Journal of ELECTRICAL ENGINEERING, VOL. 55, NO.

9-10, 2004, 225–231

POLYMER FILMS DEGRADATION AND

BREAKDOWN IN HIGH VOLTAGE AC FIELDS

Konstantinos Theodosiou* — Isidoros Vitellas**

— Ioannis Gialas*— Dimosthenis P. Agoris***

This paper examines the mechanism of electrical degradation and breakdown of polymeric insulating films as a sequence
of electron injection by Fowler-Nordheim tunnelling, and macro-ion bond scission resulting in the creation of another macro-
ion and a free radical. This is a thermo-fluctuational process which involves the action of the stretching force by the local
electric field. Breakdown initiation starts when an electron following a Poole-Frenkel hopping mechanism, is accelerated
in a hole with sufficiently large dimensions. Experiments were executed for measuring the breakdown AC voltage Vb and
dielectric strength Fb , with voltage rising rate 3 kV/sec, in order to examine their relation with specimen thickness d and
hence to derive a steady state breakdown criterion. An attempt of prediction of the polymer lifetime is also taking place.
K e y w o r d s: polymer, degradation, conduction, tunnelling, micro-voids, macromolecules.

1 INTRODUCTION 2 BRIEF REVIEW OF THE ELECTRICAL

CONDUCTION AND BREAKDOWN
MECHANISMS IN POLYMERIC INSULATORS
Polymeric insulating materials are used in solid insula-
tions of high voltage equipment such us cables, insulators, Many papers published in the area have shown that
transformers and electrical machines. In many applica- especially in high fields the shape of the conduction is fil-
tions polyethylene and its co-products are used as the amentary and starts with the electron injection from the
main insulating material, because they have fine dielec- metal electrode to the polymer[8, 9, 10]. This process is
tric properties and a low production cost. In practice, the through the Fowler-Nordheim tunneling and allows elec-
insulating materials may be arranged in one or more lay- trons to overcome the potential barrier that exists be-
ers in the insulation and usually some additives are mixed tween the metal electrode and the polymer. They have
with them for improving the insulation characteristics. also noted that the formation of the so-called low density
region [13, 16, 17, 18, 19] is the first step in the breakdown
Many models have been introduced by researchers of solid insulators. This process is more likely to increase
[1, 4, 7, 14] attempting to explain, macroscopically or mi- near the injecting electrode where the concentration of
croscopically, the aforementioned properties and charac- injected electrons is higher and the low density region ex-
teristics. More recent researches consider electrical break- pands first linearly and later in a more complex manner
down initiation as a molecular scale process, which is not forming branches[13,14,15] and tree-like structures.
the case in the classical approach. The free volume the- Once the low density region has been formed and un-
ory[5, 6, 7], which is an extreme value theory, predicts that der appropriate conditions [7](mean free path, ionization
breakdown starts when an electron having been acceler- energy Ei of the radicals and applied electric field F ), im-
ated in the longest free path, gains sufficient kinetic en- pact ionization may follow and the avalanche may propa-
ergy to hop over a potential barrier. gate creating conduction channels. This finally leads to an
increase in carrier multiplication that extends the chan-
The objective of this paper is to focus on the AC
nels and destroys the material inside them [11, 13].
field conduction and breakdown mechanism on polymeric
Electrons have been injected in the polymer, quickly
insulation, studying experimental results of breakdown
become trapped after a few scatterings [11] because of
measurements on insulating films, with PET as the main the small mean free path and consequently form a neg-
insulating material. In the next section a brief overview atively homo-space charge in the surface near to the in-
of the theory of degradation and breakdown in polymers jection contact. This trapping process takes place in lo-
is employed. calized energy states that exist in plenty in the poly-

*University of the Aegean, 82100 Chios, Greece, E-mail: k.theodosiou@physics.org

**Public Power Corporation, 9 Leontarion Street, Kantza, 15351, Athens, Greece
***University of Patras, 26500 Rio, Patras, Greece

c 2004 FEI STU

ISSN 1335-3632
226 K. Theodosiou et al: POLYMER FILMS DEGRADATION AND BREAKDOWN IN HIGH VOLTAGE AC FIELDS

mer due to many reasons such as structural or chemi- formed during manufacture by evaporation of decomposi-
cal imperfections etc. This homo-space charge creates an tion products in various chemical reactions taking place in
internal electric field opposite to the applied field that cross-linking, or by impurities and additives decomposi-
retards and may stops the injection process. According tion or by migration. Inside the hole, free electrons follow
to Liu and Kao [10], in addition to the creation of the a free path significantly larger than that in the compact
homo-space charge region, the trapping process results structure. They can gain a larger kinetic energy and, thus,
in energy transfer from the trapped electrons with two lower the potential barrier Eµ that try to overcome. The
possible ways: (a) directly to the region around the trap larger the free path and the hole volume, the more kinetic
causing structural damage breaking bonds like carbon- energy they gain [5, 6, 13]. In case the electrons acquire ki-
carbon or carbon-hydrogen (aging) and (b) to another netic energy equal or larger than (Eµ ), they overcome the
electron and making it hot due to an Auger type process
barrier and the breakdown starts. Eµ is not necessarily
[11]. This electron can also cause structural damage and
the ionization energy as it was originally supposed to be
(or) create another hot electron. The process may con-
[7], but the energy of the potential barriers in the com-
tinue and produce more radicals that form the low density
pact structure because electronic conduction is thermally
region. Unfortunately this multi-stage process has some
insuperable difficulties. In Ref. [20], this process is not re- assisted. Kao in [10, 11] had concluded that thermal en-
garded as a realistic model that describes the dissociation ergy is gained by an internal process via an Auger type
of macromolecules under the action of the electric field. phenomenon, while in Ref. [20] such a process is regarded
The reasons are mainly the lack of any calculation about to have a very low probability to happen. Some of the
the probability of energy transfer or dissipation in each previously mentioned references will be discussed later in
stage, the fact that the energy that is released in a trap- more detail.
ping event 1.5 eV is lower than the energy for the C-C In the following section the description of the sam-
bond rupture 3 eV and the absence of any experimental ples and the experimental procedure is presented. Then
results that indicate chain energy reactions in polymers. the experimental results and theoretical relations are dis-
The authors of [20] consider that the model of Ref. [12] in cussed. Finally, a few concluding remarks are presented.
which dissociation of macromolecules is a consequence of
charge recombination according to the schematic mecha-
nism
3 EXPERIMENTAL
AB + + e −→ (AB)∗ −→ Ȧ + Ḃ
PROCEDURE AND SAMPLES
where AB + is the macro-ion and (AB)∗ the excited
macromolecule, is more realistic. The reason is that the Polymer sheets are commonly used for the insulation
released energy 6 eV is higher than the C–C bond rup- of electric motor winding. The scope was the experimental
ture energy. They also propose a model in which elec- determination of dielectric characteristics of such insulat-
trical degradation and breakdown occur due ionization ing materials, especially their relation with the thickness
of macromolecules by electron tunnelling followed by of the insulating material.
thermo-fluctuational rupture of bonds according to the
scheme Specimens of high tenacity PET films with thickness of
κ1
M + e −→
κ2
Ṁ + −→ M1+ + Ṙ 23, 50, 75, 125, 190 and 350 µm and in square form with
dimensions 120 × 120 mm2 were constructed. To ensure a
Unfortunately, they do not consider the influence of good contact between specimen and test cell electrodes,
space-charge and image force potentials that is very im- additional copper electrodes were embodied on the sur-
portant especially near to the electrode. faces of the specimen, with diameters 10 and 30 mm for
According to [7, 13, 14] electrical breakdown must at breakdown measurements, and 45, 45 mm for PD mea-
least begin at a molecular level and involve molecu- surements. These electrodes were constructed in vacuum
lar scale processes, which is not part of the classical conditions with instant evaporation of a suitable amount
approaches. Because of the fact that the bulk of any of copper, avoiding air enclosure between the samples sur-
semicrystalline polymer, such as polyethylene terephtha- face and electrodes. Because of the short time duration
late (PET), contains crystalline and amorphous regions of the copper evaporation, no damage occurred on the
[2, 3] with the latter having a random structure, their vol- specimen surfaces, while a good contact was maintained
ume is not completely occupied by atoms and includes
between specimen and evaporated electrodes, as verified
empty spaces (free volume) which represent at least 2.5 %
by systematic sampling in samples cross sections with the
of the total volume. These spaces differ from the crystal
aid of a microscope. In Fig. 1 a specimen is shown.
structure defects and are always present in the bulk in the
form of empty holes between atoms. They are the reason A special experimental test cell was used, filled with
of the density decrease and specific volume increase and power transformer oil and equipped with stainless steel
show the inefficiency of molecular packing of the specific electrodes with rounding edges, as shown in Fig. 2. The
material. The hole size is of the order of the molecular size diameters of the main electrodes were equal to the corre-
but because it is a random number, it can reach the order sponding copper electrodes of the samples. The procedure
of a few nm (< 3×10−9 m) [1]. Voids and micro-voids are for putting a sample in the test cell was the following: the
 Journal of ELECTRICAL ENGINEERING VOL. 55, NO. 9-10, 2004 227

4 RESULTS AND DISCUSSION

4.1 Electrical degradation

4.1.1 Fowler-Nordheim tunnelling in the

electrode-polymer interface

As previously stated, the first step for the creation

of the low-density region is the electron tunneling from
the electrode metal to the bulk of the polymer. In high
electric fields, ionization of macromolecules can occur due
to the tunneling of electrons through reduced potential
barriers (field ionization) as have been stated in many
studies [7, 10, 25, 26]. The resulting current density due
to Fowler-Nordheim tunneling is
Fig. 1. View of the specimen inside the experimental cell.
J(t) = CFl2 (t) exp(−β/Fl (t)) (1)

with
HV
e3 4(2me )1/2
C= , β= (Φe )3/2 (2)
1 2 16π 2 ~Φe 3~e

3 4 where e and me are the charge and the effective mass of

5 electron respectively, ~ is the Planck constant, Φe is the
effective potential barrier in eV, and Fl (t) is the intensity
of the local electric field acting in the contact, which will
be discussed later.
The effective potential barrier through which electrons
Fig. 2. Experimental cell filled with power transformer’s oil. 1) In- tunnel is actually the result of other potentials that sum
ner cell and electrodes base 2) Oil can 3) PET film sample 4) High up as follows:
voltage electrode 5) Earth electrode
h e3 F (t) i1/2 e2
l
Φe = Φm − χ − P − − (3)
4πǫǫ0 4πǫǫ0 x
electrodes base (1) was taken out of the test cell, the elec-
trodes surfaces (4–5) were cleaned from the oil, the sam- with
ple (3) was placed and fixed between the main electrodes
and finally the base (1) was assembled in the bottom of Φep = Φm − χ (4)
the test cell and connected to an AC power supply. The
dielectric strength of the oil (higher than 60 kV/mm) was the potential in the electrode-polymer contact where Φm
measured regularly and was replaced when it was neces- is the metal work function and χ the electron affinity of
sary. the polymer,
h e3 F (t) i1/2
l
∆Φ(t) = (5)
3.1 Breakdown voltage measurements 4πǫǫ0
is the image force potential (ǫ the dielectric permittivity
The power supply used for the experiment was an AC of the polymer, ǫ0 = 8.85 × 10−12 F/m the permittivity
50 Hz, 100 kV, 100 mA. The voltage applied to the spec- constant of the free space), P is the polarization energy
imen increased from zero at a constant rate of 3 kV/s, produced by a pair of charges and the last term in (3) is
to the breakdown voltage. When breakdown occurred, the potential Φn of the surrounding nuclei. The electron
the voltage supply was interrupted automatically, with energy gained from the field is
the current limiting protection system adjusted at 6 mA.
The breakdown voltage was monitored using a digital We = eFl (t)x (6)
voltmeter with memory and the maximum value in each
and hence the electrode-polymer barrier transparency
specimen was recorded. For six groups of specimens, each
(probability of overcoming the barrier) is equal to
one consisting of twenty samples with the same thickness
y , the breakdown voltage Vb was measured and the di- h 2√2m Z x0 i
e
electric strength Fb of each sample was calculated. De ∼ = exp − (Φe − We )1/2 dx , (7)
~ 0
228 K. Theodosiou et al: POLYMER FILMS DEGRADATION AND BREAKDOWN IN HIGH VOLTAGE AC FIELDS

Breakdown Voltage Vb (kV) with Q(t) the total trapped charge per unit area and x0
16 the position of its center in the specimen. If we consider
the expression of electron trapping rate [10], the total
14 trapped charge will be then
12 h  σZ t i
Q(t) = ex0 Nt 1 − exp − J(t)dt . (13)
e 0
10

8
Now the variation of the intensity of the local electric field
can be evaluated by differentiating relation (11)
6
dFl (t) dV (t) dFi (t)
Polyfilm
= (α/d) − . (14)
4 dt dt dt

2 For an applied voltage with rising rate (λkV /s ), the for-

0 50 100 150 200 250 300 350 400 mer relation gives
Specimen Thickness d (µm)

Fig. 3. Breakdown Voltage in PET film sample versus Thickness d.

dFl (t)
= ρa − ρi (15)
dt

with
where x0 is the barrier width, and if we assume a trian-
gular shape, becomes ρa = (α/d)λ (16)
h 4(2m )1/2 (Φ − W )3/2 i
e e e and (because of the fact that x0 ≪ d and by differenti-
De ∼
= exp . (8)
3~e Fl (t) ating relation (13))
If we need to know the probability that electron tunnel- (
 σZ t
)
ing has to cause macromolecule ionization, equation (3) eNt h σJ(t)d dx0 i  dx
0
ρi = + exp − J(t)dt − .
should be rewritten as ǫǫ0 e dt q 0 dt
h e3 F (t) i1/2
l e2 (17)
Φei = Ig − P − − (9) It is obvious from (17) that ρi is controlled by many
4πǫǫ0 4πǫǫ0 x
parameters such as carrier mobility, distribution of traps
where Ig is the ionization energy of a macromolecule and their capture cross section, dielectric permittivity etc.
in the gaseous phase. The resulting tunneling frequency Especially for x0 , it can be said that is not a constant
(tunneling probability per unit time) is given by but moves towards the injection contact and is always
γ1 = ν0 Di (10) negative when a DC voltage is applied. This happens
because after the first group of injected carriers is trapped
where ν0 ≈ 10−16 Hz is the electron collision frequency somewhere in the bulk, it forms the first distribution
with the barrier. The last two equations give an estima- of space charge that tends to retard the next group of
tion for the tunneling macromolecule ionization rate that injected carriers. Gradually, the traps between the space
can be described by the following schematic mechanism charge center and the injection contact are filled in the
γ1 same direction. It can be seen from (15) and (1) that there
M + e −→ Ṁ + are three possible ways to describe the J − Fl or I − V
where M macromolecule and Ṁ + the resulting macro- relation.
ion. (i) ρa > ρi then Fl increases and consequently J also
At this point it is necessary to elucidate the role of the increases.
electric field acting in a specific micro-region. The local (ii) ρa = ρi thus Fl becomes time independent and J
electric field Fl (t) is becomes saturated, as a result of constant ρi .
V (iii) ρa < ρi and J after reaching a peak value, decreases
Fl (t) = α − Fi (t) (11) with Fl increasing. This will happen if Nt or σ or
d
both are large.
where V is the applied voltage, d is the polymer thick- As it was stated above, as the first group of injected
ness(electrode spacing), α the local over-stress factor carriers is trapped somewhere in the bulk, it tends to
which depends on the local morphology and Fi (t) the retard the next group of injected carriers. As the ap-
internal electric field created by the trapped electrons. plied field is increasing, the deep traps between the space
This field tends to retard the on-coming groups of in-
charge center and the injection contact are filled and the
jected electrons and its intensity is given by
further injection is enhanced. It is also very important
Q(t)  x0  to notice that each trapping event is associated with an
Fi (t) = 1− (12) energy of the order of 2 eV depending on the trap energy
ǫǫ0 d
 Journal of ELECTRICAL ENGINEERING VOL. 55, NO. 9-10, 2004 229

level, but in any case not enough to directly rupture a C– where M1+ is a new macro-ion and Ṙ a free radical. The
C bond that needs more than 3 eV to do so. This energy last schematic process represents the degradation mech-
can be also dissipated by creating thermal instability in anism of a polymer and can be used for its lifetime pre-
the micro-region around the trap. diction. When a sufficiently large amount of the bonds of
Another very important mechanism in AC fields that the polymer is destroyed, breakdown is unavoidable. The
can directly lead to macromolecule dissociation, bond left arrow and the factor r are added to stress the role of
breaking and hence formation of low density region is charge recombination in the AC voltage, emphasizing the
charge recombination (a free electron and a positive fact that although it dispossess carries, provides a large
macro-ion pair) [20]. The release of energy in this case amount of energy that can also cause bond rupture as it
is in the order of almost 6 eV and is accompanied by has been previously stated. Although an analytical solu-
electroluminescence. Contrary to the above, charge re- tion is very difficult to be found due to time variation of
combination is not significant in the DC fields [22]. In the γ1 and γ1 , for room temperatures and electric fields that
next sections we will discuss the aspects of the second are not approaching very much breakdown, the lifetime
part of the breakdown mechanism that leads to macro- τpl of the polymer its undoubtedly depends upon the elec-
ion dissociation. tric field strength with an exponential relation. As the Fl
increases, τpl decreases. Many researchers have concluded
in this result [7, 20, 30].
4.1.2 Macro-ion bond rupture
4.2 Breakdown initiation by electron hopping
After the macromolecule ionization, macro-ions are and Poole-Frenkel mechanism
subjected to the electric field which can induce mechan-
ical strain, especially in the region near the electrode
As it was stated above, PET is a partially crystalline
where the charge concentration is likely to be larger. In
insulating material with a large band gap. The movement
the the case of AC field, there will be a localized viscoelas-
of electrons in the bulk occurs via hopping between local-
tic response of the macro-ion segments and a massaging in
ized states-traps below the conduction band, in which the
the field direction. The strains induced and the viscoelas-
hopping probability is increased by the electric field, due
tic loss processes will favor bond rupture. The rupture
to Poole-Frenkel effect. It must be stated here that the
energy G+ for a C–C bond in a macro-ion is of the order
hopping process for sort time breakdown experiments, is
of ≈ 1.9 eV and is lower than the energy for breaking
taking place between shallow traps located in the region
the same bond in the macromolecule G ≈ 3.6 eV [20, 21].
of 0.3–0.6 eV [14, 20]. They may be foreign molecules,
Due to the effect of the electric field, G+ decreases and
carbonyl or other more complex groups, or even polar-
becomes
ized states and their concentration (1018 cm−3 ) is suffi-
G+ +
F = G − l0 eFl (18)
ciently larger than that of the deep traps [1, 14]. At high
where l0 ≈ 1Å is the activation length1 and Fl the electric fields the traps depth is reduced as a result of
local electric field from equation (11). This significant Poole-Frenkel mechanism and the current is increased ex-
decrease of the bond dissociation energy in the macro-ion ponentially with the square root of the electric field [24].
makes its bonds favorable to scission through a thermo- Under these conditions, electrons velocity uth and con-
fluctuational mechanism with a probability centration ne increase supra-linearly [7, 23, 27, 31]. They
 G+ − l eF  both depend on the ionization energy Ig , temperature T
Psc = exp −
0 l
. (19) and the electric field strength Fl that lowers the poten-
kT tial barrier. According to the Poole-Frenkel model [27],
The macro-ion dissociation frequency is given by the following approximation can be derived for these two
quantities:
1  G+ − l eF 
0 l
γ2 = exp − (20) uth = νx exp[−(Ig − eFl x)/kT ] , (22)
τ0 kT
1/2
where τ0 ∼= 6.25 × 10−12 s. As a result from the last equa- ne ≃ Ne exp[−(Ig − βP F Fl )/(2kT )] (23)
tion, the probability of scission is increased dramatically
when the local electric field approaches the value where Ne is the ionized center concentration, βP F =
[e3 /(πε0 ε)]1/2 is the Poole-Frenkel parameter with ε0 ε
G+ been the static permittivity. Electrons hop randomly over
Fl = . (21)
l0 e energy barriers and the velocity gained due to accelera-
tion from the applied field is added to the thermal veloc-
This value forms the maximum local electric field that the
ity. If the length of the free path inside a hole is x , the
macro-ion bond can sustain, thus the criterion for bond
kinetic energy K gained by the electron, will be equal to:
scission. At this point the second stage of the polymer
dissociation process can be added to the first one that K = eV = eFl x . (24)
described in Section 4.1 as follows
1 Itmust be stated here that activation length l0 can be changed
γ1 −r
+ γ2
M + e ⇄ Ṁ −→ M1+ + Ṙ by local thermal fluctuations
230 K. Theodosiou et al: POLYMER FILMS DEGRADATION AND BREAKDOWN IN HIGH VOLTAGE AC FIELDS

Dielectric Strength Fb (kV/mm) of the material. As we are interested in the maximum

220
(extreme) value of the free path and the hole with the
maximum size in that sample, rather than in the average
200
Polyfilm value, if we increase the volume of the sample, we actu-
180 ally increase the probability of having a bigger hole and
160 a longer free path.
140 It has been found [3, 28] that the probability of hav-
120 ing a hole with volume larger than Vh , is exponentially
100
distributed:
P (Vh ) = exp[−(Vh /Vh1 )] (29)
80
where Vh1 = 1 − e = 0.632, is the expected value. If we
−1
60
assume that hole has a spherical shape, the longest free
40 path w is proportional to the cubic root of the biggest
20 hole’s volume. Similarly, the probability of having a free
0 50 100 150 200 250 300 350 400 path larger than w , is:
Specimen Thickness d (µm)
P (w) = exp[−(w/w1 )3 ] (30)
Fig. 4. Breakdown Electric field strength in PET film sample versus where w1 = 1.443wm and P (wm ) = 0.5. The third power
Thickness d. in equation (30) is generally smaller because holes are
oblong rather than spherical, so we can use k instead of
3 with 1 < k ≤ 3.
The probability for an electron to overcome the energy If we choose a hole from a sample of n holes, taking
barrier, is given by the Frenkel formula [27]: into account equation (30), we derive the following rela-
tion about the characteristic largest value xn , of the free
Pµ = exp[−(Ig − eFl x)/(kT )] . (25) path:
Electrons are passing through a hole with an overall ve- xn = w1 (ln n)1/3 . (31)
locity (thermal and gained from the field) uo of the If we increase the sample, we take a larger value for the
p
order of 105 m/s in a time t = x/uo ≃ [3]Vh /uo free path:
were Vh the hole volume. Thus the hole is crossed by xN = xn + a ln n (32)
ne Vh /t ≃ 105 ne Vh /x ≃ 105 ne Vh 2/3 electrons per second. were a is a function of ln n [29]. The last relation clearly
This constant electron flow leads to a current density indicates that with the increase of the sample, the largest
i = ne euo (26) free path also increases.
If we assume that we have the area S of the electrode,
which is insignificant in the case of not very high applied the total volume of the specimen that is exposed under
fields, mainly due to the continuous reduction of the elec- electric strength is Vt = Sd . If there are n holes in
trons energy in every further step. At very high fields this volume, each one with a volume of Vh = cw13 the
(≃ 109 Vm−1 ) the energy gained from the field becomes total volume nVh of the holes, is a fraction θ of the Vt
large enough so the electrons have a large probability to (approximately 2.5 %). Thus we take:
overcome the barrier. Considering (Pµ = 1), we have: ncw13 = θSd (33)
Fl = Ig /ex . (27) and then
From equations (24) and (25), it is obvious that the break- xn = [(θSd ln n)/(nc)]1/3 . (34)
down voltage Fb is inversely proportional to the free path Substituting xn in equation (27), we have
x . Equation (27) is the criterion for breakdown initiation.
Ig
If rearranged as: Fb = . (35)
e[(θSd ln n)/(nc)]1/3
Fl x = Ig /e = Vµ (28) The latter relation can be expressed in a general form as
indicates that quantity Vµ depends on the polymer struc- Fb ∝ κ/(b + lnd)1/k . (36)
ture thus on the size of the holes in it [7] and generally (n ∝ d ) were κ and b are constants. The diagrams in
it is not constant. For a specific polymer, Vµ depends on Figs. 3 and 4, show results for six groups of specimens
the temperature, the relative motion between the macro- with different thickness. The error bars show the mean
molecules and the mechanical stress. error of the mean value of each group
The breakdown criterion shows that the longer the free q of twenty
Pn
specimens,
1
calculated by the formula σ n = n(n−1) i=1 (xi − x)2 .
path and the size of the hole, the lower the breakdown
voltage. Breakdown process initiates from that hole with Our percentage error values where between 2.1 minimum
the longest free path. The free path is actually the diam- and 3.9 maximum. Equation (36) is consistent with the
eter of the hole if we assume a spherical shape. The hole plot of our experimental results as it can be seen from
size is randomly distributed for a specific volume sample these figures.
 Journal of ELECTRICAL ENGINEERING VOL. 55, NO. 9-10, 2004 231

5 CONCLUSIONS [16] KAO, K. C.—XE. H. K.—TU, D. M. : J. Electrostatics 16

(1984), 115.
Electrical breakdown of polyethylene terephthalate [17] XIE, K. H.—KAO, K. C. : IEEE Trans. Electr. Insul. EI-20
(1986), 293.
(PET) films were measured under 50 Hz AC electric field.
[18] SUEDA, H.—KAO, K. C. : IEEE Trans. Electr. Insul. EI-17
Electrical degradation and breakdown is a two-stage pro- (1982), 221.
cess. Electrons are injected from the electrode to the poly- [19] BUDENSTEIN, P. P. : IEEE Trans. Electr. Insul. EI-15 (1980),
mer by Fowler-Nordheim tunnelling in micro-regions that 225.
are favorable to enhance the injection due to lowering the [20] ZAKREVSKIY, V. A.—SUDAR, N. T.—ZAOPO, A.—DUBI-
barrier profile. These electrons may cause impact ion- TSKY, Y. A. : J. Appl. Phys. 93 (2003), 2135.
ization of macromolecules transforming them to macro- [21] AKLONIS, J. J.—McKNIGHT, W. J. : Introduction to Polymer
ions. In the second stage, macro-ions are subjected to the Viscoelasticity, 1983.
[22] BAMJI, S. S.—BULISKI, A. T.—DENSLEY, R. J. : J. Appl.
electric field action and through a thermo-fluctuational
Phys. 61 (1987), 694.
mechanism, weakened carbon bonds are rupture causing
[23] ADAMEC, V.—CALDERWOOD, H. J. : J. Phys. D: Appl.
structural damage (degradation) in the polymer. After Phys. 8 (1975), 551.
the appropriate time, structural damage accumulation [24] PATSCH, R. : J. Phys. D: Appl. Phys. 23 (1990), 1497.
leads to the final breakdown. From the other hand intrin- [25] KOEHLER, M.—HUMMELGEN, I. A. : J. Appl. Phys. 87
sic breakdown initiation, in constant temperature, may (2000), 3074.
take place inside the hole with the longest free path, from [26] RAVINDRA, N. M.—ZHAO, J. : Smart. Mater. Struct. 1
an electron being accelerated by the electric field. The (1992), 197.
breakdown criterion indicates that dielectric strength is [27] FRENKEL, J. I. : Phys. Rev. 54 (1938), 647.
Kinetic Theory of Liquids, Dover, New York, 1955.
reciprocal to the longest free path and strongly related
[28] COHEN, M. H.—TURNBULL, D. : J. Chem. Phys. 31 (1959),
with the film thickness. The random nature of polymers 1164.
indicates that the longest free path and hence the largest [29] GUMBEL, E. J. : Statistical Theory of Extreme Values and
hole are more likely to exist in a thicker specimen that Some Practical Application, NBS, Washington, DC, 1954.
has a larger volume. Because the volume of insulation Statistics of Extremes, Columpia University Press, New York,
in reality is orders of magnitude larger than the speci- 1958.
men volume, it has to be noted that dielectric strength is [30] DING, H. Z.—XING, X. S.—ZHU, H. S. : J. Phys. D: Appl.
Phys. 27 (1994), 591.
much lower in this case. Another important fact is that
[31] FANJEAU, O.—MARY, D.—MALEC, D. : J. Phys. D: Appl.
in thicker films apart from bigger holes and voids , there Phys. 33 (2000), L61.
are many regions (especially in the boundaries between
amorphous and crystalline parts of the polymer) that can Received 27 April 2004
cause a local luck of homogeneity and hence a strength-
ening of the electric field. Konstantinos Theodosiou is a lecturer in the Depart-
ment of Financial and Management Engineering, University
of the Aegean, Chios, Greece. He was born in 1967 in Greece
References and has obtained an MPhys, an MSc in Electronic Engineering
and is working for a PhD in the field of electrical insulation.
[1] DISSADO, L. A.—FOTHERGILL, J. C. : Electrical Degrada- His research interests include degradation and breakdown of
tion and Breakdown in Polymers, Peregrinus, London, 1992. insulating materials, polymer physics, electromagnetics.
[2] TOBOLSKY, A. V. : Properties and Structure of Polymers, Ioannis Gialas is an Associate Professor in the Depart-
Wiley, New York, 1960. ment of Financial and Management Engineering, University
[3] BUECHE, F. : Physical Properties of Polymers, Interscience, of the Aegean, Chios, Greece. He was born in 1960 in Greece
New York, 1962. and has obtained an MPhys, an MSc in Physics and a PhD in
[4] KAO, K. C.—HWANG, W. : Electrical Transport in Solids, the field of Experimental High Energy Physics. His research
Pergamon, Oxford, 1981. interests are in the area of experimental high energy physics.
[5] ARTBAUER, J. : J. Polym. Sci. C 16 (1967), 447. Isidoros Vitellas is Director of the High Voltage Labo-
[6] ARTBAUER, J. : Kolloid Z. Z. Polym. 225 (1968), 23. ratory in DEH (Public Power Corporation) Athens, Greece.
[7] ARTBAUER, J. : J. Phys. D: Appl. Phys. 29 (1995), 446. He was born in 1954 in Greece and has obtained an Diploma
[8] KAO, K. C. : IEEE Trans. Electr. Insul. EI-11 (1976), 121. in Electrical Engineering and a PhD in the same field. His re-
[9] WATSON, D. B.—HEYES, W.—KAO, K. C.—CALDER- search interests include electrical power generation and trans-
WOOD, H. J. : IEEE Trans. Electr. Insul. EI-1 (1965), 30. fer, degradation and breakdown of insulating materials, elec-
[10] LIU, D.—KAO, K. C. : J. Appl. Phys. 69 (1991), 2489. tromagnetics.
[11] KAO, K. C. : Proc. Int. Conf. on Electrical Insulation (Canada), Dimosthenis Agoris is an Associate Professor in the De-
vol. 1, North-Holland/American Elsevier, Amsterdam, 1999, partment of Electrical Engineering and Computer Technol-
p. 517. ogy, University of Patras, Rio, Patras, Greece and Director
[12] LEBEY, T.—LAURENT, C. : J. Appl. Phys. 68 (1990), 275. of the High Voltage Laboratory in the same Department. He
[13] FOSTER, E. O. : J. Phys. D: Appl. Phys. 23 (1990), 1506. was born in 1952 in Greece and has obtained a Diploma in
[14] LEWIS, J. T. : J. Phys. D: Appl. Phys. 23 (1990), 1469. Electrical Engineering and a PhD in the same field. His re-
[15] DISSADO, A. L.—FOTHERGILL, C. J.—WISE, N.—WILL- search interests include electrostatics, electromagnetics, elec-
BY, A.—COOPER, J. : J. Phys. D: Appl. Phys. 33 (2000), trical power generation and transfer, degradation and break-
L109. down of insulating materials.