532 OPTICS LETTERS / Vol. 14, No. 10 / May 15, 1989

Picosecond pump-probe interferometric measurement of opticalnonlinearities in channel waveguides

N. Finlayson, W. C. Banyai, C. T. Seaton, and G. I. Stegeman

Optical Sciences Center, University of Arizona, Tucson, Arizona 85721

M. O'Neill, T. J. Cullen, and C. N. Ironside

Department of Electronics and Electrical Engineering, University of Glasgow, Glasgow G12 8QQ, UK

Received August 25, 1988; accepted February 8, 1989

An interferometric technique for measuring with picosecond resolution the time evolution of the real and imaginarycomponents of optical nonlinearities in channel waveguides is described. Characteristics of the technique areillustrated with measurements of band-filling optical nonlinearities in CdS.Sel1-- doped glass channel waveguides.

There is currently considerable interest in the nonlin-ear optical properties of optical waveguides, particu-larly in materials that exhibit fast, subnanosecond re-sponse times.' The attraction of guided-wave config-urations for all-optical logic and switching devices isthat interaction lengths can be absorption limitedrather than diffraction limited, and thereby smallnonlinearities can be utilized. In order to achieveuseful switching or modulation ratios in devices suchas directional couplers or integrated-optical interfer-ometers, phase shifts of 360°-720° are required, withthroughput losses being minimized.2 Semiconductor-based materials, such as CdSxSel-.-doped glasses,3-5

are considered promising since they can exhibit largeoptical nonlinearities close to the fundamental ab-sorption edge. However, nonlinear changes in the re-fractive index and absorption in semiconductors satu-rate with increasing input fluence,6 limiting the maxi-mum phase shift obtainable in a given device. It istherefore important to establish the magnitude ofthese saturated quantities and their temporal evolu-tion. In this Letter we describe a hybrid Mach-Zehnder interferometer designed to measure thesequantities and illustrate its performance with mea-surements on semiconductor-doped glass waveguides.

The technique is a modified version of an approachdemonstrated by Halbout and Tang7 and recentlyadapted for use in optical fibers by Cotter et al.8 Theexperimental apparatus is shown in Fig. 1. A mode-locked, cavity-dumped dye laser generated 2-psec(FWHM) pulses at a repetition rate of 3.8 MHz. Mostof the pulse energy was used as a pump beam to excitethe nonlinearity. A small fraction of the light (<1%)was split from the laser pulse and routed through twoarms of a Mach-Zehnder interferometer, one arm ofwhich contained the channel waveguide. The pulse inthe waveguide arm of the interferometer (designatedthe signal pulse) was end-fire coupled to the quasi-TMmode of the guide. rThe transmitted signal pulse wasthen interfered with the pulse traveling through theother arm of the interferometer (the reference pulse),and the resulting fringe pattern was imaged onto a

computer-controlled television camera and digitizedby a personal computer. The time delay between thepump and the signal pulses was varied in 0.7-psecincrements from -100 to 150 psec. Fringe patternswere recorded at each value of time delay. The digi-tally recorded fringe patterns were analyzed using afast Fourier transform to obtain the fringe contrastand phase at each time delay.

The intensity distribution at the output of the inter-ferometer may be described approximately by theplane-wave interferometer equation,

(1)

where I, and I2 are the reference and the signal beamintensities, respectively. We do not consider here thevarious convolution effects involved in the experi-ment; these will be discussed in a future publication.The signal intensity 12 and the phase difference be-tween the signal and the reference is dependent on theabsorption a and the refractive index n of the wave-guide, respectively. The quantities a and n in turn

I YAG LASER

M

BS E BS rl |MONOCHROMATOR|

C P I

CPHW IAUTOCORRELATOR|HWP

PBSI

DL

PF'

POCKELSCELL

n, r f, n I 71U LjUP P

BS

\ U ILJ

VIDEO CAMERA

|COMPU TERI

SAMPLEA . AV - VL L

DL

M

P

M

Fig. 1. The experimental apparatus. M's, mirrors; BS's,beam splitters; PBS's, polarizing beam splitters; P's, polariz-ers; HWP, half-wave plate; DL's, delay lines; L's, lenses.

0146-9592/89/100532-03$2.00/0 © 1989 Optical Society of America

I = I, + I2(a) + 2C11 �12(a) cos[o(n)],

M-PI I

;M

May 15, 1989 / Vol. 14, No. 10 / OPTICS LETTERS 533

depend on the probe delay T and, for a semiconductor,on the local density of photoexcited carriers N createdby the pump pulse. Thus a and n are functions of thespatial coordinates in the waveguide. The observedphase shift Ak is related to the nonlinear refractive-index change An by

LAO = (27r/X) fo n(z)dz, (2)

where X is the free-space wavelength.waveguide is fully saturated, then

AOsat = (27r/X)AnsatL.

When the

set at 578 nm, where the linear absorption of the chan-nel waveguide was 7 cm-'. The waveguide was 6 mmin length, approximately 10 Am 2 in cross-sectionalarea, and supported a single mode of both polariza-tions. The waveguide had been previously exposed tolarge photon fluences in other experiments and wasphotodarkened as a result. 3

The waveguide transmission was measured by usingonly the pump beam to establish that saturation wasachieved, as shown in Fig. 2. Using linear absorptionand fluorescence data we estimated the band-gapwavelength to be 566 nm for the semiconductor-dopedglass. The waveguide transmission increased with the

The normalized transmission change of the waveguideis given by

AT/TO =T-TO

= I2(ae) -2(aOt)/I2(aO) (4)

for a constant signal intensity incident upon the wave-guide. T is the transmission of the sample, To is thelinear transmission, and ao is the corresponding linearabsorption. Since the fringe amplitude A is related toI2(a) through A2 = 41112(a), the normalized transmis-sion change can be written as

AT/TO = (A2 - A02)/AO2,

60

50

40z0C) 30

C!)Z: 20

(5)

where Ao is the fringe amplitude when the probe is wellseparated from the pump in time.

The remaining light in the original laser pulse wasused to excite the optical nonlinearities in the wave-guide, giving rise to a time-dependent change in 0 and12. A delay line was included in the path of this pulse,designated the pump pulse. The pump pulse wasend-fire coupled to the quasi-TE mode of the guide.Orthogonal polarizations were required to separatethe pump and the signal pulses because spatial separa-tion was not possible in the guided-wave geometry.The polarizer at the output of the waveguide was thenused to prevent the cross-polarized pump from ob-scuring the interferometer fringe pattern.

We illustrate the technique with measurements ofelectronic nonlinearities in CdSSel-.,-doped glasschannel waveguides. The physical origin of the non-linearities in the semiconductor-doped glasses is be-lieved to be dominated by band filling.3-5 Light inci-dent upon the semiconductor microcrystallites gener-ates free carriers, occupying states that are renderedunavailable for further transitions until the carriersrecombine. The absorption coefficient of the materi-al decreases, and there is a corresponding change inthe material dispersion with increasing light intensity.At sufficiently large fluences the states are effectivelyfilled to the photon energy, and no further changes inthe refractive index or absorption can take place (inthe absence of other nonlinear processes). The mate-rial is then in the saturation regime, and the value ofthe saturated index and absorption change is impor-tant for device design.

Fabrication of these waveguides has been describedin a previous publication.9 The laser wavelength was

A -

A---F _ - _ , -

-U-3-- -IL- - -u- - ii

0- O- - -o- -101-

oL_0.01

590 nm

AA

587.5 nm

Eu-U.-

A585 nm

580 nm .

A ,-

FLUENCE (mJ/Cm2)10

Fig. 2. Transmission changes of a semiconductor-dopedglass waveguide as a function of the optical fluence.

PROBE DELAY

* - -- ~ -_ tt 46.7 psecI I I

40 psecI 3 pse

33.3 psec

I 1 I,4

I I I I

I I

I I I I- I I II

- I I I I

26.7 psec

20 psec

13.3 psec

6.7 psec

0 psec

-6.7 psec

-13.3 psec

-180 0 180 360 540(P (degrees)

Fig. 3. Fringe pattern dependence on the time delay. Eachfringe is normalized to a peak value of unity.

A- -.4-&- --A- 'A - & -

I-`-�� �1�1 �, �

534 OPTICS LETTERS / Vol. 14, No. 10

135

Ona)a)

a) 90a)

1-D

- 45

< 0I

-50 -25 0 25 50 75PROBE DELAY (psec)

100 125

Fig. 4. Phase shift and the correspondingchange as a function of the probe delay.

20

15

01- 10

H-

0

3.61 X 1 ~

2.41

1.20 <

I

0.00

I -1.20

refractive-index

-_,-50 -25 0 25 50 75 100 125PROBE DELAY (psec)

Fig. 5. Transmission change as a function of the probedelay.

fluence, eventually saturating at less than 100% trans-mission owing to residual absorption. To measure thesaturated change in refractive index with the interfer-ometer, the pump fluence was set at 24 mJ/cm2 so thatthe waveguide was driven fully into the saturationregime.

The behavior of the fringes as a function of timedelay is shown in Fig. 3. A sharp movement of thefringe is observed in the vicinity of zero delay. Acorresponding increase in the fringe amplitude occurs.As the time delay is increased the free-carrier popula-tion in the semiconductor decreases owing to recombi-nation, and the fringes return to equilibrium. Thedetailed dependence of the nonlinear phase shift onthe probe delay is shown in Fig. 4. The increase in thephase angle near zero delay corresponds to the turn onof the waveguide as the pump pulse excites the materi-al. The peak phase change at this wavelength was135°, corresponding to a saturated-index change Ansatof -3.6 X 10-5. The sign of the refractive-index

change was established by an independent adjustmentof the reference path length. Exponential relaxationwith a time constant of approximately 19 psec wasdetermined by plotting the phase change on a logarith-mic scale.

The change in transmissioin, which we calculatedfrom the change in fringe amplitude using Eq. (5), isplotted in Fig. 5. The saturated transmission at thiswavelength is 20 times the linear transmission, inagreement with transmission changes measured di-rectly. When plotted as absorption, on a logarithmicscale, exponential decay was observed with a time con-stant of approximately 22 psec, close to that measuredfrom the phase changes. The rapid relaxation ob-served in the experiment is probably due to recombi-nation enhanced by the large density. of surface defectspresent in the glasses.

Such data can be used to interpret the switchingbehavior observed in directional couplers fabricated inthese glasses.10 In Ref. 10 we presented experimentaland theoretical evidence that the switching was domi-nated by absorptive rather than dispersive effects inthe semiconductor-doped glasses. Since the maxi-mum phase change measured here is only 135°, lessthan the 720° required for refractive-based switching,the interferometric results presented here confirmthese conclusions.

This research was supported by the National Sci-ence Foundation (EET-860-4374), the U.S. Army Re-search Office (DAAG-29-85-K-0173), and the OpticalCircuitry Cooperative in the United States and theScience and Engineering Research Council in theUnited Kingdom. Collaboration between the groupswas supported by a NATO travel grant. The authorsare indebted to the referees for their helpful sugges-tions.

References

1. S. R. Friberg and P. W. Smith, IEEE J. Quantum Elec-tron. QE-23, 2089 (1987).

2. G. I. Stegeman, E. M. Wright, N. Finlayson, R. Zanoni,and C. T. Seaton, IEEE J. Lightwave Technol. LT-6,953 (1988).

3. P. Roussignol, D. Ricard, J. Lukasik, and C. Flytzanis, J.Opt. Soc. Am. B 4, 5 (1987).

4. G. R. Olbright and N. Peyghambarian, Appl. Phys. Lett.48,1184 (1986).

5. C. N. Ironside, T. J. Cullen, B. S. Bhumbra, J. Bell, W. C.Banyai, N. Finlayson, C. T. Seaton, and G. I. Stegeman,J. Opt. Soc. Am. B 5, 492 (1988).

6. E. M. Wright, S. W. Koch, J. E. Ehrlich, C. T. Seaton,and G. I. Stegeman, Appl. Phys. Lett. 52, 2157 (1988).

7. J. M. Halbout and C. L. Tang, Appl. Phys. Lett. 40, 765(1982).

8. D. Cotter, B. J. Ainslie, H. P. Girdlestone, and C. N.Ironside, in Technical Digest of Sixteenth Internation-al Quantum Electronics Conference (Optical Society ofAmerica, Washington, D.C., 1988), paper WG-3.

9. T. J. Cullen, C. N. Ironside, C. T. Seaton, and G. I.Stegeman, Appl. Phys. Lett. 49, 1403 (1986).

10. N. Finlayson, W. C. Banyai, E. M. Wright, C. T. Seaton,G. I. Stegeman, T. J. Cullen, and C. N. Ironside, Appl.Phys. Lett. 53, 1144 (1988).

/ May 15, 1989