JOURNAL OF OPTOELECTRONICS AND ADVANCED MATERIALS Vol. 9, No. 7, July 2007, p.

2110 - 2115

Mechanical and chemical damage of optical fiber

polymer coating
I. SEVERINa,b,*, M. POULAINa, R. EL ABDIa
a
LARMAUR, University of Rennes 1, France
b
detached from Politehnica University of Bucharest, Bucharest, Romania

Systematic measurements have been implemented on commercial optical silica fibers in which the polymer coating has
been damaged, either chemically or mechanically. Dynamic fatigue has been studied using a two point bending bench at
different stress rates. Small mechanical flaws, subsequently controlled by SEM observations, were created on the polymer
coating surface using a micro-hardness indenter, without reaching the glass surface. Chemical damage was implemented
by methylene chloride either as a gel or as a liquid. In this later case polymer could be removed just after a few minutes.
Fiber strength was altered when polymer – silica interface was affected.

(Received October 2, 2006; accepted June 27, 2007)

Keywords: Silica optical fibers, Epoxy-acrylate coating, Damage, Indentation, Etching, Dynamic fatigue testing

1. Introduction arises to assess to what extent fiber strength is influenced

by mechanical and chemical damages.
Optical fibers have been intensively developed [1] and In order to find an answer, fiber polymeric coatings
they make now a key component in telecommunication were damaged mechanically and chemically, then tested to
networks. Accordingly, their long-term stability, even in dynamic fatigue and finally compared to the as-received
severe environmental conditions, appears as an important commercial fibers. While one could expect that nay large
requirement. The reliability and the expected lifetime of coating degradation enhances fiber failure, experiments
the optical fibers are closely related to the chemical action have drawn some unexpected conclusions.
of the environment on silica that in turn, influences fiber
strength. 2. Experimental
To ensure the long-term mechanical strength of the
optical fibers, a polymer coating is applied onto fiber 2.1 Fibers used for testing
surface during fiber fabrication [2-4]. This external coating
is vital for the optical fiber lifetime. Its protective action Fibers used in this study are Alcatel commercial
includes several functions, such as: to protect glass fiber single mode silica fibers, 125 µm in diameter with a
from any external damage, to limit chemical attack of 62.5 µm thick epoxy-acrylate two layers polymer coating.
water and to ensure fatigue protection and bending This type of coating is the most largely used one in
insensitivity [5-7], especially during handling and in- standard optical fibers. It offers a good protection to
service installation. external mechanical damage, but it remains permeable to
It has been emphasized that failure of fibers is ruled water, playing a role of diffusion barrier. The internal
by two major factors that are micro-crack growth from the layer is soft with a low glass transition temperature and is
glass surface, “Griffith flaws” [8,9], and water activity at applied onto the glass fiber surface. It ensures protection
glass surface [10-13]. Different fatigue behaviors of silica against micro-bending and damping of the external
optical fibers have been reported as a function of glass stresses. The external layer has a higher Tg and protects
adhesion, water absorption, water permeability and the fiber against physical aggressions. To comply with
mechanical strength properties of the polymer coating process requirements, the UV-polymerization of the
[14,15]. In order to assess the influence of flaw generation acrylate resins is rapidly performed, in line, offering an
during fiber processing, some indentation tests have been excellent adhesion to glass surface and a large range of
implemented on silica surface [16,17].. They have revealed elasticity (Young modulus).
several fatigue mechanisms, in particular crack tip
blunting [18-20], and structural relaxation under static 2.2 Dynamic fatigue measurement using a
stress [17, 21]. Silica is very sensitive to environmental two-point bending testing apparatus
action, through the well-known stress corrosion
phenomenon, and it is of practical importance to assess the As-received fiber samples and as-damaged ones (see
protection ensured by the polymer and the long-term procedure described below) were subjected to dynamic
mechanical stability. As this external coating may be fatigue measurements using a two-point bending testing
damaged during handling or normal use, the question device [4, 22]. Samples, 10 cm in length, are bent and
placed between the grooved faceplates of the testing
 Mechanical and chemical damage of optical fiber polymer coating 2111

apparatus, in order to avoid the fiber slipping during the reduce significantly the fiber strength. As already
faceplates displacement and to maintain the fiber ends in mentioned, polymer coating plays a major role to protect
the same vertical plan. Several series of 30 samples were the fiber surface against mechanically induced
tested with faceplates constant velocities of 80, 150, 500 aggressions.
and 800 µm/s respectively. All measurements were In order to investigate the influence of mechanical
performed in the laboratory environment (temperature damage on fiber strength, controlled indentations using a
19°C±1ºC and relative humidity 35%±2%). The stress to Vickers micro-hardness indenter (Digital Matsuzawa Seiki
fracture applied to the fiber was calculated from the Co. MXT 70) were applied on the polymer coating in such
distance separating the faceplates, using the Proctor and a way that indenter does not reach glass surface. Different
Mallinder relation, improved by Griffioen [23]. The failure combinations of loads (10 and 25 gf) and durations (5, 10,
stress was measured for each fiber, and then the results 15 and 20 s) were chosen. The resulting mechanical flaws
were treated through a statistical approach using the were observed by SEM [22].
Weibull theory [24]. The indented samples were carefully placed between
The classical Weibull plots showing the logarithm the two point testing bench faceplates, so that the small
function of the cumulative failure probability ln(-ln(1- indented flaw should initiate fracture, which means that it
Fk)), where Fk (in %) represents the cumulative probability is centrally oriented on the sample external curvature.
to failure for each stress to fracture σ (in MPa), related to Dynamic fatigue tests were performed at a speed of 150
the logarithm of the stress to fracture ln(σ, in MPa), has µm/s and the medium stresses to fracture were determined.
allowed to find the statistical parameters, namely: the Another indented series was aged into de-ionized
medium stress value σmed, the median stress σ(50%), water for one hour (same procedure § 2.3), then dried in
corresponding to a probability to fracture Fk=50%, the air for one day prior to dynamic fatigue measurement at
Weibull slope md, and the Weibull parameter σ0. the testing velocity of 150 µm/s. Results of different
The slope p of the curve ln(σ, in MPa) as a function of groups of fibers were subsequently compared.
ln(stress strain, µm/s) is related to the stress corrosion
parameter nd, by the following relation [25]: 2.5 Chemical damage through methylene chloride
etching
nd=(1/p)-1 (1)
The chemical etching of the polymer coating was
So, from the experimental failure stress for the four performed using liquid and gel based methylene chloride.
testing speeds, by linear regression one may calculate the In the first step, a methylene chloride based gel was
straight line slope p, that finally leads to the stress used, etching the polymer coating on 1 cm in length in the
corrosion parameter, nd, to be compared to that of the as- sample’s central part for durations of 3, 5, 10, respectively,
received fibers. 15 minutes. The gel was removed by de-ionized water
abundant rinsing followed by acetone rinsing. Dynamic
2.3 Aging testing tests were performed after drying for 70 hours in ambient
environment, at a testing speed of 150 µm/s.
The fibers subjected to aging were plunged into a tank In the second step, a more severe etching was
containing de-ionized cold water for one hour. Eight series implemented: the methylene chloride based gel etched the
of 30 samples, 10 cm in length, were carefully arranged to polymer coating for 15 minutes, subsequently, for 30
float into the water tank, in zero stress conditions during minutes. Then the gel was removed through paper
the aging treatment. The number of series was adjusted in sweeping (without any friction) and dynamic fatigue tests
order to carry out the dynamic fatigue testing measurement were performed immediately. The etched region of the
at the four different testing velocities in order to calculate fiber was carefully inserted between testing bench
the nd factor. faceplates to be located on the curvature to be fractured.
After aging, fibers were removed from water and four The four selected speeds were applied. Some measurement
series were subjected immediately to dynamic fatigue problems were encountered, in particular for low speeds,
testing at the four selected speeds. The other four series because gel residues paste over bench faceplates.
were simply laid to dry into the laboratory environment on Therefore frequent cleaning was required.
absorbent paper for one day (20-24 hours) prior to similar In the third step, the most severe etching was applied,
testing. using liquid methylene chloride. Etching for 5 min.
The nd factor was calculated using the linear allowed the polymer to be removed to obtain pristine glass
regression slope that interpolates the stress to fracture σ fibers. These fibers were carefully separated prior to
corresponding to a cumulative failure probability of 40%. dynamic fatigue measurements at the four selected stress
The different group of fibers behavior was compared. rates. Practical handling difficulties were encountered due
to pristine glass fibers brittleness.
2.4 Mechanical damage through Vickers Another series of dynamic fatigue testing were
indentation performed on stripped fibers (liquid methylene chloride
for 5 min. followed by polymer detachment) subjected to
The optical fiber surface may be easily damaged ageing by immersion into de-ionized water for 30 min.,
through abrasion or indentation, and handling bare fibers then drying in air for 70 hours prior to testing. The
currently creates enough flaws on the fiber surface to
2112 I. Severin, M. Poulain, R. El Abdi

samples etched in methylene chloride were SEM superficial coating mechanical damage through Vickers
examined. The polymer damage and related viscosity indentation seems to affect fiber strength.
change were observed, sometimes even with naked eye, SEM observations, previously reported [22],
but evidenced clearly through SEM observations. confirmed the Vickers flaws size, the absence of any
fracture associated to indentations (neither to diagonals,
3. Results and discussion nor upon the indentation surface) and the “sink-in” plastic
behavior of the polymer coating [25].
3.1 Aging treatment The indented fibers were subsequently aged into de-
ionized water for one hour, then dried in ambient
The comparison between the Weibull plots drawn for environment for one day and subjected to dynamic fatigue
the four dynamic testing velocities in the case of the as- measurement at the speed of 150 µm/s, as-seen in Fig. 2.
received fibers, as-aged fibers in water for one hour, tested A slight strength increase (below 2%) may be noticed in
immediately after aging and the fibers aged and dried for the case of indented fibers aged and dried. This increase is
one day prior testing was given in Fig. 1. This comparison notable for Vickers indentations applied with a load higher
has emphasized some differences. than 10 gf and longer than 10-15 s.
By comparison to non-aged fiber, the immediate fiber
strength slightly decreased (below 10% average) following 3.3 Chemical damage of polymer coating through
aging treatment into de-ionized water. A broader gel-based etching
dispersion was noticed, too. Drying in air prior to testing
has lead to fiber strength increase to slightly higher values Fiber samples have been etched by methylene
than the non-aged fibers (below 2%). Broader dispersion chloride (CH2Cl2) based-gel for a few minutes, then
in the case of low stress rates, but slightly steeper for high abundantly rinsed in water, then in acetone, and finally
rates may be noticed. The stress corrosion coefficient nd, dried for 70 hours prior to mechanical testing. Testing
and the corresponding regression coefficients (R2) are the were performed in order to directly subject the treated
followings: as-received fibers 13.16 (0.983), as-aged zone (approx. 1 cm in length) to the stress, locating the
immediate tested 12.83 (0.985), finally, aged/dried prior to etched area centrally between the testing faceplates.
testing 12.66 (0.992). The decrease in strength may be Weibull plots for the selected speeds are shown in Fig. 3.
explained simply by the variation in water activity at glass As it may be seen, no significant differences were
surface level following aging and drying. While stress observed between these fibers and the as-received fibers at
corrosion coefficient rather decreases for aged fibers, this the testing speed of 150 µm/s.
evolution is hardly significant. The gel action on the polymer coating appeared
clearly in a second set of experiments. After etching for
3.2 Mechanical damage of polymer coating longer durations (15-30 min), gel was removed by
through Vickers micro-indentation sweeping fiber surface with absorbent paper. Dynamic
tests were performed immediately after etching and
In order to create small flaws on the polymer coating sweeping. The Weibull distribution for the different fibers
surface, controlled indentations were applied. The – etched and as-received - is shown in Fig. 4. A slightly
indentation load and duration were increased step by step, steeper dispersion is noticed for the 30 min. etching
taking care not to reach the glass surface. duration.
Using a faceplate velocity of 150 µm/s, dynamic Although some manipulation problems were
fatigue testing was performed placing the indentation encountered due to gel residues pasted on testing bench
centrally, on the external curvature of the tested sample, so faceplates, the dynamic fatigue tests were performed for
that the small flaw should act as fracture initiator. The the four selected stress rates, so the stress corrosion
fiber strength is not affected by the mechanical coefficient nd of 11.8 was determined, with a regression
indentation, as this appears in Table 2. coefficient R2 of 0.99.
As compared to the as-received fibers, the nd factor
Table 2. Vickers micro-indentation. decreases following chemical damage through gel-based
etching. In the meantime, the distribution of the Weibull
Vickers load, gf As-received 10 10 10 10 25 plots is broader and appears nearly bi-modal, meaning that
Duration, s fiber 5 10 15 20 10 an increasing number of extrinsic defects are present and
Medium strength, 5980 5998 5937 59415949 5995 active on the fiber surface. While overall fiber strength is
MPa not significantly affected by gel-based etching, especially
at short exposure times and further drying, direct
observations – visual and SEM - revealed coating
There is no significant difference between the fiber
modifications.
strength of the indented fiber and that of the as-received
After fracture, the coating surface exhibits striae
one. This suggests that the polymer coating efficiently
expressing its brittle character. Handling fibers may
protects against external mechanical aggressions as long as
damage coating, as seen in Fig. 5. In some areas the
the surface of the glass fiber is not reached. Neither the
polymer coating, already stressed and inflated, exhibits
stress at the interface induced by indentations, nor
large cracks while, in other ones, it becomes more plastic
 Mechanical and chemical damage of optical fiber polymer coating 2113

and viscous. This leads to changes in diameter, especially gel etched fibres, dried, velocity 150 µm/s
after pinching fibers for sampling before observation. All 1,5
3 min
over fiber length, fiber surface appears irregular with 5 min
frequent flakes and variations in diameter. Taking into 0 10 min

ln(-ln(1-Fk))
account the polymer fracture in some inflated zones and 15 min

diameter variations one may assume that this chemical -1,5

non-aged

treatment induces locally some residual stresses.

Non-aged fibres -3
1,5
v=80 µm/s ln (σ, MPa)
v=150 µm/s -4,5
0 8,55 8,65 8,75
v=500 µm/s
ln(-ln(1-Fk))

v=800 µm/s
Fig. 3. Fibres etched with gel for different durations, in min. (see
-1,5
legend), for clarity reasons, the abscise scale was enlarged.
gel etched fibres, tested immediately, 150 µm/s
-3

ln (σ, MPa) 1,5

gel 30 min
-4,5
gel 15 min
8,5 8,6 8,7 8,8 8,9 0 non-aged

ln(-ln(1-Fk))
aged fibres - water, 1 hour, tested immediately
-1,5
1,5
v=80 µm/s
v=150 µm/s -3
0
ln(-ln(1-Fk))

v=500 µm/s ln (σ, MPa)

v=800 µm/s
-1,5 -4,5
8,55 8,65 8,75

-3
Fig. 4. Fibres etched with gel removed by wipe drying with
ln (σ, MPa) absor bent paper prior to testing (the abscise scale
-4,5 enlarged for clarity reasons).
8,5 8,55 8,6 8,65 8,7 8,75 8,8 8,85 8,9

aged fibres - water, 1 hour, dried 1 day

1,5
v=80 µm/s
v=150 µm/s
0
v=500 µm/s
ln(-ln(1-Fk))

v=800 µm/s

-1,5

-3

ln (σ, MPa)
-4,5
8,5 8,6 8,7 8,8 8,9

Fig. 1. Weibull plot of two-point bending testing for

different faceplate velocities v, in µm/s (in axes: Fk
cumulative failure probability, in % and σ stress, in
MPa).

Indented fibres, then aged and dried - velocity =150 μm/s

1,5 Force 10gf - Time 5s

Force 10gf - Time 15s
0
ln(-ln(1-Fk))

Force 25gf - Time 10s

-1,5 Fig. 5. Polymer coating irregular viscous surface of gel-
etched fibres; inflated-stressed polymer coating cracks
-3
during handling.

ln(σ) (MPa)
-4,5 3.4 Chemical damage of polymer coating through
8,65 8,67 8,69 8,71 8,73 8,75
liquid methylene chloride etching
Fig. 2. Vickers indented fibres, testing velocity 150 µm/s;
for clarity reasons, the abscise scale was enlarged and The etching reaction was quite more severe in liquid
the non-aged fibre strength was not reported. methylene chloride. As a result, coating could be entirely
2114 I. Severin, M. Poulain, R. El Abdi

removed after etching for just a few minutes and pristine SEM observations have evidenced some features of
glass fiber carefully extracted from the liquid reagent. The polymer removal, namely: the dissolution tendency and
first series of pristine glass were subjected immediately to drop withdrawal, the viscosity change leading to an easily
dynamic fatigue testing (Fig. 6) while the other series were removed plastic layer (Fig. 8) and finally, some
aged into water for one hour and dried in air for 70 hours longitudinal polymer residues in areas under stress (cracks
prior dynamic testing. The Weibull distributions are shown on polymer drops). The pictures also show a few detached
in Fig. 7. glass fragments and fracture surfaces that underline
brittleness of pristine fiber.
etching in CH2Cl2 for 5 min, detached polymer
1,5

v=80 µm/s
0
ln(-ln(1-Fk))

v=150 µm/s
v=500 µm/s
-1,5
v=800 µm/s

-3

ln (σ, MPa)
-4,5
8,1 8,3 8,5 8,7 8,9

Fig. 6. Fibres etched in concentrated liquid CH2Cl2, polymer

coating removed, pristine fibres immediately tested.

etching in CH2Cl2, detached polymer, glass fibres aged, dried

1,5
v=80 µm/s
v=150 µm/s
0 v=500 µm/s
ln(-ln(1-Fk))

v=800 µm/s

-1,5

-3
ln (σ, MPa)

-4,5
6,9 7,4 7,9 8,4 8,9

Fig. 7. Fibres etched in concentrated liquid CH2Cl2,

polymer coating detached, pristine fibres aged for 1 hour Fig. 8. Polymer coating during removal – change of viscosity,
and dried for 70 hours prior to testing. brittle pristine glass.

The fiber strength decreased after polymer removal 5. Conclusions

following chemical etching: the medium strengths, ranging
between 5800 and 6800 MPa in the case of as-received During handling or in use, the external coating of the
fibers, decrease to values of 4400 to 5000 MPa in the case optical fibers may be damaged, naturally raising the
of etched and stripped fibers. A quite broader dispersion is question of the correlation between fiber strength and
noticed, too. An even larger decrease was observed for coating mechanical / chemical damage. Various dynamic
fibers etched for a longer time: in this case strength values fatigue testing on damaged fibers have been carried out,
range from 2700 to 4800 MPa. In all cases, failure strength using a two-point bending bench at different stress rates.
of Weibull plot had a broader distribution than non-aged Mechanically induced flaws in the coating do not
ones. significantly alter fiber strength. Surprisingly enough
Following CH2Cl2 etching, the stress corrosion mean strength slightly increases following aging in water
parameter nd increased to 16 (with R2 of 0.959) as and drying in air. However, more systematic studies are
compared to as-received fibers (13.2), suggesting either required to confirm this observation. Nevertheless the
that methylene chloride adsorbed onto fiber surface could epoxy-acrylate polymer appears to offer a good
act as a barrier against water corrosion, or it is less mechanical protection of the fiber when induced flaws do
efficient than water as a stress corrosion reagent. Due to not reach glass surface. The stresses at the interface level
aging into water, pristine glass fibers stress corrosion induced by flaw indentations do not affect fiber strength,
coefficient nd decreased drastically to a low value of 5.9 nor does superficial polymer coating mechanical damage
calculated from a plot with a poor regression coefficient through Vickers indentation. Microscopic observations
(0.86 vs 0.96). This lower stress corrosion coefficient confirm the “sink-in” polymer coating plastic behavior and
could indicate that water is much active, due to the larger the absence of flaws-associated cracks.
water diffusion into the crack tips. This observation The extension of the chemical damage depended on
confirms that polymer coating reduces the diffusion of the the reagent and the etching duration. The methylene
water molecules toward glass surface. chloride gel-based affected fiber strength when etching
 Mechanical and chemical damage of optical fiber polymer coating 2115

was prolonged and testing performed immediately after [7] S. T. Shiue, H. Ouyang, J. App. Phy. 90(11), 5759-
gel removal by paper sweeping. But fiber strength did not 5762 (2001).
change when gel was removed by abundant water rinsing [8] M. Muraoka, K. Ebata, H. Abe, J. Am. Ceram. Soc.
and fibers were dried in air, even though SEM 76(6), 1545-1550 (1993).
observations showed surface damage and polymer coating [9] J. L. Mrotek, M. J. Matthewson, J. Am. Ceram. Soc.
viscosity modifications. The chemical damage was more 83(12), 3100-3108 (2003).
severe in liquid methylene chloride that led to polymer [10] J. L. Mrotek, M. J. Matthewson, C. R. Kurkjian,
dissolution and a quite drastic decrease of fiber strength. Journal Light-wave Technol. 19(7), 988-993 (2001).
The polymer removal process arising from the chemical [11] J. L. Mrotek, M. J. Matthewson, C. R. Kurkjian,
etching may be exemplified by SEM observations. Journal Light-wave Technol. 21(8), 1775-1778
The stress corrosion coefficient slightly decreased (2003).
following aging in water and drying in air. Chemical [12] T. V. Volotinen, in Optical fiber reliability & testing,
etching in liquid reagent increased the nd factor, in the SPIE vol. 3848, 0277-786x/99/, 134-143 (1999).
case of pristine glass fibers tested immediately after [13] S. Berger, M. Tomozawa, J. Non-Cryst. Solids 324,
etching and polymer removal. But this factor drastically 256-263 (2003).
decreases after aging in water and drying. [14] T. Wei, J. Skutnik, J. Non-Cryst. Solids 102, 100-105
In summary, this study shows that optical fiber (1988).
strength has been altered only if damage affected the [15] G. M. Camillo, in Optical fiber reliability & testing,
polymer – glass interface so that polymer coating is SPIE vol. 3848, 0277-786x/99/, 55-61, (1999).
delaminated, cracked or stripped. This appears quite [16] S. L. Semjonov, C. R. Kurkjian, J. Non-Cryst. Solids
positive for reliability assessment. Insofar as residual 283, 220-224 (2001).
stresses are not high enough to affect interface, the fiber [17] N. Gougeon, J. C. Sangleboeuf, M. Poulain, R. El
strength is not significantly changed. Then, superficial Abdi, Fiber & integrated optics 24(5), (2005) (print
coating damage should not influence average fiber online).
strength. [18] K. Hirao, M. Tomozawa, J. Am. Ceram. Soc. 70(1),
43-48 (1987).
[19] K. Hirao, M. Tomozawa, J. Am. Ceram. Soc. 70(6),
References 377-382 (1987).
[20] T. Michalske, W. Smith, B. Bunker, J. Am. Ceram.
[1] J. B. MacChesney, D. J. DiGiovanni, J. Am. Ceram. Soc. 74(8), 1993-1996 (1991).
Soc. 73(12), 3537-3556 (1990). [21] I. Severin, M. Poulain, R. ElAbdi, in Photonic
[2] G. Orcel, D. Biswas, J. Am. Ceram. Soc. 74(6), 1377- Applications in Devices & Communication Systems,
1377 (1991). P. Mascher, A. P. Knights, eds., Proc. SPIE vol.
[3] A. A. Abramov, a.o., Electronics Letters 29(22), 5970, (2005).
1977-1978 (1993). [22] I. Severin, R. ElAbdi, M. Poulain, Optics & Laser
[4] I. Severin, M. Poulain, R. Elabdi, J. Optoelectron. Tech. (article in press July 2005).
Adv. Mater. 7(3), 1581-1587 (2005). [23] W. Griffioen, “Optical fiber reliability”, Thesis edited
[5] B. P. Pal, Fundamentals of fiber optics in by Royal PTT, The Netherlands NV, PTT Research,
telecommunication and sensor systems. (Wiley Leidschendam, (1994).
Eastern ltd, Delhi, 1992). [24] P. Chanticul, a.o., J. Am. Ceram. Soc. 66(7), 515-518
[6] S. T. Shiue, T. Y. Shen, H. Ouyang, K. M. Lin, (1983).
J. Mat. Chem. & Phy. 83, 239-244 (2004). [25] K. W. Peter, J. Non-Cryst. Solids 5, 103-115 (1970).

______________________
*
Corresponding author: [email protected]

View publication stats