WAVENUMBER K J

3718 IEEE TRANSACTIONS ON MAGNETICS, VOL. MAG-23, NO. 5, SEPTEMBER 1987

CHARACTERIZATION OF MAGNETOSTATIC WAVE DEVICESBY BRILLOUIN LIGHT SCATTERING

G. Srinivasan,J. G. Booth, and C.E. Patton.

Abstract - Brillouin light scattering (BLS) has been frequency profile for the scattered light, is obtained used for direct detection of backward volume wave (BVW)by analyzing the scattered light with a high contrast excitations in yttrium iron garnet (YIG) filmsin Fabry-Perot interferometer. One observes intensity magnetostatic wave devices. The measurements were peaks at frequencies corresponding to the magnetic carried outon delay line structures andon a excitations. In this study, the BLS technique was used signal-to-noise enhancer device. With an applied fieldto observe BVW excitations and to measure intensity vs. parallel to the film plane and perpendicular to the field profiles and magnon dispersion. The BVW band stripline, backward volume waves are excited in thelimits YIG and dispersion curves were found to be in film by the stripline power. Using the BLS technique,agreement with the theoretical predictions. the BVW excitations in the YIC film were observed over the frequency range2-4 GRz. The observed BVW band limits are in reasonable agreement with theory. There Backward Volume Waves is very good agreement between the experimental and the theoretical dispersion for the BVW excitations. The scattering was foundto be rather weak over the BVW There are two types of magnetic wave excitations for passband, but did reveal considerable structure. in-plane magnetized slabs, backward volume waves (BVW) and surface waves [7]. Workon surface waves has been reported previously [2]-[4]. The magnon dispersion for the volume waves is shown schematically in 1 Fig.for Introduction propagation parallel to the applied magnetic field. The frequencyfk is shown as a function of wavenumber for the various BVW dispersion branches. The upper and Magnetostatic wave (MSW) devices such as delay lower frequency limits for the BVW excitations are lines, power limiters, and signal-to-noise enhancers indicated by f and fH, respectively. These limits B are used for signal processing at microwave frequenciesare given by [l]. In these devices, magnetostatic waves are excited in a yttrium iron garnet (YIG) filman externalby microwave signal. Direct observation of such MSW excitations is possible with the technique of Brillouin light scattering (BLS). This technique has been used and to observe magnetostatic surface waves and parametric spin waves excitedin PIG films in a signal-to-noise f =yH, (2) enhancer device [2], [3]. It was possible to obtain H quantitative information on energy flow profile, magnon dispersion, and spin wave instability processes [4]. Results of recent light scattering studieson backward volume waves (BVW)in PIG filmMSW devices are presented here.

The BLS measurements were performedon (i) a signal- to-noise enhancer MSW device with a striplineon an alumina ground plane [5], [6], and(ii) delay line device structures in which stripline transducers are deposited onto silica or sapphire substrates.In both cases, one has a YIG film in contact with the stripline. With an appropriate in-plane static magnetic field applied perpendicular to the stripline, backward volume waves are excited in the YIG film by the striplinepower,

In light scattering, the interaction between the incident laser light and the magnetic excitations in the YIG film gives rise to frequency shifted scattered light. The Brillouin spectrum,i.e., the intensity vs.

Manuscript received March1, 1987. G. Srinivasan and C.E. Patton are with the Department of Physics, Colorado State University, Fort Collins, CO 80523. ,WAVENUMBER k J. G. Booth is with the Departmentof Pure and Applied Physics, University of Salford, Salford4WT, M5 UK. This work was performed while Professor Booth Fig.was 1. Schematic frequency vs. wavenumber dispersion a Visiting Professor at Colorado State University. branches for backward volume waves (BVW) This work was supported by the Electromagnetics propagating parallel to the in-plane applied fieldin Directorate of the RomeAi.r Development Center, Hanscom thin films. The frequencies fH and fg-correspond to Air Force Base,MA., Contract # F19628-85-K-0002. the lower and upper limits for BVW excltations.

0018-9464/87/0900-3718$01.00@1987 IEEE 3719 where H is the in-plane applied field,4aM is the measurements, even though there was some degradation in saturation induction, and 2ay is the gyromzgnetic the microwave impedance matching characteristics of the ratio. overall device.

The dispersion shown inFig. 1 is for the BVW The stripline dimensions and the magnetic parameters excitations with the in-plane wavevectork along fi. for the YIG films used with the various devices are For largek, the excitation profile for the volume given in TableI. The effective 4aM and y values were waves across the film thickness correspondsa toset of determined from ferromagnetic resonazce in-planefor standing waves and the magnon wavenumber perpendicularand out-of-plane fields. The film thickness were to the film plane is quantized accordingkL to= n?r/S, determined by interference techniques, courtesy of where n = 1,2..,and S is the film thickness. Each Westinghouse and Hewlett Packard. branch in Fig.1 corresponds to a particular k solution. The dispersion branches all have nekative so Table I slope, that the group velocity k af /ak is generally negative in the midband region,f < f < fB. This H. negative group velocity is the baslck reason for the Stripline width (w) and YIG film material parameters "BVW" or "backward volume wave" nomenclature. (thickness S, effective saturation induction 4aM and reduced gyromagnetic ratioy) for the signal-to-gAise The objective of this BLS study was to detect (SNE) device and the delay line (DL) device structures directly the BVW excitations which are present in YIGon sapphire and silica. films inMSW device structures and to measure the dispersion properties and scattering profiles for these excitations in actual device configurations. striplinefilm saturation reduced Device width thicknessDevicewidth induction 7 The Device Structures w (pm) S (pm) 4aM (kG)(GHz/kOe) S

The signal-to-noiseenhancer (SNE) and delay line (DL) structures which were used for the BLS investigations are shown schematically in Fig.2. The SNE device in Fig.2(a) consists ofa stripline transducer on an alumina ground plane[5], [6]. A YIG film on a gadolinium gallium garnet (GGG) substrate is in contact with the stripline.In the delay line type device in Fig. 2(b), the actual device configuration involved two different structures, one witha sapphire substrate and relatively narrow(22 pm) stripline transducers, and one with a silica substrate and In order to excite backward volume wavesin the relatively broad (560 pm) transducers. These devices, it is necessary to applya d.c. magnetic field substrates were mounted in an open support structureparallel to the YIG film plane and perpendicularto the for which the metal ground plane was removed over stripline.the For suitable bias fields, volume waves are active region of the device. This open structure was excited in the YIG film by the stripline power. These proved to be very useful in order to conveniently BVW excitations propagate parallel H toand carry power access the film with the laser beam for the BLS away from the stripline. In the SNE device, one observes attenuation of the input microwave power when BVW excitations are present,In the DL device, BVW excitations in the YIG film result in coupling of microwave power to the output transducer.

Typical dataon the applied field dependence of the measured output power are shown in 3Fig. for the DL device on sapphire, an excitation frequency of4 GHz, and an input power of40 mw. These data will also be plane useful for comparison with the results of BLS measurements described in subsequent sections. For fixed frequency measurements, as in Fig.3, it is convenient to think in terms a offield range, HB < H < -stripline HH, for which BVW excitations are supported. The lower fleld limitHB is simply the valueof H from (1) for GGG f = f H is similarly obtained from (2) with f == TheseBfie18 limits are indicated in 3 alongFig. withfH. Y IG the microwave data for theDL device.

silica or The microwave output power vs. field profileFig. in sapphire 3 does show a reasonable correspondence to the BVW I V substrate band, but with several qualifications. First, the power drops off rather rapidly forH > H Second, there is no pronounced cutoff H at= H Bas one might Fig. 2. Diagram showing(a) a signal-to-noise enhancer expect. This is due, presumably, to tEe drop in (SNE) device structure and(b) a delay line (DL) coupling for the higher wavenumber BVW modes as the magnetostatic wave device structure, The yttrium field approachesHH. Third, avery strong coupling is iron garnet film (YIG), the gadolinium gallium garnetevident for fields somewhat below HB. The large peak (GGG) substrate for theYIG, the stripline in intensity for applied fields below the low field BVW transducer(s), and the ground plane support band.edge may be due to surface waves excited in the structures are indicated. YIG film [4]. 3720 - -*O 0 -40 -t a5 5 -60 0 700 900 1100 1300 1500 APPLIEDFIELD H (Oe)

Fig. 3. Variation of the measured microwave output power with the applied field 4 atGHz in theDL device structure on sapphire. The input power was 40 mw. The bias field was parallel to the YIG film Fig. 4. Diagram showing (a) the scattering geometry plane and was perpendicular to the portion of the used for direct detection of BVW excitations in the stripline in contact with the YIG film. The field SNE device consisting of an alumina ground (A),plane , H limits for the BVW excitations B and HH, are a stripline (SL), and a yttrium iron garnet film indicated. (YIG) on a gadolinium gallium garnet (GGG) substrate; (b) the scattering geometry in terms of the incident (k. ), scattered (k ), the magnon (E) Inc It is evident that microwave data leave many open wavevectors ; and (c)'!?he diaphragm and aperture questions about the pertinent device physics. One has arrangement for wavenumber selection. The parameters B no direct way to access the MSW modes which give risek and 4 are defined in the text. to the observed device properties. Previous work[21- [4] has shown that BLS techniques becan used to DL device measurements since the substrates were of complement microwave data on MSW devicesa in transparent silica and sapphire. The limit on the meaningful way by providing specific informationon accessible k - values is determined by the sizeof the dispersion, coupling power vs. wavenumber, and othercollection lens. With the 50 mm focal length, f/1.4 properties of the magnetic signals. The results of lens used in thiswozk, ?ygnons with k - values over light scattering studies on the BVW excitations in the range 0 - 4 X 10 cm can be observed. Scattering these device structures are presented in the following due to magnons with specifick - values canbe examined sections. selectively by placing an aperture on the blocking diaphragm in Fig.4(a) [9]. The angle 4 controls the in-plane wavenumber for the magnons contributingto Experiment scattering. The direction of thein-plane wavevector relative to the applied field is controlled by the aperture orientation, as indicated Fig: in 4(c), where In Brillouin light scattering, the incident laser 8 denotes the angle between and H. k k light is focused onto the material of interest and the scattered light from the material is collected for frequency analysis. The interaction with magnons of frequency f within the material gives risea to frequency shift oft f for the scattered light. One can measure the intensityvs. frequency profile for the scattered light witha Fabry-Perot interferometer[SI. The Brillouin spectrum thus obtained shows intensity I- precision peaks at frequencies corresponding to the magnetic -& , , attenuator excitations which are present in the material. collection lens Since momentum is also conserved in the scattering process, the magnon wavevectork is defined by the scattering geometry. In the present study, we are sweep half wave oscillator interested in the direct detection4volTe of waves with (polarizer)plate polarization wavenumbers over the range 0 < k < 10 cm . A analyzer forward scattering configuration, such as shown in Fig. 4, is essential for the observation of such argon ion excitations. The scattering set-up is shown in Fig. laser 4(a) and the scattering geometry in terms of aperture wavevectors is shown in Fig.4(b). Figure 4(c) shows the details of the diaphragm and aperture arrangement for magnon wavevector selection.

Consider first the forward scatteringset-up inFig. 4(a). Incident laser light is focused onto the YIG film through a hole in the alumina ground plane in Fig.the 5. Schematic diagram showing the Brillouin light SNE device. Such an access hole was not needed for thescattering spectrometer. 3721

Figure 5 shows the overall layout of the is collected. When the collection lens focus is moved spectrometer. Laser light (X = 5145 A) is linearly from a film region near theYIG-GGG interface (as for polarized witha half wave plate and is directed Fig. 6) to a region near the YIG-air/ stripline towards the YIG film in the device. Initially,a back interface, the Stokes peak become more intense than the scattering configuration, indicated by the dotted lineanti-Stokes peak! When the collection optics is in Fig.5, is used to position the YIG atfilm the focused in the middle of the film, YIG the two focus of the collection lens. The forward scattered intensities are approximately equal. One possible light passes througha polarization analyzer and is explanation of equal intensities is that under analyzed witha high contrast multipass Fabry-Perot microwave excitation, the ground state and the excited interferometer (FPI). The photon counting system state of the magnon system are equally pop,ulated and consists ofa low dark count photomultiplier tube give rise to equal scattering intensities for magnon (PMT). The Brillouin spectrum is recorded witha creation and destruction. These points require further multichannel analyzer (MCA). The in-plane bias field investigation. is provided bya Helmholtz pair. A microwave sweep oscillator was used for the stripline excitation over the frequency range from2 to 4 GHz. Maanon Dispersion and Intensity Profiles

A representative Brillouin spectrum forSNE device structure is shown in Fig.6. These data are foran In addition to simply observing magnon peaks in the excitation frequenCy of4 GHz and input power 40of mw. BLS spectra at appropriate frequencies, it is possible An in-planestatic field of900 Oe was applied to measure the dispersion for the excitations and perpendicular to the stripline. This bias field is obtain BVW intensity profiles inMSW device configurations. Selected results for such experiments well within the field limitsHE- HH for volume waves at 4 GHz. The spectrum in Fig.6 was obtained withno section.thispresented are in ,, wavenumber selective aperture in placeso that BVW Dispersion curves ofBVW field position at fixed scattering4 due-fo all possible excitations with k < 4 X 10 cm could be detected. frequency, f= fk, vs. wavenumberk are obtained from intensity vs. field data for specific aperture selected k - values. The aperture selection scheme, following the discussion given above and the diagramsFig. of 4, utilized a 50 pm diameter aperture positionedat various places in the aperture plane, as indicated in Fig. 4(a) and 4(c). Figure 7 shows representative ). profile data for anon-axis aperture, corresponding to t k = 0, and scattering from theSNE device at 4 GHz. v) One would expecta peak in the scattering H at = HB for Z = w k 0. The observed peak, some10 Oe higher, is in I- reasonable agreement with this expectation. The field width for the "resonance" like profile in Fig.7 is z related to the spread kin - values due to the finite size of the aperture.

+15 0 -15 FREQUENCY SHIFT (GHz)

80 -

). - - The two peaks atf 4 GHz in the spectrum of Fig.6 60 correspond to theBVW excitations generated in the 5W - device. The strong central peak at"0" and the I- satellite peaks at-i 15 GHz correspond to the central and adjacent interference order peaks for the laser 40 - line frequency. Spectra similar to the example in Fig. 6, but with various wavevector selection apertures in 2 - place and asa function of field, can be used to obtain - quantitative informationon the BVW excitations in the 20 device under study. -

Some commentson the magnon peak intensities are in order here. At the outset of these investigations, it 700 740 780 820 860 900 was anticipated that theBLS intensity would be stronger for anti-Stokes scattering (destructionof APPLIEDFIELD H (Oe) pumped magnons and up-shiftan in the scattered light frequency) than for Stokes scattering (creationa of new BVW magnon anda down-shift inthe scattered light Fig. 7. Brillouin light scattering intensityvs. frequency). The relative peak intensities in Fig. 6 applied field for theSNE device at 4 GHz with the appear to support this expectation. Upon closer wavenumber selective aperture on the diaphragm in examination, however, it was found that the relative Fig. 4(c) on axis, corresponding to zero magnon intensities for Stokes and anti-Stokes scattering is wavenumber. The theoretical field position forBVW very sensitive to the way in which the scattered lightexcitations with zero wavenumber,HE, is indicate<. 3122

somewhat less rapidly forH > HB, compared to the fall off in Fig.9(b). This is attrlbuted to a stronger coupling to the highk modes for the narrow22 pm stripline transducer than for the wide pm 560 transducer. The fact that the scattering over the < < entire H H HH passband is rather weak is probably due to tie poor coupling, overall, between the transducers and the YIG. Such weak coupling is attributed to YIG - transducer mismatch, the open nature of the structure with no bona fide ground plane, etc.

600 1 I I I I I I I 0 500 10001500 2000 2500 3000 3500 WAVENUMBER k (cm?)

Fig. 8. Data on fieldvs. wavenumber for the BVW excitations propagating parallelto the applied field in the SNE device at 4 GHz. The solid lines represent theoretical field vs. wavenumber dispersion curves for the first four BVW branches.

By utilizing a series of profiles for different aperture positions, one may construct dispersion curves for the device under study. Typical results, again for the SNE device, are shown in Fig.8. The data points represent the peak positions infield, as from Fig. 7, versus wavenumber k. The k - values were controlled 100 simply by positioning the aperturea larger distance away from the optic axis[Fig. 4(c)]. The data in Fig. 8 are for a 0 - value of zero, such that L is k parallel to H. 0 The solid lines in Fig.8 show the theoretical 700 900 1100 1300 1500 dispersion curves for BVW excitations based on the well-known theory of Damon-Eshbach[7]. Curves are shown for the first four BVW branches. These dispersion curves of field vs. wavenumber are essentially "flipped over" versions of the frequencyFig. 9. BLS intensity profiles at4 GHz and 40 mw input vs. wavenumber curves in Fig.1. Figure 8 shows that power for theDL device structures on (a) sapphire the BLS measurements track the expected dispersion forand (b) silica. The field limits for BVW the lowest orderBVW dispersion branch almost excitations, H and HH, are indicated. perfectly. It was not possible to obtain data points B corresponding to the higher order BVW branches. The scattering for these modes was too weak for meaningful measurements. The profiles in Fig.9 also reveal considerable structure. It is to be noted that this structure is The final MSW device result which is unique to reallight and completely reproducible. The structure in the scattering concerns profiles of scattering intensity passband region is most likely todue the preferential vs. field for wavenumber selective aperture in the coupling to BVW excitations at specific wavelengths. collection optics. Such profiles are somewhat The structure is somewhat finer for the narrow analagous to the microwave powervs. field curve shown transducer [Fig. 9(a)] than for the wide transducer [Fig. 9(b)] as one would expect. The large intensities in Fig.3, except that one is probing the actual modes and pronounced structure belowHB are attributed to which carry the power. surface magnetostatic waves[4]. Profiles of this sort are shown in 9Fig. for the two differentDL device substrate/ transducer Summary and Conclusion configurations. The profile in Fig.9(a) is for the sapphire substrate with22 pm wide transducers, while Fig. 9(b) is for silica and 560 pm transducers. As The above results demonstrate the utility of the BLS discussed earlier, the range of MSW wavenumbers whichtechnique for the study BVWof excitations and the are accessed withnp aperture in placeis characterization of two different types of MSW devices. 0 < k < 4 x 10 cm . Both profiles in Fig.9 are for 4 40 It was shown possible to utilize open support GHz microwave excitation and an input powerof mw. structures in combination with transparent substrates The profiles were obtainedby the continuous field to observe forward scatteringon MSW signals and to sweep technique described in Ref.[lo]. measure dispersion and energy flow profiles.

The passband in field for the BVW modes,< H < HB HH, is somewhat more evident for the light scattering Acknowledgements data in Fig. 9 than for the microwave power profiles given earlier, even though the actual scattering is rather weak. The scattering in Fig.9(a) falls off The authors gratefully acknowledge Dr.J. D. Adam, 3723

Westinghouse Research and Development Center, Pittsburgh, Pennsylvania and Dr.W. Ishak, Hewlett [3] G. Srinivasan and C.E. Patton, Appl. Phys. Lett. Packard Laboratories, Palo Alto, California for -47, 761 (1985). providing the device structures and theYIG films used in thiswork, and the reviewer for his commentson the [4] G. Srinivasan, C. E. Patton, and P.R. Emtage, J intensities of Stokes/ anti-Stokes magnon peaks. Appl. Phys.6l, 2318(1987).

151 J. D. Adam, IEEETrans. Magn. l6, 1068(1980).

[6] J. D. Adam and S. N. Stitzer, Appl. Phys. Lett. -36, 485 (1980).

References [7] R. W. Damon and J. R. Eshbach,J. Phys. Chem. Solids, l9, 308(1961).

[l] J. C. Sethares, Proc.1981 Rome Air Development [8] J. R. Sandercock, Top. Appl. Phys.5l, 173(1982). Center Microwave Magnetics Technology Workshop, Hanscom, Massachusetts, June10-11, 1981, RADC-TR- [9] W. Wettling,W. D. Wilber, P. Kabos, and C. E. 83-15, January, 1983. Patton, Phys. Rev. Lett. 5l, 1680(1983).

[2] G. Srinivasan and C. E. Patton, IEEE Trans. Magn.[lo] W. D. Wilber, W. Wettling,E’. Kabos, C. E. Patton, -21, 1797 (1985). and W. Janz, J. Appl. Phys.55, 2533(1984).