956 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 47, NO.

4, AUGUST 1998

Broad-Band Characterization of Magnetic and

Dielectric Thin Films Using a Microstrip Line
Patrick Quéffélec, Marcel Le Floc’h, and Philippe Gelin

Abstract—A measurement method for the broad-band determi- strict restrictions on the dimensions of the samples since the
nation (100 MHz–10 GHz) of the permeability and permittivity test material has to entirely fill the cross section of the cell.
of thin films with thicknesses of 1–10 m has been developed. The This makes sample machining difficult. Moreover, for thin
technique is based on the measurement of the S parameters of
a microstrip line loaded with the test sample. The S parameters films this implies the propagation of the electromagnetic wave
are measured with a network analyzer. Besides its band width, along the sample thickness, thus lowering the measurement
the original feature of this method in comparison with existing sensitivity. As a consequence these methods are inadequate
techniques lies in the fact that the thin film does not entirely fill for thin-film measurements.
the cross section of the cell since it is directly laid on the substrate In this context new measurement cells have to be eval-
of the microstrip line. This leads to a simple and reproducible
measurement process. It also permits the propagation of the elec- uated for the broad-band characterization of thin films. In
tromagnetic wave along the film width of about a few millimeters, microwaves we now aim at developing a broad-band char-
thus increasing the measurement accuracy. Moreover the method acterization method which could be applied to thick samples
remains reliable for the characterization of bulk materials with as well as to thin films. The results obtained for thick samples
the same cell and data processing program. with this method have already been published in a previous
Index Terms— Microstrip, microwave measurements, perme- paper [7]. In this paper, after analyzing the main reasons for
ability measurement, permittivity measurement, thin films, scat- the selection of the measurement cell, we briefly recall the
tering parameters. general principles of the method. We present its application to
broad-band measurements of permittivity and permeability
I. INTRODUCTION of thin films. An analysis of the main error sources of the
method is done to define the domain of validity of thin-film
T ODAY in communication systems the use of magnetic
and dielectric materials exceeds the usual fields of appli-
cation (radomes, antennae, microwave circuits New com-
measurements. Finally, we give some results obtained for thin
dielectric or magnetic films with thicknesses of 1–10 m.
ponents are developed to meet the demand of leading areas.
This is the case for materials absorbing the electromagnetic
II. SELECTION OF THE MEASUREMENT CELL
energy, which are used for microwave electromagnetic com-
patibility (EMC). In microelectronics, an example of materials The starting point of our study is the method proposed
application is high-density memory, which is related to the by Barry [6]. This method permits broad-band measurements
frequency behavior of magnetic films. These specific materials of and of thick materials using the parameters
are often used in thin films for physical and technological measurements of a strip transmission line cell. The sample
reasons (circuit miniaturization, low-cost devices is placed in the middle of the line and has to fill the cross
The exact knowledge of material permittivity and per- section of the cell. In this case the dominant mode in the
meability is essential in the study of physical phenomena line is a transverse electromagnetic mode (TEM). Then the
which govern interactions between electromagnetic waves and transmission-line theory is sufficient and accurate for determin-
matter. Two different techniques have been developed for ing the material constants. This theory enables us to express
measuring the permittivity and permeability of materials: 1) and analytically using the measured parameters of
resonant methods and 2) reflection/transmission techniques. the cell. But, as previously mentioned, this method is not
In spite of their accuracy, resonant methods [1]–[3] are ap- applicable for thin-film measurements since the measurement
plicable only over a narrow frequency band since only one sensitivity is very low. To increase the measurement accu-
frequency value can be exploited with a resonant cavity. For racy we have thought of a different approach for which
measurements over a wide range of frequencies, waveguide or the microwave signal propagates along the film length. This
transmission line methods [4]–[6] are used. These methods are condition can be satisfied with a microstrip line. A strong
less accurate than the resonant cavity technique. They impose interaction between the wave and the test material is obtained
when the thin film is laid on the microstrip substrate close to
Manuscript received July 19, 1996; revised December 17, 1998.
P. Quéffélec and M. Le Floc’h are with Laboratory for Electronics and the central conductor (Fig. 1). At microwave frequencies, the
Communication Systems (LEST), 29285 Brest Cedex, France and the Univer- microstrip technology is a well-proven technology. This makes
sité de Bretagne Occidentale, U.F.R. Sciences, 29285 Brest Cedex, France. the use of a through-reflect-line (TRL) calibration procedure
P. Gelin is with École Nationale Superieure des Télécommunications de
Bretagne, 29285 Brest Cedex, France. for the network analyzer possible. This procedure is necessary
Publisher Item Identifier S 0018-9456(98)09939-2. to achieve accurate measurements in high frequencies. The
0018–9456/98$10.00  1998 IEEE

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.
QUÉFFÉLEC et al.: BROAD-BAND CHARACTERIZATION OF MAGNETIC AND DIELECTRIC THIN FILMS 957

Fig. 2. Microstrip line cross section.

Fig. 1. Microstrip cell loaded with the sample under test.

geometry of the cell is simple and requires little machining

of the samples.

III. DESCRIPTION OF THE METHOD

The determination of and calls for the knowledge
of two measurable parameters. Those are the reflection and
transmission coefficients measured with a network analyzer in
the microstrip cell loaded with the thin film and its support
(Fig. 1). Assuming that the test sample has the same cross
section as the cell, the reflection and transmission coefficients
Fig. 3. Measurement cell discontinuities. Region I: empty regions. Region II:
are calculated using the transmission-line theory [8]. Here the region partly filled with the test sample. n ; tn ; Rn ; and Tn are the coupling
thin film occupies a part of the microstrip line cross section. coefficients between the modes.
The dominant mode in the line is not a TEM mode. Then
a full-wave analysis is required to accurately determine the
In our modeling the central conductor is assumed to have
material properties. In this case the calculation of and
ideal conductivity and negligible thickness. This is a very
from the measured parameters needs the use of a numerical
realistic assumption in our structure where the strip width (645
optimization procedure (inverse problem) combined with the
m) is large compared to conductor thickness (10 m). How-
full-wave analysis of the cell (direct problem). The purpose
ever, this thickness causes an air gap between the microstrip
of this paper is not to describe thoroughly the electromagnetic
substrate and the test sample. The spectral domain approach
analysis of the cell for it has already been done in [9]. How-
enables us to take a great number of different layers into
ever, a brief recapitulation of its main features is necessary for
better understanding of our present study. account in the line cross section. An air gap with a constant
thickness over the whole microstrip line width and equal to the
strip thickness (10 m) has been introduced into the model to
A. Direct Problem obtain better results [7]. Moreover, complex values for the
The direct problem corresponds to the electromagnetic permittivity and the permeability
analysis of the cell. Its permits the determination of the of each layer are introduced in the SDA algorithm to take the
parameters of the transmission line as functions of and substrate and superstrate (thin film and its support) losses into
of the test material. The calculations are based on the mode account.
matching method [10]. The second step in the direct problem is the calcula-
First, this method implies the determination of dispersion tion of the parameters imposing continuity conditions on
characteristics of the microstrip line. The cross section of the electromagnetic fields at the edge of the cell discontinuities
cell (Fig. 2) is composed of a shielded microstrip structure, (Fig. 3). The use of the orthogonality of modes enables us
which constitutes the sample holder, a thin film, and its to determine the coupling coefficients and
support. As indicated before, the full-wave analysis must be between the modes from the continuity conditions [9]. Finally
used for the modal study of the microstrip line. We have the parameters of the microstrip cell are given by
chosen the spectral domain approach (SDA) [11], [12]. This
method requires much more calculation than the transmission-
line theory yet the domain of validity of the model used is
much more important.

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.
958 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 47, NO. 4, AUGUST 1998

where and represent the reflection and transmission

coefficients of the fundamental mode through the cell
discontinuities, is the length of the empty cell and the
propagation constant of the dominant mode. The parameters
of the test device depend on and of the test material
through the coefficients and

B. Inverse Problem
From the given geometry of the thin film and its position in
the microstrip line, the parameters are calculated using the
previously described direct analysis for assumed values of Fig. 4. Dimensions of the microstrip cell.
and This section presents computation of and of a
given thin film from the measured values of the parameters.
Electromagnetic constants of a test sample are determined
by matching calculated and measured values of the parame-
ters using a numerical optimization procedure. An optimization
problem involves minimizing a function, called the objective
function, of several variables. The objective function can be
expressed as a sum of squared functions as follows:

theoretical
measured

where
The optimization method is chosen in such a way that
it permits a fast location of the global minimum of
by avoiding local minima. We have developed an iterative
method [7] derived from the gradient method [13] to solve
the inverse problem. A certain number of tests has permitted Fig. 5. Representation on the Smith chart of the random error.
confirmation of the validity of the algorithm up to 6 GHz.
Beyond this frequency a quasi-Newton algorithm [13] based good mechanical properties. The configuration adopted here
on the calculation of the Hessian matrix of the objective for the experimental cell cross section is shown in Fig. 4.
function is used when the values of and are high The parameters measurements are performed using an
(typically and HP 8510 B network analyzer. In order to compensate for
the systematic errors caused by the network analyzer and the
measurement cell, the network analyzer is calibrated with a
C. Experimental Method specific procedure. We will see how the calibration procedure
The choice of geometry and material for the experimen- is used to measure accurately thin films further on. The
tal cell is based on three criteria. First, the microstrip line parallelepiped sample (thin film and its support) is placed on
dimensions were designed for a characteristic impedance of the microstrip substrate in the middle of the sample holder
(impedance matching with the network analyzer). (Fig. 1). For thin films with a high conductivity the sample
Secondly, we have shown that the sensitivity of the method is put onto the microstrip line, so that the thin-film support
increases when the microstrip substrate lies on the lower is adjacent to the central conductor (Fig. 2) for the line not
ground plane. We have studied the influence of the substrate to be short circuited. Insulating thin films can be adjacent
location on the ratio of the current in the strip to the power to the strip to increase the measurement sensitivity. The
flowing along the line. The highest value of this ratio, which is parameters measurements are recorded, then transferred to the
correlated with the fields magnitude in the microstrip structure, computer to be studied by the data processing program (direct
is obtained for (Fig. 2). The third criterion is low and inverse problem). The typical central unit processing time
characteristic impedance variations versus frequency to get a for 401 frequency-measured points is less than 10 min using
broad-band test device. It can be obtained with a high dielectric an IBM RS 600 computer.
constant substrate. This result is in contradiction to the high
sensitivity of the cell since much of the field is confined to the IV. THIN-FILM CHARACTERIZATION
microstrip substrate rather than penetrating the test sample. An The broad-band electromagnetic characterization of thin
alumina substrate is a reasonable tradeoff between films with thicknesses of 1–10 m confronts us with two
the second and third criterion. Moreover, alumina presents main problems. First, the thin film hardly disturbs the wave

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.
QUÉFFÉLEC et al.: BROAD-BAND CHARACTERIZATION OF MAGNETIC AND DIELECTRIC THIN FILMS 959

(a) (b)

(c)
0 0
Fig. 6. Calculated uncertainty bounds. (a) "3 = 8 j 0 and 3 = 2 j 0 length = 4 mm and thickness = 100 m; (b) "3 = 8 0 j 0 and 3 = 2 0 j 0
0 0
length = 4 mm and thickness = 5 m; and (c) "3 = 30 j 0 and 3 = 8 j 0 length = 4 mm and thickness = 5 m.

propagation in the cell for its thickness is very low, thus Random errors are caused by the noise figure of the network
limiting the measurement accuracy. Second, thin samples have analyzer used. As these errors cannot be totally eliminated, it
a bad mechanical holding. So they have to be deposited on a is necessary to analyze them in order to allow for uncertainties
rigid support for a better handling and insertion in the cell. This that are entailed on the measurements of and On
process leads to some errors in the characterization caused the Smith chart an uncertainty area around the experimental
by taking into account the thin-film support during the data results obtained from the network analyzer (Fig. 5) reveals
processing. the random error. We have calculated the upper and lower
limits of uncertainties that the random error caused on the
A. Measurement Accuracy measured and data from this model (HP 8510B). The
The measurement accuracy is affected by two main uncer- uncertainty limits versus frequency are shown in Fig. 6 for
tainty sources: random errors and systematic errors. different electromagnetic constants and thicknesses of thin

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.
960 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 47, NO. 4, AUGUST 1998

films. The uncertainties in and for thin films are more

important than for bulk materials. Indeed, when the sample
volume is small, the measured parameters magnitude is
low, and the uncertainties in and caused by random
errors are important. But these simulation results show that the
uncertainties on and decrease rapidly as the frequency
increases and are less important for high permeability and
permittivity samples (Fig. 6).
Systematic errors are caused by defects of the measurement
cell (impedance mismatch, copper losses of the conductors,
radiation from the microstrip line) and the network analyzer
(coaxial cables, electronic components). These errors can be
reduced by a calibration procedure of the network analyzer.
(a)
During the calibration, a series of known devices (standards)
are connected. The systematic errors are determined as the
difference between the measured and known responses of the
standards.
In the conventional short-open-load-through (SOLT) cali-
bration [14], three known impedance references and a sin-
gle transmission standard are required. This is an easy-to-
implement calibration procedure using a standard coaxial kit
given with the analyzer. It has been used for thick materials
measurements [7]. However, an impedance discontinuity oc-
curs at the coaxial-to-microstrip connection, and the signal is
attenuated in the microstrip line. This effect can significantly
alter the measured data especially for low permittivity and
permeability materials or for small dimensions such as thin
films. Using the SOLT calibration method, this effect can
(b)
not be significantly reduced at high frequencies (typically
above 8 GHz). Because of this limitation, an alternative Fig. 7. Measurement cell on the wafer.
method for calibration is needed that permits accurate thin-film
measurements.
Using the TRL calibration [15], [16] and a measurement cell the results from the SOLT and TRL calibrations is shown in
on wafer (Fig. 7), the impedance mismatch at the coaxial-to- Fig. 8. With the SOLT calibration applied, the measured
microstrip connection can be characterized and removed. The and data exhibit significant mismatch ripple [Fig. 8(a)].
TRL calibration relies on transmission lines rather than on a The reflection at the coaxial-to-microstrip connection is larger
set of discrete impedance standards. TRL refers to the three than that at the discontinuity created by the test sample in
the microstrip line. Using the TRL calibration, the results do
basic steps in the calibration process.
not exhibit any significant ripple up to 12 GHz [Fig. 8(b)].
1) Through: connection of ports one and two with a mi- The TRL calibration technique will be used later on for the
crostrip line. thin-film measurements.
2) Reflect: connect microstrip lines with identical high A shielded microstrip structure is considered in the elec-
reflection coefficient to each port. The simplest reflection tromagnetic analysis of the cell (Fig. 2). In this case the
is an open circuit. electromagnetic energy is not radiated from the microstrip
3) Line: insert a microstrip line between ports one and two. line. In practice, an open microstrip line (Fig. 7) is used to
Different line lengths are required for the through and line. construct each calibration standard of the TRL technique for
The TRL technique requires the use of a measurement cell mechanical reasons. Most of the microwave signal propagates
on wafer (Fig. 7) enabling us to insert calibration standards of through the microstrip substrate and the test sample. However,
different physical lengths. This cell provides a direct microstrip the errors in the measurements produced by the effect of
interface. The calibration plane is established at the end of the the radiation from the sample in an open structure must be
open circuit (reflect). As a consequence, the parameters of quantified. We have studied the influence of a ground plane
the discontinuities created by the test sample in the microstrip above the strip conductor on the measured response of the cell
line can be measured directly. Moreover, the connection for different test materials. As shown in Fig. 4 the distance
interface is repeatable since the same coaxial-to-microstrip between the upper and lower ground plane is 10 mm. The
connection is used during the calibration process. This permits effects of the upper ground plane have proved to be less than
an accurate error correction. The and measurements 1% on the parameters magnitude up to 18 GHz. Then the
for a thick dielectric sample of low permittivity is considered effect of radiation from the sample is not considered in the
to illustrate the TRL calibration procedure. A comparison of data processing program.

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.
QUÉFFÉLEC et al.: BROAD-BAND CHARACTERIZATION OF MAGNETIC AND DIELECTRIC THIN FILMS 961

(a)

Fig. 9. Vectorial components on the Smith chart of the measured S param-

eters of the cell loaded with a thin film and its support.

(b)
Fig. 8. Measured "3 and 3 data for a bulk sample of a dielectric resin
("3 = 3:3 0 j 0:2): (a) with the SOLT calibration applied and (b) with the
TRL calibration applied.

B. The Effect of the Thin-Film Support Fig. 10. Measured "3 and 3 data for a dielectric thin film of Kapton
(thickness = 10 m).
Thin films are deposited on dielectric substrates (supports)
for easier handling and insertion in the cell. Then an additional
layer between the thin film and the microstrip substrate (Fig. 2) with a thickness of 12 m and with a low dielectric constant
is introduced in the SDA algorithm. The inverse problem is Then the contribution of each material in the
solved assuming that the electromagnetic properties and measurable parameters is similar. This process significantly
of the thin film are unknown and that the dielectric constant reduces the uncertainties in the measured and data.
of the support is known. In this case the first step is to measure
the support permittivity. Measuring the parameters of the C. Results
cell containing a blank support performs this determination. In order to illustrate the experimental results obtained with
The support has a typical thickness of 500 m. Its permittivity our broad-band characterization method, we give, in Figs. 10
is measured with an accuracy better than 5% in the 100 and 11, two examples of measured and data for dielectric
MHz–10 GHz frequency range. But the measured and and magnetic thin films. The permittivity and permeability
data for the thin film strongly depend on the value given to versus frequency of a supportless dielectric film of Kapton
the support permittivity in the SDA algorithm. A change of with a thickness of 10 m are depicted in Fig. 10. The
5% in the value of the support permittivity leads to a change of permittivity measured in a wide frequency range corresponds
60% in the measured and data for the thin film. Indeed, to the value given by the supplier of the sample. As expected,
the volumes of each material are very different (ratio 1/500). was measured. The real and imaginary compo-
That makes the contribution of the support in the measured nents of the permittivity and permeability versus frequency of
parameters much larger than the one of the thin film (Fig. 9). a ferromagnetic thin film are represented in Fig. 11(a) and (b),
As a consequence a small uncertainty on the value of the sup- respectively. This film has been realized from an amorphous
port permittivity leads to a very important relative error for the cobalt layer with a thickness of 0.5 m deposited on a Mylar
measured and data of the thin film. To improve the accu- substrate with a thickness of 12 m. The thin-film thickness
racy of the results, thin films are deposited on Mylar substrates is measured with a profilometer (accuracy 1 nm). The

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.
962 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 47, NO. 4, AUGUST 1998

IEEE Trans. Microwave Theory Tech., vol. MTT-27, pp. 791–795, Sept.
1979.
[2] A. Mansingh and A. Parkash, “Microwave measurement of conductivity
and permittivity of semiconductor spheres by cavity perturbation tech-
nique,” IEEE Trans. Microwave Theory Tech., vol. MTT-29, pp. 62–65,
Jan. 1981.
[3] S. Li, C. Akyel, and R. G. Bosisio, “Precise calculation and mea-
surements on the complex dielectric constant of lossy material using
cavity perturbation techniques,” IEEE Trans. Microwave Theory Tech.,
vol. MTT-29, pp. 1041–1048, Oct. 1981.
[4] N. E. Belhadj-Tahar, A. Fourrier-Lamer, and H. de Chanterac, “Broad-
band simultaneous measurement of complex permittivity and permeabil-
ity using a coaxial discontinuity,” IEEE Trans. Microwave Theory Tech.,
vol. MTT-38, pp. 1–7, Jan. 1990.
[5] J. Baker-Jarvis, E. J. Vanzura, and W. A. Kissick, “Improved technique
for determining complex permittivity with the transmission/reflection
method,” IEEE Trans. Microwave Theory Tech., vol. MTT-38, pp.
1096–1103, Aug. 1990.
[6] W. Barry, “A broad-band, automated, stripline technique for the simul-
taneous measurement of complex permittivity and permeability,” IEEE
Trans. Microwave Theory Tech., vol. MTT-34, pp. 80–84, Jan. 1986.
(a) [7] P. Queffelec, P. Gelin, J. Gieraltowski, and J. Loaec, “A microstrip
device for the broad-band simultaneous measurement of complex per-
meability and permittivity,” IEEE Trans. Magn., vol. 30, pp. 224–231,
Mar. 1994.
[8] W.B. Weir, “Automatic measurement of complex dielectric constants
and permeability at microwave frequencies,” Proc. IEEE, vol. 62, pp.
33–36, Jan. 1974.
[9] P. Queffelec and P. Gelin, “Influence of higher order modes on the mea-
surements of magnetic and dielectric materials using a microstrip dis-
continuity,” IEEE Trans. Microwave Theory Tech., vol. 44, pp. 816–826,
June 1996.
[10] R. Mittra and T. Itoh, “A new technique for the analysis of the dispersion
characteristics of microstrip lines,” IEEE Trans. Microwave Theory
Tech., vol. MTT-19, pp. 54–63, Jan. 1971.
[11] T. Itoh and R. Mittra, “Spectral-domain approach for calculating the
dispersion characteristics of microstrip lines,” IEEE Trans. Microwave
Theory Tech., vol. MTT-21, pp. 407–409, July 1973.
[12] J. Esteban and J. M. Rebollar, “Characterization of corrugated waveg-
uides by modal analysis,” IEEE Trans. Microwave Theory Tech., vol.
39, pp. 937–943, June 1991.
[13] S. Rosloniec, Algorithms for Computer-Aided Design of Linear Mi-
crowave Circuits. Dedham, MA: Artech House, 1990.
[14] Manual Supplement—Calibration Kits (Test sets and accessories man-
(b)
ual,) Hewlett–Packard, H.P. 8510B. Network analyzer, Mar. 1988.
Fig. 11. Measured "3 and 3 data for an amorphous cobalt layer (thickness [15] G. Engen and C. Hoer, “Thru-reflect-line: An improved technique for
= 0:5 m): (a) Measured "3 data and (b) Measured 3 data. calibrating the dual six-port automatic network analyzer,” IEEE Trans.
Microwave Theory Tech., vol. MTT-27, pp. 987–993, Dec. 1979.
[16] H. P. Product Note 8510-8. Network analysis, “Applying the H.P. 8510B
T.R.L. calibration for noncoaxial measurements,” Hewlett–Packard, Oct.
measured and data agree with the results obtained for 1987.
the same material by other laboratories using characterization [17] O. Acher, P. M. Jacquart, J. M. Fontaine, P. Baclet, and G. Perrin, “High
techniques different from ours (coaxial line) [17]. impedance anisotropic composites manufactured from ferromagnetic
thin films for microwave applications,” IEEE Trans. Magn., vol. 30,
pp. 4533–4535, Nov. 1994.
V. CONCLUSION
The measurement method developed using the TRL cal-
ibration procedure for the network analyzer allowed us to
characterize thin dielectric and magnetic films in a broad
frequency band from 100 MHz to 10 GHz. The thickness of
the test samples varies from 0.5–100 m. The comparison
of our measurement results with those obtained using other
characterization techniques permitted us to validate our mea-
surements. Considering the previous results on thick materials,
we are now able to characterize with accuracy the dielectric, Patrick Quéffélec was born in France in 1966. He
magnetic, absorbing, or transparent samples with thicknesses received the Ph.D. degree in electronics from the
University of Brest, France, in 1994.
of 0.5 m to 5 mm using the same data processing program Presently, he is Maı̂tre de conférences at the
and the same cell. University of Brest, in the Laboratory for Elec-
tronics and Communication Systems, Brest, France.
His research activities deal with the electromag-
REFERENCES netic wave propagation in heterogeneous materials
and the analysis of measurement methods for the
[1] A. Parkash, J. K. Vaid, and A. Mansingh, “Measurement of dielectric
microwave characterization of materials.
parameters at microwave frequencies by cavity perturbation technique,”

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.
QUÉFFÉLEC et al.: BROAD-BAND CHARACTERIZATION OF MAGNETIC AND DIELECTRIC THIN FILMS 963

Marcel Le Floc’h was born in France in 1945. Philippe Gelin was born in France in 1948. He
He received the degree of Docteur-és-Sciences from received the Ph.D. degree in physics from the
the Université de Bretagne Occidentale, France, in Technical University of Lille, France, in 1981.
1983. He is currently a Professor of electrical
He is presently a Professor in the Electronic engineering at the École Nationale Superieure
Department, Université de Bretagne Occidentale des Télécommunications de Bretagne, Brest,
and a member of the Laboratoire d’Electronique et France. His research interests include wave-
Systèmes de Télécommunications. matter interactions and the modeling and the
characterization of materials.
Dr. Gelin is a member of the Laboratory for
Electronics and Communication Systems (LEST)
which is a research unit associated with the French National Research Council.

Authorized licensed use limited to: Universidad de los Andes. Downloaded on March 24,2020 at 15:18:26 UTC from IEEE Xplore. Restrictions apply.