1. Introduction
Optical devices based on the surface plasmon resonance (SPR) effect have been
intensively investigated for biological and chemical sensing applications. In recent
years great efforts have been made to increase the sensitivity. One of the promising
approaches is the use of not only the amplitude, but also the phase of light
reflected under SPR conditions [
1–6
H. P. Ho and W. W. Lam, “Application of differential phase
measurement technique to Surface Plasmon Resonance
Sensors,” Sens. Actuators B
96, 554–559
(2003). [CrossRef]
].
Only few authors have investigated SPR sensors for acoustic pulse recording [
7
A. Schilling, O. Yavas, J. Bischof, J. Boneberg, and P. Leiderer, “Absolute pressure measurements on a
nanosecond time scale using surface plasmons,”
Appl. Phys. Lett.
69, 4159–4161
(1996). [CrossRef]
,
8
J. Boneberg, S. Briaudeau, Z. Demirplak, V. Dobler, and P. Leiderer, “Two-dimensional pressure
measurements with nanosecond time resolution,”
Appl. Phys. A
69 [Suppl.], S557–S560
(1999). [CrossRef]
]. They could show that surface plasmons can be successfully
used for absolute pressure measurements on a nanosecond time scale with a lateral
resolution given by the probe beam diameter. Thermoelastically excited acoustic
transients in liquids are generally broadband with low pressure amplitudes and
require sensors with high temporal resolution and sensitivity. Both requirements are
fulfilled by SPR sensors.
Depending on the order of layers, SPR sensors can be arranged in the Kretschmann or
in the Otto configuration. Whereas the previous studies have exclusively employed
the Kretschmann configuration for acoustic pulse measurements, in this study it is
shown that the Otto configuration can be used as well. It offers the possibility to
use a combination of amplitude and phase of the reflected p-polarized light in the
attenuated total reflection (ATR) region to maximize the sensitivity for acoustic
pulse recording. This is not possible in Kretschmann configuration because the
acoustic wave entering the glass substrate causes an additional phase shift due to
induced birefringence. This phase shift takes place in the bulk and not on the
surface of the device and gives rise to a relatively low bandwidth acoustic signal
that is not useful for the applications discussed here. The measurement of the phase
requires some kind of an interferometer. In this study the two configurations are
compared and two methods for analyzing the reflected light, via its amplitude and
its phase, are investigated theoretically and experimentally for the detection of
acoustic transients in liquids.
2. Theory of surface plasmon resonance sensors
Excitation of surface plasmon waves (SPWs) optically in a metallic layer is only
possible by illuminating with p-polarized light, whose momentum is matched to that
of SPWs [
9
H. Raether, Surface Plasmons on Smooth and Rough Surfaces and on
Gratings , (Springer, Berlin, Heidelberg
1988).
]. Only the p-polarized electric field vector has a driving
force component to excite the electron gas of the metal. The matching condition is
given by
where k
0 = 2π/λ is the vacuum wave
vector of the incident light, nd
is the refractive index
(RI) of the dielectric medium (e.g. the glass prism) through which the light is
incident at an angle θ. The dispersion relation of SPWs
is
where ϵd
and
ϵm
are wavelength dependent complex
dielectric functions of the dielectric and metallic media respectively. When SPR
occurs, the incident light is absorbed and converted into Joule heating of the metal
layer accompanied by a strong attenuation of the reflected light in the total
reflection region.
To model an SPR sensor in ATR mode makes it necessary to calculate the reflectance
R and phase φ of the reflected
light for a three layer system. For a TM (p-) polarized wave the calculation yields
where
rkl
is the reflection coefficient at the interface
between media
k and
l,
θk
is the incident angle into the
kth
medium,
nk
is
the refractive index and
d is the thickness either of the metal
film in Kretschmann or of the air gap in Otto arrangement (
Fig. 1).
Fig. 1. (a). Layer arrangement of Kretschmann device; (b) Layer arrangement of Otto
device
3. SPR sensors for acoustic wave detection
In
Kretschmann configuration, the order of layers [
Fig. 1(a)] is a metal layer sandwiched between (glass)
substrate (from where the probe laser beam is incident) and liquid (from where the
acoustic wave arrives). An acoustic wave shifts the incident angle where SPR occurs
relative to the initial position due to changes of RI of the liquid in the range of
the evanescent SPR field. The RI is directly linked to the pressure of the acoustic
wave via the acousto-optic coupling coefficient in liquid (e.g.
dn/
dp=
1.35×
10-5Pa-1
in water). On this account in Kretschmann configuration the device acts as a
pressure sensor [
7
A. Schilling, O. Yavas, J. Bischof, J. Boneberg, and P. Leiderer, “Absolute pressure measurements on a
nanosecond time scale using surface plasmons,”
Appl. Phys. Lett.
69, 4159–4161
(1996). [CrossRef]
,
8
J. Boneberg, S. Briaudeau, Z. Demirplak, V. Dobler, and P. Leiderer, “Two-dimensional pressure
measurements with nanosecond time resolution,”
Appl. Phys. A
69 [Suppl.], S557–S560
(1999). [CrossRef]
]. Also the pressure wave entering the substrate causes a
shift of the SPR angle via the elastooptic constants of the glass. However, these
constants are at least an order of magnitude smaller than the corresponding value in
water and have a negligible influence on the recorded signal [
10
R. D. Alcock and D. C. Emmony, “Sensitivity of Reflection
Transducers,” J. Appl. Phys.
92, 1630–1642
(2002). [CrossRef]
,
11
G. Paltauf and H. Schmidt-Kloiber, “Measurement of laser-induced
acoustic waves with a calibrated optical
transducer,“ J. Appl. Phys.
82, 1525–1531
(1997). [CrossRef]
]. The dimension of the evanescent field in the liquid
determines the bandwidth of the SPR sensor. Since the evanescent field has typically
a size on the order of the light wavelength, this results in a bandwidth in the
range of GHz.
In the
Otto configuration the order of layers [
Fig. 1(b)] is glass substrate (from where the laser beam is
incident), air gap, metal layer, glass or plastic substrate, and liquid (from where
the acoustic wave arrives). This sensor acts as a displacement sensor, because the
displacement of the metal/air interface during the reflection of an incoming
acoustic wave is the main disturbing effect for the initial SPR conditions. This
follows from the fact that the evanescent wave of the plasmon is confined to the
metal layer and does not enter the adjacent substrate. The SPR conditions of the
device are therefore not affected by any changes of the substrate RI caused by the
pressure wave. The metal layer itself due to its even higher bulk modulus compared
to glass can be expected to be unaffected in its optical constants by the pressure
variations. In the Otto configuration there is a complex dependence of sensitivity
on air gap thickness and incident angle of the laser beam, mainly because an
incoming acoustic wave causes both an angular shift of the SPR peak and a change of
its amplitude.
4. Sensitivity simulations
For the design and the application of SPR-sensors it is important to find out via
theoretical, numerical calculations the conditions for the highest sensitivity with
suitable linearity. The parameters are the layer dimensions and the angle of
incidence. For all theoretical investigations silver was used as metal layer
(εAg
=-15.7+i1.0728),
BK7 was the material of the glass prism
(εP
=2.3) and the
wavelength of the probe laser beam was 632.8 nm.
4.1 Kretschmann configuration
Figure 2 shows reflectance and sensitivity versus angle
of incidence for three chosen initial RI values of the surrounding liquid. The
simulation was performed for a 50
nm silver layer that gives
under optimal SPR matching conditions a reflectance value at the ATR minimum
around zero. The position of the minimum shifts almost linearly with refractive
index. The shape of the reflectance curves remains unchanged, only the SPR
matching angle is shifted.
A simulation taking into account the RI change in glass reveals a slightly
reduced shift of the SPR angle. However, this modification is compensated by the
effect of a concomitant change of incident angle, in a similar way as it was
described for a sensor based on Fresnel reflectance [
11
G. Paltauf and H. Schmidt-Kloiber, “Measurement of laser-induced
acoustic waves with a calibrated optical
transducer,“ J. Appl. Phys.
82, 1525–1531
(1997). [CrossRef]
]. This justifies the use of the water RI variation alone
for the calculation of the sensitivity.
For the use as acoustic sensor the incident angle is set to a certain operating
point (OP). For a given OP the sensitivity Sampl
is
defined as
where R0,opt
is the reflectance at ambient pressure
and ∂Ropt
/∂p
is the derivative of the reflectance with respect to the pressure at the chosen
OP.
The highest sensitivity is obtained for an OP in the steeper slope of the ATR
peak near the reflectance minimum. The disadvantage of choosing an OP with a
higher sensitivity is a loss in linearity.
Figure 2 shows the relative change,
Ropt
/
R0,opt
, as a
function of
p for different values of
R0,opt
. There is a more or less wide linear range
around
p=0. For practical application a compromise must be
found with respect to sensitivity and linearity. The chosen OP marked in
Fig. 2 promises a sensitivity value of
0.01
bar-1, which is a relative change of one percent of the recorded signal
per 1 bar pressure amplitude.
Fig. 2. Reflectance and sensitivity curves for different initial RI conditions of
the surrounding liquid (left). OP labels the chosen operating point.
Relative change of reflectance, for RI=1.33, dependent on the pressure
amplitude for different initial reflectance values (right).
4.2 Otto configuration
4.2.1 Amplitude measurement
In the Otto configuration the sensing principle is a variation of the air gap
thickness
d during the reflection of the acoustic wave at
the metal/air interface. The sensitivity is derived from the dependence of
the optical reflectance
Ropt
on air gap
thickness
d and angle of incidence
θ of the probe laser beam.
Figure 3 (left) shows this dependence for a 100 nm
silver layer thickness displayed as an intensity plot of the reflectance
function. The quite complicated behavior becomes apparent when comparing the
slices taken out of the intensity plot [
Fig. 3 (right)]. The reflectance and the sensitivity
are strongly dependant on the initial air gap thickness. The sensitivity of
the direct amplitude measurement of the reflected light to interface
displacements is defined as
where
R0,opt
is the reflectance at initial air
gap thickness and
∂
Ropt
/∂
d
is the derivative of the reflectance R
opt with respect to the air
gap thickness. For the chosen OP, marked in
Fig. 3, follows a sensitivity of about
0.013
-1 nm or in other words, a displacement of one nanometer
causes a signal change of about 1.3 percent.
Fig. 3. Reflectance intensity plot as function of air gap thickness and
incident angle (left). Slices taken out of the reflectance intensity
plot and calculated sensitivity (right).
4.2.2 Phase measurement
Not only the amplitude but also the phase of the p-polarized component of a
laser beam reflected from a metal film under SPR conditions is dependent on
ambient conditions that are changed by an acoustic wave. The s-polarized
component is quite unaffected.
For this reason it is also plausible to use the phase shift between p- and
s-polarized portions of the probe laser beam as sensing physical parameter
for acoustic waves.
For phase sensitive measurements an interferometric setup is used where s-
and p-polarized portions of the laser beam are recombined after the SPR
device. In this setup the s-polarized component acts as a reference. The
interferometer output function is given by
where
RP
,
RS
,
φP
,
φS
are the reflectance and
phase of the two polarization components and
β
is the azimuth angle with respect to p-polarization of a polarizer used to
adjust the intensities of the two orthogonally polarized portions of the
incoming laser beam.
Φ is an additional phase
factor used to set the interferometer OP into a slope of the cosine
function.
Figure 4 (left) shows the simulated phase and
sensitivity curves versus angle of incidence for the chosen simulation
parameters: 600 nm air gap thickness,
β =
10° and
Φ = 330°. The
slope of the phase curve becomes highest at the minimum of the reflectance
curve. The sensitivity of the interferometric setup measuring the phase and
amplitude variations is defined as
with I0,out
the interferometer output at initial
air gap thickness and ∂Iout
/∂
d
the derivative of the interferometer output function with respect
to the air gap thickness.
The linear range of the sensor arranged in Otto configuration is determined
by a constant slope of the curves shown in [
Fig. 4 (right)] for different values of
R0,opt
. In practice the sensitivity and
linearity can be tuned with the angle of incidence of the probe laser beam.
For the same chosen OP as for the amplitude measurement [
Fig. 3 (right)], the sensitivity amounts to 0.022
nm
-1, almost two times higher than the sensitivity of the
direct amplitude measurement.
In summary the results of the simulations show that it is straightforward to
find the optimal sensitivity of the SPR device in Kretschmann configuration,
which acts as a pressure sensor. In Otto configuration the SPR device is
sensitive to displacements, making it more complicated to optimize it for
maximum sensitivity and linearity. A higher sensitivity at a given OP can be
achieved with the interferometric phase measurement.
Fig. 4. Phase of the p-polarized reflected light and calculated sensitivity
of the interferometric measurement method as a function of incident
angle (left). Relative change of the interferometer output function,
for d=600nm, dependent on metal/air interface displacement
(right).
5. Experimental setups and working principle
In all experiments the acoustic transients were excited thermoelastically by
irradiating an aqueous dye solution with a laser pulse from a frequency doubled
Nd:YAG laser with a wavelength of
532 nm, a radiant exposure on the
liquid surface of
0.13 J/
cm2
and a
pulse duration of
10 ns. The dye solution had an optical absorption
coefficient of
100 cm-1
. Under these conditions a
pressure wave with an amplitude of about
7 bar is generated in the
liquid.
Figure 5 shows the experimental setup for amplitude
measurements consisting of a polarized He-Ne laser, a neutral density filter for
reducing the intensity, a polarizer to adjust the portions of p- and s-polarized
light, a rotary table for setting the OP by changing the incident angle, a right
angle prism with one side acting as a mirror and a SPR device placed on the prism.
For the reflectance amplitude measurement only p-polarized light was used and the
temporal signals were directly recorded with a 125MHz bandwidth photo detector.
For the phase measurement two different interferometric setups were used. In both
setups the s-polarized component acts as an undisturbed reference. At the entrance
of the interferometer both components have the same intensity level in order to
achieve the highest modulation of the interferometric output.
Figure 6(a) shows the sketch of the used common path
interferometer (CPI) [
12
A.V. Kabashin, V. E. Kochergin, A. A. Beloglazov, and P. I. Nikitin, “Phase-polaristion contrast for
surface plasmon resonance biosensors,”
Biosens. Bioelectron.
13, 1263–1269
(1998). [CrossRef]
]. It uses a half wave plate as a polarization dependent
phase shifter, and an analyzer to get the two polarization components to interfere.
The second phase measurement setup shown in
Fig. 6(b) is a Mach-Zehnder interferometer (MZI) consisting
of a polarizing beam splitter to separate the polarization components, a half wave
plate to match the polarization state of the separated components, a phase shifter
and a non polarizing beam splitter for recombination [
1
H. P. Ho and W. W. Lam, “Application of differential phase
measurement technique to Surface Plasmon Resonance
Sensors,” Sens. Actuators B
96, 554–559
(2003). [CrossRef]
]. The two generated output beams are conjugate interference
signals, recorded with balanced photodiodes and amplified with an 80 MHz bandwidth
differential amplifier.
In comparison the MZI eliminates common mode signal perturbations such as laser noise
and doubles the output signal compared to single beam detection, whereas the CPI
represents a quite easy setup with a high stability against outside influences.
In order to compare the experimental signals between different configurations and
with the theoretical sensitivity values, all results are given as relative intensity
changes, by dividing the recorded AC voltage signal by the DC level that corresponds
to the undisturbed optical signal without pressure wave. In the case of the phase
measurement with the MZI, where the undisturbed output of the differential amplifier
is zero, this DC level was measured by blocking one of the interferometer beams.
Furthermore, the DC voltage was the same for all experiments performed in a
particular setup, enabling a direct comparison of signal to noise ratios in the
amplitude and phase measurements.
Fig. 5. Setup for excitation and measurement of acoustic transients. M: mirror, P:
polarizer, F: filter, BPF: band pass filter, PD: photo diode.
Fig. 6. CPI(left) and MZI(right) phase measurement setup. HWP: half wave plate, A:
analyzer, PBS: polarizing beam splitter, BS: beam splitter, PS: phase
shifter, BPD: balanced photo detector
6. Experimental results and discussion
Figure 7(a) shows the time resolved pressure signal recorded
with the amplitude measurement setup and the arrangement of the used SPR sensor. The
first pressure pulse arriving at the sensing area of the SPR-device is shown. From
the shape and amplitude of this peak it is possible to determine various physical
properties of the liquid, such as speed of sound, optical absorption coefficient and
thermoelastic coupling parameter [
13–15
G. Paltauf, H. Schmidt-Kloiber, and H. Guss, “Light distribution measurements in
absorbing materials by optical detection of laser-induced stress
waves,” Appl. Phys. Lett.
69, 1526–1528
(1996). [CrossRef]
]. Prerequisites for such measurements are the
wide linear sensitivity range and the high bandwidth offered by the SPR sensor.
Figure 7(b) shows the time resolved displacement signals
recorded with the amplitude and the CPI measurement setup. The applied sensor was
arranged in Otto configuration. In comparison the phase measurement shows a nearly
two times higher signal to noise ratio than the amplitude measurement. With the MZI
measurement setup a further increase of S/N ratio by at least a factor of two is
achieved due to subtraction of conjugate signals and the rejection of common mode
signal perturbations such as laser noise (
Fig. 8). The comparison of acoustic signals recorded with
Kretschmann and Otto configuration is complicated due to different measured
quantities (pressure and displacement) and layer dimensions. The sensitivity
definition allows the comparison of sensitivity at different values of
R0
.
S was always highest near a
minimum of
R. In practice, to take advantage of this fact, the
incident laser beam intensity has to be increased to compensate the losses near the
SPR minimum.
A comparison of the measured amplitudes with the predictions obtained from the
theoretical sensitivity values shows that the experimental sensitivity in the
Kretschmann configuration is quite similar to the theoretical one. For instance,
with an estimated pressure amplitude of 7 bar the experimental sensitivity obtained
from
Fig. 7(a) is ∼0.004 bar
-1. This
corresponds to a theoretical sensitivity in the falling slope of the reflectance
curve (
Fig. 2), where the actual operating point was chosen. In Otto
configuration, the experimental sensitivity is still on the same order of magnitude
as the theoretical one but lower. The estimated displacement amplitude for the
measurements is ≥10
nm , which would imply a relative
intensity change of ≥ 20% at the point of maximum theoretical
sensitivity. The lower experimental value can be attributed to the imperfection of
the real device. Factors limiting the sensitivity are the finite divergence of the
probe beam [
10
R. D. Alcock and D. C. Emmony, “Sensitivity of Reflection
Transducers,” J. Appl. Phys.
92, 1630–1642
(2002). [CrossRef]
] and most of all the fact that the surface of the substrate
carrying the metal layer was neither perfectly flat nor perfectly parallel to the
surface of the glass prism. In order to set the sensor at the optimum operating
point the air gap thickness must be controlled to a fraction of the optical
wavelength. This was not possible with our substrate that was a 4 mm thick piece of
polycarbonate (PC). This material was chosen to achieve good acoustic coupling to
water, but it has not very high optical quality. Choosing a glass substrate with
optimum optical surface quality would allow higher sensitivities but would lead to
acoustic reverberations within the glass layer.
We have compared the two possible arrangements of SPR devices for acoustic wave
detection. The Kretschmann configuration already investigated in Ref. [
7
A. Schilling, O. Yavas, J. Bischof, J. Boneberg, and P. Leiderer, “Absolute pressure measurements on a
nanosecond time scale using surface plasmons,”
Appl. Phys. Lett.
69, 4159–4161
(1996). [CrossRef]
,
8
J. Boneberg, S. Briaudeau, Z. Demirplak, V. Dobler, and P. Leiderer, “Two-dimensional pressure
measurements with nanosecond time resolution,”
Appl. Phys. A
69 [Suppl.], S557–S560
(1999). [CrossRef]
] is sensitive to RI variations of the surrounding liquid in
the range of the evanescent plasmon field and acts as a pressure sensor with high
bandwidth and a wide linear working range. The simulation shows a quite simple
dependence of the reflectance on RI and pressure, making it easy to find optimal
conditions for the sensor in terms of sensitivity and linearity.
The Otto configuration acts as a displacement sensor of the metal/air interface
during the reflection of the arriving acoustic wave. The simulated result shows a
complicated dependence of the reflectivity on incident angle and initial air gap
thickness. It could be shown that in addition to the amplitude, also the phase of
the reflected light can be used for acoustic wave detection. The S/N ratio is two
times higher by using amplitude and phase, instead of a pure amplitude measurement
of the reflected laser light. This behavior predicted by sensitivity simulations is
confirmed by experimental results. The achievable bandwidth is only limited by the
detection electronics. The sensor has a linear working range of a few tens of
nanometers. To achieve a further two times better sensitivity for the experimental
investigations a MZI was used by detecting the conjugated output beams with a
balanced photodetector.
Fig. 7. (a). Temporal signal of amplitude measurement in Kretschmann configuration;
(b). Comparison of amplitude and phase CPI measurement in Otto
configuration
Fig. 8. Comparison of amplitude and phase MZI measurement
7. Conclusion
The finding of this study is that both configurations can be used for acoustic wave
recording, with the difference that the measured quantity of an SPR sensor arranged
in Otto configuration is displacement, whereas it is pressure in Kretschmann
configuration. The combined measurement of phase and amplitude of the reflected
light is more sensitive, but only useable in combination with the Otto configuration
for acoustic wave detection.
Acknowledgment
This work has been supported by the Austrian Science Fund (FWF), Proj. Nr.
P18172-N02.
References and links
1. |
H. P. Ho and W. W. Lam, “Application of differential phase
measurement technique to Surface Plasmon Resonance
Sensors,” Sens. Actuators B
96, 554–559
(2003). [CrossRef] |
2. |
H. P. Ho, W. W. Lam, and S. Y. Wu, “Surface Plasmon Resonance Sensor
based on the measurement of differential phase,”
Rev. Sci. Instrum.
73, 3534–3539
(2002). [CrossRef] |
3. |
Y. D. Su, S. J. Chen, and T. L. Yeh, “Common-Path Phase-Shift
Interferometry Surface Plasmon Resonance Imaging
System,” Opt. Lett.
30, 1488–1490
(2005). [CrossRef] [PubMed] |
4. |
S. Y. Wu, H. P. Ho, W. C. Law, C. L. Lin, and S. K. Kong, “Highly sensitive differential
phase-sensitive Surface Plasmon Resonance Biosensor based on the
Mach-Zehnder configuration,” Opt. Lett.
29, 2378–2380
(2004). [CrossRef] [PubMed] |
5. |
C. M. Wu and M. C. Pao, “Sensitivity-tunable optical sensors
based on Surface Plasmon Resonance and Phase
Detection,” Opt. Express
12, 3509–3514
(2004). [CrossRef] [PubMed] |
6. |
Y. Xinglong, W. Dingxin, and Y. Zibo, “Simulation and analysis of surface
plasmon resonance biosensor based on phase
detection,” Sens. Actuators B
91, 285–290
(2003). [CrossRef] |
7. |
A. Schilling, O. Yavas, J. Bischof, J. Boneberg, and P. Leiderer, “Absolute pressure measurements on a
nanosecond time scale using surface plasmons,”
Appl. Phys. Lett.
69, 4159–4161
(1996). [CrossRef] |
8. |
J. Boneberg, S. Briaudeau, Z. Demirplak, V. Dobler, and P. Leiderer, “Two-dimensional pressure
measurements with nanosecond time resolution,”
Appl. Phys. A
69 [Suppl.], S557–S560
(1999). [CrossRef] |
9. |
H. Raether, Surface Plasmons on Smooth and Rough Surfaces and on
Gratings , (Springer, Berlin, Heidelberg
1988). |
10. |
R. D. Alcock and D. C. Emmony, “Sensitivity of Reflection
Transducers,” J. Appl. Phys.
92, 1630–1642
(2002). [CrossRef] |
11. |
G. Paltauf and H. Schmidt-Kloiber, “Measurement of laser-induced
acoustic waves with a calibrated optical
transducer,“ J. Appl. Phys.
82, 1525–1531
(1997). [CrossRef] |
12. |
A.V. Kabashin, V. E. Kochergin, A. A. Beloglazov, and P. I. Nikitin, “Phase-polaristion contrast for
surface plasmon resonance biosensors,”
Biosens. Bioelectron.
13, 1263–1269
(1998). [CrossRef] |
13. |
G. Paltauf, H. Schmidt-Kloiber, and H. Guss, “Light distribution measurements in
absorbing materials by optical detection of laser-induced stress
waves,” Appl. Phys. Lett.
69, 1526–1528
(1996). [CrossRef] |
14. |
A. A. Oraevsky, S. L. Jacques, and F. K. Tittel, “Measurement of tissue optical
properties by time-resolved detection of laser-induced transient
stress,” Appl. Opt.
36, 402–415
(1997). [CrossRef] [PubMed] |
15. |
A. A. Karabutov, N. B. Podymova, and V. S. Letokhov, “Time-resolved laser optoacoustic
tomography of inhomogeneous media,” Appl.
Phys. B
63, 545–563
(1996). |