ApplPhysLett_94_182505

Electric?eld tunable60GHz ferromagnetic resonance response in barium ferrite-barium strontium titanate multiferroic heterostructures Young-Yeal Song,1,a?Jaydip Das,1Pavol Krivosik,1,2Nan Mo,1and Carl E.Patton1

1Department of Physics,Colorado State University,Fort Collins,Colorado80523,USA

2Slovak University of Technology,81219Bratislava,Slovakia

?Received9March2009;accepted17April2009;published online5May2009?

A magnetic-ferroelectric?lm heterostructure with a large electric?eld tuning of the ferromagnetic

resonance?FMR?mode was fabricated.Pulse laser deposited30nm thick Pt electrodes and3?m

thick barium strontium titanate?lms on Nb-doped strontium titanate substrates were capped with an

unbonded200?m thick single crystal in-plane c-axis barium hexaferrite slab.The structure gives

a60GHz FMR frequency shift of16MHz at a bias of29V,for an average response of0.55MHz/V.

The maximum incremental tuning response at29V was 1.3MHz/V.This is a hundredfold

improvement over previous results.?2009American Institute of Physics.

?DOI:10.1063/1.3131042?

Ferromagnetic-ferroelectric?FM-FE?1–4as well as FM-piezoelectric5–9heterostructures have attracted attention because of their potential for use in multifunctional devices. The simultaneous magnetic and electric tunability in such structures can be especially useful for microwave and milli-meter wave planar devices such as tunable phase shifters,1 resonators,3and delay lines.10Enhancement in the electric tunability of the magnetic response for the small control volt-ages across the FE element is the critical factor for practical applications.

For a typical5GHz FM-FE heterostructure,the FM resonance?FMR?hybrid mode frequency can be tuned through the voltage controlled relative dielectric permittivity ?r of the FE layer.As shown in Ref.3,the tuning for a

nonmonolithic structure comprised of a nominal7?m thick yttrium iron garnet?YIG??lm in contact with a500?m thick barium strontium titanate?BSTO?slab is about100 MHz for an operating point of5GHz and a400V control voltage across the BSTO,with a corresponding FMR mode tunability of0.25MHz/V.Very recently,Das et al.4have reported on YIG/BSTO thin?lm heterostructures with a tun-ability of2MHz/V at9.5GHz.This was achieved for an applied voltage across the5?m thick BSTO layer of only 25V.

For further advances,two factors are important:en-hancement in the electric?eld tuning response for the FMR mode and extension to higher frequencies.The?rst is needed for the realization of small,light,and low power devices. The second is needed for wide bandwidths.There has been some recent work of https://www.sodocs.net/doc/bd17235939.html,tinov and Srinivasan8have made pasted structures comprised of Al substituted barium ferrite?BaM?thin?lms with the c-axis perpendicular to plane and lead zirconate titanate slabs.The tunability was 0.01MHz/V at about100GHz and800V.Das et al.11have made a high quality BaM-BSTO thin?lm structure but no electric?eld tuning of the FMR mode was reported.

This report describes substantial progress for both fre-quency and tuning.The new heterostructures are made of 200?m thick in-plane c-axis oriented single crystalline BaM?IPCA-BaM?slabs in combination with pulse laser de-posited?PLD?3?m thick BSTO?lms.The BaM slab is placed on top of the BSTO?lm with no mechanical coupling or bonding.The electromagnetic boundary problem solution for the mode of the whole structure yields the hybrid mode. This effect is present even without any direct mechanical coupling at the interface.3Intuitively,magnetic layers with an in-plane?IP?easy axis anisotropy and an easy direction IP ?eld should show larger FM-FE coupling.This is because?1?the transverse magnetization response is generally larger than for the perpendicular case and?2?the out-of-plane com-ponents produce sizeable dipole?elds that penetrate the FE layer more effectively.In the present case,the tunability at 60GHz was about0.55MHz/V with a maximum shift of16 MHz at a voltage of29V.The maximum incremental tuning response was1.3MHz/V,a factor of130improvement over previous results.8

The limiting factor for the tuning is in the relatively thin BSTO layer.Calculations based on Ref.3indicate that cou-pling between FE and FM layer scales with integral of the dynamic dipole?eld from the FM layer over the FE layer. This penetration integral increases for thicker FE layers.One can expect,therefore,that thicker FE layers will produce even larger tuning rates than achieved here.Such projected high tuning rates would represent a major breakthrough in the development of monolithic multiferroic millimeter wave devices.The present results point the way to the realization of such a new class of magnetoelectric devices.

By way of introduction,note that hybrid mode tuning depends on the effect of the voltage controlled?r value on the nominal FMR mode frequency.The FE and FM layers, taken alone,have distinct frequency?f?versus wave number ?k?dispersion curves.For the FE layer alone,one has a linear dispersion of the form?FE=ck/

??

r

.For the FM layer,the low wave number spin wave dispersion curve sits essentially at the k-independent FMR frequency?FMR.For the hetero-structure,these FE and FM dispersion curves can cross and produce a hybrid mode gap.In the usual FE system such as BSTO,an applied voltage causes?r to decrease and the slope of the FE??k?dispersion line to increase.This,in turn, causes an upward shift in the hybrid mode gap and an in-crease in the FMR operating point frequency.The gap is also higher at higher frequencies.This means that millimeter

a?Electronic mail:yysong@https://www.sodocs.net/doc/bd17235939.html,.

APPLIED PHYSICS LETTERS94,182505?2009?

0003-6951/2009/94?18?/182505/3/$25.00?2009American Institute of Physics

94,182505-1

wave frequency operation can also increase the tuning re-sponse.

As indicated above,the approach here was to use a high quality BaM single crystal slab as the FM element and a thin PLD epitaxial BSTO ?lm as the FE element.The BaM slab was in the form of a small 200?m thick 0.06?0.17cm 2rectangle.The high anisotropy BaM FM element gave access to the millimeter regime of frequencies for applied magnetic ?elds in the 3–4kOe range.The thin BSTO layer allowed the realization of a large electric ?eld tuning of ?r at rela-tively modest voltages.The BSTO layer was grown by PLD on a 0.1?0.2cm 2?100?plane Nb-doped SrTiO 3?Nb-STO ?substrate with a resistivity of about 0.005?cm.This con-ducting substrate also served as the bottom electrode.The energy ?uence and repetition rate for the 248nm KrF exci-mer laser was 1.7J /cm 2and 25Hz,respectively.The depo-sition time was 45min,the substrate temperature was 750°C,and the target substrate distance was 7.8cm.These conditions produced a nominal 3?m thick BSTO ?lm.For the top electrode,a 30nm thick room temperature PLD plati-num layer was then laid down.The IPCA-BaM slab was placed on top of the stack to complete the heterostructure.It is important to emphasize that there is no glue or epoxy at the interface that would expand the gap.This provides for the strongest possible coupling between the FM and FE layers.

Graphs ?a ?and ?b ?in Fig.1show standard ?/2?diffrac-tion scans and data on the relative dielectric constant ?r ver-sus applied voltage,respectively,for the BSTO layer.The diffraction peaks in ?a ?are from the ??00?plane scattering of the cubic structure of both the BSTO ?lm and the Nb-STO substrate.Notice that there are no second phase peaks.Note also that the ?200?peak intensities are very strong.These data show the growth of the cubic epitaxial BSTO layer on the conducting ?100?Nb-STO substrate.

The relative dielectric constant ??r ?–voltage ?V ?response of the BSTO ?lm in Fig.1?b ?was obtained from the 100kHz capacitance measurements for a range of electrode voltages from ?26to +26V.The ?r at V =0,while somewhat lower than the 2000–5000values for bulk BSTO,12,13is typical of

reported values for ?lms.13,14The tunability at the 26V limit of the data is about 25%,which appears to be somewhat smaller than that reported in Ref.13.The data also show a small amount of hysteresis.

The graphs in Fig.2show ?a ?representative room tem-perature vibrating sample magnetometry ?VSM ?hysteresis loop data in a magnetic induction ?4?M ?versus static ?eld ?H ?format and ?b ?60GHz absorption derivative versus ?eld FMR results obtained with a standard ?eld swept shorted wave guide technique.For ?a ?,data were obtained for easy and hard direction IP ?elds,as shown by the inset diagrams.For ?b ?,the data shown are for easy direction IP ?elds only.Graph ?b ?shows data for zero and 29V bias voltage ?V ?across the BSTO layer,as indicated.The inset shows the structure and bias arrangement for the dashed curve FMR response.The solid points identify the main mode FMR cen-ter ?eld positions in each case.The focus here is on the V =0FMR pro?le.The voltage driven FMR shifts will be con-sidered shortly.

The VSM data in ?a ?indicate saturation induction ?4?M s ?and effective uniaxial anisotropy ?eld ?H A ?values of about 3.9kG and 16.7kOe,respectively.These values are in nominal accord with previous work on bulk and thin ?lm BaM materials.15,16The V =0FMR pro?le in ?b ?gives main mode FMR ?eld ?H 0?and peak-to-peak derivative linewidth values of 3685and 21Oe,respectively.Based on the static 4?M s and H A values just cited,a literature value 17of the BaM gyromagnetic ratio ???,with ???/2?=2.87GHz /kOe,and a sample shape correction based on the Aharoni slab demagnetizing factor analysis,18the theoretical 60GHz uni-form mode FMR ?eld is 3730Oe.The 45Oe mismatch with the data is within the normal error limits for such compari-sons.Note further that the application of a 29V bias voltage across the 3?m thick BSTO layer shifts the FMR ?eld by about 6Oe.When converted to an equivalent frequency shift at ?xed ?eld,this corresponds to an FMR tuning factor of about 0.55MHz/V

.

FIG.1.?Color online ??a ?BSTO ?lm x-ray diffraction 2?scans.The labels for the BSTO and Nb-STO Miller indices identify the various peaks from the cubic structure.?b ?Measured relative dielectric constant ?r vs voltage V applied across the BSTO layer.The circles show the data for positive to negative and negative to positive ?eld sweeps,as

indicated.

FIG.2.?Color online ??a ?Hysteresis loops of the magnetic induction 4?M s vs IP static ?eld H .The open circles and squares show easy axis and hard axis loops,respectively,for the 200?m thick IPCA-BaM slab.?b ?Repre-sentative 60GHz FMR absorption derivative vs ?eld pro?les for the FM-FE heterostructure comprised of the IPCA-BaM slab from ?a ?placed on top of the BSTO ?lm/electrode structure used for the data in Fig.1.Data are shown for applied voltages of 0and 29V across the BSTO layer,as indicated.

Turn now to a more detailed discussion of the electric ?eld tuning response.Figure 3shows the two critical com-ponents of this response.The graph shows the down shift in the FMR ?eld ??H ?with applied voltage ?V ?and the corre-sponding ?xed ?eld FMR frequency shift ??f ?,based on the standard Kittel equation.For the current operating point pa-rameters,the ?eld-to-frequency shift conversion factor ?f /?H equivalent is very close to ???/2?.The ?H ?V ?and the ?f ?V ?data were all obtained for increasing voltage,as indi-cated by the arrows beside the curves.Because of the small amount of hysteresis evident in Fig.1?b ?,the corresponding data for a decreasing voltage would be slightly shifted.The fact that the ?xed frequency FMR ?eld decreases with de-creasing ?r ,along with the corresponding increase in the ?xed ?eld FMR frequency,is consistent with the qualitative scenario given in the introduction.It is noteworthy that the tuning response is not linear in the ?f ?V ?and ?r ?V ?response.That is,the ?H ?V ?and the equivalent ?f ?V ?response are nonlinear while the ?r ?V ?in Fig.1?b ?is essentially linear.

These results show that the equivalent ?xed ?eld FMR frequency shift is 16MHz at 29V .This ?f value is a factor of two larger than the shift from Ref.8that was obtained for a voltage that was a factor of 25larger.This 16MHz shift at 29V corresponds to an average frequency tuning response of 0.55MHz/V .Note that the incremental frequency tuning re-sponse,e.g.,the slope of the response,is about 1.3MHz/V at 29V .In absolute terms,the current tuning response in MHz/V is 130times the value from Ref.8.This large tuning response is even more remarkable if one considers the theo-retical prediction noted in the introduction that thick FE lay-ers yield a higher tunability.The current results,moreover,also indicate very good magnetoelectric coupling,even for these relatively thin FE layers.This means,by implication,

that optimized structures in terms of thickness for the FM and FE layers as well as extensions to monolithic multilayer con?gurations should be able to give incredible levels of magnetic tuning through the application of modest electric ?elds.

This work was supported in part by the ARO-MURI Grant No.W911NF-04-1-0247,the ONR Grant Nos.N00014-07-1-0597and N00014-08-1-1050?through Sub-award No.PT103701-SC101157from Virginia Common-wealth University ?,and seed funding from the ARO-DARPA Grant No.W911NF-06-1-0163.The authors gratefully ac-knowledge Philips Research Laboratory,Hamburg,Germany and Dr.W.Tolksdorf for providing the single crystal BaM materials and Dr.P.McCurdy of the Department of Chemis-try at Colorado State University for the scanning electron microscopy ?lm thickness measurements.

1

W.J.Kim,W.Chang,S.B.Qadri,H.D.Wu,J.M.Pond,S.W.Kirchoe-fer,H.S.Newman,D.B.Chrisey,and J.S.Horwitz,Appl.Phys.A:Mater.Sci.Process.71,7?2000?.2

H.Zheng,J.Wang,S.E.Lo?and,Z.Ma,L.Mohaddes-Ardabili,T.Zhao,L.Salamanca-Riba,S.R.Shinde,S.B.Ogale,F.Bai,D.Viehland,Y .Jia,D.G.Schlom,M.Wuttig,A.Roytburg,and R.Ramesh,Science 303,661?2004?.3

https://www.sodocs.net/doc/bd17235939.html,tinov,V .S.Tiberkevich,G.Srinivasan,A.N.Slavin,A.A.Semenov,S.F.Karmanenko,B.A.Kalinikos,J.V .Mantese,and R.Ramer,J.Appl.Phys.100,093905?2006?.4

J.Das,Y .Y .Song,N.Mo,P.Krivosik,and C.E.Patton,Adv.Mater.?unpublished ?.5

Y .Jia,H.Luo,X.Zhao,and F.Wang,Adv.Mater.?Weinheim,Ger.?20,4776?2008?.6

Y .Chen,J.Wang,M.Liu,J.Lou,N.X.Sun,C.Vittoria,and V .G.Harris,Appl.Phys.Lett.93,112502?2008?.7

J.Ryu,S.Priya,K.Uchino,and H.E.Kim,J.Electroceram.8,107?2002?.8

https://www.sodocs.net/doc/bd17235939.html,tinov and G.Srinivasan,Appl.Phys.Lett.93,142503?2008?.9

S.Shastry,G.Srinivasan,M.I.Bichurin,V .M.Petrov,and A.S.Ta-tarenko,Phys.Rev.B 70,064416?2004?.10

Y .K.Fetisov and G.Srinivasan,Appl.Phys.Lett.87,103502?2005?.11

J.Das,B.Kalinikos,A.R.Barman,and C.E.Patton,Appl.Phys.Lett.91,172516?2007?.12

H.Zhang,X.Yao,and L.Zhang,J.Am.Ceram.Soc.90,2333?2007?.13

L.C.Sengupta and S.Sengupta,IEEE Trans.Ultrason.Ferroelectr.Freq.Control 44,792?1997?.14

J.Im,O.Auciello,P.K.Baumann,S.K.Streiffer,D.Y .Kaufman,and A.R.Krauss,Appl.Phys.Lett.76,625?2000?.15

R.Karim,K.D.McKinstry,J.R.Truedson,and C.E.Patton,IEEE Trans.Magn.28,3225?1992?.16

Y .Y .Song,J.Das,Z.Wang,W.Tong,and C.E.Patton,Appl.Phys.Lett.93,172503?2008?.17

C.R.Buf?er,J.Appl.Phys.33,1360?1962?.18

A.Aharoni,J.Appl.Phys.83,3432?1998?

.

FIG.3.?Color online ?The 60GHz FMR ?eld shift ?H and corresponding constant ?eld FMR frequency shift ?f vs voltage V applied across BSTO layer in the FM-FE heterostructure,as indicated.The solid lines are guide to the eye.All data shown are for increasing voltage.