You are on page 1of 11

54 IEEE Transactions on Electrical Insulation Vol. 28 No.

1, February 1993

The Structural Characteristics of


Water Trees in Power Cables and
Laboratory Specimens

E. Moreau, C. Mayoux, C. Laurent


Laboratoire de Genie Electrique, CNRS,
UniversitC Paul Sabatier, Toulouse, France

and A. Boudet
Centre d’Elaboration de MatCriaux et d’Etudes Struc-
turales, LOE, CNRS, Toulouse,
France

ABSTRACT
The physical structure of water trees grown in power cables
and in laboratory specimens made of crosslinked polyethylene
are studied and compared. Detailed observations are made by
fluorescence microscopy. This technique, used for the first time
in this type of research provides images with good contrast and
high resolution. It is completed by 3D views acquired by confo-
cal laser scanning microscopy. In both cases t h e tree structure
is a network of continuous submicroscopic branched and zigzag
tracks. The previous published observations are critically ex-
amined. Moreover, t h e self-luminescence of water trees and its
neighborhood is studied, and is shown to be localized in t h e
tree tracks.

1. INTRODUCTION The tree-like figures appearing in the water-


impregnated polymer insulating a power cable are
named ‘water trees’. This phenomenon is particularly
insulation aging to be discussed in this paper con-
T HE
cerns power cables buried underground. Polymer in-
sulated power cables up to 400 kV are now in use. One of
associated with buried distribution cables. It presumably
leads to polymer breakdown and is due to the combined
effects of the electric field and moisture.
the objectives of the current research and development
is to increase the working stress by reducing harmful
imperfections inside the insulation and a t its interfaces. A survey of the literature on the subject of treeing in
Solid insulation is usually extruded over a semiconduct- solid dielectrics was presented in 1976 by Eichhorn [l].
ing shield on the conductor. Many different materials More recently a review discussed water tree growth in
are used, including polyethylene, butyl rubber, ethylene- polyethylene cables [2]. Water trees are diffuse structures
propylene copolymer, or ethylene-propylene-diene ter- resembling a bush or a fan. Two types of water trees
polymer. This study deals with crosslinked polyethylene have been recognized: the ‘bow-tie’ trees and the ‘vented’
(XLPE). trees, according to where these trees start growing. Bow-

0018-9367 $3.00 @ 1993 IEEE


6

IEEE !Ikansactions on Electrical Insulation Vol. 28 No. 1 , February 1003 55

tie trees are initiated in the bulk of the insulating polymer blue. However, because of light scattering, this meth-
and vented trees grow a t the insulation surfaces. od does not allow one to distinguish very small cavities
against the background.
A precise description of the morphology of water trees
is essential in order to understand the mechanisms of wa- Both in scanning and transmission electron microscopies,
tertreeing and the subsequent breakdown process. Al- it is difficult to differentiate, in the cross section of a wa-
though water trees are generally accepted as the ma- ter tree, a spherical cavity from a cylindrical channel. If
jor cause of failure of polyethylene insulated cables, the water trees are filament-like, it would be rather difficult
mechanism of water treeing is not agreed upon. More- to differentiate the cross sections. Indeed the section of a
over, it has been proved that there is a relationship be- cylinder structure reveals a n elongated shape if the sec-
tween the presence of water trees and the development tion is parallel to the cylinder axis and a circle if perpen-
of an electrical tree [l,3,4,5] but the exact nature of this dicular. Since the observation is done in a plane, it is not
relationship and the process leading to the inception of an easy to conclude on the original shape. Moreover, it has
electrical tree are not known. What is known is that an been proved that subsequent treatment with etchants and
electrical tree starts to propagate from a cavity or a vol- contrastants (permanganic, nitric, chromic, and chloro-
ume in which ionization can happen. Detailed knowledge sulfonic acids, osmium tetroxide, carbon tetrachloride)
of the morphology of water trees is then very important. which are applied t o these sections, creates many arti-
Therefore, we undertook a study to investigate the fine facts [ l o , 131.
morphology of the defects either in cables (bow-tie tree)
or in laboratory specimens (vented tree). Both specimens Most authors from their observations through LM [3-5,
are made of crosslinked polyethylene. 71, SEM [3,9-131 or TEM [13], deduced that water trees
are made of a distribution of small non-connected cavi-
2. LITERATURE SURVEY ON ties of variable sizes, ranging from l to -
3 pm. This
model has received wide acceptance and has constituted
WATER TREES
the base for models of water tree growth and calculation
of conductivities and diffusion [18,20,21].
is no general agreement on the morphology of
T HERE
water trees and two models are competing. In one,
water trees form continuous paths such as microchan-
In contrast to this model, some authors have observed
hollow microchannels in SEM [3,14] or TEM [8,14-171,
nels while in the other model the tree has a high den-
while Rose, de Bellet et al. [15-171 in TEM saw either
sity of microvoids that are not connected. These models
voids, or chains of voids, or pieces of channels. Ross and
were suggested by observations with different types of
Smit [8,19] described hydrophilic tracks.
microscopy: light microscopy (LM) [3-5,7,8], scanning
electron microscopy (SEM) [3,9-141, or transmission elec-
tron microscopy (TEM) [8,13-171. They were supported In this study, we used fluorescence light microscopy
by comparative measurements of electrical conductivities (FLM) after staining of water trees with a fluorescent
[18,19], electrical strength [7], diffusion of ions [7,19] or dye. Although this method could not allow the separa-
gases [18] made on treed and untreed regions. tion of two fluorescent points < 0.22 p m apart with the
best objectives available [22], it offers the possibility to
Many different observations have led to opposite mod- detect very small objects far below the resolution of the
els because the interpretation of images is particularly microscope if they are separated by > 0.22 pm. Thus,
critical. An image is the combined effect of an incident observation of isolated objects ranging from 30 to 5 nm
ray (light, electron beam), a mechanism of interaction in diameter is routinely achieved.
with the object, a mechanism of image formation involv-
ing optical characteristics of the microscope and the re- We also used epifluorescence light microscopy without
duction of a 3D object to a 2D image. Moreover in order staining to detect the natural luminescence of the water
to be observed, the object undergoes a specific prepa- trees, as already done by Ross et al. [23]. These authors
ration that often leads to artifacts. A detailed discus- could detect the luminescence of water trees and ascribed
sion about the validity of the observations should then it to oxidation of the polyethylene. They compared it to
include the influence of the microscope and of the speci- the results obtained by infrared spectroscopy. The analy-
men preparation. sis of this luminescence brought valuable information on
the chemical changes in the material, like double bonds
In light microscopy, resolution is limited by the low nat- or carbonyl and carboxyl groups formation. However, the
ural contrast of the water trees. Contrast is generally en- following results will show the advantages of staining to
hanced by staining the polymer cavities with methylene visualize the finest branches of water trees.
56 Moreau et al.: Characteristics of Water Trees in Power Cables

As the out-of-focus fluorescence impairs the contrast 3.2 DESCRIPTION OF


of the images, we also used a confocal laser scanning mi- MACROSCOPIC OBSERVATION
croscope (CLSM) in epifluorescence mode [24]. Recently TECH NlQU ES
other authors have used CLSM and have observed non-
connected cavities [25], but there was no attempt to ob-
Epifluorescence microscopy was operated with a stan-
tain enlarged images with a good resolution and no dis-
dard video enhanced light microscope, i.e. a microscope
cussion to justify their conclusion. We present here the
equipped with a video camera and a computer used to
first images in CLSM with the maximum resolution of the
digitize, record and process the images. A Zeiss Axio-
apparatus.
phot microscope was used in both bright field and epiflu-
orescence mode (excitation filter band pass: 546 ~ t nm;
6
emission filter low pass > 590 nm) equipped with high
3. EXPERIMENTAL numerical aperture (NA) objectives 4 0 x (NA: 0.90), 63x
(NA: 1.4), 1OOx (NA: 1.30). The images were recorded
E observed water trees grown in crosslinked poly-
with a Nocticon camera (Lhesa), digitized and processed
ethylene (XLPE) insulated cables and in laborato- with an image enhancing treatment (Quantel, image aver-
ry specimens with a needle test [26]. The details of both aging, histogram and stretch functions), carefully avoid-
specimens will be given in Sections 4 and 5. ing alteration of the information present in the primary
image.

3.1 SAMPLE PREPARATION The confocal laser scanning microscope used was a Zeiss
LSM. Schematically, the 543 n m wavelength laser source
is focused inside the specimen where the fluorescence is
emitted by a punctual source. Then fluorescence of this
Specimens containing water trees were prepared for ob-
source is collected on a photomultiplier, recorded in a
servation in epifluorescent light microscopy (FLM) and
computer, and the %dimensional image is reconstructed
in confocal laser scanning microscopy (CLSM) in fluo-
by scanning the specimen. This method improves the
rescent mode. They were first sectioned into thin slices.
contrast and allows the superposition of optical sections
Then water trees were contrasted by a fluorescent dye as
of 2 0.3 p m depth field each. The CLSM is equipped
explained below. with an emission filter low pass > 590 nm. The objec-
tives used were a 6 3 x , and a 1OOx the characteristics of
For observations of laboratory specimens in FLM and which are identical to those of the Axiophot microscope.
CLSM, 10 p m thick sections were produced by cryo-ultra-
microtomy performed at -1OO"C, so that the specimens 3.3 FLUORESCENCE MICROSCOPY
were rigid enough to be easily sectioned. In order to avoid EFFlClEN CY
the possibility that remaining water may fracture the ma-
terial during the subsequent freezing procedure, the spec- In order to understand the efficiency of the fluorescent
imens were first dried a t 60°C for 30 min to remove water. dye in revealing the structure of water trees, we carried
Sections were cut both along the axis of the needle and out some preliminary observations on a water tree grown
perpendicular to it. in a laboratory specimen (see aging test conditions in
Section 5). We collected three adjacent 50 p m sections
100 p m thick sections from cable specimens were ob- from the same water tree which underwent three differ-
tained by microtomy and subsequently observed in CLSM. ent treatments, and we compared the images obtained.
The first one (Figure l(a)) was obtained in bright field
Some sections were observed in FLM without staining light microscopy after staining the sample with methy-
in order to detect the natural luminescence of the water lene blue, following the staining procedure by Ashcraft
tree. All other sections were immersed in a 70 mg/ml [27]. The second section was stained by the rhodamine
solution of rhodamine, used as fluorescent dye, a t 60"C, treatment described above, then observed in bright field
for 2 4 h and then washed in water before observations. microscopy where rhodamine is visible as a dye (Fig-
The sections were then mounted between a microscope ure l(b)). Finally, the same rhodamine stained water tree
slide and cover glass in a non-fluorescing 50/50 solution of was observed in fluorescence microscopy (Figure l(c)).
glycerin and ethanol, in order t o achieve optimum optical
It is striking to notice that the fluorescence microscopy
performance.
technique (Figure l(c)) revealed a much larger altered
IEEE Transactions on Electrical Insulation Vol. 28 No. 1, February 1993 57

er than the one stained by methylene blue and it seems


20um a evident that the methylene blue has not revealed the to-
tal extension of the altered polyethylene. This should
be taken into account when measuring, for instance, the
rate of water tree growth. Another advantage of fluores-
cent microscopy is that it offers a much higher contrast
than bright field microscopy, so that a higher resolution
is achievable. Figure l ( c ) shows much more details of the
internal structure than the other two images. Further-
more, the CLSM provides precious information in three
dimensions.

4. OBSERVATION OF WATER
TREES IN CABLE INSULATION
4.1 SAMPLES

ATER trees were grown during accelerated aging of


cables . The insulation is a dry-cured chemically
crosslinked polyethylene (crosslinking agent 2% dicumyl
peroxyde containing 3000 ppm of antioxidant, Irganox
1081). The insulation thickness was 6 mm. Water was
injected into the cable along the conductor a t room tem-
perature. The polyethylene was then submitted to an
average field of 4 kV/mm a t 50 Hz for two months.

Figure 2.
General shape of a XLPE cable bow-tie tree. Rho-
damine staining. Confocal laser scanning mi-
Figure 1. croscopy (CLSM) in fluorescence mode.
Cross-section of a water tree. (a) Stained by
methylene blue. Bright field light microscopy. 4.2 LARGE SCALE OBSERVATION
(b) Stained by rhodamine. Bright field light mi-
croscopy. (c) Stained by rhodamine, and observed
in fluorescence microscopy. Confocal laser scanning microscope (CLSM) allowed
us to observe even thick sections in fluorescence mode.
area than the image in bright field with rhodamine (Fig- Among many other images, Figure 2 is an example of a
ure l ( b ) ) . The extension of the visible area is also lare- 1bow-tie tree located in the middle of the insulation. This
-
\ I , Y
58 Moreau et al.: Characteristics of Water Trees in Power Cables

Figure 3.
Part of a water tree showing the fluorescent tracks in a bundle. Images obtained in CLSM, (a) by the
addition of 10 0.5 pm thick sections, (b) by the addition of 20 sections. Spots A, B, C are isolated
in (a) and become connected to the main network in (b). The arrows point to quasi-circular figures.
Observation in CLSM by rhodamine staining.
image has been obtained by addition of 40 optical slices fan of branches. It must be pointed out that this is not
of 0.5 p m depth field each. So it represents the bow-tie a characteristic figure, but a casual arrangement due to
tree over a total depth of 20 pm. Its axis is radial, i.e. in the particular plane of view. The vertical view along the
the direction of the field. axis of the tree has shown other branches starting from
the bottom of this track close to the central spot.
The general tree shape is very similar to what is usually
observed using bright field light microscopy. The outline At this small scale, the tree appears filamentous, but
is very well defined and has the shape of a double fan. valuable conclusions can only be drawn from microscopic
The total axial length is 200 p m and the maximal width observations.
is 100 pm. In this particular image, the upper left side of
the tree, indicated by an arrow, does not seem to be con-
nected to the main tree. In fact, a change of observation
4.3 MICROSCOPIC SCALE
plane reveals the existence of a connection on another OBSERVATION
level. The two bundles constituting the tree seem to have
been initiated a t the central white spot which looks like
a cavity. From this spot, a broad and long fluorescent At higher magnification, the objects seen in CLSM mi-
track (- 37 p m long) goes down before spreading into a croscopy in fluorescence mode offer enough contrast to
IEEE Transactions on Electrical Insulation Vol. 28 No. 1, February 1993 59

reach the resolution limit which is 0.2 p m in our con- rhodamine, either because it is less concentrated or be-
ditions. We recorded several optical sections of 0.5 p m cause it is less extended in depth. The width of the faint
field depth each. Images of thicker portions of the water or bright portions of tracks varies from 0.2 to 1 pm. But
tree were reconstructed by the addition of several of these as discussed in Section 2, the real values may be much
sections. lower.

Figure 3(a) is an example of such an image obtained 5. OBSERVATION OF WATER


by the addition of 10 sections a t moderate magnification.
In Figure 3(b), the same image is completed by adding
TREES IN LABORATORY
10 more sections. We can see that the largest part of the SPECIMENS
bundles is constituted by a network of thin fluorescent
tracks where rhodamine has concentrated. Some spots
seem to be isolated (letters A, B, C, on Figure 3(a)). 5.1 SAMPLES
However, a 3-dimensional view of the network shows that
they are connected by paths going outside the visible vol- E carried out the growth of water trees in the same
ume. On Figure 3(b), the spots A, B, C are now seen
linked to the network. By inspection of the different op-
W crosslinked polyethylene as used for cables, though
some specimens were free of antioxidant. All the spec-
tical sections, it was evident that the fluorescent tracks imens were molded, as described below, and polymer-
go up and down, and link through the polyethylene. As ized, then the samples were heated to 70°C in vacuum for
the track grows and progresses through the polyethylene, 10 days in order to evacuate the residual by-products of
its orientation often changes and follows an erratic path. the crosslinking reactions and to release residual strains.
During the same time it gives rise to many branches giv- Then they were immersed in water a t room temperature
ing the appearance of a tree. The orientation mostly re- for 24 h. They were artificially aged by a water needle
mains within an angle of 45" from the principal axis, but experiment [26]; the water reservoir was obtained by com-
it can sometimes be larger, forming a sort of ring or, in pression molding at 180°C of polyethylene pellets around
three dimensions, of a helix (arrows on Figure 3(a)). a metallic needle. The liquid electrode was constituted
by a 0.1 M/1 NaCl solution (curvature radius 10 pm, gap
3 mm) [28]. An electrical stress of 7 kV a t 1.5 kHz was
applied. Three aging periods were used: 1, 20 and 200 h.
Water trees were dyed with rhodamine and observed by
epifluorescence microscopy.

Figure 4.
Detail of tracks in a XLPE cable water tree. Rho-
damine staining. CLSM, addition of five 0.5 pm
thick sections.
Figure 4 was recorded a t a higher magnification and
therefore provides details on the structure of the tracks.
It was obtained by the addition of 5 successive optical Figure 5.
sections so that the total depth of this image is 2.5 pm. General shape of a water tree, grown during 20 h
The tracks are made of large bright areas M 1 p m wide, in a laboratory XLPE, from a water needle. Rho-
connected by fainter tracks, indicating that there is less damine staining. Fluorescence Microscopy.
60 Moreau et al.: Characteristics of Wate;. n e e s in Power Cables

5.2 LARGE SCALE OBSERVAT ON the bundles. Figure 6 is the image obtained in CLSM
RESULTS of a 0.5 p m optical section. It presents the edge of one
bundle. We can see that a bundle is formed by a network
of erratic and entangled tracks. As in cables, most of
Water trees grow from the tip of the nee' le and from the isolated spots are linked to the principal network by
its side. An example of a water tree grown from the paths going through other planes of focus (see [28]).
tip after 20 h aging is shown in Figure 5. It consists
of several well oriented and well outlined bundles that
seem to follow the electric field lines. The general shape
resembles the well known images obtained by the usual
methylene blue dye through bright field light microscopy,
but here the contrast is improved. The appearance is
bushy. Other water trees on the side are more elongated
[28]. The axial length is 100 p m and the maximum width
is 125 pm. The bundles constituting the trees exhibit an
identical filamentary morphology.

The water trees after 200 h aging exhibit similar struc-


tural features, but they are longer.

Measurements of water tree lengths have been carried


out in the research laboratory of Atochem (Groupement
de Recherche de Lacq) directly on 26 specimens during Figure 7 .
their growth on the aging bench. The mean length after Enlargement of a track in a laboratory XLPE wa-
20 h aging was 75 pm along the needle axis, and 200 p m ter tree grown during 200 h. Rhodamine staining.
perpendicular to this axis. After 200 h aging, their mean Fluorescence Microscopy.
length was 230 p m along the axis, and 600 p m perpen-
dicular to the axis.

Figure 8.
Laboratory XLPE water tree grown during 1 h
Figure 6. at the electrolyte/polyet hylene interface. Rho-
Network of tracks forming a water tree grown damine staining. Fluorescence Microscopy.
during 200 h in a laboratory XLPE. Rhodamine
staining. CLSM, 0.5 pm thick optical section. In the close-up of another zone of the water tree shown
in Figure 7 we clea.rly observe the detailed shape of a
5.3 MICROSCOPIC SCALE track. Its width varies from 0.2 to 0.6 pm.

In order to understand how a water tree is initiated


Higher magnification on large water trees after 20 and on the surface of the electrode, we observed specimens
200 h aging leads t o the observation of the structure of aged for 1 h. An example is shown on Figure 8. On the
IEEE Transactions on Electrical Insulation Vol. 28 No. 1, February 1993 61

intensity: it is convenient to distinguish three main areas


-
left we can see the liquid electrode, and on the right the
polyethylene. The water tree length is 12 p m and was
between 10 to 20 p m on other images. The interface,
according to the intensity of luminescence: a high lumi-
nescent central area, its intensity arbitrarily labeled a t
where a water-tree has begun to grow, appears fluores-
cent, because of the accumulation of rhodamine on the
rugged surface. The tree is made of a network of entan-
mentary structures a t an --
100 units, a larger well outlined area composed of fila-
35 units, and a background
with a low luminescence of 25 units.
gled tracks and seems to have the same structure as the
tip of the larger trees. We compared the image of the water tree obtained from
its own luminescence (Figure 9) to the image of the same
We can notice that the network is linked to the elec- water tree obtained after staining by rhodamine (Fig-
trolyte surface by only one 0.2 p m wide track. That is the ure 5). Both images reveal the same extension of the
case with all very young water trees we have observed. water tree, indicating that the location of luminescent
substances is the same as the location of rhodamine. We
have now to discuss the relationship of these two phe-
6. WATER TREE nomena. However, the intensity of the luminescent signal
LUMINESCENT PROPERTIES with rhodamine is 15x higher, and the obtained images
display a much better contrast. With the rhodamine flu-
orescence, we could visualize the fine network of tracks
ATER trees were grown during 20 h in a crosslinked
W polyethylene laboratory specimen, and the speci-
men was sectioned into 10 p m thick slices which were
through the water tree.

observed in fluorescence light microscopy without stain- 7. DISCUSSION


ing. The polyethylene was to be free of antioxidant in
order to avoid the fluorescence from this additive. 7.1 TRACK NETWORK STRUCTURE

observations on water trees in cables and in labo-


T HE
ratory specimens a t medium magnification (Figures 3
and 6 ) and high magnification (Figures 4 and 7), show
similar structures. In both cases rhodamine makes visi-
-
ble 1 p m areas, connected by thinner pathways where
fluorescence is fainter. This is the evidence of a change
in property of this volume of polyethylene, which now al-
lows rhodamine to penetrate by diffusion to reveal tracks.
<
These tracks have a diameter from 1 pm to 0.2 pm, the
resolution limit of the microscope. The variations of the
intensity of the image of a track along its path is due ei-
ther to the variation of its depth or to the variation of
the density of rhodamine inside this track. The tracks do
not follow the orientation of the electric field, but have an
erratic path. Therefore, we think that water trees consist
of a network of entangled tracks constituting continuous
pathways. Although some spots remain isolated, it is pos-
sible that pathways going through upper or lower sections
could link them to the principal network.

Figure 9. These results contradict those of Shinozaki et al. [25],


Self-luminescence in the same laboratory XLPE for instance, who observed water trees in cables by confo-
water tree as that shown in Figure 5. Fluores- cal laser scanning microscopy in fluorescence mode. They
cence microscopy. concluded that the water tree consisted of non-connected
cavities. However their observations are not provided
Figure 9 shows a water tree sectioned along the nee- through high magnification views, but are only a general
dle axis. The luminescence of the water tree decreases overview a t moderate scale. Furthermore, the use of the
from the root of the water tree to its extremity. The computer contrast enhancement is complex. A miscon-
video equipment provides relative measurements of this trolled filter can erase significant signals. The high-pass
62 Moreau et al.: Characteristics of Water Trees in Power Cables

filter and the neighborhood filter by matrix convolution contains 0.25/1000 C=C bonds that explains the lumi-
may cancel the fainter signals, masking the thinner parts nescent background displayed in Figure 9. The fact that
of the tracks and making them discontinuous. Finally, water trees are more luminescent indicates that polyeth-
the three dimensional views of the water tree presented ylene has been chemically modified. This higher intensity
by Shinozaki et al. are incomplete as they recorded and proves that there are more C=C bonds or new unsatura-
added successive optical sections with 1 or 2 pm steps us- tions like C=O bonds due to a chemical modification of
ing a 40x objective with the field depth < 0.6 pm. Thus the polymer. This has been already pointed out by local
some information is lost a t each record, and in their case, infrared spectrography which revealed that the treed area
even a submicrometer track could appear as a succession comprises many oxidation products, such as carbonyl and
of cavities. carboxylate groups [23].

Another experiment noted to support the non- It has been suggested that oxidation would occur in
connected structure of water trees is the gas permeation a volume larger than the water tree [30]. However, our
measurement through a tree structure experiment observations show that the extension of the water tree
presented by Cross et al. [18], who could not detect any is the same by self-fluorescence or by rhodamine fluores-
helium gas diffusion through the treed polyethylene. cence. Thus oxidation is limited to the altered volume
They also calculated that a 1 p m diameter channel of visualized by rhodamine. This indicates that oxidation
200 p m length would have allowed detectable diffusion is one of the mechanisms active during the change of the
of helium gas through the treed polyethylene. So they structure, but not preceding it. To further confirm this
concluded that ‘clearly the dry water trees are not statement it would be necessary to know how rhodamine
composed of interconnected channels’. However, when diffuses into the water tree and is fixed in the deteriorated
the calculations are repeated using 0.1 p m channels (as region.
suggested by the smallest sizes of our tracks), the helium
flow rate is 1Ox lower than the detection threshold of
the gas detector used. 8. CONCLUSION

7.2 WATER TREE GROWTH BSERVATIONS with fluorescence microscopy after dif-
0 fusion of rhodamine into water trees has allowed us
to present a fine description of the structure of water trees
The comparison of laboratory specimens aged during grown in XLPE from cables and from laboratory speci-
1, 20 and 200 h provides us with information on water mens.
tree growth. Even though the extension of the water tree
logically increases with the aging time, the local structure
of the water tree remains the same. The track network in
the young water tree (Figure 8) is analogous to the edge -
In both cases, they appear as a network of entangled
tracks, with diameters varying from 1 p m to < 0.2 pm.
The tracks do not follow a precise direction, whereas on
of the oldest one (Figure 6). That means that the mech-
anisms of the growth of large water trees are operating a larger scale they constitute bundles with a well defined
already a t the beginning of the tree growth. When the outline.
aging stresses are applied for longer periods the tracks
progresses by zigzaging and branching, and increasing in These features have been found even for newly grown
lengths. The higher luminescence of the water tree core water trees and in laboratory specimens, suggesting that
with respect to the edge is due to large thickness of the the mechanism of deterioration is the same a t all growth
core as seen in single optical sections in CLSM. The vol- stages.
ume density of tracks seems to remain constant inside the
network during the growth. Moreover, we have observed self-luminescence of water
trees. It is believed that oxidation responsible for such lu-
7.3 LUMINESCENCE minescence is localized in the tracks described above, and
that oxidation does not precede the deterioration process.

Polyethylene is only weakly luminescent in this range We have shown that this new technique of microscopy,
of excitation wavelength. It is known that this weak lu- and the use of rhodamine as fluorescent dye, reveal a
minescence is due to unsaturation in the molecular back- larger extension of the treed polyethylene than the usual
bone [29] and to residual byproducts. Our polyethylene methylene blue stain, and offer a higher contrast.
IEEE !lYansactions on Electrical Insulation Vol. 28 No. 1 , February 1993 63

ACKNOWLEDGMENT [9] H. Orton, K. Abdolall, S. Bamli, A. Bulinski, J .


Densley and A. Garton, “An Investigation into the
Morphology of XLPE Cable Insulation”, Ann. Re-
confocal microscopy has been kindly performed by
T HE
Mr. Freiss and Mr. Gonadard, in the laboratory of
Dr. Dussourd d’Hinterland and Mrs. Pine1 (Pierre Fabre
port Conf. on Elec. Insul. and Dielec. Phenom., CEI-
DP, pp. 306-313, 1981.
Mkdicament) who are gratefully acknowled&d. w e also [lo] S. Bamji, A. Bulinski, J . Densley and A. Garton,
thank Mr. Saur for his help in preparing the specimens. “Etching and the Morphology of Crosslinked P E Ca-
The authors are indebted to Atochem and Cables Pirelli ble Insulation”, IEEE Trans. Elec. Insul., Vol. 18, pp.
for their financial support. 32-41, 1983.

[ll] S. Bamji, A. Bulinski, J . Densley, A. Garton and N.


REFERENCES Shimizu, “Properties of Water Treed and Non Treed
XLPE Cable Insulation”, Ann. Report Conf. on Elec.
[I] R. M. Eichhorn, “Treeing in Solid Extruded Elec- Insul. and Dielec. Phenom., CEIDP, pp. 141-147,
trical Insulation”, IEEE Trans. Elec. Insul., Vol. 12, 1984.
pp. 1-18, 1976.
[12] J. Muccigrosso and P. J. Phillips, “The Morphol-
[2] E. F. Steennis and F. H. Kreuger, “Water Treeing in ogy of Crosslinked Polyethylene Insulation”, IEEE
Polyethylene Cables”, IEEE Trans. Elec. Insul., Vol. Trans. Electr. Insu!., Vol. 13, pp. 172-178, 1978.
25, pp. 989-1028, 1990.
[13] C. W. Melton, D. Mangaraj and M. M. Epstein,
[3] W. Kalkner, V. Muller, E. Peschke, H. J . Henkel and “Morphology of Thermoplastic and Crosslinked Pol-
R. Von Olshausen, “Water Treeing in Polyethylene yethylene Cable Insulation”, Ann. Report Conf. on
and XLPE Insulated HV Cables”, Proceedings of the Elec. Insul. and Dielec. Phenom., CEIDP, pp. 299-
CIGRE Intern. Conf. on Large HV Electric Systems, 305, 1981.
Paris 21-07, 1982.
[14] S. M. Moody, V. A. A. Banks and A. S. Vaughan,
[4] J . C. Filippini, J . Y. Koo, Y. Poggi, C. Laurent, C. “A Preliminary Study of the Relationship between
Mayoux and S. Noel, “Water Trees and Electrical Matrix Morphology and Water Treeing in Medium
Trees”, Proceedings of the Intern. Conf. on Polym. Voltage Polymeric Insulated Cable”, Ann. Report
Insul. Power Cables, JICABLE, pp. 87-92, 1984. Conf. on Elec. Insul. and Dielec. Phenom., CEIDP,
pp. 352-360, 1991.
J . L. Chen, P. C. Filippini and Y. Poggi, “Distri-
bution of Microcavities and Prebreakdown in Water [15] L. J. Rose, V. Rose and J . J . De Bellet, “The Effect of
Treeing” , Intern. Conf. Properties and Applications Polymer Microstructure on Water Tree Performance
of Dielectric Materials, ICPADM, Xian, pp. 366-369, in Cross Linked Polyethylene for Cable Insulation”,
1985. Froceedings of the second Intern. Conf. on Cond. and
Breakd. in Solid Dielec., ICSD, Erlangen, pp. 237-
E. Moreau, A. Boudet, C. Mayoux and C. Laurent, 241, 1986.
“A Comparison between Water Tree Structure in
Laboratory Specimen and in Cable”, Proceedings of [16] J . J. De Bellet, G. Matey, L. J . Rose, V. Rose, J .
the third Intern. Conf. Properties and Applications C. Filippini, Y. Poggi and V. Raharimalala, “Some
of Dielectric Materials, ICPADM, Tokyo, pp. 232- Aspects of the Relationship between Water Treeing
235, 1991. Morphology and Microstructure of Polymers”, IEEE
Trans. Elec. Insul., Vol. 22, pp. 211-217, 1987.
K. Abdolall, H. E. Orton and M. W. Reynolds, “An
Investigation into Water Trees: Existence of Chan- [17] L. J . Rose, V. Rose and J. J. De Bellet, “Microscopic
nels and the Influence of Crosslinking” , Ann. Report Failure in Water Treeing and Fatigue as Revealed
Conf. on Elec. Insul. and Dielec. Phenom., CEIDP, by Transmission Electron Microscopy”, Proceedings
pp. 302-311, 1985. of the third Intern. Conf. on Polym. Insul. Power
Cables, JICABLE 87, pp. 283-289, 1987.
R. Ross and J . J . Smit, “The Chemical and Physical
Composition of Water Trees in XLPE”, Intern. Conf. [18] J. D. Cross and 3. Y. Koo, “Some Observations on
Properties and Applications of Dielectric Materials, the Structure of Water Trees”, IEEE Trans Electr.
ICPADM, pp. 156-159, 1991. Insul., Vol. 19, pp. 303-306, 1984.
84 Moreau et al.: Characteristics of Water Trees in Power Cables

[19] R. Ross and J. J . Smit, “Water Tree Growth Pro- [25] D. M. Shinoeaki, P. C. Cheng, S. Haridoss, R.
cesses in XLPE”, Proceedings third Intern. Conf. Mitchell and A. Fenster, “Three-dimensional Op-
Properties and Applications of Dielectric Materials, tical Microscopy of Water Trees in Polyethylene”,
ICPADM, pp. 214-217, 1991. Journal of Material Science, Vol. 26, pp. 6151-6160,
1991.
[20] J. Sletback and E. Ildstad, ‘(The Validity of the Me-
chanical Damage Theory of Water Treeing Tested [26] M. Saure and W. Kalkner, “On Water Tree Test-
against Experimental Result” , Conf. Record of IEEE ing of Materials”, Proceedings of the CIGRE Sym-
Internat. Symp. on Elect. Insul., Montreal, pp. 29- posium SO 5-87, Wien, 15/21-03, Section 6.2, No.
32, 1984. 620-10, 1987.

[27] A. C. Ashcraft and R. M. Eichhorn, “Method for Vi-


[21] A. Bulinski and R. J . Densley, “The Voltage Break- sualization of Water Trees by Staining”, IEEE Trans.
down Characteristics of Miniature XLPE Cables on Elec. Insul., Vol. 13, pp. 198-199, 1978.
Containing Water Trees”, IEEE Trans. on Elec.
Insul., Vol. 16, pp. 319-326, 1981. [28] E. Moreau, A. Boudet, C. Mayoux, C. Laurent and
M. Wright, “Fine Structure of Defects in Polyeth-
[22] S. Inoue, Videomicroscopy, Plenum Press, New- ylene used for Power Cable Insulation Observed by
york, 1986. Fluorescence Microscopy”, to be published in Jour-
nal of Material Science.
[23] R. ROSS,W. S.M. Geurts, J . J. Smit, J . H. Van Der
[29] L. Zlatkevich, Luminescence Techniques in Solid
Maas and E. T. G. Lutz, “Watertrees in Polyethyl-
State Polymer Research, Marcel Dekker, Inc. New
ene Studied by Luminescence and FTIR Microspec-
York, Basel, 1989.
trometry”, Proceedings of the third Intern. Conf. on
Cond. and Breakd. in Solid Dielec., ICSD, Trond- [30] A. Garton, J . H. Groeger and J. L. Henry, “Ionic
heim, pp. 512-516, 1989. Impurities in Crosslinked Polyethylene Cable Insu-
lation”, IEEE Trans Electr. Insul., Vol. 22, pp. 427-
[24] T. Wilson, and C. J. R. Sheppard, Theory a.nd 434, 1990.
Practice of Scanning Optical Microscopy, Academic Manuscript was received on 19 February 1992, in revised form
Press, New-York, 1984. 2 November 1992.

You might also like