∫ |𝐸⃗ | + |𝐸⃗ | 𝑑 𝑟

(S3)

𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ 𝑑 𝑟

(S4)

In Eq. S3, the contributions of the image dipoles in the upper half-space (𝑧 > 0) are the same as those of the real dipoles would be in the lower half space (𝑧 < 0), if the metal would not be present. The lhs. of Eq. S3 therefore is the same as that of just the two dipoles without metal, which is what the rhs. of Eq. S3 describes with the integral going over the full space. 𝑉

thus represents the infinite (in the

case of point dipoles) but distance-independent self-energy of the two dipoles, which can be subtracted out. The six terms of 𝑉 , on the other hand, describe the pairwise interactions between the four dipoles, as indicated by the six arrows in Figure 4a. We group these terms as follows: 𝑉, +𝑉

,

≡𝜀 ∫

(

𝐸⃗ 𝐸⃗ + 𝐸⃗ 𝐸⃗ 𝑑 𝑟 = 𝜀 ∫ 𝐸⃗ 𝐸⃗ 𝑑 𝑟 =

≡ 𝑉,

)

(S5)

where the integral in the rhs. goes over the full space for the same reason as in Eq. S3, and describes the normal dipole-dipole coupling without the presence of any metal surface. Here, r is the distance between the two dipoles on the surface (see Figure 4 (b)). The coupling of each dipole with its own image dipole is given by: 𝑉, +𝑉 , ≡𝜀 ∫ where the two terms 𝑉 ,

𝐸⃗ 𝐸⃗ +𝐸⃗ 𝐸⃗ 𝑑 𝑟 = 𝜀 ∫ 𝐸⃗ 𝐸⃗ 𝑑 𝑟 = −2 and 𝑉 ,

(

(

)

≡ 𝑉,

)

(S6)

are equal due to the symmetry of the arrangement, and at the

same time only half of the size of the usual dipole-dipole coupling, since the integrals go over the SI 5

upper half-space only. The two terms thus can be lumped together into one term with the integral going over the full space (rhs.), i.e., as if the metal surface was not present. Finally, the coupling to the corresponding other image dipole is given by: 𝑉, +𝑉 , ≡𝜀 ∫

𝐸⃗ 𝐸⃗ +𝐸⃗ 𝐸⃗ 𝑑 𝑟 = 𝜀 ∫ 𝐸⃗ 𝐸⃗ 𝑑 𝑟 =

) /

(

(1 − 2cos 𝜙) ≡ 𝑉 (,

)

(S7) with tan 𝜙 = 𝑟/2𝑅. As in Eq. S6, both terms are identical and can be lumped together. We see that the configuration in Figure 4 (a) can be reduced to that of Figure 4 (b), removing the (

metal and having only three effective interaction terms 𝑉 , (

and 𝑉 ,

)

)

(

, 𝑉,

)

(

, and 𝑉 , (

contribute to the transition dipole coupling of Eq. 1, while 𝑉 ,

)

)

(

. The terms 𝑉 ,

)

does not. However, since

the distance between the two dipoles r (4-5 Å) is much closer than that to the image dipoles (

√4𝑅 + 𝑟 (20 Å) in our concrete experiment, 𝑉 ,

)

dominates and the presence of the metal

surface hardly affects the vibrational energy transfer rate. It is important to keep in mind that the image dipoles are only a construct that enforce that the electric field lines enter the metal surface perpendicularly. What is happening in reality is surface charges that are induced at the metal/liquid interface. They generate an electric field that equals that of the image dipoles in the upper half-space and at the same time cancel out the field of the real dipoles in the lower half-space. The derivation given above is therefore not affected by that change of perspective.

SI 6

References (1)

Donaldson, P. M.; Hamm, P. Gold Nanoparticle Capping Layers: Structure, Dynamics, and Surface Enhancement Measured Using 2D-IR Spectroscopy. Angew. Chemie-International Ed. 2013, 52, 634–638.

(2)

Kraack, J. P.; Kaech, A.; Hamm, P. Surface Enhancement in Ultrafast 2D ATR IR Spectroscopy at the Metal-Liquid Interface. J. Phys. Chem. C 2016, 120, 3350–3359.

(3)

Kraack, J. P.; Hamm, P. Vibrational Ladder-Climbing in Surface-Enhanced, Ultrafast Infrared Spectroscopy. Phys. Chem. Chem. Phys. 2016, 18, 16088–16093.

(4)

Kraack, J. P.; Hamm, P. Surface-Sensitive and Surface-Specific Ultrafast Two-Dimensional Vibrational Spectroscopy. Chem. Rev. 2017, 117, 10623–10664.

(5)

Jackson, J. D. Classical Electrodynamics; John Wiley & Sons Inc.: New York, 2007.

SI 7