Optical Fibre Technology Centre and Department of Physical Chemistry, University of Sydney, Sydney, NSW 2006, Australia. Received November 12, 1993.
August 1, 1994 / Vol. 19, No. 15 / OPTICS LETTERS
1119
Spatial distribution of 650-nm luminescence in UV-processed germanosilicate preforms J. Canning and M. G. Sceats Optical Fibre Technology Centre and Department of Physical Chemistry, University of Sydney, Sydney, NSW 2006, Australia Received November 12, 1993 Broadband luminescence at -650 nm obtained on excitation at 633 and 647.5 rm is shown to be markedly increased in a UV-processed germanosilicate preform. The luminescence profile across the core is found to follow that of the germanium concentration.
Photosensitivity in optical fibers is of prime importance in the creation of all-fiber optical devices, the most obvious example being narrow-linewidth filters that use periodic phase structures written interferometrically with UV light at -240 nm.' Such devices have been used as the feedback reflectors in Er 3 +-doped optical fibers. 2
To date, practical de-
vices have used Ge-doped silica fibers, the presence of Ge increasing significantly the number of oxygendeficient centers identified to be the starting point for the photolytic reaction that gives rise to changes in the refractive index. It is especially important therefore that some understanding of these processes be gained, as there is ample evidence to suggest that the techniques are far from optimum. For example, it has recently been demonstrated that H2 processing before photolysis greatly enhances the rate and size of refractive-index changes able to be achieved through the formation of SiOH and GeOH and possibly of GeH and SiH.3 Changes of the order of 6 x 10for low-concentration (3 wt.%) GeO2 fibers have been reported.4 There is little doubt that the photolytic process is triggered by the excitation of oxygen-deficient defects that absorb near 240 nm, leading to a rise in the refractive index through the Kramers-Kronig relationship5 as well as a restructuring of the local environment through stress relaxation. The actual steps involved in the subsequent mechanism are less clear. This band at 240 nm, which follows the Ge radial profile, is assigned to the wrong bond Ge-X (X = Ge or Si).6 These bonds are photoionized to give GeE' centers that release a large number of local electrons, which, together with some structural rearrangement, trigger a range of processes not fully understood. That these electrons are free, possibly to wander some distance from their source, is supported by the dramatic increase in the rate of index change with the use of an electron acceptor such as B.7 Structural rearrangement is also believed to prevent the recombination of the resultant Ge and X radicals, and as a consequence a diradical species [Ge&X-assigned to the Ge drawing-induced defect (Ge-DID)8 ] forms. We have observed that when a UV-photolyzed fiber is irradiated by visible light an intense red luminescence 0146-9592/94/151119-03$6.00/0
centered at 650 nm is produced. This luminescence is assigned to the Ge-DID diradical.7 Cathodoluminescence of unprocessed modified-chemical-vapordeposition fibers has shown that the profile of this emission does not follow the Ge profile but rather exhibits a peak at the core-cladding interface,9" 0 and evidence suggests that it even lies outside this interface.'0 This is explained as trapping of the defects in the preform and fiber fabrication stage by thermal quenching." However, to our knowledge there has been no characterization reported of photolytically processed fibers or fiber preforms. Here we present what are to our knowledge the first such results, using excitation at 633 nm to observe the photoluminescent spectrum of both a UV-processed and an unprocessed fiber preform. The spectrum's profile across the core is also obtained and compared with the Ge profile reflected in the refractiveindex profile. An intense red emission was observed by visible laser excitation from lengths of Ge-doped silica fiber that had been processed with UV light. To investigate this further, we carried out studies on germanosilicate preforms. Two preform slices, -6 mm thick, were obtained from the same preform frabricated by the modified-chemical-vapordeposition technique (-12 wt. % Ge in the core). Figure 1 shows the refractive-index profile of the preform, which maps the Ge profile. One of these slices was photolytically treated with 240-nm light from an excimer laser incident upon the polished surface that uniformly excited the core region (30 min, 10-ns pulse, 10-kHz repetition rate, 50-mJ/cm2 peak power density). The emission spectra of Fig. 2 were obtained with a Raman microscope (Renishaw Ramascope) with an excitation source at 633 nm for both preform slices, with the microscope focused a few mi-
crometers below the surface in an area of -100 ,ttm2 . In the core of the unprocessed preform the visible emission from the Ge-DID's was sufficiently weak that the characteristic Raman spectrum of germanosilicate glass was observed [Fig. 2(a)]. On the other hand, for the processed fiber the red spectrum peaking at 650 nm attributed to the Ge-DID was observed [Fig. 2(b)] with an intensity several orders of magnitude larger than the germanosilicate © 1994 Optical Society of America
OPTICS LETTERS / Vol. 19, No. 15 / August 1, 1994
1120
Fig. 1. Refractive-index profile of the germanosilicate preform. wavelength(nm) 562 9000
633
595
677
725
that produced an approximately collimated beam of -120-Am diameter that passed through the core center. A collection lens focused an enlarged image normal to the entrance slit of a monochromator (0.01-nm resolution) to give a spatial filter of 45-tm resolution across the 400-,tm core. A mechanical translator was used to scan the image across the entrance slit to produce an intensity profile I(z) of the emission, where z is the propagation axis of the excitation laser. Figure 4 shows the typical profile obtained with the monochromator fixed at 650 nm. The slight asymmetry of the scan was associated with the weak divergence of the laser beam within the preform. The beam diameter is not insignificant compared with the core dimensions, so the scan I(z) profile of Fig. 4 is not a measure of the defect profile P(r) but is related to it by convolution:
I(z) =
(a) .
8000 7000 2
>, 5000 e
4000 3000 2000 1 000 0
P(r)I(z,y) - dr,
(1)
y
where I(z, y) is the Gaussian profile of the laser beam. The expected convolution profile is also shown in Fig. 4, and the good agreement demonstrates that the Ge-DID profile induced by UV photolysis followsthe Ge profile. It was noted that when a profile was taken farther down the core (roughly a
6000
.S
f
to computer
-2000
0
-1000
1 000
2000
wavenumber (cm-1)
doublemonochromalor
547.5nmKr-Ionline
j
fIAT
wavelength (nm)
562
595
633
677
725
focusinglens I
XYZposllioner
:t, q
steppingmotor
Fig. 3. Experimental setup for profile measurements across a preform core. PMT, photomultiplier tube. 0 wavenumber (cm-I)
core.
profile of beer end refractve-index
'! 4 . . I..tl:. I N
3000 0
rIt", .
2500
p.
Raman intensity. In the silica cladding, which was also irradiated, there was a small increase in the intensity that could be attributed to the production of Si-DID defects.
Visual observations of the emission profile under the Raman microscope indicated that the emission of the Ge-DID's followed the Ge profile of Fig. 1
rather than the cathodoluminescence profile of thermally produced Ge-DID's. This radial profile was quantitatively investigated with a different fluorometer
(Fig. 3). The probe was a Kr+-ion laser
profile.
J: -.: 494'..,' "..
Fig. 2. (a) Spectrum of an unprocessed germanosilicate preform core, (b) spectrum of a processed germanosilicate preform
*expermental --convoluo
3500 F
a
I I 16A. i I
2000
,I
i 1500 1000
500_ -400
't
-300 -200 -.1 )0
0
1(00
200
300
400
Position ainm)
Fig. 4. Luminescence profile across the UY-processed germanosilicate core. The line scan was taken just below the surface.
August 1, 1994 / Vol. 19, No. 15 / OPTICS LETTERS '30 120 110 100 :i
so 90
p.
80
. I
70 60 50 40
E
S,
.11 7-11 ,111
row1M
30
20 L
*
0-0-0-0
0
0 100 -600 -400 -300 -200 -100 position (jpn)
2200 3300 400 400
Line scan of UV-processed germanosilicate fiber prefor m 3-4 mm from the surface. Fig. 5.
few millimeters), the intensity dropped significantly and the line shape approached two convoluted peaks at and slightly beyond the core-cladding interface ends (Fig. 5). Again the asymmetry of the line scan is a consequence of the slight beam divergence of the laser within the preform. This agrees with the results of Atkins et al.9 and Williams et al.' 0 and is further evidence for the weak penetration into the glass by UV light at 240 nm. We have demonstrated an increase of several orders of magnitude in the red luminescence associated with the Ge-DID's after UV processing. These defects have also been shown to follow the Ge con-
centration profile, unlike the results reported for unprocessed preforms using cathodoluminescence. However, our results away from the UV-affected region of the core are in qualitative agreement with a nonlinear dependence on Ge concentration. This is expected because the linear UV absorption follows the Ge profile, whereas for unprocessed preforms the thermal effects, dependent on the glass viscosity, during fabrication play a significant role in generating a nonuniform distribution of defects. The linear relationship of the convolution with the Ge concentration profile appears to give further credence to the presence of Ge-Si wrong bonds, 7 as opposed to a
quadratic dependence for Ge-Ge bonds. This is not expected on thermodynamic grounds.' 2
1121
It is clear that an understanding of the mechanism involved in defect formation is essential for tailoring future designs of photosensitive fiber. For example, the red luminescence can be expected to be less for a H2 treated fiber, since the H2 will attach itself to the negatively charged GeE' centers, preventing the formation of diradical species. This will further complicate any direct relationship between the red luminescence and the wrong bond photoionization. We gratefully
thank Renishaw
PLC for the use
of their Ramascope, Renishaw representative Brian Smith for his assistance, Ron Bailey and Andrew Higley for fabricating the preform, and Jocelyn Lauzon for the UV processing.
This research was carried
out under the auspices of the Australian Photonics Cooperative Research Centre.
References 1. G. Meltz, W. W. Morey, and W. H. Glenn, Opt. Lett. 14, 823 (1989).
2. J. L. Zyskind, V. Mizrahi, D. J. DiGiovanni, and J. W. Silbert, Electron. Lett. 28, 1385 (1992). 3. A. Iino, M. Kuwabara, and K. Kokura, J. Lightwave Technol. 3, 1675 (1990).
4. P. J. Lemaire, R. M. Atkins, V. Mizrahi, and W. A. Reed, Electron. Lett. 29, 1191 (1993). 5. D. P. Hand and P. St. J. Russell, Opt. Lett. 15, 102 (1990). 6. M. G. Sceats, G. R. Atkins, and S. B. Poole, Annu. Rev. Mater. Sci. 28, 381 (1992).
7. D. L. Williams, B. J. Ainslie, J. R. Armitage, and R. Kashyap, Electron. Lett. 29, 45 (1993). 8. H. Kawazoe, Y. Watanabe, K. Shibuya, and K. Muta, Mater. Res. Soc. Symp. Proc. 61, 350 (1986). 9. G. R. Atkins,
S. B. Poole, M. G. Sceats,
H. W.
Simmons, and C. E. Nockolds, IEEE Photon. Technol. Lett. 4, 41 (1992).
10. D. L. Williams, M. J. Wilson, and B. J. Ainslie, Electron. Lett. 28, 1744 (1992). 11. M. G. Sceats, Workshop on Materials Science of Glasses (Australian Photonics Cooperative Research Centre, University of Sydney, Sydney, Australia, 1993).
12. C. A. Angell, Department of Chemistry, Arizona State University, Tempe, Ariz. 85287 (personal communication, 1993).
