Plasmonic sensing of heat transport and phase change near solid-liquid interfaces

David G. Cahill and Jonglo ParkDepartment of Materials Science and Engineering

University of Illinois at Urbana-Champaign

supported by ONR

Collaboration with Patrick Hopkins and Pam Norris, U. Virginia

Motivation (big picture): Improve experimental methods for probing heat transfer and phase transformations at solid/liquid and vapor/liquid interfaces

• Prior study of thermal conductance of hydrophobic and hydrophilic interfaces with water

Ge et al., Phys. Rev. Lett. (2006)

Motivation (big picture): Improve experimental methods for probing heat transfer and phase transformations at solid/liquid and vapor/liquid interfaces

• Prior study of fast water desorption from a hydrophilic surfaces by time-resolved ellipsometry

Min et al., J. Phys. Chem. C (2012)

Outline

• Thermal conductance of interfaces

• Nanodisk sensors (prepared for us by Insplorion) and characterization of their sensitivity and depth resolution.

• Application to interface thermal conductance and thermal diffusivity of fluids by separating the response from the Au temperature and the index change in the adjacent fluid. (unpublished)

• Initial application to fast condensation and evaporation of a refrigerant (R124) from interfaces with controlled chemistry. (work in progress)

Thermal transport coefficients

• Thermal conductance (per unit area) G is a property of an interface

• Thermal conductivity is a property of the continuum

Interface conductance spans a factor of 60 range at room temperature

nanotube/alkane

W/Al2O3

Au/water

PMMA/Al2O3

Lyeo and Cahill, PRB (2006)

Insplorion nanodisk plasmonic sensors

• Fabricated by “hole-mask colloidal lithography”

• Au adhesion to SiO2 substrate is good even though they avoided the use of a conventional Cr or Ti adhesion layer (Cr or Ti would damp the plasmon)

Au disk diameter 120 ± 10 nm, height 20 ± 2 nm

SEM AFM

Sensitivity d(Tr)/dn (change in transmission coefficient with respect to optical index) approaches unity

• Coat with PMMA and take difference spectra of the absorption.

• Noise floor of pump-probe measurements should be

1/2

1/2

13 1/2

0.3 ppm Hz3 mK Hz

10 m Hzliquid

liquid

nT

h

Tr=transmission

Sensitivity to dn is localized to within 13 nm of the Au surface

• Atomic-layer deposition of alumina

• Assuming constant deposition rate per cycle

Long-pass optical filter

Short-pass optical filter

Kang et al., RSI (2008)

Time-domain thermoreflectance and transient absorption

Signal is a combination of the temperature change of the Au and the temperature (or pressure or density) change of the surroundings

• Isolate the two terms using a linear combination of the response at two wavelengths.

• “breathing mode” acoustic oscillation is minimized at the same wavelength that minimizes the sensitivity to fluid temperature.

( ) ( )

Au fluidAu fluid

d Tr d TrTr T TdT dT

Vary the contributions from TAu and Tfluid by varying wavelength of the probe light

0 1000 2000 3000 4000100

101

102

103

780 nm 790 nm 800 nm 810 nm 820 nm

Vin (

V)

time (ps)

Signal from the lateral “breathing mode” acoustic oscillation is minimized at the same wavelength that minimizes the sensitivity to fluid temperature

0 100 200 300 400 500-120

-80

-40

0

40

80

Vin(d

ata)

-Vin(th

erm

al)

time (ps)

780 nm 790 nm 800 nm 810 nm 820 nm

Data after subtracting thermal response

Modeling of heat transfer builds on our standard methods of analyzing TDTR data.

• Model the 120 nm diameter, 20 nm thick heat source as a 120 nm diameter uniform intensity laser beam heating a blanket 20 nm Au film that has zero in-plane thermal conductivity.

• dn/dT of the fluid dominates over dn/dT of the glass substrate so model the signal as a weighted average of the temperature of the fluid within 13 nm of the Au

Control the surface chemistry using self-assembled monolayers (thiol bond to Au surface)

• Plasmon resonance is a built-in diagnostic for what is happening near the interface.

1.32 1.34 1.36790

800

810

820

Pea

k w

avel

engt

h (n

m)

Refractive index (n)

hydrophilic

bare

hydrophobic

Water Ethanol1.0 1.2 1.4 1.6

720

740

760

780

800

820

840

Pea

k w

avel

engt

h (n

m)

Refractive index (n)

hydrophilic

bare

hydrophobic

Hydrophilic HS(CH2)3SO3 and hydrophobic HS(CH2)9CH3

Data acquisition and analysis for interfaces with fluid mixtures

• First step: select wavelength that minimizes contribution from fluid temperature

• Compare to thermal model with interface conductance as a free parameter using literature values for fluid thermophysical properties

100 101 102 103101

102

data (785 nm) 80 MW m-2 K-1

160 MW m-2 K-1

240 MW m-2 K-1

Vin (

V)

time (ps)100 101 102 103101

102

data (790 nm) 40 MW m-2 K-1

80 MW m-2 K-1

160 MW m-2 K-1

Vin (

V)

time (ps)

0% Ethanol (Water)

100% Ethanol

Data acquisition and analysis for interfaces with fluid mixtures• Next step: shift wavelength to provide greater sensitivity to

fluid temperature near the interface.• Subtract breathing mode acoustic signal by fitting to a

damped oscillator• Compare to thermal model with interface conductance as a

free parameter

100 101 102 103101

102

data (810 nm) data (810 nm) 80 MW m-2 K-1 (A = 0.65, B = 0.063) 120 MW m-2 K-1 (A = 0.65, B = 0.063) 180 MW m-2 K-1 (A = 0.65, B = 0.063) 240 MW m-2 K-1 (A = 0.65, B = 0.063)

Vin (

V)

time(ps)100 101 102 103101

102

data (810 nm) data (810 nm) 40 MW m-2 K-1 (A = 2, B = 0.25) 80 MW m-2 K-1 (A = 2, B = 0.25) 160 MW m-2 K-1 (A = 2, B = 0.25)

Vin (

V)

time(ps)

100 % Ethanol

0% Ethanol (Water)

Vary liquid composition between pure ethanol and pure water for hydrophobic SAM, hydrophilic SAM, and “bare” Au.

• Data for pure water and pure ethanol are in agreement with prior work for planar interfaces and supported nanoparticles.

• Data for pure ethanol are relatively insensitive to the interface chemistry.

• Competitive adsorption of water at the hydrophilic interface (?)

0.0 0.2 0.4 0.6 0.8 1.00

50

100

150

200

G (M

W m

-2 K

-1)

Ethanol / Water

hydrophilic

Bare

Hydrophobic

Kapitza lengthLK = /G ≈ 3 nm

Initial experiments on a refrigerant as a function of pressure

plasmonic substrate

R124 gas chromatograph analysis1-chloro-1,2,2,2-tetrafluoroethane (99.79%)1,1,1,2,2 – pentafluoropropane (0.21 %)

• Vapor pressure is ≈40 psi at lab temperature

200 300 400 500

10-1

100

101

102

Pre

ssur

e (p

si)

Teperature (K)

Data from NIST webbook (http://webbook.nist.gov/chemistry/)

wikipedia.org

Initial experiments are “frequency domain” (not “time-domain”) to access longer time scales

• Fixed negative delay time (probe pulse arrives 20 ps before pump pulse)

• Vary frequency of pump over a wide range (10 Hz to 10 MHz)• Correct for the phase and amplitude of the system response

using pump beam directly incident on the fast photodiode.• In vacuum, the signal is due to the Au temperature

100 101 102 103 104 105 106 107 108

-50

0

50

in-phase out-of-phase (negative)

V in, V

out (V

)

Frequency (Hz)100 101 102 103 104 105 106 10710-2

10-1

100

101

102

in-phase out-of-phase (negative)

V in, V

out (V

)

Frequency (Hz)

vacuum vacuum

Only at the highest frequencies (>1 MHz) is the temperature not laterally homogeneous on the length scale of the separation between Au nanodisks (200 nm)

100 101 102 103 104 105 106 107

10-2

10-1

100

101

102

Single nanodisk (r = 60 nm) r = 10 m

T

(K)

Frequency (Hz)100 101 102 103 104 105 106 107

-80

-60

-40

-20

0

Single nanodisk (r = 60 nm) r = 10 m

Pha

seFrequency (Hz)

Calculated thermal response, 3 mW pump

Now add R124 at 17 psi (approximately ½ of the vapor pressure)

• Difference between vacuum and 17 psi only appears at f<10 kHz.• Work in progress. Results are not what we were expecting. (Much

smaller and much slower).• Au surface is coated by hydrophobic SAM. Maybe not completely

stable in contact with refrigerant vapor• Will need to consider Maragoni effects (fluid flow) in addition to

evaporation/condensation

100 101 102 103 104 105 106 107

-20

0

20

40

in-phase out-of-phase

V in, V

out (V

)

Frequency (Hz)100 101 102 103 104 105 106 107

0.01

0.1

1

10

100

in-phase out-of-phase (negative) out-of-phase (positive)

Vin, V

out (V

)

Frequency (Hz)

Comparison of vacuum and 17 psi of R124

100 101 102 103 104 105 106 107

-20

0

20

40

in-phase out-of-phase

Vin, V

out (V

)

Frequency (Hz)100 101 102 103 104 105 106 107 108

-50

0

50

in-phase out-of-phase (negative)

V in, V

out (V

)

Frequency (Hz)

vacuum 17 psi

Summary

• Plasmonic nanodisks are powerful platform for probing small changes in index of refraction due to temperature excursions in liquids or evaporation/condensation of thin layers (or changes in density/pressure) near an interface.

• Signal due to the index of refraction of the fluid provides greater sensitivity to interface conductance (needed when the thermal effusivity of the fluid is small).

• Kaptiza length for ethanol in contact with hydrophilic and hydrophobic SAMs is ≈3 nm.

• Experiments on refrigerant (R124) as a function of pressure, temperature, and surface chemistry are in progress.