You are on page 1of 7

1208

S. Hol: Behind Space Charge Distribution Measurements

Behind Space Charge Distribution Measurements


Stphane Hol
Laboratoire de Physique et d'tude des Matriaux (LPEM, CNRS 8213)
Universit Pierre & Marie Curie (UPMC) - ESPCI-ParisTech
10, rue Vauquelin - 75005 Paris - France

ABSTRACT
Through an analysis of different space charge measurement methods, the paper makes
recommendations for ensuring optimal implementations and measurements. The
thermal, pressure-wave-propagation and pulsed-electro-acoustic methods are considered.
From the underlying physics of these methods, the effect of impedance mismatches and
measurement conditions are discussed and simple signal treatments are presented taking
into account the case of non uniform materials.
Index Terms - Space charge distribution, thermal method, pressure-wave-propagation
method, pulsed-electro-Acoustic method, measurement conditions, signal treatment.

1 INTRODUCTION
THE measurement of space charge distributions gives valuable
information in the study of the electric properties of dielectrics.
Indeed, it shows in a direct and non-destructive way where charges
are trapped without requiring any electrical model, and allows step
by step charge buildup and/or migration to be followed. Using a
direct and non-destructive space charge distribution measurement
technique is however not as simple as it seems. Though they are
often shown as a turnkey system, many parameters may be
involved in the generation of the signal. As a consequence, a good
signal analysis often requires a complete knowledge of the system,
from the physical interaction during the measurement to the signal
processing where it is expressed in charge units.
In this paper I first recall the signal origin of thermal [1-3],
pressure-wave-propagation (PWP) [4, 5] and pulsed-electroacoustic (PEA) [6] methods in planar materials. It is worth noting
that PEA method should not be understood as restricted to
electrical pulsed excitation only but rather to any kind of rapid
electrical excitation. Uniform and non-uniform materials are taken
into account. Then the measurement method implementation is
discussed in terms of electrical, acoustic and thermal impedance
mismatches and measurement conditions. In the last section before
conclusion, the basic signal treatment is described. The special case
of heterogeneous materials is also discussed.

2 SIGNAL GENERATION
2.1 THERMAL AND PRESSURE WAVE
PROPAGATION METHODS
In the case of the thermal and PWP methods, an electrical
signal is produced by applying either a thermal or a
mechanical perturbation to the sample. The evolution of that
perturbation in the sample, a diffusion or a propagation
Manuscript received on 15 November 2011, in final form 22 March 2012.

depending on the nature of the perturbation, is relatively slow


compared to the time required by the electromagnetic state to
stabilize. Therefore it can be said that electrostatic equilibrium
is reached at any time during the measurement and thus that
the Gauss equation applies.
Due to the perturbation, the charge density becomes
, the permittivity becomes

and the electric


field E becomes in turn E E , E being the source of the
measured signal.

It is possible to characterize the variation E by the


difference of Gauss equations between during and before the
perturbation. That difference gives in effect what has changed
due to the perturbation and is then directly connected to the
generated signal. One has

div(( )( E E )) div( E ) 0 (1)

During perturbation

Before perturbation

which results at first order in

(2)
div( E E )

Therefore E is equivalent to the electric field that would


be generated in the sample by the charge
density and the

permanent dipole distribution E . In addition electrodes


would be grounded in short-circuit conditions or would hold
no charge in open-circuit conditions. Indeed the voltage does
not vary during the measurement in short-circuit conditions
therefore the difference in equation (1) results in a potential
equal to zero on the electrodes. Similarly the charges on the
electrodes do not vary during the measurement in open-circuit
conditions therefore the difference in equation (1) results in no
charges on the electrodes.
Physically speaking, the measured current in short-circuit
conditions comes from the physical displacement of charges
depending on and from the displacement current
depending on .

1070-9878/12/$25.00 2012 IEEE

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 19, No. 4; August 2012

The charge density variation can be obtained using the


charge conservation law

div( u )
(3)

with u the material displacement produced by thermal


expansion or mechanical deformation. Permittivity variation
depends however on the nature of the perturbation. In the
case of a mechanical perturbation, the electrostrictive
coefficient a links, by definition of the electrostrictive effect,

with the material deformation S div(u ) as

a div(u )
(4)
In most materials, the electrostrictive coefficient a is of the
order of / 2 [7]. In the case of a thermal perturbation, it is
necessary to take into account the primary and secondary
pyroelectric effects. The primary pyroelectric effect is defined as
the variation of polarization with temperature at constant volume,
that is to say with no material deformation. In the case of induced
polarization, the case for most insulators, a coefficient b links the
permittivity variation with the temperature variation T as

bT

a div(u ) a* div(u )
L

(6)

The charge variation Q on the electrode from which the


measurement is carried out, called for instance the measurement
electrode, can be calculated from equation (2) taking into
account equations (3), (4) and (6). The details are given in [8].
In short-circuit conditions and in planar geometry, one has

Q C0 1

a* u z
dz
E
z

(7)

where C0 is the sample capacity at rest and u z is the


component of material displacement along the sample thickness
axis. For a thermal perturbation, the measured current im (t ) is then
a* T
im (t ) C0 L 1 E
dz
t

(8)

and for a mechanical perturbation, the measured current


im (t ) is
a P
im (t ) C0 1 E
dz
t
with P the pressure and the elastic compressibility.

It should be mentioned that coefficients L and in (8)


and (9) are the exact coefficients for a free material, that is
to say a material that is not constrained at boundaries. Most
of the time, the thermal or mechanical perturbation is not
imposed over the whole surface of the material so the
material deformation is laterally constrained leading to a
slightly different coefficient taking into account Poisson
ratio. That small correction is fortunately not necessary for
the analysis.
2.2 PULSED-ELECTRO-ACOUSTIC METHOD
In the case of the PEA method,
an elastic signal is produced
by a rapid variation of field E . That rapid variation of field
is generated by a rapid voltage variation V on one
electrode, called for instance active electrode. If charges have
no time to move during the perturbation, the difference of
Gauss equations between during and before the electric
perturbation is

div( ( E E )) div( E ) 0
(10)

(9)

During perturbation

(5)

The secondary pyroelectric effect is defined as the variation


of polarization due to thermal expansion. In the case of
induced polarization, the permittivity variation is therefore
given by equation (4) where material deformation is produced
by thermal expansion.
Because thermal expansion is proportional to temperature,
it is possible to combine equations (4) and (5) to obtain an
equivalent electrostrictive coefficient a in the case of a
thermal perturbation and thus to treat both methods similarly.

One has in a planar geometry div(u ) L T , with L the


linear thermal expansion coefficient, so that

1209

Thus

Before perturbation

div(E ) 0

(11)

The field E is then the electric field that would be in the


sample in absence of charges when the active electrode is held
to V . Any other constant voltage applied to the sample
cancels in the difference (10).
The variation of the electric field modifies the electrostatic
force density acting inside the sample. Three contributions
have to be taken into account at least, (i) the coulombic force
density which variation is

f E
(12)
(ii) the force density acting on induced dipoles which
variation is [9]


f E E grad( ) 12 E 2 grad( )
(13)

and (iii) the correction due to electrostrictive effect since


permittivity slightly changes with material deformation. One
has


f a grad(aE E ) 12 grad(aE 2 )
(14)
The total force density variation is the source of the
generated elastic waves, and the material deformation
produced at a given position in space, in our purpose at the
transducer position, can be calculated by solving the second
Newton's law

2u

mv 2 grad(C div(u )) f f f a
(15)
t
where mv is the mass density and C is the elastic stiffness
constant. The general solution of equation (15) for the
material deformation at the transducer position is described in
detail in [8]. It is useful to introduce a Green's function to
solve equation (15). In the case of measuring a material

1210

S. Hol: Behind Space Charge Distribution Measurements

deformation Sm at the transducer position, one obtains in


planar geometry
a g
S m C0 1 E z V dz
z
C0 2
2A

1 a g z
V 2 dz
1
z

(16)

where A is the sample tested area and stands for


convolution product. Using the reciprocity theorem, it can be
shown that g z / z V is equivalent to the material
deformation produced inside the sample by the transducer if it
were supplied by the voltage V . That shows the great
similarity between equations (7) and (16) [8]. In absence of
attenuation and dispersion of sound, one has

g z
z

V 12 V t t
z
z
vs

reduces at the same time electric pulse attenuation and


dispersion in the cable in case of thick samples.

(17)

where is the transmission coefficient of elastic wave at the


interface of the sample, vs is the sound velocity in the sample
and t is the time required by elastic waves to reach the
transducer position from the sample interface.
The second line of equation (16) contains a square term that
cannot be neglected at first glance since E is of the order of
kV/mm. However, if E is much larger than E , the square
term can be neglected. Otherwise subtracting for instance a
signal obtained with the voltage variation V with one
obtained with the voltage variation V eliminates the
influence of that square term. It can then be omitted in the
following discussions.

Figure 2. Electrical matching of the pulse generator by an impedance Z to


avoid electrical reflections. The voltage on the active electrode (graph) shows
the incident pulse but no longer echo for two cable lengths.

However, the rise time of the voltage on the active


electrode depends on the cable impedance and on the sample
capacitance (see Figure 3). The presence of a resistor in
parallel with the sample reduces the equivalent resistance, but
also the pulse amplitude.

Figure 3. (a) Equivalent circuit at the active electrode and corresponding rise
time. (b) Reduction of a coax cable impedance.

If the sample capacitance is too large, it can be useful to


reduce the impedance of the coax cable. That impedance
depends on the ratio between external electrode radius rint and
internal electrode radius rext as
1
2

Figure 1. The multiple reflections of the electric pulse in the cable may
induce superimposed signals. The voltage on the active electrode (graph)
shows the incident pulse and the echo for two cable lengths.

3 IMPLEMENTATION
3.1 ELECTRICAL IMPEDANCE MISMATCH
In the case of the PEA method the voltage pulse V
experiences electrical reflections in the cable between the
generator and the sample. These reflections may produce
multiple mixed signals as illustrated in Figure 1.
The usual work around is to lengthen the cable in order to
delay the reflections. As a consequence the signal generated by
each electrical pulse reflection is sufficiently delayed from the
preceding pulse so as not to superimpose. This is shown for two
cable lengths in Figure 1. Another possibility is to match the
pulse generator to the cable impedance for obtaining equivalent
results with a shorter cable length as illustrated in Figure 2. That

rext
ln

rint

(18)

where and are the permeability and permittivity of the


cable insulator. A simple way to reduce the cable impedance
is to add a shield electrode on a coax cable or to use a triax
cable, the external shield electrode being connected to ground
and the two internal electrodes being connected to the active
electrode. As an example this divides by about 4 the
impedance of a RG58 coax cable and thus makes it possible to
obtain 4 times shorter electric pulse. Another way to reduce
the cable impedance is to use numerous 50 cables in
parallel, each being matched at the pulse generator side by a
resistor. The overall impedance is thus divided by the number
of cable in parallel.
3.2 PHYSICAL IMPEDANCE MISMATCH
Even if the sample is completely uniform, reflections can
occur at interfaces, leading to a modification of temperature or
elastic wave profile through the sample. The reflection
coefficient of a heat flux or pressure wave can be estimated by

Z E ZS
with Z cmv or Z mv vs
ZE ZS

(19)

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 19, No. 4; August 2012

where subscripts E and S respectively refer to as electrodes


and sample, is the thermal conductivity, c is heat capacity,
mv is mass density and vs is sound velocity. Figure 4 shows
the effect of a mismatch. In the case of thermal diffusion, the
impedance, also known as effusivity, is larger for medium 1.
As a consequence the temperature increase at the back
interface and a back diffusion has to be taken into account
since it superimposes on the useful signal. In the case of an
elastic propagation, the acoustic impedance is larger in
medium 2. The pressure is larger at the interface and the
displacement is smaller (smaller signal). The back reflection
can be taken into account but it is not absolutely necessary
since it does not superimpose on the useful signal.

1211

Figure 5. (a) Typical calibration measurement. (b) Pulse spectra and matching
correction.

Fortunately low frequencies do not attenuate and disperse,


therefore the ratio of the back to the front peaks at low
frequencies in the spectrum should be unity in case of perfect
matching, larger than one for a back material with lower
acoustic impedance and lower than one otherwise. This is
illustrated in Figure 5b for a back material with lower
impedance. Low frequencies also correspond to the mean
peak amplitude in the time domain. Therefore the ratio of the
back to the front peaks at low frequencies in the spectrum can
be approximated using the back peak amplitude multiplied by
its width divided by the front peak amplitude multiplied by its
width in the time domain. In the case of Figure 5, one has
1.8810 in the time domain and 1.8816 in the spectrum. The
error of approximation is thus negligible and the second peak
in Figure 5 should be understood as 1.88 times smaller than it
is in the signal.

Figure 4. (a) Effect of a thermal mismatch at different times, a back diffusion


as to be taken into account for the temperature profile (dotted line). (b) Effect
of an elastic mismatch at different times, pressure (dotted line) is modified at
interface. All profiles are decomposed in incident (black), reflection (light
gray) and transmission (dark gray) components.

A matching material is often used at the back interface, for


instance black carbon polyethylene or a material similar to the
sample. In the case of elastic waves, it is possible to estimate
the reflection coefficient at the back electrode by comparing
the amplitudes at low frequencies of the front and back
electrode peaks of a calibration measurement. In a calibration
measurement, of which an example is shown in Figure 5a, the
two peaks correspond to the signal generated at each interface
of the sample and are an image of the pressure at these
interfaces. The difference between these two peaks, excluding
peak polarity which is due to the charge polarity on the
electrodes, is that one is the image of the pressure before
traveling through the sample whereas the other is the image of
the pressure after having traveled through the sample. That
difference is particularly useful in the deconvolution
procedure for determining elastic attenuation and dispersion
[10]. The direct consequence is that it is not possible to use
the peak amplitudes directly to estimate the matching, mainly
because of dispersion.

Figure 6. (a) Equivalent measurement circuit. (b) Estimation error in almost


open-circuit conditions. (c) Loss of resolution in almost short-circuit
conditions.

3.3 MEASUREMENT CONDITIONS


Measuring the signal for the thermal or PWP methods
simply consists in connecting an amplifier to the sample.
Depending on the sample capacitance C0 and the amplifier
input impedance R , a cutoff frequency appears (see Figure
6a). In the case of the PEA method it is not the sample
capacitance that is important but the transducer capacitance. If
all signal frequencies are above the cutoff frequency, in other
words if RC0 is larger than the measurement duration, opencircuit conditions apply. If all signal frequencies are under that
cutoff frequency, in other words if RC0 is smaller than signal
rise time, short-circuit conditions apply. Otherwise, in the case
of a mixed condition, estimation errors (Figure 6b) or a loss of
resolution (Figure 6c) may result.
To analyze a signal, it is very important to know which
measurement conditions apply, especially when C0 is

1212

S. Hol: Behind Space Charge Distribution Measurements

unknown. This can be estimated by introducing an additional


resistor or capacitor between the sample and the amplifier as
shown in Figure 7.

adjusted so that the tails of v(t ) and v(t ) are of the same
amplitude. Subtracting these two signals will result in a signal
with more symmetrical and shorter peaks.
Besides being simple, the advantage of such procedure is
that noise increases by a limited factor, at least 1 . The
example in Figure 8 shows a direct spatial resolution
improvement by a factor 1.58 with a signal to noise ratio
reduced by a factor 1.39 only.

Figure 7. (a) Frequency response with a resistor in series. (b) Frequency


response with a capacity in series. SC stands for short-circuit, OC for opencircuit, and mixed for mixed measuring conditions.

4.2 INTERFACIAL CHARGES


Charges at the interfaces are commonly encountered,
through injection or accumulation for instance. The signal
they produce is then mixed with the one produced by the
charges on the electrodes and the spatial resolution [11] is not
sufficient to distinguish between them. In the case of homocharges, the PWP and PEA interfacial peak slightly broadens
but its amplitude remains a good indication of the interfacial
field (Figure 9a). In the case of hetero-charges, the peak
becomes bipolar and the maximal and minimal values shift
(Figure 9b). It can be shown that interfacial field is better
approximated using the difference between the maximal and
minimal values.

Adding a resistor in series reduces the cut-off frequency


and high frequency amplitudes but does not change low
frequency amplitudes. As a consequence the signal would
reduce in amplitude without deformation in open-circuit
conditions, remain unchanged in short-circuit conditions or
deform in mixed conditions (Figure 7a). Conversely adding a
capacitor in series increases the cut-off frequency, reduces
low frequency amplitudes but does not change high frequency
amplitudes. In this case, the signal would remain unchanged
in open-circuit conditions, reduce in amplitude without
deformation in short-circuit conditions or become deformed in
mixed conditions (Figure 7b).

Figure 8. (a) Simple treatment in case of mixed measuring conditions. (b)


The resulting signal has more symmetrical peaks.

4 SIMPLE SIGNAL TREATMENT


4.1 MIXED MEASUREMENT CONDITIONS
When measurement conditions imposed by the sample
capacitance and amplifier input impedance are in a mixed
situation, adding a resistor or a capacitor can be a solution to
shift the cutoff frequency sufficiently but it is at the expanse
of the signal amplitude and thus detectability in other words.
A simple work around without making complicated
calculations consists of subtracting from the signal an
amplitude-adapted delayed-copy of that signal. One has for a
signal v(t )
v '(t ) v(t ) v(t )

(20)

where and are respectively the coefficients for


amplitude and delay. In terms of signal processing, equation
(20) almost corresponds to a convolution with a 1 s
where s is the Laplace variable. A good choice for is the
rise time of the signal. Once has been chosen, is

Figure 9. (a) Interfacial peak in presence of homo-charges. (b) Interfacial


peak in presence of hetero-charges. (c) Position of the zero (solid line) and of
the maximal value (dashed line) with respect to the original position of the
maximal value as a function of the hetero-charge width. (d) Hetero-charge
estimation error as a function of the hetero-charge width.

Large errors remain however when charges are very close


to the electrode especially when hetero-charges are of the
same order of magnitude than capacitive charges. In such a
situation the zero in the signal is exactly positioned where was
the original maximum. Figures 9c and 9d are examples of the
shifts (Figure 9c) that can be observed in a measurement and
charge estimation errors (Figure 9d) when there are as much
hetero-charges in the material as capacitive charges on the

IEEE Transactions on Dielectrics and Electrical Insulation

Vol. 19, No. 4; August 2012

1213

electrode. Calculations are made as a function of the heterocharge extent. In any cases the charge estimation error is the
largest when considering only the maximal value of the
measurement. Less than 12.5% error is made however with
the difference between the maximal and the manimal values
down to a hetero-charge extent which is 33% the peak width.
That corresponds to a zero positioned in the measured signal
at 30%-peak width on the right of the original peak position or
a peak shift of 20%-peak width on the left. For hetero-charge
extents larger that 70% the peak width, it is even better to use
the difference between the maximal value and half the
minimal value for the estimation of the hetero-charges.
Owing to its better spatial resolution at the interface, the
thermal method gives a better estimation of the field at the
front electrode.
4.3 ELECTRIC FIELD PROFILE
In the case of the PWP and PEA methods the electric field
distribution can easily be obtained in true units. The procedure
uses a calibration signal obtained from the sample without
internal charges when submitted to a moderate voltage, and
any other signal. If the sample already contains charges at the
origin, it is possible to estimate the calibration measurement
from the difference between two successive measurements,
one obtained under voltage and the other obtained under
short-circuit. The signal due to the internal charges then
naturally cancels.
For a pulsed perturbation, the ratio between the signal
integral and the calibration integral derived from equations (9)
and (16) is approximately
t

Figure 10. Simple calculation of the electric field profile in true units. (a)
Calibration measurement, (b) amplitude coefficient H estimation by an
integration over time and (c) electric field distribution deduced by the signal
integration multiplied by Vc / dH .

The spurious signal deviation observed between the peaks


is sometimes explained as a conductivity mismatch
between the materials. The charges moved by the conduction
current would then accumulate on each side of the defect. This
is however not generally consistent with the time constant
/ required for charges to build up. In conventional
insulators, the time constant can indeed be hours. On the
contrary the effect of a spatial variation of permittivity appears
instantaneously in the signal.

im (t ')dt ' Sm (t ')dt ' A E ( z v t )


s
t
t
ic (t ')dt ' Sc (t ')dt ' C0Vc

(21)

where the subscript c denotes the calibration, A is the


sample area, Vc is the calibration voltage and E is the mean
electric field over an extent corresponding to the spatial
resolution at the considered position z . The term A / C0 is
equal to the sample thickness in the case of a uniform
material. Notice that with PWP method, if signals are obtained
in open-circuit conditions, no integrals over time are required
in the left hand side of equation (21). Figure 10 illustrates the
procedure for a uniform material.
First, the calibration signal is integrated over time. The
amplitude coefficient H is estimated. The following signals
are then integrated and corrected by the amplitude coefficient
and mean electric field during the calibration measurement.
4.4 HETEROGENEOUS MATERIALS
Charges are often thought of as the only source of the
signal. However equations (7) and (16) clearly show that a
variation of permittivity inside the sample also produces a
signal in the presence of an electric field. Figure 11a shows
such a signal in a material due to a permittivity defect in its
center. The superimposed X-ray image of the sample crosssection clearly shows the defect.

Figure 11. Signal in presence of a heterogeneity in the middle of the sample.

The spurious signal can be confused with a charge induced


signal. Expression (21) can be used to correct the signal and
analyze it as if the sample were uniform [12]. Indeed the
spatial derivative of equation (21) gives the real charge
distribution since AC0 / Vc is a constant.

5 CONCLUSION
Measuring space charge seems relatively simple, but the
complexity of the sample structure and interfacial conditions
may result in biased analysis. In this paper, some clues are
given to test which factors are important to check before
analyzing measurements and simple calculation procedures
are given to test the consistency of measurements and to
obtain reliable preliminary results.

1214

S. Hol: Behind Space Charge Distribution Measurements

REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]

[7]
[8]
[9]

R.E. Collins, Analysis of spatial distribution of charges and dipoles in


electrets by a transient heating technique, J. Appl. Phys., Vol. 47, pp.
4404-4408, 1976.
S.B. Lang and D.K. Das-Gupta, Laser-intensity-modulation method: a
technique for determination of spatial distributions of polarization and space
charge in polymer electrets, J. Appl. Phys., Vol. 59, pp. 2151-2160, 1986.
A. Toureille and J.P. Reboul, The thermal step technique applied to the
study of charge decay in polyethylene thermoelectrets, IEEE 6th Intl.
Sympos. Electrets, pp. 23-27, 1988.
P. Laurenceau, G. Dreyfus and J. Lewiner, New principle for the
determination of potential distributions in dielectrics, Phys. Rev. Lett.,
Vol. 38, pp. 46-49, 1977.
C. Alqui, G. Dreyfus and J. Lewiner, Stress wave probing of electric
field distributions in dielectrics, Phys. Rev. Lett., Vol. 47, pp. 14831487, 1981.
T. Maeno, H. Kushiba, T. Takada and C.M. Cooke, Pulsed electroacoustic method for the measurement of volume charge in e-beam
irradiated PMMA, IEEE Conf. Electr. Insul. Dielectr. Phenomena , pp.
389-397, 1985.
K. Nakamura and Y. Wada, Piezoelectricity, pyroelectricity and
electrostriction constant in PVF, J. Polym. Sci., Vol. 9, pp. 161-173, 1971.
S. Hol, T. Ditchi and J. Lewiner, Non-destructive methods for space
charge distribution measurements: what are the differences? IEEE
Trans. Dielectr. Electr. Insul., Vol. 10, pp. 670-677, 2003.
S. Hol, T. Ditchi and J. Lewiner, Influence of divergent fields on
space charge distribution measurements using elastic methods, Phys.
Rev. B, Vol. 61, pp. 13528-13539, 2000.

[10] T. Ditchi, C. Alqui and J. Lewiner, Broadband determination of


ultrasonic attenuation and phase velocity in insulating materials, J.
Acoust. Soc. Amer., Vol. 94, pp. 3061-3066, 1993.
[11] S. Hol, Resolution of direct space charge distribution measurement
methods, IEEE Trans. Dielectr. Electr. Insul., Vol. 15, pp. 861-871,
2008.
[12] S. Hol, Space Charge Distribution Measurement in Complex
Insulating Structures, IEEJ Trans. Electr. Electron. Eng., Vol. 5, pp.
405-409, 2010.
Stphane Hol, was born in 1968 at Pontoise
(France).
He
studied
electronics
and
instrumentation at Universit Pierre et Marie CurieParis6 (Paris, France). He joined Laboratoire
d'lectricit Gnrale of cole Suprieure de
Physique et de Chimie Industrielles (Paris, France)
to study an instrument for measuring fast
development of space charges in insulators under
rapid voltage variations. It was the topic of his PhD
he received in 1996. Since 1997 he has been Matre
de Confrences (assistant professor) at Universit Pierre et Marie Curie in
Laboratoire des Instruments et Systmes d'Ile de France from 1997 to 2007
and in Laboratoire de Physique et d'tude des Materiaux since 2007 where he
leads the Instrumentation Group. He teaches solid state physics, electronics
and sensor physics. He conducts researches on various topics such as space
charge in insulators and semiconductors (main topic), electrostatic and
magnetostatic sensors. He received the Jack Hollingum Award in 2002 and
2004, and obtained his Habilitation in 2007.

You might also like