Three-dimensional (BiY) 3Fe5O12 waveguide witha load layer made of SiO2

Kaoru Matsuda and Satoshi Ishizuka

We have developed a loaded (BiY) 3Fe 5O1 2 waveguide for integrated magnetooptic devices. The load layermade of SiO2 was fabricated with a reactive ion etching using CHF 3 gas. The waveguiding loss was measuredby the streak method and by the cutback method. The measured values were <7 cm', and the waveguidingloss caused by the lateral confinement was deduced to <1 cm 1. Key words: Magnetooptic device, garnetwaveguide.

1. Introduction

Recently, magnetooptic (MO) devices, such as opti-cal isolators' and optical switches,2 have been devel-oped using magnetic garnets since they have largespecific Faraday rotations and low insertion losses forthe 1.3-Am wavelength range. There have been manystudies of magnetic garnet waveguides to integratethese MO devices monolithically with other devices.3-7Among them, an improvement of the TE to TM modeconversion coefficient was reported,3 where the stressof the waveguide layer caused by the mismatch of thelattice constant between the waveguide layer and thesubstrate was used to control the birefringence. Tocouple these MO devices to an optical fiber, a 3-Dwaveguide is necessary. In other studies,4 5 fabrica-tion methods for the 3-D magnetic garnet waveguidewere reported. However, the ion-beam etching usingAr4 causes damages of the garnet waveguide layer. Onthe other hand, the wet chemical etching using phos-phoric acid at 1000C5 lacks the controllability for finepatterning.

We fabricate (BiY)3Fe5O1 2 3-D waveguides with aload layer made of SiO2. The SiO2 layer was etchedwith a reactive ion etching (RIE) using CHF3 as anetching gas. Although RIE is a dry etching, it is basedon both physical and chemical etching and causes

Both authors are with Matsushita Electric Industrial Ltd., 3-15,Yagumonakamachi, Moriguchi, Osaka 570, Japan; K. Matsuda is

with the Optoelectronics Laboratory of the Semiconductor ResearchCenter, and S. Ishizuka is with the Networks Development Promo-tion Center.

Received 5 February 1990.0003-6935/91/151963-04$05.00/0.© 1991 Optical Society of America.

slighter damage than the ion-beam etching. We havedemonstrated 3-D guidance of light beam etching.The waveguiding loss of this loaded waveguide hasbeen measured by the streak and cutback methods.

II. Fabrication

Figure 1 shows the fabrication steps of the wave-guide. (BiY)3Fe5012 garnet film (5 Asm thick) was firstgrown on the Ca-Mg-Zr-substituted Gd3Ga5012(GGG) substrate by liquid phase epitaxy. Then SiO2(-3000 A thick) was deposited by rf sputtering usingAr gas. The SiO2 was etched by RIE with the etchinggas of CHF3 using a photoresist (Micro Posit MP1400-31) as a mask. It was patterned into a stripe shapewith the conventional photolithography technique.The etching was done in the following condition. Theetching gas flow was 40 cc/s, the etching gas pressurewas 6 X 10-2 Torr, and the microwave power was 50 W.The etching rates of SiO2 and the photoresist areshown in Fig. 2. As shown in the figure, the etchingrate of SiO2 is 0.015 ,gm/min, and that of the photore-sist is 0.002 Am/min. The magnetic garnet was notetched at all. The etching ratio of SiO2:resist:magne-tic garnet is 7.5:1:0, which is enough to fabricate theload layer. The photoresist stripe pattern was re-moved with acetone by an ultrasonic rinse after etch-ing. Stripe shaped SiO2 is used as a load layer for a 3-Dwaveguide.

After the wafer process shown in Fig. 1, the waferwas cut into some pieces a few millimeters square insize. Then the ends of the waveguide were polished tobe optically flat to prevent input and output light frombeing scattered. Two wafers were contacted to eachother carefully to protect the corner of the end facetsduring polishing. The scanning electron microscope(SEM) photograph of the load layer is shown in Fig. 3.Since (BiY)3FeO 12 is an insulator, the SEM photo-

20 May 1991 / Vol. 30, No. 15 / APPLIED OPTICS 1963

1ffiZZZ/// 7-/m///X BiYIG

GGG

SiO 2

LIIII7I SiO2 deposition

MP 1400

-.... resist coatingUV exposuredevelopment

reactive ion etchingCHF3 gas

resist removed

Fig. 1. Process flow of the (BiY)3 Fe5O12 3-D waveguide with a loadlayer made of SiO 2 .

0.15 I 1 1

E I/

0.10 SiO2

I 0

uJ M P1400-i

0 -

0 2 4 6 8 10ETCHING TIME ( min)

Fig. 2. Reactive ion etching rate of SiO2 and photoresist (MicroPosit MP1400-31) using CHF3 as an etching gas in the conditions ofthe etching gas flow of 40 cc/s, the etching gas pressure of 6 x 10-2

Torr, and the microwave power of 50 W.

graph was taken after Au deposition. As can be seenfrom the figure, the damage to the sidewall of the loadlayer is very small.

11111. Characteristics

We have observed the near field pattern of the lightguided in the loaded waveguide using the experimentalsetup shown in Fig. 4. A laser diode (LD) with a lasingwavelength of 1.3 ,um was used as alight source. The

Fig. 3. SEM photograph of the SiO 2 load layer fabricated using thereactive ion etching process.

sample

LD Vlens lens tens IR camera

Fig. 4. Experimental setup to measure the near field pattern of theguided mode in the (Biy) 3Fe5 0 12 waveguide with a load layer made

of SiO2.

Fig. 5. Near field pattern of the guided mode in the loaded(BiY) 3Fe5O12 waveguide with a load layer width of 7 um and a

waveguide length of 2 mm.

diverged laser beam was collimated by a lens which wasset in front of the LD. The collimated light was fo-cused on the end facet of the waveguide by an objectivelens with a magnification of 20. The output lightbeam from the waveguide was collimated again by alens and led to an infrared video camera to observe thenear field pattern of the guided mode. The near fieldpattern of the guided mode is shown in Fig. 5 for aloaded waveguide with a load layer width of 7 ,um and awaveguide length of 2 mm. As shown in the figure, thelight beam from the LD is guided three-dimensionally.The near field pattern and its intensity distribution ofthe guided mode corresponding to the near field pat-tern are shown in Fig. 6. The loaded waveguide has aload layer width of 20 ,gm and a waveguide length of 4mm. The results for a slab waveguide are also shownin the figure. The vertical axis shows the optical in-tensity in an arbitrary unit, and the horizontal axisshows the location within the near field pattern. Ascan be seen in the figure, the light was guided three-dimensionally in the waveguide as long as 4 mm.

1964 APPLIED OPTICS / Vol. 30, No. 15 / 20 May 1991

M......NOWROM

Fig. 6. Near field patterns and optical intensitydistributions of the guided mode in the loaded(BiY)3 Fe5O12 waveguide with a load layer width of20 ,um and a waveguide length of 4 mm. The nearfield pattern for a slab waveguide is also demon-

strated.

0I-s

0) -. 2

-c

LU Vlens lenv V lens IR camerasample

Fig. 7. Experimental setup and system to measure the waveguidingloss of the loaded waveguide by the streak method.

We have measured the waveguiding loss of the load-ed waveguide with a load layer width of 7 gm by astreak method using the experimental setup shown inFig. 76 In addition to the setup shown in Fig. 4,another infrared video camera was set above the wave-guide to observe streak light from the waveguidethrough the load layer. The streak light was moni-tored by a video camera and a camera controller withan analog-to-digital converter. A sampling line wasset perpendicular to the streak light of the guidedmode. Signals corresponding to the light intensitydistribution along the sampling line were sent to amicrocomputer and were integrated along the sam-pling line. Then the sampling line was moved alongthe streak light, and the same measurement was re-peated for each sampling line. As a result, powers ofthe streak light in relation to the guiding length wereobtained as shown in Fig. 8. As the power of the streaklight is proportional to the power of the guided mode, awaveguiding loss coefficient a, which consists of anabsorption loss of the (BiY)3Fe5O1 2 crystal and a losscaused by the lateral confinement, can be obtainedfrom a gradient of the streak light to the guidinglength. The power of the streak light I is

I = Io exp(-at), (1)

where Io is the streak light power at the initial pointand t is the guiding length from the initial point. In

a)I.co

a)

0_

- 3

-. 5- - u wu on M `- Il LO n (

guiding length (a.u.)Fig. 8. Power of streak light in relation to the change in the wave-guiding length obtained by the streak method. Measurementpoints are shown by an asterisk (*) in the figure. Notice that thewaveguiding length is plotted in the direction opposite to the hori-zontal axis, because the streak light at the point farthest from the

incident point was chosen as a standard point.

the measurement system shown in Fig. 7, the micro-computer also calculated the gradient by the least-mean-square approximation. In this measurement,the absorbent was used in the interface of the substrateand a sample holder to obtain the power of streak lightcorrectly since the Gd3Ga5O12 substrate is transparentfor the wavelength of 1.3-Arm light, and there was irreg-ular refracting light transmitted through the sub-strate. The waveguiding loss measured by the streakmethod was 7 cm-' on the average of several mea-surements at different parts of the waveguide.

We have also measured the total loss of the loadedwaveguide by the cutback method by the experimentalsetup shown in Fig. 9. The output light beam from thewaveguide was divided into two beams by a beamsplitter. One beam was led to an infrared video cam-era to observe the near field pattern of the guidedmode. Another beam was focused on the photodetec-tor. Under observation with the infrared video cam-era, a pinhole was inserted just behind the sample to

20 May 1991 I Vol. 30, No. 15'/ APPLIED OPTICS 1965

t:z-z) z

z

I= 10 exp (-t)

I: power of streak light

Io: power ofstandard streak light

t: guiding lengtha: waveguiding loss

-. 4F

IRcamera

pinhole t o

LX - beamlens lens splitter

Fig. 9. Experimental setup to measure the waveguiding loss by thecutback method.-5

i4

-J<0

00WAVEGUIDE

2LE NGTH t (mm)

3

Fig. 10. Output power from the loaded (BiY)3 Fe5O12 waveguidewith a load layer width of 7jum as the change in the waveguide length

to evaluate the loss of the guided mode by the cutback method.

measure correctly output optical power from the wave-guide by excluding the light except for the guidedmode. We measured two samples which have a wave-guide length of 1 and 2 mm. The measured result isshown in Fig. 10. In the figure the perpendicular axisshows the output power from the waveguide, and thehorizontal axis shows the waveguide length. This isthe same relation as in Eq. (1) for the optical outputpower from the waveguide I, the waveguide length t,and the waveguiding loss coefficient a. From the gra-dient of optical output power to the waveguide lengthin Fig. 10, a waveguiding loss of 6.6 cm-' was obtained.

The (BiY)3Fe5O12 crystal has absorption loss of -6cm'1. Since the waveguiding loss was -7 cm-' by the

streak method and 6.6 cm-' by the cutback method,the waveguiding loss caused by the lateral confinementis estimated to be <1 cm-1.

The absorption loss of the magnetic garnets in thenear infrared region is caused by Fe2+ and Fe4+ ionsassociated with impurities which are incorporatedfrom the flux during growth.7 To reduce the nearinfrared absorption loss, Ca2+ and Si4+ were added tocompensate valences.7'8 Therefore, the waveguidingloss of the (BiY)3Fe5Ol2 loaded waveguide may bedecreased by using this technique.

IV. Conclusion

We have fabricated (BiY)3Fe5O1 2 3-D waveguideswith a load layer made of SiO2 with reactive ion etchingusing CHF 3 gas. The waveguiding loss was measuredby the streak and cutback methods. The waveguidingloss caused by the lateral confinement estimated fromthe waveguiding loss was <1 cm- 1 .

We would like to thank T. Onuma and T. Kajiwarafor their continuous encouragement and valuable dis-cussions.

References1. K. Matsuda, H. Minemoto, 0. Kamada, and S. Ishizuka, "Tem-

perature-Stabilized Optical Isolator for Collimated Light Using(BiLuGd) 3 Fe5 Ol 2 /(BiGd) 3 (FeGa)50 12 Composite Film," Appl.Opt. 27, 1329-1333 (1988).

2. M. Shirsaki, H. Nakajima, T. Obokata, and K. Asama, "Nonme-chanical Optical Switch for Single-Mode Fibers," Appl. Opt. 23,4229-4234 (1982).

3. K. Ando, N. Takeda, N. Koshizuka, and T. Okuda, "AnnealingEffects on Growth-Induced Optical Birefringence in Liquid-Phase-Epitaxial-Grown Bi-Substituted Iron Garnet Films," J.Appl. Phys. 57, 1277-1281 (1985).

4. Y. Okamura, M. Ishida, and S. Yamamoto, "Magnetooptic RibWaveguide in YIG: an Experiment," Appl. Opt. 23, 124-126(1984).

5. E. Pross, W. Tolksdorf, and H. Dammann, "Yttrium Iron GarnetSingle-Mode Buried Channel Waveguides for Waveguide Isola-tors," Appl. Phys. Lett. 52, 682-684 (1988).

6. A. Miki, Y. Okamura, and S. Yamamoto, "Measurement of Prop-agation Properties of Optical Waveguide Circuits Using a TVCamera," Trans. IEICE J71-C (3),453-460 (1988) (in Japanese).

7. D. L. Wood and J. P. Remeika, "Effect of Impurities on theOptical Properties of Yttrium Iron Garnet," J. Appl. Phys. 38,1038-1045 (1967).

8. G. L. Nelson and W. A. Harvey, "Optical Absorption Reduction inBiLu 2Fe5Ol2 Garnet Magneto-Opticdl Crystal," J. Appl. Phys.53, 1687-1689 (1982).

1966 APPLIED OPTICS / Vol. 30, No. 15 / 20 May 1991

I I

I = Ie-' 0

a= 6.6 cm1 a .