August 1, 1989 / Vol. 14, No. 15 / OPTICS LETTERS 823
Formation of Bragg gratings in optical fibers by a transverse
holographic method
G. Meltz, W. W. Morey, and W. H. Glenn
United Technologies Research Center, East Hartford, Connecticut 06108
Received February 6, 1989; accepted April 28, 1989
Bragg gratings have been produced in germanosilicate optical fibers by exposing the core, through the side of the
cladding, to a coherent UV two-beam interference pattern with a wavelength selected to lie in the oxygen-vacancy
defect band of germania, near 244 nm. Fractional index perturbations of approximately 3 X 10-5 have been written
in a 4.4-mm length of the core with a 5-min exposure. The Bragg filters formed by this new technique had
reflectivities of 50-55% and spectral widths, at half-maximum, of 42 GHz.
In 1978, Hill et al. 1' 2 reported the formation of refrac-
tive-index gratings in germanosilicate fiber by sus-
tained exposure of the core to the interference pattern
of oppositely propagating modes of 488- or 514.5-nm
argon-ion laser radiation. Subsequent investigations
by Lam and Garside3 showed that the grating strength
increased as the square of the writing power, which
suggested a two-photon process as the cause of the
index changes. This Letter presents the first results
to our knowledge that show that in-fiber Bragg grat-
ings can also be formed by illuminating the core from
the side of the fiber with coherent UV radiation that
lies in the 244-nm germania oxygen-vacancy defect
band.4-6 This intense absorption band, which is -35
nm wide, coincides with the second harmonic of both
blue-green argon-ion laser lines used in previous re-
search.
The index modulation, which can be selected to
correspond to a desired Bragg wavelength, is written
within the core by exposing it to a two-beam interfer-
ence pattern. The grating period is determined by the
incident wavelength and the included angle between
the beams. This transverse holographic method of
forming gratings proves to be much more efficient and
flexible than the previously reported technique.
Gratings that are formed in this manner are not length
limited by saturation effects3 and can be tailored to a
desired transmission or reflection filter characteristic
by shaping and tilting the writing pattern through
control of the included angle and divergence of the
beams.
A grating is formed by exposing a short length of
bare, photosensitive, germanosilicate fiber to a pair of
overlapping coherent UV beams. The experimental
arrangement is shown in Fig. 1. A tunable excimer-
pumped dye laser, operated at a wavelength in the
range of 486-500 nm, is used with a frequency-dou-
bling crystal to provide a UV source that lies in the
244-nm band and has an adequate coherence length.
The UV radiation is split into two equal-intensity
beams and then recombined to produce an interfer-
ence pattern within the core, normal to the fiber axis.
The intensity of the pattern is increased by focusing
the beams on the fiber with a pair of cylindrical lenses.
The resulting focal spot is approximately rectangular,
approximately 4 mm long by 125 tm wide.
A filtered mercury arc source is used with a high-
resolution monochromator to measure the reflection
and transmission spectra of the grating. The reflect-
ed signal is monitored by inserting a beam splitter at
the fiber input, and the reflectivity is measured by
comparing the reflected signal level to the power re-
flected, at a wavelength near but out of the filter band,
from a mirror placed at the output end of the fiber.
The strongest gratings were written with 244-nm
pulsed radiation that had an average power of 4-20
mW. Several different fibers were used, with core
diameters of 2.2-2.6gum and N.A.'s of 0.17-0.24, corre-
sponding to GeO2 doping of 5-12.5 mol%. Bragg grat-
ings were formed with center wavelengths of 577-591
nm in (i) commercial (Spectran) 6.6-mol % germanosili-
cate-core, silica-clad fiber; (ii) fiber similar to that
used by Hill et al. 1' 2 ; (iii) elliptical-core, polarization-
MERCURY ARC LAMP
MICROSCOPE OBJ.
PHOTOSENSITIVE FIBER
YELLOW FILTER REFLECTOR
BRAGG GRATING >o
PMT ¢' I ENCLOSURE
BEAM SPLITTER
UV LASER BEAM-.- 1
XY RECORDER I
Fig. 1. Diagram of the experimental setup. A beam split-
ter (not shown) at the fiber input end is used with the
monochromator to measure the reflection spectrum of the
Bragg grating. PMT, photomultiplier tube.
0146-9592/89/150823-03$2.00/0 -© 1989 Optical Society of America
824 OPTICS LETTERS / Vol. 14, No. 15 / August 1, 1989
1.0 -
wJ
w
Cc,
| TRANSMISSION
FWHM 42 GHz
0.5 F
o _
575
VMIRROR IN
BACKGROUND
576 577
WAVELENGTH IN nm
Fig. 2. Transmission and reflection spectra for a 4.4-mm-
long Bragg grating filter. A 1-m narrow-band monochroma-
tor with a resolution of 0.02 nm was used with a filtered arc
lamp source to measure the in-fiber filter characteristics.
The measured FWHM is corrected for the monochromator
spectral response broadening.
maintaining fiber (Andrew); and (iv) in fiber with a
high germania content (N.A. = 0.24) containing a
small amount (0.5-1 mol %) of phosphorus.
As the periodic index modulation develops in the
fiber core, a narrow notch (or peak) appears in the
transmission (reflection) spectrum. The center of the
peak or notch occurs at the predicted Bragg wave-
length X = 2nA, where A is the grating period and n is
the mode index. Figure 2 shows the reflection and
complementary transmission spectra of a grating
formed in a 2.6-gim-diameter core, 6.6-mol % GeO2 -
doped fiber after 5-min exposure to a 244-nm interfer-
ence pattern with an average power of 18.5 mW. The
two spectra have similar line shapes and complemen-
tary values of transmittance and reflectance. The
length of the exposed region is estimated to be be-
tween 4.2 and 4.6 mm, as deduced from inspection of a
witness burn spot in a paper target. The FWHM of a
uniformly exposed region of this length should be
about 26 GHz (Refs. 3 and 7); however, the observed
linewidth shown in Fig. 2 is 42 GHz, suggesting that
the intensity pattern is tapered. The lack of pro-
nounced sidelobes also supports this conclusion.
The gratings are observed to form quickly at power
levels of 10 mW and higher. For example, after 10 sec
of exposure to an average pulse power of 23 mW the
measured transmittance at the center of the Bragg
filter decreases to 0.65, and after 30 sec of exposure it
decreases to 0.55. Exposure to the UV flux in some
fibers causes an immediate broadband drop in trans-
mission, which then gradually recovers. In the 6.6-
mol % GeO2-doped commercial fiber the transmission
returned to within 6% of its previous level within 1
min.
Because the spectral width of the grating filter is
narrow and the index perturbation extends across the
entire core cross section, it can be used to separate the
fundamental HE,, mode from the higher-order
modes. Figure 3 shows the measured transmission
spectrum of a slightly multimode fiber. The fiber
used in this experiment had a cutoff wavelength of 632
nm, corresponding to a N.A. of 0.22 (11 mol % GeO2
doping), and a core diameter of 2.2 gim. The Bragg
wavelength of the fundamental occurs at 581.5 nm.
At this wavelength, the value of the normalized fre-
quency V is 2.62; the fiber just supports the first set of
higher-order modes. Under these conditions the sep-
aration of phase indices, and therefore the Bragg
wavelengths, of the two modes is greatest and the
individual peaks in the spectrum are easily resolvable
by the in-fiber grating filter. The measured separa-
tion is within 10% of the predicted value as computed
from the Bragg condition XB = 2nA and the dispersion
relation for step-index fibers.8
The Bragg wavelengths for the principal modes in a
polarization-maintaining fiber will also be separated
by the difference in their axial wave numbers, or the
fiber birefringence. To show this, a weak grating was
written in a commercial elliptical-core, germania-
doped polarization-maintaining fiber (Andrew). The
transmitted line shape, measured without a polarizing
filter at the output of the fiber, consisted of the super-
position of two lines. By use of a polarizer at the
output, these lines could be identified with the princi-
pal horizontal and vertical modes of the fibers.
We can estimate the strength of the index perturba-
tion (An/n) by comparing the measured peak reflectiv-
ity of a grating of known length L with a prediction of
the efficiency of a volume hologram within the core of
the fiber. 3'7 It can be shown3 by solution of the cou-
pled mode equations for the forward- and backward-
traveling waves in a fiber containing a Bragg filter that
the reflectivity at the Bragg wavelength is given by
R = tanh2 Q, (1)
where
Q= irn(L/X)(An/n)rq(V). (2)
The factor ti(V) a 1 - 1/V2, V > 2.4, is the fraction of
the integrated fundamental mode intensity contained
in the core.
The measured peak reflectivities of two Bragg grat-
ing filters, written in different fibers with different
-J
z
I-c
> 0
r
_J
578.25 581.5
!- - 3.25 nm -
577 578 579 580
WAVELENGTH IN nm
581 582
Fig. 3. Transmission spectrum of a Bragg filter in a multi-
mode fiber. The fundamental mode i8 reflected by -30% at
a wavelength of 581.5 nm. The next set of higher-order
modes appears at a wavelength that is 3.25 nm shorter than
the notch at the fundamental.
August 1, 1989 / Vol. 14, No. 15 / OPTICS LETTERS 825
6 60 -
W
-J 40 -
LU-
a: 1 ~~~~~~X 1 0
20
0
0 2 4 6 8 10
LENGTH, mm
Fig. 4. Computed (solid curves) and measured reflectivity
for Bragg gratings of various strengths as a function of
length. Experimental points are shown for a grating written
with an average power of 18.5 mW at a wavelength of 244 nm
(filled square) and with an average power of 4.5 mW at a
wavelength of 257.3 nm (filled circle). Two different fibers
were used.
power levels, are compared in Fig. 4 with theoretical
predictions for gratings of various strengths as a func-
tion of the length of the exposed region. The weaker
7% reflectivity filter was written with a UV laser beam
at a wavelength of 257.3 nm, just on the edge of the
oxygen-vacancy defect absorption band, using pulses
from a mode-locked argon-ion laser and a KDP sec-
ond-harmonic generator. The core was exposed for 20
min to an average power of 4.5 mW. The fiber was
similar to those used by Hill et al.1,2 (core diameter 2.2
gim, N.A. = 0.22) in the first experiments on photore-
fractive effects in germanosilicate fiber. Much stron-
ger gratings of about the same length with reflectivi-
ties of 50-55% were written in commercial germanosi-
licate fiber (core diameter 2.61 gum, N.A. = 0.17) using
a 5-min exposure to a pulsed crossed-beam pattern
with an average power of 18.5 mW. In this case a dye
laser was used with a /3-BaB204 crystal to generate
second-harmonic UV radiation at 244 nm, which is
close to coinciding with the center of the defect ab-
sorption band. Based on the measured reflectivity,
the fractional index change is estimated to be 2.8-3 X
10-5, assuming a grating of uniform strength. The
peak index perturbation could be somewhat larger,
however, since the linewidth measurements are wider
than expected for a uniform grating.
We can compare the efficiency of writing in-fiber
gratings with coherent UV radiation at 244 nm to the
two-photon process at 514.5 nm. To obtain an index
perturbation of 3 X 10-5 using cw argon-ion radiation
at 514.5 nm required a writing power of 90.7 mW with
an exposure of approximately 6 min.3 This is equiva-
lent to exposing the core (diameter of 2.5 gim) to an
energy flux of 665 MJ/cm 2 . A grating of similar
strength is obtained with an energy flux of only 1 kJ/
cm2 at a wavelength of 244 nm by directly bleaching
the absorption band, an improvement of 6.7 X 105 in
writing efficiency!
The Bragg gratings formed by our holographic tech-
nique with 257- or 244-nm radiation appear to be per-
manent and stable at high temperatures. A grating in
the commercial fiber was heated to 5000C and main-
tained at that temperature for 18 h without a change in
its reflectivity or line shape. The only variation was
an expected shift in the line center due to the combi-
nation of thermal expansion and a change in refractive
index with temperature and stress relief.
The mechanism that forms the gratings is not fully
understood; all we can say is that it is related to the
bleaching of the oxygen-vacancy defect band in ger-
mania or germania-doped silica. Normally, germani-
um is incorporated in the silicate glass in the Ge+4
oxidation state, i.e., as GeO2 ; however, Ge+2 can occur
if GeO2 is dissociated into GeO and 02 in the formation
of the glass, say, during the preparation of a modified
chemical-vapor-deposition process preform.5'6 This
process, whereby the reduced Ge+2 species is formed,
is favored if the processing temperature is raised to
16500C and the molten glass is cooled quickly.9 Pre-
liminary measurements of the UV absorption spectra
of germanosilicate preforms suggest that the drawing
process can cause the Ge+2 defect band to form.
In summary, a new method for forming in-fiber
Bragg gratings has been demonstrated. The grating is
formed in photorefractive germanosilicate fiber by ex-
posure to a coherent two-beam UV interference pat-
tern. This technique provides a new means for mak-
ing quasi-distributed measurements of temperature
and strain by monitoring the shift in the Bragg wave-
length of the sensing regions, each being individually
tuned to a distinct wavelength, or by forming pairs of
independent Fabry-Perot cavities. Possible applica-
tions include high-efficiency distributed-feedback re-
flectors, wavelength-selective couplers and taps, and
dispersion-compensating filters.
We acknowledge the important contributions of R.
M. Elkow and J. D. Farina and the skillful technical
assistance of A. L. Wilson. We also thank R. A. Weeks
(Vanderbilt University), K. 0. Hill (Canadian Optical
Communications Research Center), and E. Snitzer
(Rutgers University) for valuable discussions of the
phenomenology. This research was supported in part
by the U.S. Department of the Air Force.
References
1. K. 0. Hill, Y. Fujii, D. C. Johnson, and B. S. Kawasaki,
Appl. Phys. Lett. 32, 647 (1978).
2. B. S. Kawasaki, K. 0. Hill, D. C. Johnson, and Y. Fujii,
Opt. Lett. 3,66 (1978).
3. D. K. W. Lam and B. K. Garside, Appl. Opt. 20, 440
(1981).
4. A. J. Cohen and H. L. Smith, J. Phys. Chem. Solids 7,301
(1958).
5. P. C. Schultz, in Proceedings of the Eleventh Interna-
tional Congress on Glass (North-Holland, Amsterdam,
1977), pp. 155-163.
6. M. J. Yuen, Appl. Opt. 21, 136 (1982).
7. H. Kogelnik, Bell Syst. Tech. J. 48, 2909 (1969).
8. A. W. Snyder and J. D. Love, Optical Waveguide Theory
(Chapman and Hall, New York, 1983).
9. J. M. Jackson, M. E. Wells, G. Kordas, D. L. Kinser, and
R. A. Wecks, J. Appl. Phys. 58, 2308 (1985).