Low-loss GeO 2 optical waveguide fabrication using lowdeposition rate rf sputtering

Zhong-Yi Yin and B. K. Garside

Low-loss GeO2 thin-film optical waveguides have been prepared using rf reactive sputtering with a GeO2tar-get, and the propagation properties of the waveguides prepared over a wide range of fabrication conditionshave been investigated. We have found that the waveguide attenuation dramatically decreased when a verylow deposition rate of rf reactive sputtering in an argon-oxygen atmosphere was used in conjunction withappropriate annealing. In particular, 3800-A thick GeO 2 thin-film optical waveguides have been preparedwith propagation losses <0.7 dB/cm for the TEo mode at a wavelength of 0.63 ,Am. The average refractiveindex of the GeO2 films was measured to be 1.6059 at X = 5461 A by an ellipsometer technique in good agree-ment with measurement on bulk materials. Propagation losses have also been measured at different wave-lengths, which shows that GeO 2 thin-film optical waveguides could be used over a very wide wavelengthrange from the visible to the near infrared.

1. IntroductionConsiderable interest has been evidenced recently in

the use of GeO2 for optical waveguides.1- 3 This interestis based on the possibility of fabricating devices such asdeflectors, filters, distributed feedback devices, multi-plexers, and demultiplexers for use in integrated opticalsystems by exploiting the photoinduced refractive-indexchange in this material.- 7 This effect has already beenexploited in single-mode optical fibers based on GeO2-doped SiO2 to produce narrowband optical filters.8However, the optical properties of GeO2 are not well-known for bulk material3 and have not been investi-gated in thin films. As a result, we report here on workaimed at producing high-quality thin films of GeO2using rf reactive sputtering techniques with a GeO2target and thereafter determining some of the opticalproperties of the optical waveguides. It was found thata very low deposition rate of sputtering, together withpostfabrication annealing, permitted the production ofvery good quality waveguides by reducing both bulk andsurface scattering in optical waveguides. The fabri-cation techniques of the GeO 2 thin-film waveguides aredescribed below, and, subsequently, measurements ofthe attenuation of the waveguides and the refractiveindices of the GeO 2 thin films are reported.

The authors are with McMaster University, Department of Engi-neering Physics, Hamilton, Ontario L8S 4M1.

Received 3 June 1982.0003-6935/82/234324-05$01.00/0.© 1982 Optical Society of America.

II. GeO 2 Thin-Film Waveguide FabricationA large variety of possible thin-film optical wave-

guides for integrated optics have been reported. 9"10These optical waveguides have been fabricated in dif-ferent ways, such as out-diffusion, in-diffusion, chemicalvapor deposition (CVD), and rf sputtering methods. Asmentioned by Tien,10 most of the low-loss films used inthe light-guide experiments are amorphous. This isalso true in our case, where the rf reactive sputteringwith a GeO2 target is used. The distance between tar-get and substrate is -9 cm. The diameter of the targetis -6.5 cm. The target and substrate are both cooledby flowing water to produce amorphous films. This wasascertained by x-ray diffraction techniques used fortesting the crystalline state of the films. Generally,GeO2 thin films were deposited onto clean microscopeslides to fabricate the thin-film waveguide. Gases Arand 02 are introduced into the vacuum chamberthrough two gas flowmeters. Three valves are used forthis system to change the rate between Ar and 02-Different ratios of Ar to 02 in our experiment weretested. It was found that if we used Ar only, the colorof the deposited thin film became a dark yellow eventhough a GeO2 target was used rather than using a Getarget and relying on reactive sputtering in an argon-oxygen atmosphere. Rutherford backscattering wasused to analyze the composition of these fabricatedfilms. When the deposited rate was 80 A/min and onlyAr was present in the chamber, the ratio of germaniumto oxygen in the film was 1.90, with an accuracy of 1%.This means that this film was deficient in oxygen, pro-ducing absorbing centers active in the visible wave-length range. To eliminate this problem, we used a rf

4324 APPLIED OPTICS / Vol. 21, No. 23 / 1 December 1982

wo70

(86oW0.1

wo4

,,

<I-

woO3

(a)

I I I

I I I I0.0 1.0 2.0

RELATIVE DISTANCE (cm)(b)

Fig. 1. (a) Side-view of a thin-film waveguide. (b) Typical resultsat = 0.63 m for TEo-mode waveguide loss measurements, where

length is measured relative to the first test point.

reactive sputtering technique in which oxygen is in-troduced into the sputtering chamber. Once bothoxygen and argon are present in the vacuum chamberin the right proportions, the films deposited onto themicroscope slides become colorless. In the case ofthin-film waveguides, the ratio of 02 to Ar in thechamber was 1:1 and the total gas pressure was 2 X 10-3Torr. We found that the propagation properties ofoptical waveguides were strongly dependent on thedeposition rate of GeO2 films and the postannealingmethod played a significant role for producing low-lossGeO 2 thin-film waveguides. The preparation proce-dure is as follows. First, the GeO2 films are depositedonto the microscope slides at a rate as low as 12.7 A/min.Next, the GeO2 waveguides are heated in a furnace inthe ambient atmosphere for 6 h at 250°C. Films fab-ricated by this method, i.e., with very low deposition rateof rf reactive sputtering and with subsequent annealing,are of excellent quality over the range of thickness in-vestigated (up to 3800 A). The waveguide attenuationfor a sample film fabricated in this way was determinedto be <0.7 dB/cm for the TEo mode at X = 6328 A.Ill. Optical Experimental Methods

Light can be coupled into the waveguide by variousbeam-coupling techniques such as head-on, grating,prism, and taper couplers.11 In our case, the light wascoupled into the waveguide with a glass prism coupler

whose refractive index was 1.805 measured at X = 5461A. The GeO2 thin-film waveguide was pressed againstthe base of the prism by a stainless steel bar and thenmounted on a turntable so that a laser beam could enterthe input prism at any selected angle. Both a He-Nelaser and an Ar laser were used as sources, and only theTEO mode in the waveguide was excited in the experi-ment. To measure losses in the GeO2 waveguides, themethod of the transmission measurement10 using twoprism couplers was used [see Fig. 1(a)]. Light wascoupled into the GeO 2 thin film with one prism, andanother prism coupled the light wave back out of thefilm. Such an output coupler is always an effectivelyperfect output coupler provided the coupling length issufficiently long. In other words, it can easily be made100% efficient. Thus, for measuring attenuation, theoutput coupler was applied at different points along thelight streak. At each point, the output coupler wasadjusted so that effectively all the light could be coupledout. The light from the output prism was detected witha stop, a lens, and a photodiode. The measurementsthus obtained at three selected points along the lightstreak were plotted and used to determine the wave-guide attenuation. An example of the results obtainedin this way, together with an indication of the associatederrors, is shown in Fig. 1(b). Only three points wereused since the waveguide length was relatively small andthe measurement technique only permitted three in-dependent points to be accessed. However, each pointwas remeasured several times, and the final measure-ment precision is estimated to be 0.12 dB/cm. Thethickness and the average refractive indices of GeO2thin films and microscope slides were measured by anellipsometer.

IV. Experimental Results and DiscussionThe rf reactive sputtering method was used to pre-

pare GeO2 optical waveguides as we mentioned before.First, the sputtering system was pumped to a pressurein the region of 4 X 10-6 Torr, and then backfilled withgases of argon and oxygen. We used two flowmeters asmonitors of Ar and 02- In sputtering the GeO 2 film, agermanium dioxide target was employed with micro-scope slides as substrates. The deposition time vsthickness of GeO2 film for an input power of 180 W isshown in Fig. 2. We found that the relationship be-tween thickness and deposition time was linear whenthe thickness was increased from -500 to 5000 A. Notethat the plot is linear to within 3% of the thickness.From Fig. 2, one can easily obtain a selected thicknessof a film by choice of the appropriate deposition time.As we know, the thickness of the GeO2 thin film is veryimportant for preparing GeO 2 waveguides of prede-termined characteristics. So, the method of controllingthe thickness with the definite deposition time was usedin our experiments to obtain the selected thickness ofthe GeO2 film.

Figure 3 shows the sputtering rate vs the sputteringinput power. This curve was obtained under conditionswhere the pressures of both Ar and 02 were 1 X 10-3Torr. It is clear that when the sputtering input power

1 December 1982 / Vol. 21, No. 23 / APPLIED OPTICS 4325

5000

4000

.cR

N

Li.00)0)wz0UI-

3000_

2000_

1OOOH

0 10 20 30 40 50 60 70DEPOSITION TIME (min)

Fig. 2. Deposition time vs thicknesses of GeO2 films, P = 180 W,pressure of Ar and 02: 1.87 X 10-3 Torr and 0.53 X 10-3 Torr,

respectively.

C

I-

0zw

a.0)

50[-

25

nL0 50 100 150

SPUTTERING INPUT POWER (W)200

Fig. 3. Sputtering rate vs sputtering input power. The pressure ofAr is 1 X 10-3 Torr, and 02 1 X 10-3 Torr. The substrates are mi-

croscope slides.

is <180 W, the sputtering rate is <60 A/min. Thelowest sputtering rate in our experiment is 12.7 A/minwhen the sputtering input power is -40 W. From Fig.3, we found that the relationship between the sputteringrate and the sputtering input power was approximatelylinear. We can then determine the average sputteringrate/sputtering input power (R,) to be 0.33 A/min W.The value R, is very useful for fabrication of GeO2thin-film waveguides in our experiment. It is clear thatwe can use different deposition time and sputteringinput power to obtain different thicknesses of the films.It means the definite thickness of the GeO2 thin film willbe obtained by the methods of the sputtering timecontrol and input power control. For example, if theinput power is 40 W, the deposition time is 288 min, andthen the thickness of the film h is -3800 A. Thethickness h is given by

h = t -P-R8, (1)

where t is deposition time, and P is sputtering inputpower.

Next, let us consider the optical waveguides. Forthin-film waveguides [see Fig. 1(a)], we have9

2Knfh cosO - 2q5 -. 2q5 = 2mr, (2)

where K = 27r/X, X is the free-space wavelength, 0 is theincident angle between the wave normal and the normalto the interface in the thin film, 2, and 2X0 are thephase shifts at the film-substrate and film-cover in-terfaces, nf is the refractive index of the film, and m isan integer (0,1,2,. .. ). For the TE mode, suppose thatonly one mode (TEO) in the waveguide can be propa-gated. In this case, the thickness of the GeO2 thin filmh is given by an inequality as follows:

tan1 /a/(K Iny- n2) < h < (tan- x/; + r)/(K N/2 ),

(3)

where a = (n 2- n2)/(n2 - n2), n,, and n, are the re-fractive indices of the substrate and cover, respectively.For the situation of GeO2 and microscope slide glass, thethickness limit at X = 5461 A lies between 1860 and 7009A for the TEO mode only, and in our experiments, the3800-A thickness of GeO2 film was chosen to be wellwithin the limits specified by the inequality (3) usingHe-Ne and Ar lasers.

Now we wish to report the losses of GeO2 waveguidesin our experiments. As reported previously12-"4 thesurface and bulk scattering of light in optical wave-guides represent loss mechanisms. The low-losswaveguides are desirable to make integrated opticaldevices more useful in laser communication and opticalprocessing. The different sputtering rates were usedto achieve this purpose. We found the low depositionrate was helpful for this aim. Films fabricated at arelatively high deposition rate of 62.9 A/min were foundto have a very large loss at X = 0.63 Am. No guidedmode could be properly launched into this waveguide,whereas at a rate of 12.7 A/min the light could belaunched satisfactorily into the GeO2 film, and the losswas found to be 18 dB/cm. Note that the loss was stillhigh. To decrease the waveguide attenuation further,we put the sample into a furnace, and then it was heatedat 250'C in the ambient atmosphere. After annealing,the waveguide attenuation significantly decreased.Figure 4 shows the relationship between waveguideattenuation and the annealing temperature. As can beseen in the figure, the loss of the waveguide reduceddramatically from 18 dB/cm before annealing to 0.7dB/cm after annealing at a temperature of 2500C. Thewaveguide attenuation vs the rate of the sputtering,measured at X = 0.63 ,um after annealing at a tempera-ture of 250'C and X = 0.63 m, is shown in Fig. 5. FromFig. 5, it is clear that the smaller the rate of the sput-tering, the smaller the waveguide attenuation. Theattenuation varied from 17 dB/cm at the rate of 43.5A/min to 0.7 dB cm at the rate of 12.7 A/min. So, whencombining the annealing temperature of the waveguideswith a very low sputtering ratio, the waveguide loss of<0.7 dB/cm was obtained. In other words, both surface

4326 APPLIED OPTICS / Vol. 21, No. 23 / 1 December 1982

I I I I iT

/I I I I I

T

71-

us

us - -

n

mN.

z0

.4

z

1-1

I-.4

I0

10

5

I I I

zI.4

z

C]

0

Li.4

20 ' 200 300 400ANNEALING TEMPERATURE (PC)

Fig. 4. Waveguide attenuation vs the annealing temperature at X= 0.63 Am. The sample was prepared with a deposition rate of 12.7A/min, the pressure of Ar 1 X 10-3 Torr, and 02 1 X 10-3 Torr.

20 | 1 l

q0 15

I/

1

0.4

z

10 20 30 40SPUTTERING RATE (min)

Fig. 5. Waveguide attenuation vs the rate of the sputtering at X =0.63 m. The sample was prepared with an annealing temperatureof 250'C, the pressure of Ar 1 X 10-3 Torr, and 02 - 1 X 10-3

Torr.

and bulk scattering losses were reduced by very lowdeposition rate and annealing.

In addition, the refractive indices of GeO2 thin filmand the substrate microscope slide were also measuredby an ellipsometer. The average index of microscopyslides was 1.5158 at X = 5461 A. However, the index ofthe film was slightly changed by the sputtering condi-tions and the annealing temperature. Under the con-ditions of a deposition rate of 12.7 A/min at the an-nealing temperature of 2500C the average refractiveindex was 1.6059 at X = 5461 A. This value is similar

_ ~~I I I

2 -

0 I I I450 500 550 600

OPTICAL WAVELENGTH (nm)650

Fig. 6. Waveguide attenuation vs the optical wavelength. Thesample was prepared with a deposition rate of 12.7 A/min, the pressure

of Ar 1 X 10-3 Torr and 02 - 1 X 10-3 Torr.

to the index of GeO 2 reported by Devyatykh et al.3 So,one can use a formula of the type,

n2-1= 3E 2i=1l --bi (4)

to estimate the dispersion n (X), which is useful for de-signing the GeO2 waveguides in the different wave-lengths. The appropriate values of the parameters aiand b are given in Ref. 3. The measured waveguideloss vs optical wavelength is shown in Fig. 6. We foundthat as the wavelengths became shorter, the losses be-came larger. Similar behavior has been found withTa2O5 waveguides as reported by Tien,10 even thoughthe material is different. Figure 6 implies that thewaveguide loss will be much <0.7 dB/cm when the op-tical wavelength become longer. The absorptionspectra and the dispersion of the refractive index ofGeO2 and of fused quartz have been reported by Dev-yatykh et al.

3 This showed that GeO 2 bulk materialhad low absorption, and the refractive index of GeO 2was greater than that of SiO 2 from the visible into thenear infrared. Consequently, if one uses fused quartzas a substrate, the GeO2 thin-film waveguide can beused over this same range of wavelength.

V. ConclusionsLow-loss GeO2 optical waveguides of <0.7 dB/cm for

TEO mode at X = 0.63 um have been fabricated for thefirst time by combining the techniques of very low de-position rate of rf reactive sputtering (12.7 A/min) andpostannealing (250'C). These techniques reduce bothbulk and surface scattering in optical waveguides. Theactual value may be even somewhat lower than the valuewe reported above because this measurement is closeto the limit of our experimental system. If we use fusedquartz as a substrate, the GeO2 thin-film opticalwaveguide has potential application over a very widewavelength range from the visible into the near infrared.In addition, if the sputtering system can be operated atthe rate of <12.7 A/min and the surface of a substrateis prepolished, the propagation property of the wave-guide can possibly be further improved.

1 December 1982 / Vol. 21, No. 23 / APPLIED OPTICS 4327

20 I. ...

I" ,,

Although many different thin-film waveguides havebeen reported since 1969,15 waveguide losses of <1dB/cm have not been observed very frequently. Somebulk materials have very good optical properties, butwhen the thin-film waveguides were made using thesematerials, they were found to produce very big lossescreated by bulk and surface scattering. 10 Therefore,the method we present here might be useful for modi-fying fabrication procedures for other optical thin-filmwaveguides.

Since the GeO2 low-loss waveguides have now beenfabricated, the possibility exists of using the photoin-duced refractive-index change to develop some usefuloptical devices for optical information processing.

The authors are grateful to many of their colleagues.Special thanks are owed to K. B. Sundaram for somesuggestions and useful discussions.

Zhong-Yi Yin is a visiting scholar from ZhongyuanResearch Institute of Electronics Technology,Zhumadian, China.

References

1. C. R. Hammond, Opt. Quantum Electron. 10, 163 (1978).2. E. M. Dianov, Sov. J. Quantum Electron. 10, 259 (1980).3. G. G. Devyatykh et al., Sov. J. Quantum Electron. 10, 900

(1980).4. A. M. Glass, Opt. Eng. 17, 470 (1978).5. A. M. Glass et al., Appl. Opt. 19, 276 (1980).6. C. M. Verber et al., Appl. Phys. Letters 30, 272 (1977).7. K. 0. Hill, Y. Fujii, D. C. Johnson, and B. S. Kawasaki, Appl.

Phys. Lett. 32, 647 (1978).8. D. K. W. Lam and B. K. Garside, Appl. Opt. 20, 440 (1981).9. R. V. Schmidt and I. P. Kaminow, Appl. Phys. Lett. 25, 458

(1974).10. P. K. Tien, Appl. Opt. 10, 2395 (1971).11. T. Tamir, Integrated Optics (Springer, Berlin, 1979).12. M. Gottlieb et al., IEEE Trans. Circuits Syst. 26, 1029 (1979).13. S. Dutta, H. E. Jackson, and J. T. Boyd, Appl. Phys. Lett. 37,512

(1980).14. M. Imai et al., J. Appl. Phys. 52, 6506 (1981).15. S. E. Miller, Bell Syst. Tech. J. 48, 2059 (1969).

10th International Optical Computing Conference

April 6-8, 1983

Massachusetts Institute of Technology

Cambridge, Massachusetts

"Unconventional Imaging and Unconventional Transformations"

Sponsored by

IEEE Computer Society and International Commission for Optics

Contact:

Cardinal Warde

Department of Electrical Engineering & Computer Science

Massachusetts Institute of Technology

Cambridge, MA 021.39

4328 APPLIED OPTICS / Vol. 21, No. 23 / 1 December 1982