1126 OPTICS LETTERS / Vol. 18, No. 14 / July 15, 1993

Dichroism in the absorption spectrum of photobleachedion-implanted silica

J. Albert, B. Malo, D. C. Johnson, F. Bilodeau, and K. 0. Hill

Communications Research Center, P.O. Box 11490, Station H, Ottawa, Ontario K2H 8S2, Canada

J. L. Brebner and G. Kajrys

Groupe de Recherche en Physique et Technologie des Couches Minces, D6partement de Physique,Universit6 de Montr6al, Montr6al, Qu6bec H3C 317, Canada

Received January 21, 1993

The optical absorption spectrum of ion-implanted silica is shown to become dichroic when it is photobleachedwith polarized 248-nm light. The dichroism is modeled by a photobleaching process that bleaches color centerspreferentially depending on their local orientation with respect to the polarization of the photobleaching light.A similar photobleaching process may occur in photosensitive fiber, thus providing a possible explanation for thebirefringence that is observed in the photoinduced index change.

Photosensitivity in optical fibers' is an interestingnonlinear optical phenomenon that has many po-tential applications.2 An unusual characteristic ofoptical fiber photosensitivity is that the photoinducedrefractive index is anisotropic,3 i.e., the photoinducedrefractive-index change is birefringent.4 Photosen-sitivity has also been detected in silica glass that hasbeen implanted with germanium and silicon ions.5 6

Although the photosensitivity in ion-implanted silicahas slightly different characteristics from that inoptical fibers, it is believed that the two phenomenaare related: In both cases that photosensitivity isassociated with the bleaching of UV color centers near240 nm, and both types of center are thought to arisefrom oxygen vacancies. The main difference is thatbleaching increases the index in fibers as a resultof an associated increase of absorption in adjacentUV bands and reduces the index in implanted silica,in which all UV absorption bands decrease afterbleaching. The advantage of studying photosensitiv-ity in ion-implanted silica is that, unlike for opticalfibers, the change in the optical absorption of theion-implanted layer in the UV spectral region that isdue to irradiation with photobleaching light is readilydetectable. In this Letter we show that photobleach-ing of ion-implanted silica glass with polarized lightresults in a dichroic optical absorption spectrum.

Samples of silica were implanted with germaniumions as described in Ref. 5: 1-in- (2.54-cm-) diameterdisks of Suprasil 2 synthetic fused silica (availablefrom Heraeus Amersil, Inc.) were implanted at 5MeV with 1013 germanium ions/cm2 over an areaof -2 cm2 . As a result of the implantation, strongUV absorption bands appear near 212 and 244 nmowing to color centers created by displacement dam-age in the silica. The damage occurs irrespective ofthe nature of the implanted ions, and the numberof color centers created is greater for heavy ionsand/or higher energies. In the conditions used here,the damage extends to a depth of 3.7 /utm from the

surface. Since the measurements presented in thisLetter consist of integrated values over the thicknessof the implanted layer, it is not necessary to take intoaccount the fact that the damage, and consequentlythe density of the color centers, is not uniformlydistributed in depth.

The 244-nm band is photobleached permanently byirradiation from a pulsed KrF excimer laser operatingat 248 nm.5'6 In this experiment, we carried outthe photobleaching of the band by using a linearlypolarized light beam from the excimer laser inci-dent normal to the sample surface with a fluenceof 20 mJ/(pulse cm2) at 2 pulses/s. We carried outabsorbance measurements with a polarized probebeam over the wavelength range 230-400 nm, us-ing a Cary 3 spectrophotometer equipped with apolarizer. The absorbance of an irradiated sampleis measured with the probe light polarized eitherparallel or perpendicular to the direction of polariza-tion of the bleaching laser beam. Figure 1(a) plotsthe change in absorbance at 244 nm for the twoorthogonal polarization states of the probe beam as afunction of the cumulative UV irradiation dose. Forprobe light polarized parallel to the polarization ofthe photobleaching light, the absorbance change thatwe measure is greater than for perpendicularly polar-ized probe light, i.e., the optical absorbance becomedichroic. The dispersion of the absorbance changeafter the final bleaching is shown in Fig. 1(b).

The experimental results show that the op-tical transmission of a germanium-ion-implantedsilica sample becomes dichroic when the sampleis photobleached with high-intensity polarized UVlight. This dichroism can be explained in terms ofpreferential photobleaching of the absorbing centersin the glass. We assume that the color centersabsorb like dipoles that are randomly orientedthroughout the implanted region of the glass. Thatis, the probability of absorption of a polarized UVphoton by a color center depends on the orientation

0146-9592/93/141126-03$6.00/0 © 1993 Optical Society of America

July 15, 1993 / Vol. 18, No. 14 / OPTICS LETTERS 1127

0.00m-o

E -0.08

. -0.1 6(D0C

-o0 -0.24.0

Ce -0.32a,aC

0

-0.40

0.1

m 0.0

arU

a -0.10-

0.0

c -0.2

C

= -0.3

-0.4 '230

0 3 6 9 12Cumulative UV dose (Joules/cM2)

(a)

264 298 332Wavelength (nm)

(b)

15

366 400

Fig. 1. (a) Change in absorbance of a germanium-ion-implanted silica layer as a function of the 248-nm irra-diation dose. The absorbance changes are for the twopolarization states measured at a wavelength of 244 nm.(b) Dispersion of the absorbance change with probingwavelength after the final bleaching.

of the color-center dipole axis with respect to thedirection of polarization of the UV light. In thecase of a color center's having a dipole absorptiontransition, the probability of absorption of a photonat the resonant wavelength varies as cos2 0, where6 is the angle between the dipole axis and thedirection of polarization of the light beam. Theselective photobleaching of color centers results inthe transmission of the photobleached sample'sbecoming dichroic, as in the case of crystalline7 andoriented polymer8 materials.

The above photobleaching process can be modeledmathematically as follows. Figure 2 shows the coor-dinate system used in the modeling. The coordinateaxes have been selected so that the polarization direc-tion for the photobleaching light is along the z axisand the beam propagation direction is along the xaxis normal to the sample surface. Thus the y axislies in the sample plane. The angular orientationof an absorbing dipole is specified by its coordinates(0, 4'), where 0 is the angle between the direction ofpolarization of the photobleaching light (z axis) andthe axis of the absorbing dipole and 0 is the anglebetween the direction of incidence of the light and theprojection in the (x, y) plane of the axis of the dipole.The probability per unit time that a polarized UV

photon from a light beam with intensity I will beabsorbed by a color center having orientation (0, 'k)is

P(0, I) = AoI cos2 6. (1)

The constant AO is a proportionality constant (insquare centimeters per joule) that gives a measureof strength of the absorbing transition. We are as-suming here that the absorption is linear, i.e., theprobability for absorption is a one-photon process.Let N(6, 0, t) represent at time t the density perunit volume of color centers with their axes orientedalong (0, 4'). Initially, at time t = 0, the densityN(6, X', 0) = No is assumed to be independent of(0, 0) because the dipoles are randomly oriented.Irradiation of the sample for a time tb with UV lightat an intensity Ib photobleaches some of the colorcenters. The absorption of a photon erases one colorcenter. Then the rate of change in the density ofcolor centers with orientation (0, 4) is given by

dN(6, X, t) = -N(, 4, t)P(0, Ib).

dt

This equation is solved at t = tb to yield

N(6, 0', tb) = No exp(-AoIbtb COS2 0).

(2)

(3)

In going from Eq. (2) to Eq. (3), we assume that thebleaching light intensity does not decrease with time.This is valid as long as the changes in absorptionduring bleaching remain small. After the samplehas received a radiation dose of Ibtb J/cm

2, prefer-

ential bleaching of the color centers occurs, and thecolor-center density as a function of angle is given byEq. (3).

The transmission of a polarized low-intensity probebeam through the sample will now be dichroic, i.e., itwill depend on the polarization of the probe beam.In this case, the probe light intensity is so low thatfurther changes in N(6, X, t) through Eq. (3) are neg-ligible. The linear dichroic absorption coefficientscan be calculated as follows. Assume that a probebeam with low intensity I, is propagating along thex axis with its polarization parallel to the polarizationof the photobleaching light, i.e., polarized along the

z

dipole axis

00 I

Fig. 2. Coordinate system used in modeling the photo-bleaching process. The bleaching and probe lights areincident along x, and the implanted surface lies in thez-y plane.

1128 OPTICS LETTERS / Vol. 18, No. 14 / July 15, 1993

z axis. The intensity of polarized probe light ab-sorbed by a dipole is proportional to the square ofthe projection of the dipole axis on the polarizationaxis of the probe light.' 0 For light polarized alongz, parallel to the polarization of the bleaching light,the projection is proportional to cos(6); for light po-larized along y, perpendicular to the polarization ofthe bleaching light, the projection is proportional tosin(6)sin(o). The total absorbed light in each caseis obtained by integrating over all possible dipoleorientations. Therefore

I7b,(tb) =Co |2 d f 2 sin a d 6 No COS a

X exp(-AoIbtb cos2 0), (4)

I =(tb) Co f do sin 0 dO No sin2 0

x sin2 0 exp(-Aolbtb cos 6) , (5)

where C0 is a proportionality constant. Equations (4)and (5) depend only on two parameters: AO andan amplitude factor given by CoNO. Therefore weneed to use only two data points from Fig. 1(a) tomodel the process. We chose to use the values ofparallel and perpendicular absorption at the largestdose, obtaining AO = 1.434 cm2 /J and CoNo = 0.0377.With these values, we used Eqs. (4) and (5) to plotthe solid curves in Fig. 1(a). The fact that boththeoretical curves predict the correct behavior nearthe origin supports the validity of the model. At highdoses, the model also predicts that the dichroism willdisappear.

The observation of photoinduced dichroism ingermanium-ion-implanted silica provides a possibleexplanation for the photoinduced birefringence thathas been observed in optical fibers.3 4 The expla-nation is as follows. Photobleaching of the UV ab-sorption bands results through the Kramers-Kronigcausality relationship in a refractive-index change.The change in refractive index extends to allwavelengths on the long-wavelength side of theabsorption bands. If photobleaching with polarizedlight results in an absorption change that dependson polarization (i.e., dichroism), the resulting pho-toinduced refractive-index change will also dependon polarization, i.e., exhibit birefringence. Thismodel is consistent with the observation in opticalfibers that the polarization of the photobleachinglight determines the orientation of the photoinducedbirefringent axes. It is also consistent with theobservation3 that for an optical fiber under prolongedexposure to polarized UV light the birefringencewill initially increase and then decrease to zero.Furthermore, the basic assumption used to explainour measurements, i.e., that the color centers absorb

as dipoles and that bleaching with polarized lightrearranges the distribution of the dipoles withorientation, is consistent with the model of Anand Sipe" for polarization effects in photosensitivityand has recently been proposed as the mechanismresponsible for photoinduced birefringence in opticalfibers. 12,13

Further experiments are needed to verify this ex-planation for the origin of the photoinduced birefrin-gence. In the case of the germanium-ion-implantedlayer in silica, the photoinduced dichroism shouldresult in a photoinduced birefringence. Owing to thesmall thicknesses of the ion-implanted layers, themagnitude of the photoinduced birefringence is toosmall to be observed with the measurement tech-niques available to us. In the case of optical fibers,the photobleaching of the UV absorption bands in thecore of the optical fiber has been detected directly."4Thus it would be interesting to look for photoinduceddichroism in the UV absorption bands of an opticalfiber irradiated with polarized UV light.

In conclusion, we have observed photoinduceddichroism in germanium-ion-implanted silica sam-ples that have been irradiated with polarized UVlight. The result has led us to propose photoinduceddichroism as the origin of the photoinduced birefrin-gence that has been observed in optical fibers.

References

1. K. 0. Hill, Y. Fujii, D. C. Johnson, and B. S. Kawasaki,Appl. Phys. Lett. 32, 647 (1978).

2. K. 0. Hill, B. Malo, F. Bilodeau, and D. C. Johnson,Annu. Rev. Mater. Sci. 23, 125 (1993).

3. M. Parent, J. Bures, S. Lacroix, and J. Lapierre, Appl.Opt. 24, 354 (1985).

4. K. 0. Hill, B. Malo, K. A. Vineberg, F. Bilodeau, D.C. Johnson, and I. Skinner, Electron. Lett. 26, 1270(1990).

5. J. Albert, K. 0. Hill, B. Malo, D. C. Johnson, J. L.Brebner, Y. B. Trudeau, and G. Kajrys, Appl. Phys.Lett. 32, 148 (1992).

6. J. Albert, B. Malo, D. C. Johnson, K. 0. Hill, J. L.Brebner, and R. Leonelli, Opt. Lett. 17, 1652 (1992).

7. I. Schneider, Phys. Rev. Lett. 16, 743 (1966).8. Y. Shi, W. H. Steier, L. Yu, M. Chen, and L. R. Dalton,

Proc. Soc. Photo-Opt. Instrum. Eng. 1559, 118 (1991).9. L. I. Schiff, Quantum Mechanics, 2nd ed. (McGraw-

Hill, New York, 1955), p. 253.10. D. S. Kliger, J. W. Lewis, and C. E. Randall, Polarized

Light in Optics and Spectroscopy (Academic, NewYork, 1990), p. 185.

11. S. An and J. E. Sipe, Opt. Lett. 17, 490 (1992).12. S. Bardal, A. Kamal, and P. St. J. Russell, Opt. Lett.

17, 411 (1992).13. J. Lauzon, D. Gagnon, S. LaRochelle, A. Blouin, and

F. Ouellette, Opt. Lett. 17, 1664 (1992).14. R. M. Atkins and V. Mizrahi, Electron. Lett. 28, 1743

(1992).