1
Hogan, N. J., Urban, A. S., Ayala-Orozco, C., Pimpinelli, A., Nordlander, P., & Halas, N. J.
(2014). Nanoparticles Heat through Light Localization. Nano Letters. doi:10.1021/nl5016975
Nanoparticles Heat through Light Localization
Nathaniel J. Hogan
1, 2
Alexander S. Urban
2, 3
Ciceron Ayala-Orozco
1, 5
Alberto Pimpinelli
4
Peter Nordlander
1, 2, 3, 4
Naomi J. Halas
1, 2, 3, 4
1
Department of Physics and Astronomy
2
Laboratory for Nanophotonics
3
Department of Electrical and Computer Engineering
4
Rice Quantum Institute
5
Department of Chemistry
Rice University, Houston, Texas 77005, United States
2
Abstract
Aqueous solutions containing light-absorbing nanoparticles have recently been shown to produce
steam at high efficiencies upon solar illumination, even when the temperature of the bulk fluid
volume remains far below its boiling point. Here we show that this phenomenon is due to a
collective effect mediated by multiple light scattering from the dispersed nanoparticles.
Randomly positioned nanoparticles that both scatter and absorb light are able to concentrate light
energy into mesoscale volumes near the illuminated surface of the liquid. The resulting light
absorption creates intense localized heating and efficient vaporization of the surrounding liquid.
Light trapping-induced localized heating provides the mechanism for low-temperature light-
induced steam generation and is consistent with classical heat transfer.
Keywords: Plasmon; nanoscale heating; radiative transport
3
In a world of ever-increasing energy demand, the harvesting of sunlight has the potential
to provide a useful energy source for much of the planet. In addition to the solar energy
harvesting technologies currently under development,(1) such as photovoltaics(2) and solar
thermal electricity generation,(3) there is avid interest in novel effects that may lend themselves
to entirely new approaches. Recently it was reported that light-absorbing nanoparticles,
immersed in water and illuminated by sunlight, are capable of generating steam without the
necessity of heating the entire fluid volume, resulting in remarkably high steam generation
efficiencies. It was initially observed that over 80% of the absorbed energy was utilized for
steam generation.(4) Because of the large interparticle distances of ten microns or more in these
solutions, the particles were assumed to be both optically and thermally isolated. A phase change
localized around the individual particles, an effect well-established at higher illumination
levels,(5) was proposed as a possible explanation for this process. However, it has been noted
that at such illumination intensities a temperature increase at the nanoparticle surface
commensurate with vapor nucleation is not in agreement with other single particle measurements
and would require nonclassical effects.(6) It has been suggested that collective thermal effects
are important in explaining the response of such systems.(7-11) Here we show, however, that
collective thermal effects alone are not sufficient to explain the thermal response of these
systems. Rather, light trapping by solutions of particles that simultaneously absorb and scatter
light results in highly localized heating, an unanticipated effect that, when combined with
classical heat transfer, provides an accurate theoretical description of the system.
When light interacts with an ensemble of randomly dispersed particles, as in a fluid
environment, the light can be either scattered or absorbed, or both, depending on the properties
of the particles. When the average distance between particles in a solution is smaller than the
wavelength of light, the wave nature of light must be explicitly considered, and multiple
scattering events can lead to phenomena such as weak localization or Anderson localization of
light.(12-16) For the relatively low nanoparticle concentrations used in this study, the wave
nature of light is only implicitly used to determine the optical properties of a single particle. Each
photon is considered independently in a ballistic transport formulation. Conservation of energy
leads directly to an equation of transfer, which in the absence of emission terms is
where I is the specific intensity, c is the speed of light, Ω is the solid angle, k
s
and k
a
are the
scattering and absorption coefficients, respectively, and p is the phase function of scattered
radiation.(17, 18) The first two terms on the left-hand side stem from a continuity equation that
corresponds to the conservation of photons. The third term represents the exponential attenuation
of intensity due to scattering and absorption in the medium and, together with the previous two
terms, constitutes the Beer–Lambert (BL) law. The BL law provides a good approximation of
light transfer in systems where absorption is much stronger than scattering. The right-hand side
represents the contribution to the intensity from photons either scattered prior to absorption or
scattered out of the system (Figure 1A). With this term taken into account, approximations are
required for the equation to be solvable analytically. When scattering is much stronger than
absorption, it reduces to a diffusion equation.(17, 19) However, neither the strongly absorbing
4
nor strongly scattering approximations apply when particles scatter and absorb with comparable
probabilities, and the full equation must be solved. This more general case has been of particular
interest for applications in biomedical imaging and for unconventional therapeutics, such as
photothermal ablation of tumors.(20-24) To numerically solve this for the case of absorber–
scatterers, we simulate the fate of each incident photon passing through a random ensemble of
particles.(25, 26) In the limit of pure absorbers or pure scatterers this approach agrees with both
the Beer–Lambert law and the diffusion approximation, respectively (Figure S1, Supporting
Information).
Figure 1. (A) Schematic illustrating characteristic experiment (left) where a dense solution of nanoparticles
contained in a cuvette is illuminated with 808 nm laser light; multiparticle optical interactions in such nanofluids
5
(right) where photons are scattered and/or absorbed. (B) Experimentally obtained (left) and Monte Carlo (MC)
simulated (right) scattered light as viewed from the side of cuvettes containing nanoshell solutions of the indicated
concentrations. Integration times are not the same for all three experiments.
To study light propagation through a solution of nanoparticles that both scatter and
absorb light, we fabricated solutions of two different metallo-dielectric nanoparticles: nanoshells
and nanomatryoshkas.(27-29) Each type of particle has the same absorption cross section of 1 ×
10
–10
cm
2
for resonant excitation at 808 nm but substantially different scattering cross sections of
1 × 10
–9
and 6 × 10
–11
cm
2
, respectively (determined by comparing Mie theory for the two
particle geometries with experimentally obtained extinction spectra; Figure S2 A,B, Supporting
Information). Nanoshells are primarily scattering particles with a ratio of scattering to absorption
efficiency of 10, while nanomatryoshkas are predominantly absorptive, with a scattering to
absorption efficiency of 0.6. These particles not only offer tunability of the scattering and
absorption efficiencies but also have well-known scattering phase functions,(30-32) a feature not
accessible in biological systems. We do not consider nonmetallic absorbers such as carbon
nanoparticles in this study due to the inability to accurately measure or control the cross sections.
The ability to control all of the experimental parameters allows us to compare experimental
results with theoretical predictions rigorously.
First we examine the light-scattering properties of solutions of nanoshell particles,
illuminated from above with resonant light at 808 nm wavelength, as a function of particle
concentration (Figure 1B). The experimental images of the light scattered through the side face
of the cuvette reveal a decreasing amount of light as the incident beam propagates into the
solution. For higher particle concentrations the light does not penetrate as deeply, and strongly
increased backscattering off the fluid surface at the highest concentration is apparent. The
observed scattered light distributions closely resemble those predicted directly from the multiple-
scattering simulations (Figure 1B), which incorporate the dipole scattering distribution of the
resonant particles.
On the basis of this approach, we examine the effect of light scattering on the fraction of
light absorbed by particle solutions over a broad range of concentrations (Figure 2). Here we
compare the fractional light absorption, η, as predicted from a single-scattering corrected BL law
calculation(33) with the results of the multiple-scattering simulations (Figure 2A). Multiple
scattering begins to affect the fractional light absorption dramatically for nanoshells and
nanomatryoshkas at concentrations of 1 × 10
9
and 1 × 10
10
NP/milliliter (ml), respectively.
The Beer–Lambert law estimates that η saturates at 0.18 for nanoshells; when multiple
scattering is accounted for, this value is 0.70. For nanomatryoshkas, η saturates at 0.60 in the
single-scattering regime but is >0.90 when multiple scattering events are taken into account.
These very large discrepancies result from the inherent assumption in the Beer–Lambert law that
all scattered photons are lost from the beam as it is transmitted through a solution of particles.
For absorber–scatterers, this approximation is justified only at low particle concentrations.
Multiple scattering events increase the average path length of the photons, which increases the
average absorption probability. When the scattering length is comparable to any of the linear
dimensions of the solution, a scattered photon is likely to traverse the entire length of the
solution without any subsequent interactions. Therefore, at low concentrations (or small
volumes), the Beer–Lambert law and the multiple-scattering simulations should agree (Figure 2A
(inset)). The concentrations for which the multiple scattering regime begins to deviate strongly
