E ec o annealing on conduc i i y in XLPE mid- ol age
cable insula ion
F. F u os
a
, M. Acedo
a
, M. Muda a
b
, J. Belana
b,
,J.O
` i
b
, J.A. Diego
b
,
J.C. Can
˜adas
b
, J. Sella e
`s
b
a
Dp o. de Fı
´sica Aplicada I, ETSII, Uni e sidad de Se illa, A da. Reina Me cedes s/n, 41012 Se illa, Spain
b
Dep . de Fı
´sica i Enginye ia Nuclea , ETSEIAT, Uni e si a Poli e
`cnica de Ca alunya, c. Colom 11, 08222 Te assa, Spain
1. In oduc ion
Polye hylene (PE) is a non-pola polyme . I is ound in
semi-c ys alline o m a oom empe a u e. The c ys alline
ac ion is o med when he ma e ial is cooled down
om he mel . The lamellae s uc u e o med du ing
solidifica ion depends on he cha ac e is ics o he he mal
ea men and he condi ions o c ys alliza ion. In he
usual con o ma ion, lamellae g ow adially, esul ing in
sphe uli es, which a e composed o bo h c ys alline and
amo phous egions.
Due o he non-pola cha ac e o PE, conduc ion
p ocesses should be associa ed wi h he p esence o ee
cha ge ca ie s. Ne e heless, pola con ibu ion may a ise
om he exis ence o ca bonyl g oups CQO ha a e
p oduced by oxida ion o PE, al hough he e may also exis
pola impu i ies.
In indus ial applica ions, PE is commonly used as an
elec ical insula o . Specifically, i cons i u es he base
compound o cable insula ion manu ac u ing due o i s
ou s anding insula ing cha ac e is ics and i s low p ice. In
he s anda d configu a ion o mid- ol age (MV) cables (up
o 25 kV se ice ol age), he PE insula ion is co e ed by
wo coaxial cylind ical semiconduc ing (SC) sc eens which
a e bo h ex uded simul aneously du ing ab ica ion,
esul ing in a concen ic h ee laye sys em: SC–insula-
ion–SC. The composi ion o hese sc eens ( ipically a
blend o PE, e hylene inyl ace a e and ca bon black)
esul s in a low- esis i i y ma e ial ha p e en s discha ges
ARTICLE IN PRESS

Co esponding au ho .
E-mail add ess: [email p o ec ed] (J. Belana).
on he su ace o he insula ion ha may damage i . In
addi ion, he PE insula ion is subjec ed o a c osslinking
p ocess so ha i finally esul s in c osslinked PE (XLPE).
This p ocess imp o es i s p ope ies; pa icula ly i
enhances i s chemical and mechanical esis ance, allowing
a highe se ice empe a u e.
I is well known ha PE conduc i e p ope ies a e
condi ioned by i s mo phology [1–8] and ha insula ion
deg ada ion can be ela ed o olume space cha ges [9].
Pa icula ly, conduc ion and space cha ge o ma ion in
low-densi y PE we e s udied including he in e al o
empe a u es o in e es o cable diagnos ics [10]. In he
las ew yea s, he e has also been a g owing in e es in he
s udy o insula ion p ope ies unde he applica ion o DC
ol age, he co esponding o ma ion o space cha ge
dis ibu ions and i s ela ion o he emaining li e ime o
he cable [11–13]. Al hough i is gene ally conside ed ha
ca ie s esponsible o PE conduc ion a e basically
elec ons [13,14], ions coming om addi i es and c oss-
linking by-p oduc s should be conside ed also.
Cu en e sus ol age (I–V) cha ac e is ics s udied in
LDPE by S e e [15] a 42, 82 and 110 1C, indica e ha o
low elec ic fields (E) an ohmic beha iou is obse ed; o
highe elec ic fields, conduc ion can be in e p e ed on
he basis o a quad a ic law in Ewhich depends on
empe a u e and e eals he exis ence o aps. Finally, o
highe elec ic fields, a hi d egion can be ound whe e
cu en ollows Child’s law wi h ull aps, and conduc ion
can be in e p e ed om gene al c i e ia o he heo y o
space-cha ge-limi ed cu en s (SCLC) [16]. Mo e ecen ly,
se e al au ho s ha e also obse ed ha cu en – ol age
cha ac e is ics in PE can be explained by SCLC model
[17,18].Pe
´lissou e al. [17] s udied he isoch onal cu en-
– ol age cha ac e is ics in LDPE o a wide empe a u e
ange below and abo e i s mel ing poin . They concluded
ha a low a e age elec ical fields he beha iou is
ohmic, bu be ween 5 and 50 MV/m i becomes highly
non-linea , ollowing JpV
n
wi h n42. This esul
sugges s ha an exponen ial dis ibu ion o aps may be
p esen [18]. In ano he s udy o I–Vcha ac e is ics,
Mizu ani shows ha cu en is limi ed by he elec ode,
ollowing a Richa dson–Scho ky law o high elec ic
fields (E44105V=cm) [19].
Na h e al. [20] apply he band heo y o LDPE by
de eloping a ma hema ical model based on he hopping o
ca ie s which a e injec ed in o amo phous egions. They
conside a SCLC and a p ocess o cha ge- apping a
amo phous-c ys alline bounda ies. In addi ion, hey sup-
pose ha he densi y o apping cen es is high enough so
ha he in e ac ion be ween hem esul s in an e ec i e
lowe ing o he ap dep h (Poole–F enkel e ec ). In his
la e model, he ‘‘dis ance be ween aps’’ pa ame e is
in oduced and i has a cons an alue. E en mo e ecen ly,
ollowing he same esea ch heme, i is concluded ha
elec onic anspo is bound o he mally ac i a ed
hopping, which is assis ed by elec ic field wi h a e y
low ac i a ion ene gy [21]. F om cha ge dis ibu ion
s udies and, specifically, by applying he elec oacous ic
pulse echnique (PEA), he p esence o pe iodical cha ge-
packe s be ween elec odes can be de ec ed o high elec ic
fields [22].
Re e ing o he na u e and dep h o aps, i has been
p oposed ha impu i ies and/o chain de ec s could be
esponsible o hei exis ence in he ma e ial. Ieda [23] has
pe o med a de ailed s udy on PE inqui ing in o bo h
heo ies.
The mally s imula ed depola iza ion cu en echnique
(TSDC) has been applied o XLPE cable elec e s [24,25].
The spec um ob ained (in ensi y o he depola iza ion
cu en e sus empe a u e) is complex and i is o med by
se e al peaks associa ed wi h he elaxa ion p ocesses o
he conduc ion mechanisms ha a e p esen in he
ma e ial. A b oad cu en peak can be obse ed a
105 1C, which has been associa ed wi h cha ge in he
c ys alline olume [24]. The combina ion o TSDC and
in a ed spec oscopy (IR) s udies [25] p o ed ha anneal-
ing p ocedu es in MV cables lead o he di usion o
componen s om he SC sc eens in o he XLPE insula ion
bulk. Mo eo e , he de ec s associa ed wi h di used
pa icles ac as apping cen es o injec ed cha ge om
elec odes, by ioniza ion. This p ocess is clea ly p ominen
a empe a u es abo e he c i ical empe a u e, T
c
,
app oxima ely equal o 80 1C, because TSDC cu en
e e ses i s pola i y ( om he e opola o homopola ).
Fu he annealing esul s in new cu en pola i y in e sion,
and he cu en egains i s he e opola cha ac e [24]. These
changes a e in ag eemen wi h he ac ha di used
componen s con inue o pene a e he insula ion olume.
In addi ion, he p og essi e inc ease o he he e opola
TSDC cu e a ea indica es a he mally ac i a ed gene a-
ion o ions om an in insic sou ce (ions esul ing om
dissocia ion o by-p oduc s and/o o he species p esen in
he insula ion).
The aim o his pape is o show he di e ences in
beha iou in conduc i e p ocesses ha ake place in XLPE
insula ion o wo di e en cases: (i) when he insula ion
e ains i s o iginal SC sc eens and, (ii) when he o iginal
sc eens a e eplaced by me allic elec odes. In his way, we
e alua e, by using complemen a y ime/ equency domain
echniques, he incidence o componen di usion om SC
sc eens owa ds he olume o insula ion. The empe a u es
used in his s udy a e close o hose o a ypical powe
dis ibu ion cable in se ice condi ions. The esul s
ob ained may be use ul o de eloping a cable li e ime
model on he basis o he moelec ic b eakdown.
2. Expe imen al
2.1. Abso p ion/ eso p ion cu en s
Cable was supplied by Gene al Cable SA (se ice ol age
up o 25 kV, 4.5 mm insula ion hickness, 1 mm SC sc eens
hickness). Samples we e cable sec ions o 7 cm leng h. In
he case o XLPE cylinde s, inne and ou e SC sc eens
ARTICLE IN PRESS
we e emo ed by using a mechanical la he and hey we e
eplaced by adap ed coppe elec odes o an op imal
fi ing o he inne and ou e su aces o he XLPE
insula ion. This p ocedu e allowed us o ca y ou
measu emen s o he elec ical p ope ies o he XLPE
insula ion, a oiding he e ec s o he SC sc eens. In he
case o cable samples only he ou e SC sc een was pa ially
emo ed, so ha a cen ed ing o 2 cm wid h was le . This
ing and he inne sc een we e used as elec odes. I is well
known ha eliable measu ing o high impedance insula-
ion equi es e y ca e ul gua ding and shielding o he
measu ed objec . This is o pa icula impo ance when
measu ing small samples [26] o sho cables [27]. One
me allic ing on each side o he ou e elec ode, 2 mm aside
o he la e , we e used as gua d elec ode in o de o
a oid he e ec o su ace conduc i i y and o a oid he
dispe sion o he field lines. The compa ison o he esul s
ob ained om bo h kinds o sample allowed us o s udy he
e ec o he SC sc eens du ing he ea men s.
Conduc i i y (s) was de e mined by means o equa ion
[28,29]
sð Þ 0
C0UðIað ÞþI ð ÞÞ, (1)
whe e C
0
¼4.4 pF is he geome ical capaci ance o he
sample (de e mined by a Hewle -Packa d impedance
analyse model HP-4192a LF), I
a
( )andI
( ) a e he
abso p ion and eso p ion cu en s (ARC), espec i ely, U
is he ol age applied o each cable sample and e
0
is
he acuum pe mi i i y. The ARC me hod p o ides a
con enien way o ob ain sbecause measu emen s equi e
less ime o be pe o med han wi h he usual cu en – ol-
age cha ac e is ics me hod since he e is no need o each a
s a iona y cu en .
P e iously o any measu emen , he su ace o he
sample was cleaned wi h e hanol in o de o a oid he
e ec o emains ha may come om he mechaniza ion
p ocess. The MV cable samples and XLPE cylinde s o be
measu ed we e placed inside a Ca boly e ype PF60 o en
(Fa aday cage). The o en was u nished wi h sui able
connec ions o elec ical measu emen s, using low-noise
coaxial connec o s and cables, sui able o high ol age.
Cable and connec o junc ions we e sil e solde ed. XLPE
cylinde s and cable samples we e placed in a measu ing cell
inside he o en desc ibed abo e. A K- ype he mocouple
(Kei hley 6517-TP) placed inside he insula ion o ano he
iden ical sample, which was loca ed close o he sample
unde es , was used o measu e he sample empe a u e.
The DC ol age sou ce o a Kei hley 6517 A elec ome e
was used o pola ize he samples. To measu e he
abso p ion cu en a ol age o 1 kV (co esponding o a
mean field o 0.22 MV/m) was applied o he inne
elec ode o he samples and he ou e elec ode was
g ounded h ough he a o emen ioned elec ome e . A e
a pe iod o ime, he inne elec ode was swi ched o
g ound and he eso p ion cu en was eco ded. The
cha ging pe iod (abso p ion cu en ) was abou he double
o he discha ging pe iod ( eso p ion cu en ), 2000 and
1000 s espec i ely, so ha Eq. (1) can be used o
conduc i i y calcula ions as a good app oxima ion [30].
The backg ound noise was lowe han 20 A in all
measu emen s. The en i e se up o measu ing ins umen
and sc eened o en was also placed inside ano he Fa aday
cage in o de o p e en measu emen fluc ua ions due o
ex e nal pe u ba ions. A ske ch o he se up can be seen in
Fig. 1.
A new sample was used in each se o ARC measu e-
men s in o de o a oid he e ec s o he p e ious
measu emen s. Each sample was subjec ed o annealing
a a empe a u e T
a
(50 1CoT
a
o100 1C) o a ime
a
.
Se e al annealing imes we e used o each empe a u e.
The longes annealing imes used a each empe a u e we e
de e mined by he kine ics o he changes obse ed in he
conduc i i y p ope ies and hei o de o magni ude was
ARTICLE IN PRESS
(1) (3)
(11)
(2)
(5)
(4)
(6)
(7)
(9)
(10)
(8)
K6517A
CONTACTORS
KK
I( )
U
S1
S2
U
discha ging
cha ging
Iabs
Ic
c
I es
(a)
(b)
Fig. 1. (a) Expe imen al se up o abso p ion/ eso p ion cu en measu e-
men (block diag am): (1) pe sonal compu e (Pen ium IV p ocesso ); (2)
GPIB cables; (3) Kei hley 6517A elec ome e ; (4) DC sou ce ou pu o
Kei hley 6517A; (5) box o con ac o s; (6) con ol o con ac o s by digi al
ou pu o Kei hley 6517A; (7) high ol age ou pu om con ac o s; (8)
ca boli e PF60 o en; (9) coaxial measu ing cable; (10) sample; (11)
Fa aday cage (2 m 1m1 m). (b) Simplified ske ch o expe imen al
se up o abso p ion/ eso p ion cu en measu emen and example o
measu emen : cha ging pe iod, 0o o c(S1 ON, S2 OFF) and discha ging
pe iod, 4 c(S1 OFF, S2 ON); I
abs
, abso p ion cu en ; I
es
, eso p ion
cu en ; I
c,
conduc ion cu en ;
c,
cha ging ime; K, elec ome e ; U,
cha ging ol age.
ipically 10
2
–10
3
h. Using he p ocedu e desc ibed in
Fig. 1(b) o ob ain he ARC alues and Eq. (1) o calcula e
he conduc i i y, conduc i i y measu emen s we e pe -
o med o se e al annealing imes, in o de o de e mine
he e olu ion o cable conduc ion up o an e en ual
s abiliza ion. Ne e heless, a o al s abiliza ion was ne e
obse ed, al hough, in i s place, we could confi m he
exis ence o oscilla ions a ound a quasi-s a iona y alue.
2.2. Dynamic elec ical analysis
F equency domain cha ac e iza ion was pe o med ia
dynamic elec ic analysis (DEA), using a dielec ic spec o-
me e BDS40 wi h a No o he m empe a u e con ol
sys em manu ac u ed by No ocon ol, in condi ions o
iso he mal annealing. The samples we e 150 mm- hick
XLPE films o 2 cm diame e , which we e cu om a
MVC using a la he wi h a ool designed o his pu pose, so
ha a 2 cm wid h ibbon could be ob ained. We pe o med
equency sweeps be ween 0.01 Hz and 1 MHz o di e en
annealing pe iods o ime anging om 0 o 72 h. Real and
imagina y pa s o he conduc i i y we e eco ded as a
unc ion o he equency o each annealing ime and
annealing empe a u e alue in o de o ca y ou he
discussion o he esul s.
3. Resul s and discussion
3.1. Time domain measu emen s (abso p ion/ eso p ion
cu en s)
Expe imen al esul s ha e been g ouped by empe a u e
o measu emen and annealing, i.e., whe he hese we e
ca ied ou a empe a u es T
a
below o abo e he c i ical
empe a u e, Tc¼80 1C. We ha e a anged expe imen al
da a his way in o de o emphasize he no iceable obse ed
di e ences in he e olu ion o conduc i i y e sus anneal-
ing ime [31].
3.1.1. Annealing below Tc
Fo bo h cable samples and XLPE cylinde s (samples
wi hou SC sc eens), he gene al beha iou o conduc i i y
e sus annealing ime consis ed in a con inuous dec ease.
This dec easing end was associa ed wi h di e en le els
o conduc i i y oscilla ions un il eaching an appa en
quasi-s a iona y s a e a e a pe iod o se e al days.
Fluc ua ions in conduc i i y con inued a leas o ou
o al pe iod o annealing ime (
a
41 mon h, in he case
o Ta¼50 1C, o a XLPE cylinde , as i is shown in
Fig. 2), leading o a quasi-s a iona y s a e wi h fluc ua ions
a abou he 4 h day o measu emen s. These fluc ua ions
a e mo e no iceable in he case o XLPE cylinde wi h
coppe elec odes, especially a low empe a u es (Fig. 2).
They could be a ibu ed o p oblems a he elec ical
con ac s. E en conside ing he sligh a ia ions in he
na u e and geome y o ou samples, i is nea ly impossible
o achie e a pe ec fi ing be ween coppe elec odes and
XLPE. This ac p obably leads o cha ge injec ion (co ona
e ec s, pa ial discha ges, e c.) ha occasionally may cause
sha p fluc ua ions in he ins an aneous alues o ou
measu emen s.
In Fig. 3, also o Ta¼50 1C, conduc i i y measu e-
men s o a cable sample (cable sample wi h SC sc eens) a e
shown. Due o he clea homogenei y in he dec easing
conduc i i y end, he measu emen pe iod was es ic ed
o less han 4 days. Only du ing he e y fi s hou s o
measu emen could conduc i i y fluc ua ions be de ec ed.
The a ia ions o conduc i i y in bo h ypes o samples
(cable samples and XLPE cylinde s) o annealing em-
pe a u es o 60 and 70 1C a e depic ed in Figs. 4 and 5.
Thei beha iou is no e y di e en om he conduc i i y
e olu ions desc ibed o 50 1C. In conclusion, o TaoTc,
ARTICLE IN PRESS
Fig. 2. E olu ion o conduc i i y wi h annealing ime o a XLPE cylinde
(cable sample wi hou semiconduc ing sc eens) a 50 1C in iso he mal
condi ions, co esponding o cable C5.
Fig. 3. E olu ion o conduc i i y wi h annealing ime o a cable sample
(‘‘comple e’’ cable sample wi h semiconduc ing sc eens) a 50 1Cin
iso he mal condi ions, co esponding o cable C5.
we de e mine ha he measu ed conduc i i y when using
SC sc eens (cable samples) is clea ly smalle han
conduc i i y ob ained when using coppe elec odes
(XLPE cylinde s). I can be no ed ha he highe he
annealing empe a u e, he lowe he ela i e a ia ion o
he conduc i i y o he ma e ial du ing he annealing
p ocess.
3.1.2. Annealing abo e Tc
In Figs. 6–8, we depic he a ia ions in conduc i i y (s)
as a unc ion o he annealing ime o empe a u es
anging om 80 o 100 1C. Ini ially, o bo h sample ypes,
s¼ (
a
) dec eases quickly un il a minimum alue is
measu ed, and he ea e , sbegins o inc ease. F om hen
on, conduc i i y e ol es e y di e en ly o he cable
sample wi h SC elec odes and o he XLPE cylinde wi h
coppe elec odes.
In he case o coppe elec odes (XLPE cylinde s),
conduc i i y inc eases up o a maximum alue, and hen
a mono onically dec easing end is obse ed ha is
accompanied by di e en le els o fluc ua ions un il a
quasi-s a iona y alue is a ained. The highe he annealing
empe a u e, he sho e he ime equi ed o eaching his
maximum alue o conduc i i y. As an example, a 80 1C, i
can be obse ed ha his ime is se e al days and
fluc ua ions a e e y significan , whe eas a 100 1C, his
ime is only o a ew days and oscilla ions a e almos
impe cep ible.
In he case o SC elec odes (cable samples), a mono onic
inc ease in conduc i i y eaches a quasi-s a iona y max-
imum final alue. Highe he annealing empe a u e, he
highe is he ela i e inc emen o conduc i i y in
compa ison wi h i s ini ial alue. This pe cen age inc e-
men is app oxima ely 450% o T
a
¼100 1C(Fig. 8).
ARTICLE IN PRESS
Fig. 4. E olu ion o conduc i i y wi h annealing ime o a cable sample
(&) and a XLPE cylinde (K)a 601C in iso he mal condi ions.
Fig. 5. E olu ion o conduc i i y wi h annealing ime o a cable sample
(&) and a XLPE cylinde (K)a 701C in iso he mal condi ions.
Fig. 6. E olu ion o conduc i i y wi h annealing ime o a cable sample
(&) and a XLPE cylinde (K)a 801C in iso he mal condi ions.
Fig. 7. E olu ion o conduc i i y wi h annealing ime o a cable sample
(&) and a XLPE cylinde (K)a 901C in iso he mal condi ions.

3.2. F equency domain measu emen s (DEA)
Two sepa a e beha iou s could be clea ly dis inguished
o he equency and annealing empe a u e anges we
used du ing he conduc i i y measu emen s. Some ep e-
sen a i e esul s a e shown in Figs. 9–11.
Fo equencies highe han 1 Hz, we obse ed a
educ ion in he eal pa o complex conduc i i y e sus
annealing ime o all annealing empe a u es anging om
80 o 105 1C.
Fo equencies lowe han 1 Hz, wo di e en beha-
iou s we e obse ed. Fi s , o empe a u es below 95 1C,
he eal pa o he complex conduc i i y dec eases
mono onically wi h annealing ime, and his e olu ion
becomes as e as empe a u e g ows. Fu he mo e, he
ange o a ia ion o he conduc i i y alues is also smalle .
Secondly, o empe a u es abo e 95 1C, he eal pa o he
complex conduc i i y ini ially inc eases ( o example, a
Ta¼100 1C, i inc eases o app oxima ely 12 h), bu
he ea e dec eases o p e ious alues. Rising and
dec easing pe iods become sho e as empe a u e g ows.
3.3. Discussion
In o de o p ope ly in e p e he esul s, i is e y
impo an o ake in o accoun ha DEA measu emen s
a e pe o med only in XLPE samples, bu no di ec ly in
sec ions o cable. These samples we e cu om a ibbon
ha had been p e iously ob ained om he insula ion o a
MV cable, as desc ibed in Sec ion 2.1. The e o e, hese
XLPE DEA samples do no ha e SC sc eens as elec odes
on hei su aces. Fo his eason, hei esul s should only
be compa ed wi h hose co esponding o XLPE cylinde s
(cable samples wi hou SC sc eens) measu ed by using he
ARC echnique.
Secondly, we can confi m ha e ec i ely, ARC and
DEA esul s on XLPE samples a e consis en based on he
conduc i i y ends wi h inc easing annealing imes, bo h
o low and high empe a u es. In he case o low
empe a u es, conduc i i y always dec eases wi h annealing
ime and pa icula ly, e e ing o DEA measu emen s, his
educ ion is obse ed o he eal pa o complex
conduc i i y along he en i e equency ange ha was
examined (Fig. 9). Fo highe annealing empe a u es, we
ha e also ound ha expe imen al esul s ob ained by
using bo h echniques a e compa able. A low equencies
(Figs. 10 and 11), he eal pa o conduc i i y ini ially
inc eases wi h annealing ime bu a e wa ds, i dec eases
again. Pa icula ly, o Ta¼100 1C, an abso p ion peak a
0.03 Hz could be p esen (Fig. 11). Ne e heless, o highe
equencies, he e is a con inuous educ ion o he eal pa
o complex conduc i i y, which p oduces a c osso e
equency o pa ame ic isoch one cu es (cu es ob ained
o di e en annealing imes). This c osso e equency
inc eases wi h T
a
.
A global explana ion o he di e en beha iou s and
a ia ions o conduc i i y ha ha e been desc ibed in
p e ious sec ions can be es ablished on he basis o he
exis ence o wo p e ailing ypes o conduc ion mechan-
isms in he ma e ial. Bo h mechanisms a e associa ed wi h
elec onic ca ie s. On he one hand, he e is a band
conduc ion mechanism, sligh ly dependen on empe a u e
(in ou ange o empe a u es), bu which is e ficien o e
he en i e s udied domain o equencies and annealing
measu ing imes. On he o he hand, he e is a ‘‘hopping’’
conduc ion mechanism [32] be ween localized s a es, in he
mobili y gap, which is bo h empe a u e and elec ic field
dependen and which is con olled by a Poole–F enkel
mechanism, as we will see sho ly. This hopping p ocess is
only e ficien om a c i ical empe a u e alue, due o he
ARTICLE IN PRESS
10-2 10-1 100101102103
1x10-11
1x10-12
1x10-13
1x10-14
1x10-15
1x10-16
1x10-17
σ′ (S/cm)
e
q
uenc
y
(
Hz
)
Fig. 9. Va ia ions o he eal pa o complex conduc i i y e sus
equency by using isoch one pa ame ic cu es ob ained by DEA
echnique, o an annealing empe a u e o Ta¼80 1C: (’)
annealing
¼0h; (K)
annealing
¼1h; (m)
annealing
¼3h; (.)
annealing
¼12 h; (E)
annealing
¼24 h; (b)
annealing
¼36 h; (c)
annealing
¼72 h.
Fig. 8. E olu ion o conduc i i y wi h annealing ime o a cable sample
(&) and a XLPE cylinde (K) a 100 1C in iso he mal condi ions.
inc emen o de ec concen a ion associa ed wi h di used
componen s om he SC sc eens [25]. Dissocia ion o
ypical species p esen in a MV cable insula ion, such as
c osslinking by-p oduc s, may also play a ole in his
p ocess.
Hopping conduc ion is g ea ly enhanced by he di usion
o di e en componen s om SC sc eens o XLPE
insula ion which implies ha he e a e wo de e mining
condi ions o hopping conduc ion o be a p e ailing
mechanism: (a) he exis ence o SC sc eens and (b)
empe a u es abo e T
c
(i.e. se ice empe a u es o powe
dis ibu ion MV cables) o inc ease he di usion o he
e e enced componen s [24,33].
3.3.1. Tempe a u es below Tc
The assump ions made abo e can explain he di e ences
in he beha iou o conduc i i y be ween XLPE cylinde s
and cable samples as de ec ed by ARC measu emen s.
ARTICLE IN PRESS
103
102
101
100
10-1
10-2
1x10-12
1x10-13
1x10-14
1x10-15
1x10-16
σ′ (S/cm)
equency (Hz)
Fig. 10. Va ia ions o he eal pa o complex conduc i i y e sus equency by using isoch one pa ame ic cu es ob ained by DEA echnique, o an
annealing empe a u e o T
a
¼95 1C: (’)
annealing
¼0h; (K)
annealing
¼1h; (m)
annealing
¼3h; (.)
annealing
¼12 h; (E)
annealing
¼24 h; (b)
annealing
¼36 h; (c)
annealing
¼72 h.
10-2 10-1 100101102103
1x10-12
1x10-13
1x10-14
1x10-15
σ ′ (S/m)
e
q
uenc
y
(
Hz
)
Fig. 11. Va ia ions o he eal pa o complex conduc i i y e sus equency by using isoch one pa ame ic cu es ob ained by DEA echnique, o an
annealing empe a u e o Ta¼100 1C: (’)
annealing
¼0h; (K)
annealing
¼1h; (m)
annealing
¼3h; (.)
annealing
¼12 h; (E)
annealing
¼24 h; (b)
annealing
¼36 h; (c)
annealing
¼72 h.
Below T
c
, elec onic band conduc ion is he p e ailing
mechanism, as di usion a hese empe a u es can be
neglec ed du ing annealing p ocedu es [23]. To explain he
di e ences o conduc i i y obse ed be ween he wo kinds
o samples, we assume ha bo h SC elec odes (cable
sample) and coppe elec odes (XLPE cylinde s) ha e a
blocking cha ac e and hey limi elec ic cu en . This
e ec is e en s onge in he case o SC elec odes (Figs. 4
and 5). Fo bo h ypes o elec odes, hei blocking
beha iou could be associa ed wi h he p esence o
componen s di used om SC sc eens du ing he manu-
ac u ing p ocess. These componen s ac as apping
cen es, and apped cha ge limi s he flux o elec ic
cu en . In he case o coppe elec odes (XLPE cylinde s),
mechanical p ocedu es de eloped o elimina e ou e and
inne SC sc eens also emo e he mos ex e nal laye s o
he insula ing XLPE ma e ial—whe e ap concen a ions
a e highes —consequen ly leading o an impo an de-
c ease in e ec i e apping and hen esul ing in highe
alues o conduc ion cu en and conduc i i y.
The gene al end o educ ion o conduc i i y e sus
annealing ime—especially o sho annealing imes—can
be explained on he basis o a ec ys alliza ion p ocess ha
akes place o his ange o empe a u es [34]: XLPE
c ys als su e a econfigu a ion, app oaching pe ec
c ys als and possibly o la ge dimensions ha esul s in
a educ ion o conduc i i y.
These hypo heses also explain he beha iou obse ed by
DEA. We mus emembe ha , by using his echnique, we
dispose o only XLPE samples (wi hou SC laye s), and
consequen ly DEA esul s should be compa ed only wi h
ARC measu emen s pe o med on XLPE cylinde s (wi h
coppe elec odes and wi hou SC sc eens). As we ha e
al eady shown (Figs. 2, 4 and 5), hese measu emen s
clea ly exhibi ed a educ ion in conduc i i y wi h inc eas-
ing annealing ime ha can also be associa ed wi h he
educ ion in he eal pa o he complex conduc i i y in
DEA measu emen s (Fig. 9).
3.3.2. Tempe a u es abo e Tc
Abo e he c i ical empe a u e and a he final s age
(long annealing imes), cable sample conduc i i y is
no iceably highe han conduc i i y o XLPE cylinde s
(coppe elec odes). Wi hin his ange o empe a u es,
hopping conduc ion p ocesses be ween aps a e supe -
imposed on hose based on elec onic band anspo . The
inc emen in he concen a ion o aps has wo di e en
sou ces. On he one hand, we mus conside ions di used
om SC sc eens and, on he o he hand, ions p o ided by
c osslinking by-p oduc s decomposi ion.
Conce ning coppe elec odes samples, he fi s —and
mos impo an —sou ce o ions is no a ailable, which
explains qui e ob iously hei co esponding smalle
alues o conduc i i y as obse ed in Figs. 6–8. Fu he -
mo e, cu es depic ed in hese figu es show ha , ini ially,
du ing he e y fi s hou s o annealing, he a ia ion o
conduc i i y e sus annealing ime is simila o bo h ypes
o elec odes: i begins wi h a dec ease. We should fi s ake
in o conside a ion ha significan di usion o componen s
om semiconduc o h ough insula ion olume equi es a
ce ain ime and, secondly, ha he highe he empe a u e,
he mo e e ficien he di usion p ocess becomes. In his
way, he p ocess ha ini ially appea s is elec onic band
conduc ion and consequen ly, as we ema ked in las
sec ion (T
a
oT
c
), he conduc i i y diminishes wi h anneal-
ing ime. A e a ce ain annealing ime, in he case o cable
samples, ion gene a ion—bo h om c osslinking by-
p oduc s and di used componen s om SC sc eens—
becomes an e ec i e p ocess and a no iceable inc ease o
conduc i i y wi h annealing ime esul s. This end is
indeed due o he inc emen o ions inside he XLPE
insula o , which enhances hopping conduc ion be ween
hem. Finally, his e e enced conduc i i y inc ease does
no con inue indefini ely, bu a he i eaches a sa u a ion
le el. Ou expe imen al esul s a e compa ible wi h
p e ious publica ions in he li e a u e [21] and wi h he
heo e ical model o Na h e al. [20], al hough ou esul s
indica e ha he pa ame e associa ed wi h he dis ance
be ween localized s a es should be a ime and empe a u e-
dependen one, because i is condi ioned by he densi y o
de ec s, which changes wi h he di usion p ocess du ing
annealing p ocedu es.
In he case o XLPE cylinde s (coppe elec odes), he
gene a ion o ions is es ic ed o he sou ce om c oss-
linking by-p oduc s, whe eas his is he p ocess esponsible
o conduc i i y g owing a e i s ini ial dec ease. Once his
sou ce o ions is exhaus ed, and aking in o accoun ha
many o hese gene a ed species a e ola ile, he con ibu-
ion o he conduc i i y by hopping be ween aps
p og essi ely diminishes un il a nea ly cons an alue is
a ained. This final egime is based on elec onic conduc-
ion o e ex ended s a es and is ma ked by he ma e ial
equilib ium s uc u e a he co esponding empe a u e.
The p eceding explana ions can jus i y o a g ea ex en
wo aspec s clea ly defined in he cu es in Figs. 6–8.On
he one hand, he highe he empe a u e, he as e he
minimum alue in conduc i i y is eached a e he ini ial
dec ease. This is due o he ac ha high empe a u es
enhance he e ec i eness o di usion and decomposi ion o
c osslinking by-p oduc s, making he p ocess o hopping-
conduc ion be ween aps mo e e ficien o sho e imes.
On he o he hand (in he case o coppe elec odes), he
highe he empe a u e, he as e he conduc i i y egime
o elec onic-band anspo is achie ed (due o he
deple ion o ions p oceeding om c osslinking by-p o-
duc s).
This explana ion is also alid o expe imen al alues
ob ained by DEA. A low equencies (unde ¼0:1Hz
o Ta¼95 1C and unde ¼1 Hz o Ta¼100 1CFigs.
10 and 11) we can obse e ha ini ially, he eal pa o
conduc i i y ends o inc ease, hen i acqui es a maximum
alue and a e wa ds i diminishes wi h annealing ime.
This beha iou is in acco dance wi h he elaxa ion
associa ed wi h ions p esen inside XLPE. They a e
ARTICLE IN PRESS
gene a ed by dissocia ion o by-p oduc s so ha , empo a-
ily, he con ibu ion o ions is enhanced while annealing
ime g ows. As hese elemen s a e mainly ola ile, howe e ,
hey disappea g adually. As a consequence, hei con-
ibu ion o he eal pa o conduc i i y p og essi ely
dec eases and finally, elec onic anspo p ocesses p e ail
again, as one can easonably in e p e om Fig. 11 (see
cu e co esponding o he la ge annealing ime,
a
¼3
days). Unde his model, he c osso e equency o
isoch one cu es is p oduced by he di e ence in he
e olu ion o conduc i i y, a high and low equencies, wi h
annealing ime. Fu he mo e, his c osso e equency is
also dependen on T
a
. The highe he empe a u e, he
highe he c osso e equency. Fo his eason, a p ecise
de e mina ion o T
c
by measu emen s in he equency
domain is much mo e di ficul han i s de e mina ion in
he ime domain. As annealing empe a u e dec eases, his
c osso e equency dec eases as well, and, eg e ably, his
equency e en ually is lowe ha he equencies ha can
be eached expe imen ally in ou se -up. In Fig. 12,we
ha e ep esen ed as a Mo plo (ln (s) e sus T1=4) he
con ibu ion o conduc i i y o he hopping mechanism
be ween aps in he case o he ‘‘comple e’’ cable samples
wi h sH¼ssEX, whe e sis he o al conduc i i y and
s
EX
is he con ibu ion o conduc i i y by ex ended s a es.
The alues o conduc i i ies we e ob ained om he quasi-
s a iona y egimes (sa u a ion alues), co esponding o
e y long annealing imes, o di e en annealing empe a-
u es. A linea beha iou o ln (s
H
) e sus T1=4can be
es ima ed, which implies ha he anspo mechanism is
basically ia he mally assis ed hopping conduc ion [32].
The esul ing alue o conduc i i y by ex ended s a es is
sEX ¼1:65 1017 S=m. Hill [35] also ound a simila
ela ion be ween conduc i i y and annealing empe a u e,
by conside ing a field-assis ed hopping conduc i i y a he
limi o low elec ic fields, which is e iden ly one o he
impo an expe imen al condi ions obse ed in ou p esen
wo k.
4. Conclusions
We ha e s udied he conduc i e p ope ies o eal MV
dis ibu ion cables wi h XLPE insula ion and ex uded SC
sc eens o a ange o annealing empe a u es ha includes
he se ice empe a u e ange o powe dis ibu ion cables
(E90 1C). We obse ed significan di e ences in he
beha iou o cable samples wi h SC sc eens and wi hou
SC sc eens (XLPE cylinde s). SC sc eens condi ion e y
much he elec ical beha iou o he en i e cable. We
conclude ha in o de o unde s and he he moelec ical
aging o powe cables, i is necessa y o ake in o accoun
he SC sc eens in bo h heo e ical and expe imen al
esea ch. Fu he mo e, esea ch de o ed bo h o unde -
s and aging/b eakdown p ocesses and o imp o e insula-
ion quali y in MV and HV powe cables should be ca ied
ou on ‘‘comple e’’ cable sec ions, i.e., including ex uded
SC sc eens.
Measu emen s ha allowed us o de e mine he e olu-
ion o conduc i i y wi h annealing ime we e pe o med by
using wo di e en me hods: abso p ion/ eso p ion cu en
echnique ( ime domain), and dynamic elec ical analysis
( equency domain). We ob ained good co ela ion be-
ween esul s o measu emen s ca ied ou in bo h domains
( ime and equency), as well as a plausible explana ion o
hem, based on he coexis ence o wo conduc ion
mechanisms. The fi s mechanism in ol es ex ended s a es,
being sligh ly dependen on empe a u e (a leas o ou
empe a u e in e al o in e es ). The second mechanism
ope a es by ca ie hopping be ween aps and is he mally
assis ed. This la e mechanism is e ficien om a ce ain
c i ical empe a u e T
c
, and i is associa ed wi h he
c ea ion and di usion o componen s, essen ially coming
om he SC sc eens. These SC sc eens gene a e an
inc emen o ion concen a ion o empe a u es abo e T
c
ha in ol es a significan inc ease in conduc i i y wi h
annealing ime. The la e can be jus ified by his second
conduc ion mechanism.
Acknowledgemen s
This wo k has been pa ially suppo ed by he Spanish
Minis y o Science and Technology (P ojec MAT
2001-2338-C02-01). Au ho s acknowledge collabo a ion
o company Gene al Cable S.A. h ough he de elopmen
o he en i e wo k, in pa icula by p o iding powe cables
and pe o ming se e al ypes o ea men on hem.
Re e ences
[1] P.H. Lindenmeye , V.F. Holland, Rela ionship be ween molecula
weigh , adial g ow h a e and he wid h o he ex inc ion bands in
polye hylene sphe uli es, J. Appl. Phys. 35 (1964) 55–58.
[2] P.J. Phillips, Mo phology—elec ical p ope y ela ions in polyme s,
IEEE T ans. Elec . Insul. 13 (1978) 69–81.
[3] J. Muccig osso, P.J. Phillips, The mo phology o c oss-linked
polye hylene insula ion, IEEE T ans. Elec . Insul. 13 (1978)
172–178.
ARTICLE IN PRESS
0.226 0.228 0.230 0.232 0.234 0.236 0.23
8
-42
-40
-38
-36
-34
-32
-30
T-1/4
(
K -1/4
)
ln (σH / Sm-1)
Fig. 12. Mo diag am o ep esen ing he con ibu ion o conduc i i y
by hopping p ocesses be ween localized s a es.