Experiments and simulations for optical controlled thermal

management on the nanometer length scala

F.Garwe
1*
, U.Bauerschfer
2
, A. Csaki
1
, A. Steinbrck
1
, A. Weise
3
, W.Paa
1
, W.Fritzsche
1

1
Institute of PhotonicTechnology Jena, PO Box 100239, D-07745 Jena
2
GmBU, Erich-Neuss-Weg 5, D-06120 Halle/S.
3
Friedrich Schiller University Jena, Institute of Human Genetics and Anthropology, D-07702 Jena

*Corresponding author: frank.garwe@ipht-jena.de


Abstract

We show experimentally that the energy of
femtosecond laserlight pulses with specific
wavelength can be very efficiently absorbed by 30
nanometer gold nanospheres without ablation or
destruction [10]. In this way nanospheres are usable
as heat conversion tools on the nanometer length
scale. The reason is that excited plasmon-polariton
resonances are damped by electron-electron-,
electron-phonon-, electron-defect-, and electron-
surface-scattering, which will cause the formation
of a new thermal equilibrium. These sub-
picosecond relaxation dynamics of the
nonequilibrium electrons and the electron-phonon
interaction can be described by the electron-phonon
coupling model.
Here we simulate a number of fs laser pulses with
very high repetition rates by numerical coupling of
the electron-phonon coupling model, which
describe the lattice temperature of the metallic
nanosphere after excitation with a femtosecond
laser pulse. Additionally the heat equation, which
models the heat transfer from the nanosphere into
the complex nonsymmetric surroundings is taken
into account.
As a result we show the possibility for control of
the temperature (300 K-420 K) of the nanosphere in
a very small space-time region (15 nm - 30 nm
nanosphere diameter, 5 ps - 100 ps) and its sub-
wavelength sourrounding (10 nm - 50 nm, 1 ns - 1
ms). Parameters for optimal control are the
femtosecond laser parameters, the nanoparticle
material and its geometry parameters as well as the
environment with its geometry parameters.

1. Introduction

The surface plasmons (SP) at the interface between
the gold- or silver nanoparticle and the dielectric
surrounding show pronounced optical absorption
resonances due to coupling of the light to the
collective electronic excitation [1]. Width and
spectral position of the resonance are well
described by the Mie theory [2]. The spectral
position depends on particle material, size, shape
and dielectric environment [3,4]. The size
dependence is introduced by assuming a size-
dependent material dielectric function (,R),
which can be written as a combination of an
interband term, accounting for the response of the d
electrons and a Drude term considering the free
conduction electrons only. The phenomenological
damping constant in the Drude term is related to
inhomogeneous line widening and the lifetimes of
all electron scattering processes in the nanoparticle
that are mainly due to electron-electron, electron-
phonon, electron-defect scattering, and (in the range
of tens of nanometers of the nanoparticle diameter)
electron-surface scattering. The damping
mechanism is often described in terms of a loss of
phase coherence of the collective electron
oscillation due to scattering of single electrons [1].
Using laser irradiation, this phenomenon can lead to
a temperature rise of electrons with respect to the
ionic lattice. Electron-electron und electron-phonon
collisions will cause the formation of a new thermal
equilibrium [5]. This dynamic was studied by pump
probe experiments [6]. In [7] the subpicosecond
time response of the optical excitation with
femtosecond laser pulses and the relaxation
dynamics of the nonequilibrium electrons by the
electron-phonon interaction was studied
theoretically by the electron-phonon coupling
model.

2. Experiments

For thermal manipulation of technical and
biological materials or control of chemical reactions
near the gold nanoparticles, we have to find a
combination of moderate intensity of laser pulses
and necessary fluence, so that on the one hand the
desired material manipulation or chemical reaction
nearby the nanoparticles takes place, on the other
hand no destruction of nanoparticles, glass slide or
other carier material would occur. First of all, to get
optimal conditions for heat conversion, an
excitation of the nanoparticle with the resonance
wavelength (plasmon or interband) is necessary. In
the case gold nanoparticles with 30 nm diameter we
used 530 nm. In addition, the pulse length and the
pulse repetition rate has to be in a range, where
nanoparticles absorb effectivly a well determined
Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
dose of the incoming light wich is converted into
heat. We used gold nanoparticles on a thin PMMA
layer to test several pulse length and pulse
repetition rate at 530 nm wavelength of sub-
wavelength light to heat conversion effectivity [8].
To prevent fragmentation and ablation during laser
irradiation we used 100 fs laser pulses with fluences
below the ablation threshold of 2,8 mJ/cm
2
but with
high pulse repition rates of 76 MHz. Because
PMMA at temperatures nearby T
g
= 372 K will
have a high viscosity, the PMMA layer serves as
thermosensitive reporter to the temperature. So the
hot nanoparticle can sink into the PMMA and this
can be detected by AFM scans (see Fig. 1c, dark
regions). Using continuous wave irradiation with
the same dosis (Fig.1a) or 3 ps laser pulses with the
same fluence (Fig.1b) the nanoparticle will not sink
into the the PMMA. These particle are shown as
bright regions. To test the prevention of ablation
during the fs laser irradiation we put gold
nanoparticles on pure glass. As you can see in
Fig.1d ablation does not occur. A section analysis
prove this too (not shown here).



Fig.1: AFM images (4 m x 4 m) of different laser pulse experiments on 30 nm gold nanoparticles on a 50 nm PMMA
layer. Here the laser parameter were: wavelength = 530 nm and a) continuous wave (cw), 2,8 mJ/cm
2
, b) pulse length = 3 ps,
pulse repetion rate = 76 MHz, fluence = 2,8 mJ/cm
2
, c) pulse length = 100 fs, pulse repetition rate = 76 MHz, fluence = 2,8
mJ/cm
2
, d) the same as in c) but the gold nanoparticles are provided on pure glass

In the next section we simulated this complex
behavior of the nanoparticle, taking into account the
environment and the interaction with low power fs
laser pulses with high repetition rate.
Femtosecond laser irradiation with wavelength of
800 nm (two photon excitation of gold core silver
shell nanoparticles) does not destroy chromosomes
or other biological tissue unspecifically. The picture
in Fig.2b shows an AFM zoom into a chromosome
to two DNA sequence specific bound silver
enhanced gold nanoparticles. The same
chromosome was imaged after laser treatment
(Fig.2c). Now the particle have disappeared and
cavities are detected at the former particle positions
and indicated by arrows (The arrows indicate the
dark field of the former particle postion). The laser
induces sub-wavelength damage to the chromosom.



Fig.2: Metaphase chromosomes labeled with silver enhanced gold nanoparticles shown in optical dark field contrast
microscopy (a). AFM images of nanoparticles coupled to chromosome surface before (b) and after (c) laser irradiation.
Pictures below (b), (c) are zooms of marked particle on chromosome and section analysis.
Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble

However, the fluences (28 mJ/cm2) are above the
ablation threshold (2,8 mJ/cm2). The results also
show structural changes of nanoparticles located
outside the chromosomes. This observation leads us
to the assumption of partial ablation of the
nanoparticles in combination with a significant
sinking of the remaining particle structure into the
chromosome material.

3. Simulation methods

We examined the relaxation dynamics of the
electron temperature by the electron-phonon
coupling model [3]. The time evolution of the
electron temperature T
e
and the effective lattice
temperature T
l
, is described by the coupled
differential equations

) (
) ( ) ( ) (
l e
l
l
l e e
e
e e
T T G
t
T
c
t P T T G T
t
T
T c
=

+ =

(1)

Here is the temperature-dependent
electronic heat capacity, is the lattice heat
capacity,
) (
e e
T c
l
c
is the thermal conductivity, is the
electron-phonon coupling constant, is the
excitation energy density per unit time and unit
volume in a particle. The first term on the right-
hand side representing thermal-conductivity losses
from a particle to the matrix can be neglected for
few picoseconds after laser irradiation. The value of
G is 310
G
) t
=
J
6
/
( P
e
c
10
4
J/(ps m
3
K) [5]. We used the parameters
of c and with ,
[4] and
[5]. As transient absorption a value of 27% was
taken [9].
) (
e e
T
3
/ K m J
l
c
e
T c
0
K m
3 2
0
66 c = c
l
3 =
The conductive heat transfer from the gold
nanosphere to the PMMA-matrix starts after the
equilibration of the electron gas with the lattice.
The model for heat transfer by conduction is
described by the heat equation
q T k
t
T
C
p
= +

) ( (2)
with the following material properties: is the
density, C is the heat capacity, k is the thermal
conductivity, is the heat source. The two
subdomains are a gold particle modeled as a
truncated sphere with a radius of 15 nm without a
cup of 5 nm height and the PMMA matrix as a
cuboid 200 nm x 200 nm x 100 nm. The gold
properties are the density
p
q
=
Au
19300 kg / m
3
, the
heat capacity = 129 J / kg K, and the thermal
conductivity = 317 W / mK, while the PMMA
properties are
Au
p
C
Au
k
=
PMMA
1190 kg / m
3
,
=1420 J / kg K, = 0,19 W / m K.
The initial temperature in the gold particle is
(the lattice temperature after the fs-pulse)
and the initial temperature of the PMMA matrix is
. At the boundary between the two
subdomains the continuity interior boundary
condition is valid:
pPMMA
C
K 344
K 300
(
PMMA
k
( )
PMMA
k
K 300
(
inf
T h
Au
T
( T k
Au
n n
) =
25 . 0

L
T
54 . F
lam

h
25 . 0
3

T
Ra

L
= k
lam
F

0 ) =
PMMA PMMA Au
T k

(3)

As boundary condition of the PMMA cuboid we set
on all surfaces (without the top surface) the
prescribed temperature of , the heat bath of
the surrounding. The top boundaries are described
by external natural cooling conditions

) T n (4)
with the horizontal surface upside
0 = (5)
with
(6)
with the Rayleigh number Ra and the characteristic
length L.


4. Results

The electron-phonon coupling model simulation
shows a high electron temperature near 2000 K
with relaxation times of 5 ps ( see Fig.1a,
absorption = 0,27) for 30 nm gold nanoparticles
femtosecond laser radiation (pulse length = 100 fs,
wavelength = 530 nm, repetition rate = 76 MHz)
and dielectric surrounding (PMMA) parameters. At
the same time scale the temperature of the
nanoparticle lattice increases from room
temperature of 300 K to 344,5 K (see Fig.1b,
Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
absorption 0,27). 10 ps after the first femtosecond
laser pulse, the particle temperature T
l
is near below
the glass transition temperature of PMMA (372 K).
Therefore the nanoparticle can sink into the PMMA
after the first femtosecond laserpuls.

Fig.3: a) Time dependence of the electron temperature T
e
for different transient absorptions. b) Lattice temperature T
l
of 30
nm gold spheres, also for different absorptions.



Fig.4: a) 3D FEM simulation of the temperature distribution of a 30 nm gold sphere on PMMA 12ns after the femtosecond
laser pulse. b) temperature for different times after the laser pulse at different vertical postions under the nanoparticle. The
particle surface is here at the position of 30 nm.



Fig.5: a) 3D FEM simulation of the temperature distribution of a 30 nm gold sphere on PMMA after 5 femtosecond laser
pulses with 80 MHz pulse repetition rate. b) Temperature of the nanosphere after different numbers of femtosecond laser
pulses with 80 MHz pulse repetion rate.


To simulate the heat transport into the non-uniform
PMMA and air environment we numerically solved
the heat conduction equation by FEM methods
(Fig.4a). The most prominent result is, that the
temperature increase of the PMMA environment is
very local (only 40 nm). This is caused by the low
Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
heat capacity of the small nanoparticle (see Fig.4b).
Secondly and very important, at the end of the
period between two femtosecond laser pulses (12
ns) there is only a very small temperature increase.
This encourages to increase the temperature by
additional femtosecond pulses, each of them after
12 ns. Further simulations show the results after a
number of ultrashort pulses in a time distance of 12
ns. Every pulse increases the temperature of the
particle (see Fig.5b). Fig.5a shows the temperature
in the PMMA environment after 5 pulses. After
only 5 pulses a balance between cooling of the air-
and PMMA environment (convection and heat
conduction) and the periodical heating of the
nanoparticle was reached. Here we find that the
maximal temperature change, which can be
obtained, is only 3,5 K.



Fig.6: a) 3D FEM simulation of the temperature distribution of a 30 nm gold sphere on PMMA after 12 femtosecond laser
pulses with 1 GHz pulse repetition rate, the red line shows the heat flux b) Temperature of the nanosphere 1 ns after different
numbers of femtosecond laser pulses with 1 GHz pulse repetition rate. (for two different laser pulse intensities and therefore
two different lattice temperatures (44 K, 22 K))



Fig.7: a) Simulation of nanoparticle heating by a pulse repetitation rate of 1GHz (from 0 ns - 10 ns), than a temper zone (10
ns 20 ns) and at last a cooling with several cooling rates, realized by laser pulses with varion pulse repetition rates (0,2
GHz, 0,5 GHz) and cooling without laser pulses. b) the same as in a) but for the temperature in PMMA 10 nm below the
nanoparticle (z = -10 nm) and 30 nm below the nanoparticle (z = -30 nm).


By increasing the femtosecond laser pulse
repetition rate (time between two pulses is only 1
ns), it should be possible to increase the
temperature change further. The simulation in Fig.
6b show a nanoparticle temperature of 420K after
12 ns.
In principle any temperature increase, any constant
temperature or any temperature decrease (in a
temperature range between 20C and 150C) should
be possible by controlling the pulse repetition rate
and the laser pulse intensity below a maximal
fluence for ablation of nanoparticles, which is 2,8
mJ/cm
2
(Fig. 7a). In PMMA in a distance of 10 nm
and 30 nm below the nanoparticle the temperature
can be controlled, too. Here, the temperature
increase after the pulse is not so significant. (Fig.
7b).



Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble
References
[1] Kreibig K, Vollmer M, 1995 Optical properties of Metal Clusters Springer
[2] Mie G 1905 Ann. Phys. 25 377
[3] Gotschy W, Vonmetz K, Leitner A, Aussenegg F R 1996 Appl. Phys B 63 381
[4] Kelly K L, Colronado E, Zhao L L, Schatz G C 2003 J. Phys. Chem. B 107 668
[5] Groeneveld R, Sprik R 1995 Physical Review B 51 (17) 11433
[6] Ahmadi T, Logunov S L, El-Sayed M A 1996 J. Phys. Chem. 100 8053
[7] Tokizaki T, Nakamura A, Kaneko S, Uchida K, Orni S, Tanji H, Asahara Y, 1994 Appl. Phys. Lett. 65 8
[8] Garwe F, Bauerschaefer U, Csaki A, Steinbrueck A, Ritter K, Bochmann A, Weise A, Akimov D, Maubach G,
Knig K, Huettmann G, Paa W, Popp J, Fritzsche W 2007 Nanotechnology, submitted
[9] Logunov S L, Ahmadi T S, El-Sayed M A, Khoury J T, Whetten R L 1997 J. Phys. Chem. B 101 3713
[10] Csaki A, Garwe F, Steinbrck A, Maubach G, Weise A, Knig K, Fritzsche W 2007 Nano Lett. 7, No.2 247
Excerpt from the Proceedings of the COMSOL Users Conference 2007 Grenoble