1232 IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-20, NO. 5, SEPTEMBER 1984

MSSW RESONATORS WITH STRAIGHT EDGE REFLECTORS

Ernst Huijer and Waguih Ishak

ABSTRACT

The performance of a novel magnetostatic surface wave (MSSW) resonator structure is presented. The resonator uses rectangular YIG film to propagate MSSWS where the straight edges serve as reflectors. The resonator was tunable from 3 to 9 GHz by varying the DC bias field from 500 to 2500 Oe and exhibited very high Q ranging from 900 to 4000. Through a careful parameter choice, the spurious modes were suppressed by more than

coupling to width modes and their effect on the main 15 dB from the main resonance. Problems arising from

resonance are discussed.

INTRODUCTION

During the early investigations of MSW it was found that resonances occur in rectangular slabs of YIG [I]. MSW resonators, however, have been built using thin films of YIG with grating reflectors [2],[3]. In addition to the complicated processing required to build the gratings, The resonators Q did not exceed 1000. This paper reports on MSSW resonators containing rectangular pieces of YIG film where the straight edges, produced by cutting the YIG films with a wafer saw, serve as reflectors for MSSWs. These devices are tunable between 3 and 9 GHz and exhibit Qs of 900 to 4000. Detailed discussion of spurious modes, interfering with the main resonant peak, is given.

DESCRIPTION OF THE RESONATOR

A typical configuration of the straight edge MSSW resonator, shown in Fig 1, consists of a ferrimagnetic resonant cavity, placed on a thin film transducer Structure. The resonant cavity is made of a piece of YIG film epitaxially grown on a GGG substrate and cut into a rectangle by a wafer saw. The transducers are microstrips patterned in gold on a dielectric substrate. This substrate is mounted on a block of aluminum, thus providing a ground plane to the microstrips. This structure is placed between the pole pieces of an electromagnet, such that the magnetic field is parallel to the transducers. Such an orientation of the field permits the coupling to MSSWs in the magnetic film by one or two transducers. These waves propagate along the surface of the YIG film and are reflected onto the other surface at the straight edges. A circulating wavepattern

results that is resonant if the following condition is met:

( k+ + k- 11 = 2nn, n=1,2, ... (1)

where k, and k- are the wave numbers for the top and bottom surfaces, respectively, and 1 is the distance between the straight edge reflectors. The frequency response of such a device shows a family of resonant modes that tunes with the applied magnetic field. Such a device can be applied to make a tunable oscillator, in which case it will be desirable to have only one dominant resonant mode. Higher mode suppresion is achieved by spacing away the transducers in one or two dimensions, causing a wavelength dependent coupling between the transducers and the YIG film. An interesting aspect of this device is its reciprocity. The amplitudes of S-+2 and S 2 1 are nearly the same, but the phase may be

and k- especially at longer wavelengths. signlficantly different due to the difference between k,

EXPERIMENTAL RESULTS

Frequency responses were measured using an HP-8409C Network Analyzer System under the control of an HP-9826A Desktop Computer. Tuning characteristics are illustrated by the results obtained with a 4x1.4x0.0217 nun YIG film between 4x0.03 nun transducers on a sapphire substrate at 3 nun disctance. The vertical spacing is the 0.33 nun thickness of the GGG substrate. With a variation of the bias field from 497 to 2490 Oe, The resonant frequency was tuned from 3 to 9 GHz. Fig 2 shows the frequency response at an applied field of 1793 Oe. The main resonance occurs at 6.987 GHz, with an insertion loss of 23 dB and a Q of 3995. The second mode has 15 dB more insertion loss than the main mode. At the low end of the tuning range, power limiting occurs due to coinicidence limiting. The 1-dB power compression points at 3 and 3.5 GHz were -20 and +4.5 dBm, respectively. The current distribution in the transducers and the interference of spurious modes limit the tuning range at the high frequency end. The frequency response of the resonator is shown in Fig 3 at 3 and 9 GHz. At 9 GHz the transmission amplitude has decreased to 31 dB. The performance of this resonator is summarized in Fig 4 where the resonance frequency, Q, and the transmission amplitudes of the main and second modes are plotted versus the applied bias field.

ezIIz3 TOP VIEW

Fig. 1: Straight edge MSSW resonator.

E. Huijer is with Thomson-CSF, Orsay, France. Fig- 2: Resonator frequency response at 7 GHz. W. Ishak is with Hewlett-Packard Labs, Palo Alto, Ca. 94304.

0018-9464/84/0900-1232$~1.00~1984 IEEE

1233

FREQUENCY IMHZI AMPLITUDE (dB1 0 10000 1 I 1 I 1 ] 0 sow

-33

- 4 0

-50

- 4 0

-50

-60

Fig. 3: Resonator response (a) 3 GHz and (b) 9 GHz.

SPURIOUS MODES

One of the effects limiting the tuning range Of these resonators is the occurence of spurious modes. These modes tune with the surface wave frequency band but at a slightly different rate than the principal resonant modes. At certain field values, the two types of modes cross over and may interfere with each other. A principal resonant mode may exhibit a split peak due to this effect. The most probable explanation of these observations is the occurence of width modes. O'keefe and Patterson [4] have analyzed width modes in YIG film delay line configurations and reported a procedure to find the dispersion equations for these width modes. These equations were obtained by modifying equations derived for films of infinite extent. Following their procedure, dispersion diagrams were calculated for comparison with measured .results for both delay lines and resonators. In the former case, k, is determined whereas in the latter case kt a 1 = k, + k- is determined. The dominant width mode in both types Of devices is expected to be the ktrans = "/w mode, where w is the film width, if the excltation is uniform over the width of the YIG film. Fig 5 shows the caculated dispersion curves for an MSSW delay line configuration. The dispersion is rather sensitive to varaitions of the YIG-ground plane spacing at lower values of k and curves for three different values of this spacing are shown.

The dispersion of a film of matching parameters was

4000

3000

2000

1 wo

0 I I 1 I 0 500 1000 1500 2M)o 2500

I -50 0

APPLIED FIELD (04

Fig. 4 : Resonator response summary.

FREQUENCY (MHz)

Fig. 5: Dispersion curves for an MSSW delay line

shift along the frequency axis was allowed to obtain a best fit. A discrepancy of the frequency between measured and calculated values can be ascribed to the demagnetizing effect of the YIG film and to errors in the values of the applied bias field and the saturation magnetization. The agreement is best for a spacing of 300P m, although the actual spacing was 254p m. NO explanation of this discrepancy has been found. Experimental determination of kt otal was done by using the actual length of the YIG film. The measured data together with cutves calculated for three different values of YIG-ground plane spacing for a resonator structure are shown in Fig 6. Agreemeilt is poorer than

the uniformity of the effective applied bias field due in the case of delay lines. One possible explanation is

to the demagnetizing effect of the finite film. Adam [ 5 1 repdrted resonances in slabs of YIG and found discrepancy between observed and calculated resonant frequencies unless he used experimental arrangements to suppress the, demagnetizing field.

determined by extracting values for k from the phase after eliminating the 2 X phase jumps, as recorded from The above noted discrepancies make it difficult to a transmission measurement on the network analyzer. The make a very exact Of $purious modes with resulting data points are also shown in Fig 5, where a calculated higher order width modes. However, a

1234

Fig. 6: Dispersion curves for a resonator structure.

R m p h t u d e ( d B )

-10

-20

-30

- 4 0

-50

F r e q u e n c y ( M H z )

R m p I I t u d e ( d B 1

- I O

/bl

-20

-30

- 4 0

-50

Frequency (MHz1

LC) R m p l ~ t u d e ( d B 1

-10

-20

-30

- 4 0

-50

qualitative comparison shows very similar features. In effect, the crossing over of the modes, as mentioned above, was also found to occur for calculated modes. However, the field and frequency at which the modes Cross over are different in the experimental and analytical cases. Fig 7 shows a sequence of measured frequency responses of the resonator at three different values of the applied field. A spurious mode is seen to interfere with the principal mode such that it splits its peak. Fig 8 shows the calculated dispersion characteristics of the first and some higher modes at two different applied field values. Principal resonances occur where the curve of the n=l mode crosses the constant ktotal lines. Spurious modes occur when the curves for higher n modes cross constant k lines. In both diagrams, an n=3 mode is close in frequency to the main resonant mode . The frequency of this n=3 mode at H=1100 Oe (point A) is above the main mode frequency

the main mode resonance (point D). (point B), however, for H=1500 Oe, It is (point C) below

FREOUENCV IMHz)

Fig. 8: Dispersion curves for main and higher order modes at (a)H=1100 Oe and (b)H=1500 Oe.

CONCLUSIONS

A very high Q MSSW resonator was demonstrated using a simple planar structure. The resonator exhibits Qs up to 4000 and the off resonance isolation is better than 15 dB. Width modes interfere with the main resonant peak and methods to eliminate them are under investigation. This MSSW resonator could be used to construct low phase noise oscillators.

ACKNOWLEDGEMENTS

The authors like to thank Elena luiz for circuit assembly.

REFERENCES

Brundle, L.K. : Elect. Lett.,1968,4,p. 132. Castera, J.P. : Proc RADC Workshop,l98l,p. 218. Owens, J.M. : IEEE Ultrasonic Symp. Proc..1978, p. 440. O'Keefe, T.W.: J. Appl. Phys., 1978,49,p. 4886. Adam, J.W. : Elect. Lett.,1970,6,p. 434 .

Fig. 7 : Spurious mode interference with the main mode. (a)H=1086 Oe, (b)H=1103 Oe and (c)H=1128 Oe.