Infrared spectral dependencies of pulsed X-ray radiation-induced attenuation in single-mode optical fibers

Abstract

We investigate the changes of optical absorption at infrared wavelengths induced by pulsed X-ray irradiation on single mode fibers. We focus on the influence of the germanium (Ge) and phosphorus (P) dopants. To do that, we characterize the changes of the radiation-induced attenuation (RIA) at 1310 nm and 1550 nm for shorter times after X-ray pulse (10^-7-10^-1s). We also measure the spectral dependence of the RIA (500-2000 nm) from 10^-1 s to several hundreds of seconds. We show that absorption bands peaking at infrared wavelengths mainly explain the RIA at 1310 nm and 1550 nm. In a Ge-doped fiber two infrared absorption bands, not previously reported in literature, are shown to substantially contribute to infrared RIA. In a P-doped fiber, the infrared RIA is explained by using the absorption bands of P-defects previously reported in literature with an un-reported absorption band peaking around 1.43 eV (FWHM=0.41 eV).

Full text

Page 1

Infrared Spectral Dependencies of Pulsed X-Ray
Radiation-Induced Attenuation in Single-Mode
Optical Fibers


J. Bisutti1,2, Student Member, IEEE, S. Girard1, Member, IEEE, C. Marcandella1 and J. Baggio1, Member, IEEE


.Abstract – We investigate the changes of optical absorption
at infrared wavelengths induced by pulsed X-ray
irradiation on single mode fibers. We focus on the influence
of the germanium (Ge) and phosphorus (P) dopants. To do
that, we characterize the changes of the radiation-induced
attenuation (RIA) at 1310 nm and 1550 nm for shorter
times after X-ray pulse (10-7-10-1s). We also measure the
spectral dependence of the RIA (500-2000 nm) from 10-1 s
to several hundreds of seconds. We show that absorption
bands peaking at infrared wavelengths mainly explain the
RIA at 1310 nm and 1550 nm. In a Ge-doped fiber two
infrared absorption bands, not previously reported in
literature, are shown to substantially contribute to infrared
RIA. In a P-doped fiber, the infrared RIA is explained by
using the absorption bands of P-defects previously reported
in literature with an un-reported absorption band peaking
around 1.43 eV (FWHM=0.41 eV).


I. INTRODUCTION

HE optical reponse of single-mode optical fibers
submitted to a radiative pulse has already been
characterized at 1310 nm and 1550 nm Telecom
wavelengths in previous studies [1-7]. Pure-silica-core
fibers have been shown to exhibit very high transient
radiation-induced attenuation (RIA) when subjecting to
an high-dose rate pulse (see e.g. [4]). For
germanosilicate fibers, it was shown that the P addition
into the fiber core or fiber cladding substantially
decreases the transient RIA but leads to very high
permanent induced-losses. The RIA results from the
creation of point defects, also called color centers in the
silica glass constituting the fiber core and the fiber
cladding. Spectral absorption measurements can give
information on the nature of the generated point defects.
Indeed, by decomposing the radiation-induced
attenuation (RIA) spectra with the help of Gaussian
absorption bands associated to the created point defects,
it is possible to evaluate the contribution of each point
defect to the induced losses. In this paper we study the

Manuscript recieved September 7, 2007.
J. Bisutti is with the CEA DIF, F91680 Bruyères-le-Châtel, France, and
also with the Laboratoire TSI, Université de Saint-Etienne, F42000
Saint-Etienne, France (e-mail: jean.bisutti@cea.fr).
S. Girard, C. Marcandella and J. Baggio are with the CEA DIF,
F91680 Bruyères-le-Châtel, France (e-mail:
sylvain.girard@cea.fr;claude.marcandella@cea.fr;jacques.baggio@cea.
fr).

influence of Ge and P dopants on the transient RIA of single-
mode (SM) fibers at infrared (500-1900nm) wavelengths. For
this, we perform two types of transient RIA measurements:
short time-resolved measurements (10-8-0.1 s) at single
wavelengths (1310 nm and 1550 nm) and in-situ spectral
measurements (500-1900 nm, ~0.1s to 1000 s). A Super
Continuum source, which has an optical power much greater
than that of an usual halogen source, has been used, to
perform in-situ spectral RIA measurements on single-mode
fibers. With this new tool, we will be able to improve our
knowledge about the origins of fiber degradation at infrared
wavelengths.

II. EXPERIMENTAL PROCEDURE
A. Tested Optical Fibers
The characteristics of the tested SM optical fibers are given
in Table I. The #Ge and #Ge/P are commercial off-the-shelf
fibers. The #P is a prototype fiber (SR-00-12-28 preform
manufactured by the Fiber Optic Research Center, Moscow).
TABLE I.
CHARACTERISTICS OF SINGLE-MODE TESTED FIBERS
Fibers Manufacturer
Core/cladding
diameter
Core dopants
Cladding
dopants
#Ge Corning 8.2 /125 µm Ge (~4%mol) none
#P FORC 4/125 µm P (~12-13%mol) none
#Ge/P Alcatel 9/125 µm Ge Ge, P, F


B. Experimental Setup
The tested optical fibers were subjected to transient exposure
using the ~1 MeV pulsed X-ray generator ASTERIX from
the Centre d’Etude de Gramat in France [8]. The dose rate is
>109 rad s-1. All the measurements were performed at room
temperature. The tested fibers were coiled to diameters of
60 mm and mated to optical sources and detectors through
30-m fiber pigtails. The radiation responses of the fibers were
successively characterized with the two setups represented in
Fig. 1. The lengths of the fiber samples were adapted to
measure the fiber response on several attenuation decades.
The RIA uncertainties are mainly attributed to the dosimetry
and are estimated within ±20%.
T
978-1-4244-1704-9/07/$25.00 ©2007 IEEE

Page 2

INSTRUMENTATION ZONEIRRADIATION ZONE
X-ray pulse
Tested
Fiber
20 m
Photodiode
Laser diode
1.31/1.55 m
Scope (ns-s)
Scope (s-0.1s)
a)
20 m
b)
INSTRUMENTATION ZONEIRRADIATION ZONE
Tested
Fiber
SC source
X-ray pulse
Visible
spectrometer
Infrared
spectrometer

Fig. 1. Schematic diagram of the experiments: (a) Temporal
measurement of the radiation-induced attenuation using laser diodes:
1310 or 1500 nm, 0.1s-0.1s. (b) Spectral and temporal measurement
of the radiation induced-attenuation using a Super Continuum (SC)
source: 500-1900 nm, times from 0.1 s to 1000 s.
The first experimental setup allows to characterize the
radiation-induced attenuation at 1310 and 1550 nm for
times from 0.1 s to 0.1 s after an X-ray pulse. Stable
laser diode power light (~2 mW) is injected into the fiber
sample. The temporal evolution of the light power is
measured with Allegro-20 photodiode (20 GHz) from
ALS-Gmbh. Spectral measurements extended to the IR
(500-1900 nm) for times from 0.1 s to hundreds of
seconds after an X-ray pulse have been performed with
the second experimental setup. The spectral dependence
of the transmitted power is measured with OceanOptics
spectrometers: HR4000 (500-1100 nm), NIR256 (800-
2100 nm) or NIR512 (900-1700 nm). The three
spectrometers have a ~ms time resolution. As the optical
power injected from a tungsten-halogen source into a
SM fiber (core diameter <10m) is too weak to perform
measurements with a low temporal resolution, we have
used a Super Continuum (SC) light source. In that
source, a laser pulse is spectrally broadened through its
propagation inside a photonic crystal fiber [9]. Our SC
source was the SuperK Red from Koheras (wavelengths
ranging from 575 nm to more than 2000 nm, Popt~100
mW, ~40 kHz) [10].

III. EXPERIMENTAL RESULTS
A. Kinetics of the RIA
The kinetics of the RIA at 1310 nm and 1550 nm after
the high-dose rate pulse are given in Fig. 2. For times
shorter than 10-2 s, the RIA measurements were
performed with the laser diodes and photodiodes. For
times longer than 10-2 s, the single-wavelength kinetics
data are extracted from the RIA spectral measurement
(SC source and spectrometers). The RIA has been normalized
with respect to the deposited dose. For fibers doped with Ge
or P and submitted to an high-dose rate pulse, studies have
shown the linearity of the RIA at ~850 nm, ~1300 nm and
1550 nm with respect to the deposited dose [1, 2, 11-13]. For
the #Ge sample, the attenuation quickly and continuously
decreases at both wavelengths for times from 10-7 to 100 s
after the pulse. For the #Ge/P sample and times ranging from
10-7 to 10-2 s, the RIA quickly decreases and is lower than
the RIA of the P-free fiber at both wavelengths. However, for
longer times, the RIA decay kinetic of the P-codoped fiber
slows down and becomes stable. In this way, the addition of
P in the cladding of a germanosilicate fiber decreases the
transient RIA, but leads to higher permanent damage, as
previously shown in [14]. The #P sample, only P-doped,
exhibits no RIA recovery after an X-ray pulse: the RIA
increases with time.

B. Spectral Dependence of the RIA
For each sample, Fig. 3 collects the RIA spectra recorded at
different times after an X-ray pulse.
For the #Ge fiber (core Ge-doped) the RIA decreases ffrom
shorter to longer wavelengths at a given time after the pulse.
For this fiber, the RIA decreases with time after the pulse in
the whole spectral range of measurements. Thus, the point
defects responsible for the IR absorption in this fiber seem
rather unstable at room temperature.
For the #P fiber (core P-doped) the RIA decreases from
shorter wavelengths to ~1000 nm and increases from ~1000
nm to greater wavelengths. For wavelengths shorter than
~1000 nm the RIA decreases with time after the pulse, while
the RIA increases for greater wavelengths. The increase of
RIA after the pulse suggests that at least one defect is
converted into an other defect after the pulse. At 0.1 s after
the pulse, the RIA of the #Ge/P fiber (core Ge-doped and
cladding containing Ge, P and F) seems to decrease from
shorter to longer wavelengths. The RIA shape of the #Ge/P
sample changes with time: for greater wavelengths the RIA
seems stable while the RIA decreases for shorter
wavelengths. In this way, e.g. 10 s after the pulse, the RIA
decreases from shorter wavelenghts to ~1200 nm, while the
RIA increases for greater wavelengths.

Page 3

10
-9
10
-7
10
-5
10
-3
10
-1
10
1
10
3
10
-4
10
-3
10
-2
10
-1
10
0
(iii)
(ii)
(i)
RIA (dB km
-1 rad
-1)
#Ge, 4 m, 9.7 krad, LD
#P, 50 m, 2.7 krad, LD
#Ge/P, 4 m, 33.6 krad, LD
#Ge/P, 100 m, 2.9 krad, LD
b)1550 nm
Time after X-ray pulse (s)
SC
LD
10
-4
10
-3
10
-2
10
-1
10
0
(iii)
(ii)

(ii) #P, 10 m, 22.6 krad, SC
(iii) #Ge/P, 100 m, 12.6 krad, SC
(ii) #P, 10 m, 22.6 krad, SC
(iii) #Ge/P, 100 m, 12.6 krad, SC
(i)
(i) #Ge, 1 Km, 7.5 krad, SC
(i) #Ge, 1 km, 7.5 krad, SC
#Ge, 4 m, 11.4 krad, LD
#P, 50 m, 3.7 krad, LD
#Ge/P, 4 m, 10 krad, LD

1310 nm
a)
SC
LD

Fig. 2. Kinetics of radiation induced attenuation (RIA) after X-ray
pulse at a) 1310 nm and b) 1550 nm. The results, normalized with
respect to the deposited dose, have been obtained at room temperature
for times from 10-8s to 10- 2 s using laser diode (LD), and for times
from to 10-1 s to 1000 s with a Super Continuum (SC) source.

500700900
Wavelength (nm)
110013001500 17001900
10
0
10
1
10
2
10
3
0.1 s
0.3 s
1 s
3 s
10 s
300 s
Ge/P fiber - 50m - 11 krad

10
0
10
1
10
2
10
3

RIA (dB km
-1)
1 s
10 s
100 s

#Ge fiber - 50 m - 9 krad
10
2
10
3
0.1 s
0.1 s
1 s
10 s
100 s
0.1 s
1 s
10 s
100 s
#P fiber - 10m - 22.6 krad

.

Fig. 3. Spectral dependence of radiation-induced attenuation (RIA)
after an X-ray pulse in the three tested single-mode fibers. The
experiments have been performed at room temperature.



IV. DISCUSSION
A. Radiation-induced attenuation in the #Ge sample
In a recent paper, we have studied the spectral response
following an X-ray pulse of a multimode Ge-doped fiber in
the 300-900 nm spectral range (1.37-4.13 eV) [15]. For that
fiber, we have been able to fit the RIA spectra using the
Ge(1), Ge-X bands previously given by Anoikin et al. in
[16]. The addition of an absorption band peaking at ~3.0 eV
(FWHM=0.6 eV) was also necessary the fit the RIA spectra
at short times after the pulse. This band was tentatively
ascribed to [GeO4]- defects on the basis of an allowed
transition calculated at a similar energy by Dianov et al. [17].
Recently, Régnier et al. have reported an IR absorption band
peaking at 0.54 eV (FWHM=0.3 eV) in a Ge-doped fiber
[18]. This band has been ascribed to a germanium self-
trapped hole species (Ge-STH) on the basis of the previously
proved existence of STH in Ge-doped glasses and the
existence of small polaron SP(Ge) in -quartz having similar
properties that STH. We do not succeed to fit the RIA
spectrum of the Ge-doped fiber using the absorption bands of
the Ge(1), Ge(X), [GeO4]- and Ge-STH defects in the range
of energies measured: ~0.7-1.9 eV. At least two absorption
bands peaking at 0.98 eV (FWHM=0.29 eV, related defect
called Ge-A) and 1.35 eV (FWHM=0.34 eV, related defect
called Ge-B) are necessary in order to obtain a good fit. Fig.
5 shows the RIA spectrum of the Ge-doped fiber obtained 1 s
after an X-ray pulse. That spectrum was decomposed into the
bands of the Ge-STH, Ge-X, Ge-A and Ge-B defects whose
characteristics are summarized
decomposition is valid for the RIA spectra recorded between
0.1 s and ~1000 s after the pulse. The absorption bands of the
Ge(1) and [GeO4]- defects do not contribute.
in Table II. This

0.60.81.01.21.41.61.82 0
1E-7
1E-6
1E-5
1E-4
Ge-STH
Ge-A
Ge-B
RIA (dB/m/rad)
Energy (eV)
Ge-X

Fig. 4. Spectral decomposition of the attenuation (RIA) induced in the Ge-
doped fiber #Ge at 1 s after an X-ray pulse of 9 krad. The experimental
curves have been fitted using the Gaussian bands given in Table II.

Page 4

TABLE II.
ABSORPTION BANDS USED FOR THE
DECOMPOSITION OF THE RIA SPECTRUM OF THE #GE FIBER
Ge-X Ge-STH Ge-A Ge-B
Spectral
position
2.61 ev
(~475 nm)
0.54 eV
(~2300 nm)
0.98 eV
(~1265 nm)
1.35 eV
(~918 nm)
FWHM
0.82 eV 0.3 eV 0.29 eV 0.34 eV


B. Radiation-induced attenuation in the #P sample
Griscom et al. have performed in [19] a thorough
study on the nature of the phosphorus color centers
together with their optical properties. In their study,
Griscom et al. have performed optical measurements in
two types of sample. The first one is a bulk sample glass
of 9.5 P2O5-90.5 SiO2. This sample has been used for
optical measurements in the 2-6 eV spectral range. The
second sample is a P-doped fiber with a slight boron (B)
doping for which optical measurements has been
performed in the 0.7-1.1 eV range. Griscom et al. have
decomposed the absorption spectra of these two types of
sample into absorption bands. After that, the evolutions
of the heights of these bands during isochronal annealing
have been correlated with ESR measurements giving the
point defect related to each band. Griscom et al. have
decomposed the RIA spectra of the fiber sample with the
help of three bands. A band peaking at 0.79 eV
(FWHM=0.29 eV) has been ascribed to the P1 related
defect [19, 20], the phosphorus equivalent of the SiE’
center. The height of an other band peaking at 1.07 eV
(FWHM=0.32 eV) was virtually constant. This second
band is possibly due to the P2 or P4 defects, as these
two bands are also constants in the same temperature
range of isochronal annealing. A third band centered at
0.45 eV (FWHM=0.8 eV) was necessary to correlate
ESR and optical measurements. However, this third band
has not been observed in an other paper in which optical
measurements have been performed at far infrared
wavelengths in a fiber containing only P [21]. Moreover
it is known that boron induces a wide absorption band at
energies shorter than ~0.9 eV [22]. The RIA spectra of
the bulk sample have been found to mainly consist of
absorption bands due to P2, P4 and phosphorus-hole
centers (POHC) defects. The structure of the POHC
defect is rather near of that of the Si-NBOHC and Ge-
NBOHC defects [19, 23]. We have tried to use the
absorption bands of P defects reported by Griscom et al.
in order to decompose the RIA spectra of the #P fiber.
The best fit has been obtained by using the P1 band, the
band peaking at 1.07 eV (possibly due to optical
transitions of the P2 or P4 defects) and a band centered
at 2.5 eV (FWHM=0.63 eV) associated with POHC
defects. However these three bands were not sufficient to
fit our whole RIA spectra. The addition of a fourth band
peaking at 1.43 eV (FWHM=0.41 eV) was necessary to
perform a good fit. The optical characteristics used for
the P1 band (Ec=0.78 eV, w=0.28 eV) are very slightly
different from that of those reported by Griscom et al..
With this set of four bands, whose characteristics are
summarized in Table III, it is possible to decompose the RIA
spectra of the #P fiber for times ranging from ~0.1 s to
~1000 s. Griscom et al. have not performed measurements in
the ~1.1-2eV energy range, this could explain why the band
peaking at ~1.43 eV has not been observed by these authors.

0.60.81.01.21.41.61.82.
102
103
104
0
200015001000500
POHC
Griscom
JAP 1983
P2 or P4 ?
Griscom
JAP 1983
unknown band
Ec=1.43 eV
w=0.41 eV
RIA (dB km-1)
Energy (eV)
P1
Griscom
JAP 1983
1 s
Fig. 5. Spectral decomposition of the attenuation (RIA) induced in the P-
doped fiber #P at 1 s after an X-ray pulse of 22.6 krad. The experimental
curves have been fitted using the Gaussian bands given in Table III.

Fig. 5 gives the RIA curve of the #P fiber recorded 1 s
after an X-ray pulse of 22.6 krad together with the fit
achieved by using the four Gaussian bands presented
discussed earlier. This figure suggests that the RIA at 1550
nm (0.8 eV) is mainly due to the P1 band, while the RIA at
1310 nm (0.95 eV) mainly results from the P1 band with the
band peaking at 1.07 eV. The RIA at 1550 nm increases from
shorter to longer times after the pulse (see Fig. 2). This result
suggests that a part of the P1 defects are not created during
the irradiation pulse, but results from the conversion of a
defect into P1 defects. Griscom et al. has proposed the
conversion of POHCs into P1 defects during isochronal
annealing experiments. Fig. 6 shows the time evolution of the
loss ratio between the RIA (dB m-1 rad-1) at 1310 nm and
1550 nm (laser diode measurements). The ratio decreases
with time. This result suggests that the contribution of the
band peaking at 1.07 eV to the RIA at 1310 nm decreases
with time.

TABLE III. ABSORPTION BANDS USED FOR THE
DECOMPOSITION OF THE RIA SPECTRUM OF THE #P FIBER
P1 P2 or P4 POHC
Spectral
position
0.78 eV
(~1590 nm)
1.07 eV
(~1160 nm)
2.5 eV
(~510 nm)
1.43 eV
(~860 nm)
FWHM
0.28 eV 0.32eV 0.63 eV 0.41 eV

Page 5

1E-61E-51E-41E-30.01
0.66
0.68
0.70
0.72
0.74
0.76
0.78
0.80
RIA ratio (1310nm/1550nm)
Time after X-ray pulse

Fig. 6. Time evolution of the radiation-induced attenuation loss
ratio (1.31/1.55 lm) after pulsed X-ray irradiation for the #P fiber.

C. Radiation-induced attenuation in the #Ge/P sample
At several times after the pulse Fig. 7 compares the RIA
for the three samples. From 0.1 s to 10 s after the pulse
and for wavelengths shorter than ~1200 nm, the #Ge/P
fiber exhibits RIA lower RIA than that the #Ge fiber.
Thus, P addition seems to reduce the radiation sensitivity
of a Ge doped fiber in this range of wavelengths. This
result seems consistent with results obtained by Friebele
et al. in [14]. These authors have shown that the addition
of P in the core of a fiber doped with Ge and B decreases
the RIA in the visible range. To explain such a behavior,
Friebele et al. have proposed that P addition provides
deep alternate trapping sites that compete with the Ge
trapping sites. P addition in a Ge-doped fiber also
reduced the RIA at 1310 nm and 1550 nm for shorter
times after the pulse (see Fig. 2). (This result confirms
measurements previously performed on the same type of
fiber but with an other test-bench [5].) West has
favoured in [24] the hypothesis that the improve
resistivity for short times of fiber containing P may be
due to the reorientation of GeO4 tetrahedrons into GeO6
groups [25]. The power of light propagating into the
fiber cladding depends on the wavelength. Thus, for the
#Ge/P fiber it seems difficult to decompose the RIA
spectra into Gaussian bands since, at a given wavelength,
the proportion of power of light is differently “seen” by
the Ge-related defects (mainly in the fiber core) and the
P-related defects (in the fiber cladding). For the #Ge/P,
around 90% of the power of the light propagates into the
fiber core at 1310 nm (standard Telecom fiber).
However, the RIA at 1310 nm of the #Ge/P fiber is
around two times lower at 10-2 s than that of the #Ge
sample (Fig. 2). Thus, we suggest that P doping during
the fabrication process of the fiber may rather decreases
the number of precursors of the Ge-related defects.

500 700 900
Wavelength (nm)
1100 130015001700 1900
10
-4
10
-3
10
-2
10
-1
(iv) (iii)
(ii)

(i ) #Ge (50 m, 9 krad) (iii) #Ge/P (50 m, 11 krad)
(ii) #P (10m, 22.6 krad) (iv) #Ge/P (100 m, 12.6 krad)
10
-4
10
-3
10
-2
10
-1
(i) #Ge (50 m, 9 krad)
(i)
(i)
(ii) #P (10m, 22.6 krad)
(iii) #Ge/P (50 m, 11 krad)
(ii)

100 s after X-ray pulse
10 s after X-ray pulse
10
-4
10
-3
10
-2
10
-1
(i) #Ge (50 m, 9 krad)
(i)
(ii) #P (10m, 22.6 krad)
(iii) #Ge/P (50 m, 11 krad)
(iii)
(iii)
(ii)

1 s after X-ray pulse
10
-4
10
-3
10
-2
10
-1
(iii)
(ii)

(ii) #P (10m, 22.6 krad)
(iii) #Ge/P (50 m, 11 krad)
0.1 s after X-ray pulse
RIA (dB km
-1 rad
-1)
(i)
(i) #Ge (50 m, 9 krad)

Fig. 7. Comparison of the spectral variation of radiation-induced attenuation
(RIA) in the three tested single-mode fibers at 0.1 s, 1 s, 10 s and 100 s after
X-ray pulses. The RIA spectra, normalized with respect to the deposited
dose, have been obtained at room temperature.

IV. CONCLUSION
In this paper we have studied the changes of optical
absorption at infrared wavelengths induced by pulsed X-ray
irradiation on single mode fibers doped with germanium or
phosphorus. We have tried to identify the absorption bands
responsible for the induced attenuation by performing in-situ
spectral measurements using a Super-Continuum source. We
have shown that the infrared induced-loses are mainly due to
infrared absorption bands. To decompose the IR spectra of
the Ge-doped fiber it has been necessary to use two un-
reported absorption bands. The RIA spectra of the P-doped
fiber was found to mainly result from the phosphorus
absorption bands previously reported by Griscom et al. with
an un-reported absorption band peaking around 1.43 eV
(FWHM=0.41 eV) in the P doped fiber. The addition of
phosphorus in the cladding of a fiber Ge-doped in its core
leads to a decrease of attenuation for shorter times after the
pulse. The previous hypothesis proposed in literature to