1 Magnetoelectric Effects in Ferromagnetic/Piezoelectric Multilayer Composites

MAGNETOELECTRIC EFFECTS IN FERROMAGNETIC/PIEZOELECTRIC MULTILAYER COMPOSITES

G. Srinivasan, C. P. DeVreugd, R. Hayes, M.I. Bichurin* and V.M. Petrov*

Physics Department, Oakland University, Rochester, MI 4830, USA *Novgorod State University, B. S. Peterburgskaya St. 41, 173003 Veliky Novgorod, Russia

ABSTRACT

The observation of strong magnetoelectric effects is reported in thick film bilayers and multilayers of ferrite-lead titanate zirconate (PZT) and lanthanum nanganite-PZT. The ferrites used in our studies included pure and zinc substituted cobalt-, nickel- and lithium ferrites. Samples were prepared by sintering 10-40 µm thick films obtained by tape-casting. Measurements of ME voltage coefficients at 10-1000 Hz indicated a giant ME effect in nickel ferrite-PZT, but a relatively weak coupling in other ferrite-PZT and manganite-PZT systems. Multilayers prepared by hot pressing was found to show a higher ME coefficient than sintered samples. Evidence was found for enhancement in ME coefficients when Zn was substituted in ferrites. The Zn-assisted increase was attributed to low anisotropy and high permeability that resulted in favorable magneto-mechanical coupling in the composites. We analyzed the data in terms of our recent comprehensive theory that takes into account actual interface conditions by introducing an interface coupling parameter. Theoretical longitudinal and transverse ME voltage coefficients for unclamped and clamped samples are in general agreement with data. From the analysis we inferred excellent interface coupling for nickel zinc ferrite-PZT and weak coupling for other layered systems.

1. INTRODUCTION films made by tape casting and characterized by This work is concerned with the fabrication of structural, magnetic, and electrical measurements. novel ferromagnetic-ferroelectric thick film Magnetoelectric measurements were made both at multilayers and studies on the nature of low (10 Hz–1 kHz) and high frequencies (9-10 GHz). magnetoelectric interactions. In such two-phase Key findings and accomplishments are as follows. (i) composites, the magnetoelectric (ME) coupling is The first observation of ME coupling in lanthanum mediated by mechanical stress [1]. An applied ac strontium manganite-PZT [7]. (ii) A giant low magnetic field produces dynamic deformation in frequency ME interaction in ferrite-PZT [8-11]. (iii) ferromagnets due to magnetostriction and results in Ultrahigh ME effects at electromechanical resonance an induced electric field due to piezoelectric effect. (EMR). (iv) Analysis of low frequency ME data The systems of interest in the past were bulk samples using a model for a bilayer that allows the estimation of nickel or cobalt ferrite with BaTiO3 or lead of an all-important interface coupling constant zirconate titanate (PZT) that showed ΜΕ coupling [12,13]. (v) Theoretical models for resonance ME much smaller than predicted values [2-5]. The main coupling: at EMR for the piezoelectric phase and cause is low resistivity for ferrites that gives rise to a ferromagnetic resonance for the ferromagnetic phase leakage current and limits the electric field for [14-16]. Our efforts so far have resulted in orienting the dipoles, leading to loss of charges and considerable progress toward an understanding of poor piezoelectric coupling. Such problems could be ME interactions in layered systems [8-19]. The eliminated in a layered structure [6]. composites are candidate materials for We recently initiated studies on layered magnetoelectric memory devices, smart sensors, and heterostructures. The main emphasis of our work has electric or magnetic field controlled signal-processing been on ferromagnetic manganite-PZT and ferrite- devices [20]. PZT. Samples were synthesized by sintering thick 2

In this article, first we provide a brief review of coefficient α′E was measured for two conditions: (i) low frequency ME interactions in two-phase product transverse or α′E,31 for H and δH parallel to each property composites. Section 2 deals with ME other and to the disk plane (1,2) and perpendicular to effects in bulk composites. In Section 3, our δE (direction-3) and (ii) longitudinal or α′E,33 for all investigations on thick film layered systems and the the three fields parallel to each other and observation of giant ME effects are described. perpendicular to sample plane. Theoretical models for bilayers and comparison with Figure 1 show representative data on H data are considered in Section 4. dependence of α′E at 300K and 100 Hz for bulk composites of 50 wt.% ferrite and 50% PZT 2. MAGNETOELECTRIC EFFECTS IN [11,19,25]. The sample with NFO shows a higher ME BULK COMPOSITES voltage than for the composite with CFO. A general We provide here a review of past efforts by increase in α′E is observed with H to a peak value, other groups and our current investigations on bulk followed by a rapid drop. The coefficients are ferromagnetic-ferroelectric oxides. directly proportional to the piezomagnetic coupling The magnetoelectric effect is defined as the q=δλ/δH, where λ is the magnetostriction, and the H- dielectric polarization of a material in an applied dependence tracks the slope of λ vs H. Saturation of magnetic field or an induced magnetization in an λ at high field leads to α′ =0. For most ferrites, λ = external electric field [21]. The induced polarization E // P is related to the magnetic field H by the expression, 2 λ⊥ [19,26] and one expects α′E,33=2 α′E,31. The α′E values are an order of magnitude smaller than P = α H, where α is the second rank ME- theoretical estimates due to poor piezoelectric susceptibility tensor. A sample of piezomagnetic- coupling and leakage current resulting from low piezoelectric phases is expected to be composite resistivity [6]. In summary, the low magnetoelectric since α = δP/δH is the product of the composite resistivity reduces the strength of ME piezomagnetic deformation δz/δH and the interactions in bulk samples. piezoelectric charge generation δQ/δz [1]. We are interested in the dynamic ME effect; for an ac 6 CFO-PZT

) Bulk magnetic field δH applied to a biased sample, one e 4 O e g a t cm / l 2 V measures the induced voltage δV. The ME voltage o V (m E 1 3 M E, 0 ' e α coefficient α′E = δV/t′δH and α = εo εr α′E where t′ is s t er v en i

s -2 c fi the composite thickness and εr is the relative an r ef T o -4 permittivity. The (static) effect, first observed in C -6 antiferromagnetic Cr2O3, is weak in single-phase -15 -10 -5 0 5 10 15 compounds [22-24]. Static Magnetic Field (Oe) )

e 10

Bulk composites of interest in the past were O 50 wt.% NFO-50% PZT

m Longitudinal c

/ Bulk NiFe2O4 (NFO) or CoFe2O4 (CFO) with BaTiO3 [2- V 8 (m E ' α

5]. We synthesized similar bulk composites, but with t 6 en i c i f

PZT [11,19]. The two oxides were mixed in a ball- ef 4 o e C mill and disk shaped samples were prepared by g a t

l 2

o Transverse V

traditional sintering at 1400-1500 K or microwave E M 0 sintering at 1200 K. X-ray diffraction showed no 0 500 1000 1500 2000 2500 3000 impurities. Magnetic and electrical characterization Bias Magnetic Field H (Oe) Fig.1: Bias field dependence of ME voltage yielded parameters consistent with expected values coefficients in bulk composites of 50 wt.% ferrite and for a simple mixed phase. Silver electrodes were 50% PZT. deposited on the samples for poling with an electric field perpendicularly to its plane. Samples were 3. LAYERED COMPOSITES placed between the pole pieces of an electromagnet The strongest ME coupling is expected in a fitted with Helmholtz coils for ME studies and were layered structure due to (i) the absence of leakage subjected to a dc field H and ac field (10-1000 Hz) current and (ii) ease of poling to align the electric δH=1 Oe. The resulting voltage δV=t′δE was dipoles and strengthen the piezoelectric effect. measured across the sample thickness. The ME 3

Harshe, et al., proposed such structures, provided a expected values due to either higher than expected theoretical model for a bilayer and prepared conductivity of PZT films or the presence of “shorts” multilayers of CFO-PZT or BaTiO3 that showed very in the PZT films since the porosity is on the order of weak ME coupling. 5-10%. The ferroelectric-to-paraelectric transition Our efforts have been mainly on bilayers and temperature of 600 K agreed with expected values multilayers (MLs) of lanthanum manganites-PZT and [10]. ferrite-PZT [8-14]. Layered composites were synthesized using thick films of manganites, ferrites 3.1. LANTHANUM MANGANITE-PZT and PZT obtained by tape-casting [27]. The Our studies on manganite-PZT resulted in the ferrite/manganite powder necessary for tape casting first report on ME coupling in the system. Lanthanum was prepared by the standard ceramic techniques that manganites with specific concentration of divalent involved mixing the oxides or carbonates of the substitutions such as Ba, Ca, or Sr show metallic constituent metals, followed by presintering and final conduction and ferromagnetism due to the double sintering. A ball-mill was used to grind the powder exchange interactions [28]. The oxides are of interest to submicron size. For PZT films, we used for studies on ME coupling because of high λ, low commercially available powder. The fabrication of resistivity, and structural homogeneity with PZT. thick films contained the following main steps: a) Bilayers and multilayers of La0.7Sr0.3MnO3 (LSMO)- preparation of cast of constituent oxides; b) PZT and La0.7Ca0.3MnO3 (LCMO)-PZT were deposition of 10-40 µm thick films tapes by doctor synthesized. Since the ME voltage is generated blade techniques; and c) lamination and sintering of exclusively in the piezoelectric phase in layered composites. Ferrite or PZT powders were mixed with samples, we measured αE for unit length of the a solvent (ethyl alcohol), plasticizer (butyl benzyl piezoelectric phase and is related to α′E through the phthalate), and binder (polyvinyl butyral) in a ball expression αE=α′E(t′/t) where t′ and t are the mill for 24 hrs. The slurries were cast into 10-40 µm composite and PZT thickness, respectively. tapes on silicon coated mylar sheets using a tape Representative results on H-dependence of αE for caster. The films were dried in air for 24 hrs, manganite-PZT are shown in Fig.2. The data at 120 removed from the mylar substrate and arranged to K is for a LSMO-PZT bilayer. The room temperature obtain the desired structure. They were then value was a factor of 2 smaller than at 120 K [7,10]. laminated under high pressure (3000 psi) and high temperature (400 K), and sintered at 1375-1475 K. Bilayers were made with 200 µm thick ferrite or 60 120 K

) 40 e manganite and PZT. Multilayers consisted of (n+1) e O ag t l layers of ferrites/manganite and n layers of PZT (n = o 20 cm. / V

v E

5-30). m ( M 0 e s Structural characterization was carried out on ent er ci i

f -20 f sv

powdered multilayers and layered samples using an n a LSMO-PZT Bilayer Coe x-ray diffractometer. Two sets of well-defined peaks, Tr -40 corresponding to the magnetic and piezoelectric -60 phases, were present. The data implied a strain-free -400 -200 0 200 400 spinel structure for the ferrites. But the manganite Static Magnetic Field (Oe) and PZT layers had a strained tetragonal lattice, with a volume reduction of 1-2% [9]. Magnetic Fig.2: Transverse ME voltage coefficient vs H for characterization included magnetization with a lanthanum strontium manganite-PZT bilayer with Faraday balance and a vibrating sample 200 µm thick manganite and PZT layers. magnetometer, ferromagnetic resonance at x-band and magnetostriction with a strain gage. The Strong ME interactions are evident in Fig.2 and one saturation magnetization agreed with bulk values. observes unique hysteresis and remanance. The Measurements of electrical resistance R and longitudinal coefficient (not shown here) was a factor capacitance C were carried out to probe the quality of of 2-3 weaker than the transverse effect. We the composites. The R and C were smaller than 4

measured a stronger ME coupling in LSMO-PZT largest ever measured [29-31]. In than in LCMO-PZT. Although XRD shows no impurity phases, the 400

) Transverse sample parameters were found to be very sensitive to e e g a O 200

t sintering temperature. In addition, the ME effect was l o cm / Longitudinal V weaker in 10-40 micron thick multilayers compared c V i r m t ( c 0 to 200 micron thick bilayers. It further weakens when t e l n e e o the layer thickness was reduced or the number of ci i f et f e layers was increased. These observations point to the gn o -200 a C interface diffusion as the possible cause of poor ME M effects, a serious problem that needs to be resolved -400 [7,10]. -2000 -1000 0 1000 2000

3.2. FERRITE-PZT AND GIANT Static Magnetic Field H (Oe)

MAGNETOELECTRIC EFFECTS We succeeded in achieving giant ME effects Fig.3 Magnetoelectric (ME) voltage coefficient αE = predicted by theory in bilayers and multilayers of δE/δH versus bias magnetic field H for a multilayer ferrite-PZT. A series of oxides, including pure and consisting of 15 layers of NFO and 14 layers of PZT. Zn substituted nickel-, cobalt, and lithium ferrites The data at room temperature and 100 Hz are for was used for the ferromagnetic phase. transverse (out-of-plane δE perpendicular to in-plane δH) and longitudinal (out-of-plane δE and δH) field 3.2.1. NICKEL ZINC FERRITE-PZT orientations. There is 180 deg. phase difference We found evidence for a giant ME effect in between voltages for +H and –H. bilayers and multilayers of nickel zinc ferrite (NZFO), Ni1-xZnxFe2O4 (x=0-0.5), and PZT [8,9]. summary (i) the ME coupling is enhanced in layered samples compared to bulk and (ii) a giant ME effect Representative data on H dependence of αE are shown in Fig.3 for a multilayer with 15 layers of is evident for NFO-PZT. We recently developed a theoretical model that facilitates quantitative NFO and 16 layers of PZT. As H is increased, αE,31 increases, reaches a maximum value of 400 mV/cm information on interface bonding and is considered in Oe and then drops rapidly to zero. One observes a Section 4. noticeable hysteresis in the field dependence. The

variation of αE,33 with H is linear up to 1000 Oe. The ) 1000 e e g a O longitudinal ME effect is almost an order of t l m o c / magnitude weaker than the transverse effect. We V V E m

observed a phase difference of 180 degrees between ( t 100 se M the induced voltages for +H and -H. As discussed en er ci i f later, the magnitude and the field dependence in Fig. f e ansv r Co 3 are related to variation in q with H. The key T observation in Fig.3 is the giant ME coupling that is 10 0246 any order of magnitude stronger than in bulk NFO- PZT (Fig.1). A similar ME coupling, but 10% higher Volume Ratio R than in multilayers was measured for bilayers of Fig.4: Variation of the transverse coefficient αE,31 m p NFO-PZT [8]. The variation in ME coupling with the with R = v/ v, forNFO-PZT multilayer. The solid volume for the two phases was studied in NFO-PZT. line represents theoretical values for a two-layer

Figure 4 shows the measured dependence of αE on structure. Values of αE,31 are for the bias field H the ratio of volumes v of the magnetostrictivve (m) corresponding to maximum value in the ME effect at and piezoelectric (p) phases, R = mv/pv, for NFO- 100 Hz.

PZT. An exponential increase in αE,31 occurs with v and shows a maximum of 1500 mV/cm.Oe for R = Similar ME studies were performed on nickel zinc ferrite-PZT samples with x = 0-0.5 [9]. Figure 5 2.2. The αE-values in NFO-PZT are one of the 5

shows representative data on the H-dependence of αE Finally, the results discussed so far are for layered for NZFO-PZT samples with x = 0-0.4. The data samples made by traditional sintering. Since the were obtained on samples with n=10-15 at room interface bonding is a critical factor that determines temperature for a frequency of 100 Hz. For NFO- the strength of ME coupling, we prepared similar

PZT (x=0), αE,31 vs H shows a resonance-like samples by hot-pressing. The procedure involved character with a maximum centered at H = 400 Oe. sintering at high pressure (5000 psi) and high When Zn is substituted for Ni, we notice an increase temperature (1300 K) of preheated multilayers. ME in the peak value of αE,31 for low x-values. As x is data for such samples are shown in Fig.7 and increased, a down-shift is observed in H-value compared with data for traditionally sintered samples. One observes doubling of peak ME voltage corresponding to maximum αE,31. The magnetic field range for strong ME effects decreases with increasing coefficient for hot-pressed case compared to sintered Zn content. Data on the longitudinal coupling samples. This study is in progress. showed a similar behavior, but the ME coupling is ) realized over a wide field range. The variations of e 250 O m maximum α with x are plotted in Fig.6. Average c Hot-pressed NZFO (x=0.2) - PZT

E / V values of αE and the spread are shown. The data

(m 200 31

reveal a 60% increase in the transverse ME voltage α

t coefficient as x is increased from 0 to 0.2, followed n e i

c 150 fi by a reduction in αE,31 for higher x. The longitudinal f coefficient also shows a similar character. e Co e g

a 100 t l ) o e 700 V ic O E r x = 0.2 Sintered t m c c M / 50

V 525 ele se r o m e ( x = 0.4 x = 0 NZFO-PZT v s

gnet 350 an cient r Ma fi 0

f T e

s 0 200 400 600 800 1000 175 er Coe

sv Bias Field H (Oe)

n age a lt r 0 o 0 200 400 600 800 1000 T V Fig.7: Low frequency ME voltage coefficient vs H Static Magnetic Field (Oe) data for hot-pressed and sintered multilayers of NZFO (x-0.2)-PZT. Fig.5: ME voltage coefficients versus H data for multilayer samples of Ni1-xZnxFe2O4 (NZFO) – PZT Now we comment on the results for NZFO-PZT with R=1. layered composites. The most significant inferences in Fig.2-6 are: (i) for equal volume of ferrite and e

e g a

800 100 t ) ) ag l PZT, the maximum αE,31 ranges from 400 mV/cm Oe t e e l NZFO-PZT o o O

O V V

in multilayers to 460 mV/cm Oe in the bilayer, (ii) E m m E c c M / / 600 75

M V V

e αE,31 increases with increasing volume of nickel n s i (m (m d 3 1 er 3 3 u t

E, 400 50 E, ferrite and the largest measured value is 1500 mV/cm i α α

t t m p n ansv r Oe for v/ v =2.2 and (iii) Zn-substitution in NFO en Lo cien ci i i 200 25 m T f f m results in the strengthening of ME coupling with the u ef ef mu o o m C C xi xi

a maximum occuring for x=0.2. These ME coefficients 0 0 a M M 0 0.1 0.2 0.3 0.4 0.5 0.6 must be compared with 20 mV/cm Oe for Cr2O3, the Zinc Concentration x best single phase ME material [22-24]. It is more Fig.6: Zinc concentration dependence of maximum than an order of magnitude higher than reported transverse (circles) and longitudinal (squares) ME values for ferrite – BaTiO3 bulk composites, and and coefficients in NZFO – PZT layered samples with a factor of five larger than in laminated composites of R=1. The line is guide to the eye. Ni(Co,Cu)-Mn ferrite – PZT [5,30].

3.2.2. COBALT ZINC FERRITE – PZT 300 ) e

O CZFO-PZT Next we consider studies on samples of CZFO- e

g m a t c

l transverse PZT [9,25]. Figure 8 shows representative data on / o

V 200 the H dependence of α and α for a multilayer V

E,31 E,33 E (m E α M sample in which 40% Co is replaced by Zn in cobalt t m u ferrite. The data at room temperature and 100 Hz en 100 m ci i f axi are for a sample with n = 10. As the bias field is ef longitudinal M o C increased from zero, αE increases rapidly to a peak 0 value. With further increase in H, the ME coefficients 0 0.1 0.2 0.3 0.4 0.5 0.6 drop to a minimum or zero value. There was no Zn concentration x noticeable hysteresis or remenance in αE vs. H. Consider the H dependence of transverse and Fig.9: Variation of peak transverse and longitudinal longitudinal coefficients. Although overall ME voltage coefficients with zinc concentration x in layered CZFO – PZT. The data at room temperature ) e and 100 Hz were obtained from profiles as in Fig.8. O 300 m

c The bars indicate the range of measured values. The /

V 200 CZFO-PZT transverse x = 0.4 solid lines are guide to the eyes. (m E

α 100

en 0 measured on several samples and the figure shows ci i f f the range of measured values and their average. As e

o -100

C the Zn

e longitudinal -200 ag

t substitution is increased, one observes a sharp l o -300 V increase in αE,31, from 50 mV/cm Oe for x = 0 to 280 E -3000 -2000 -1000 0 1000 2000 3000 M mV/cm Oe for x = 0.4. Further increase in x is Static Magnetic Field (Oe) accompanied by a substantial reduction in αE,31. The longitudinal coupling parameter is very weak for the Fig.8:Static field dependence of room temperature entire series. One needs to compare the results in transverse and longitudinal ME voltage coefficients, Figs.8-9 with past studies on bulk samples of CFO- αE,31 and αE,33 respectively, for a multilayer BaTiO3 and CFO-PZT and multilayers of CFO- 16 composite of Co1-xZnxFe2O4 (CZFO) (x=0.4)–lead PZT. Bulk samples showed very weak ME zirconate titanate (PZT). The sample contained 11 interactions (see Fig.1) but layered CFO-PZT showed layers of CZFO and 10 layers of PZT with a αE of 75 mV/cm Oe, comparable to values in Fig.9 thickness of 18 µm. [6]. It is obvious from the present study that Zn substitution in cobalt ferrite is a key ingredient for features in Fig.7 are similar for both cases, one finds strong ME coupling in multilayers [9]. We attribute the following differences. (i) The initial rate of the efficient field conversion properties to increase in αE with H is much higher for the modification of magnetic parameters due to Zn transverse case than for longitudinal orientation for (Section 4). the fields. (ii) The peak-αE,31 is a factor of five higher than αE,33. (ii) The peak value in αE,33 3.2.3. LITHIUM ZINC FERRITE - PZT occurs for a higher bias field than for the transverse case. These observations could be understood in Finally, we consider studies on lithium zinc terms of H variation of parallel and perpendicular ferrite-PZT composites [32]. Samples with n=10 and magnetostriction for the ferrite. 15, a layer thickness of 15 micron and ferrites of

Similar αE vs H data were obtained for composition Li0.5-x/2ZnxFe2.5-x/2O4 for x=0-0.4 were samples with x-values varying from 0 to 0.6. Both synthesized. Figure 10 shows data on H-variation of

αE,31 and αE,33 were measured. Figure 9 shows the the transverse ME voltage coefficient for x = 0-0.3. variation of maximum αE with x for both transverse The data are for a frequency of 100 Hz at room and longitudinal cases. The peak coefficients were temperature. To our knowledge, these data constitute 7

the first report of strong ME coupling in lithium necessary to quantify less-than-ideal interface ferrite – PZT composites of any kind. Important coupling [6]. We present here a comprehensive observations are as follows. (i) Data show features theory in which the composite is considered as a similar to the other two systems. (ii) A factor of five homogeneous medium with piezoelectric and increase in the peak αE,31 value is evident when x is magnetostrictive subsystems. Important aspects of increased from 0 to 0.3. (iii) An up-shift in the H- the model are as follows. (i) We take into account value corresponding the peak αE,31 is seen as x is less-than-ideal interface conditions by introducing an increased from 0. Recall that for CZFO-PZT and interface coupling parameter k. (ii) Expressions for NZFO-PZT samples, a down-shift in H for maximum longitudinal and transverse ME coefficients are

αE,31 is observed for increasing x. (iv) The H-interval obtained using the solutions of elastostatic and for strong ME effects is essentially independent of x. electrostatic equations. (iii) We consider a third field

The inset in Fig.10 shows data on peak values of αE,31 orientation of importance in which E, δE, H, and δH vs x. One notices a rapid increase in the ME voltage are in the sample plane and parallel to each other and coefficient with x for x = 0- 0.3, followed by a sharp is referred to as in-plane longitudinal ME coupling decrease for x=0.4 [32]. (αE,11 ). (iv) The theory is developed for two types of measurement conditions: unclamped and clamped 120 120 LZFO-PZT

100 bilayers. (v) The ME voltage coefficients are ) e

e x = 0.3 80 g 31 O α a

t 60 estimated from known material parameters l eak o

P 40 cm

/ 80 V

(piezoelectric coupling, magnetostriction, elastic V 20 E

m 0 M ( 00.10.20.30.4

t constants, etc.,) and are compared with data. e n

s x

e x = 0.2 er

ci 40 We consider a bilayer in the (1,2)-plane i f sv f n e

a consisting of piezoelectric and magnetostrictive r Co T x = 0 phases as shown in Fig.11. Since most studies have 0 0 200 400 600 dealt with polycrystalline thick films for the two Static Magnetic Field (Oe) constituents, we do not assume any epitaxial characteristics for the layers. An averaging method is Fig. 10: Transverse voltage coefficient versus H used for deriving effective composite parameters and 17-19 profiles for multilayer samples of Li0.5- is carried out in two stages. In the first stage, the x/2ZnxFe2.5-x/2O4 (LZFO) - PZT. The inset shows peak sample is considered as a bilayer. In the second stage, value of αE,31 vs x. the bilayer is considered as a homogeneous medium.

4. Theory 3

We developed a model for ME effects in E, δE piezoelectric phase layered samples for an understanding of the giant ME magnetostrictive phase effects [12,13]. Previous attempts just considered H, δH 2 longitudinal ME voltage coefficient αE,33 for ideal coupling at the interface [6]. The major deficiencies 1 of the earlier model are as follows. (i) For the longitudinal case, influence of the finite magnetic Fig.11: Schematic diagram showing a bilayer of permeability for the ferrite was ignored. A reduction piezoelectric and magnetostrictive phases in the (1,2) in the internal magnetic field and weakening of ME plane. Field orientations for longitudinal interactions are expected due to demagnetizing fields. magnetoelectric (ME) effect are also shown. (ii) The model did not consider ME coupling under transverse field orientations for which studies on ferrite-PZT show a giant ME effect. (iii) It is

The theory yields the following expressions for the longitudinal and transverse voltage coefficients for unclamped bilayers [12,13].

p m -2µ0k(1-v) d13 q13 α = ------× (1) E,33 p 2 p T p p m m 2 ( d13) (1-v)k + ε 33 [( s11 + s12)(v-1)- k( s11 + s12)v]

p p m m [( s11 + s12)(v-1)- ( s11 + s12)kv] × ------m m m p p m 2 2 [µ0(v-1)- µ33v][kv( s11 + s12)-( s11 + s12)(v-1)]+ 2 ( q31) kv

p m m -k(v-1) d13 ( q11 + q12) α = ------. (2) E,31 m m p T p p p T p 2 ( s11 + s12) ε 33 kv + ( s11 + s12) ε 33 (1-v) –2 ( d13) k(1-v)

p -12 2 p -12 2 m Here d and q are the piezoelectric and piezomagnetic s11 = 15*10 m/N, s12 =-5*10 m/N, s11 = -12 2 m -12 2 coupling coefficients, respectively, s is the 6.5*10 m/N; s12 =-2.4*10 m/N, and ε33/ε0 = T compliance coefficient, ε is permitivity at constant 1750 [6,9,33]. The measured value for d33 in the bulk p p stress, µ is the tensor permeability and v = v/( v + and layered samples was 250 pm/V, corresponding to m v). The parameter k is the interface coupling d31 = 0.5 d33 = 125 pm/V. parameter, with k=1 for ideal coupling and k=0 for Theoretical estimates of ME coupling for NZFO the case with no friction. – PZT indicated very good agreement with data for We now use the model for theoretical estimates the entire series of Zn substitution [9]. of αE for comparison with the data. Calculated αE,33 Representative results are shown in Fig.13 for x=0. are expected to be quite small due to weak q13 and The estimates are for a series of interface coupling k demagnetizing fields. Such a prediction is in and q-values obtained from magnetostriction data in agreement with observations in all of our systems Fig.12 and other material parameters mentioned discussed in Sec.3. Discussion to follow is therefore earlier for ferrites. The data were obtained on a restricted to transverse ME coupling. We estimated multilayers with n=15 and a layer thickness of 15

αE,31 vs H for manganite-PZT and ferrite-PZT using µm. One observes good agreement between theory bulk values for s, ε and d, and q=q11+q12 obtained for k=1 and data. There is excellent agreement in the from λ vs H data as in Fig. 12. The following values magnitude of αE,31, but the theory predicts a sharper were used in Eq. (2) for the composite parameters: drop in αE,31 at high fields than observed experimentally.

10 400

) k = 1

0 e NFO-PZT perpendicular NFO O

-10 m c / ) parallel V 300 0.8 m (

ppm -20 t ( n o en i i t

-30 c 0.6 c i fi

r 0 500 1000 1500 2000 t ef s 200

40 o o t C e

perpendicular E

gn 0.4

a 0 M M se r

-40 CFO e 100 v 0.2 s n

-80 a r

parallel T -120 0 1000 2000 3000 4000 5000 0 0 500 1000 1500 Static Magnetic Field (Oe) H (Oe) Fig.13: Theoretical (lines) and measured (circles)

Fig.12: Room temperature in-plane parallel (λ11) αE,31 versus H for NFO – PZT composites. and perpendicular (λ12) magnetostriction versus H for NFO and CFO bulk samples made from thick A similar comparison of the data and films. theoretical values of αE,31 for CZFO-PZT, however, revealed a poor interface coupling [9]. 9

Representative results are shown in Fig.14 for x = 0. 1 We observe a substantial disagreement between NZFO-PZT theory and data. Neither the magnitude of αE,31 for 0.75 k

k=1 nor its H dependence agree with the data. The nt a

predicted values for perfect interface coupling are an t s n

order of magnitude higher and H-values for o

C 0.5 maximum αE,31 are a lot smaller than measured ng i l values. A weak interface coupling, with k on the p CZFO-PZT u o order of 0.1, is evident for x=0. We noticed a similar C 0.25 behavior for x=0.2. Magnitudes of theoretical αE,31 for k=0.2 were are in agreement with the data and was indicative of a stronger interface coupling than in 0 x=0. A general improvement in the ME coupling is 00.10.20.30.40.5 thus accomplished with Zn substitution in the ferrite. Zn Concentration x It is appropriate to compare the results of theory for the two ferrite-PZT systems. In order to Fig. 15: Interface coupling k as a function of Zn facilitate such a comparison, the interface coupling substitution x in ferrites. obtained from theory are shown in Fig.15. One draws the following inferences from Fig.13-15. First, inferred from the calculations. Possible causes of the for CZFO–PZT, there is total lack of agreement wide variations in k are discussed in the following between theory for k=1 and data. Although section. The key accomplishment here is the ability magnetostriction data implies strong piezomagnetic to obtain information on interface coupling [9]. Analysis of the low frequency ME data was also 700 k = 1 ) performed for LZFO-PZT composites [32]. A weak e CFO-PZT O interface coupling with k = 0.2 was obtained for cm

/ 0.8 V 525 LFO-PZT, and k increased with Zn substitution in the m (

t ferrite. n e i

c 0.6 A similar theoretical analysis was carried out for i f

ef 350 o LSMO-PZT and the results are shown in Fig.16. The C 0.4 E magnetostriction data needed for q-values is also M e

s shown. The calculated values for k=1 are an order of 175 ver 0.2 magnitude higher than the data [7]. ns a

r T

) e O

0 1000 2000 3000 4000 5000 m 25 25 250 c

/ λ 20 12

H (Oe) V λ ) 15 11

m m

( 20 200 p 10 t

(p λ

nt 5 en e i

0 y c 15 150 r m Fig.14: Comparison of theoretical and measured i i f o r

f -5 e

e 0 500 1000 1500 2000 2500 e p values of the transverse ME voltage coefficient αE,31 o H (Oe) 10 100 Th C

E for layered samples of CFO– PZT. The solid curves - Ex 5 50

are theoretical values for a series for interface se M Transverse er coupling parameter k. v s 0 0 n

a 0 100 200 300 400 500 r T H (Oe) coupling, the estimated αE,31 are factor of 2-10 higher than measured values. Second, the introduction of Zn leads to an enhancement in the Fig.16: Comparison of theoretical estimates and data strength of interface coupling k and αE, in particular for transverse ME voltage coefficient in LSMO-PZT for CZFO-PZT. The constant k increases from 0.1 to bilayers. 0.6 when Zn progressively replaces 40% of Co. A near perfect coupling for NZFO-PZT is In summary, the theory and analysis presented here 10

provide an elegant means to quantify an otherwise dependence of αE,31 on H. There are two types of complex parameter k. The analysis reveal excellent magnetostriction in a ferromagnet: (i) Joule interface coupling only for NZFO-PZT layered magnetostriction associated with domain movements systems. We need to point out an important limitation and (ii) volume magnetostriction associated with of the model used here. The model is valid for a magnetic phase change. The volume simple bilayer structure. It is necessary to extend the magnetostriction is not important in the present theory to include a multilayer consisting of n- situation since it is significant only at temperatures interfaces in a structure with (n+1)-ferrites and n-PZT close to the Curie temperature. In ferrites domains layers. One expects a stronger interface coupling in are spontaneously deformed in the magnetization multilayers than in bilayers. direction. Under the influence of a bias field H and ac field δH, domain wall motion and domain rotation 4.1 ORIGIN OF GIANT ME EFFECTS contribute to the Joule magnetostriction. Since the Next we comment on the possible causes of ME coupling involves dynamic magneto-mechanical giant ME effects and the inferred excellent coupling coupling, key requirements for the ferrite are parameter k only in NZFO-PZT. One expects k to be unimpeded domain wall motion, domain rotation and dependent on a variety of factors including structural, a large λ. A soft, high initial permeability (and low mechanical, chemical and electromagnetic anisotropy as indicated by FMR studies) ferrite, such parameters. X-ray diffraction data indicated the as NFO, is the key ingredient for strong ME effects. absence of any structural abnormalities. Mechanical In magnetically hard cobalt ferrite, however, one has bonding at the interface that arises due to the high the disadvantage of a large anisotropy field that limits temperature processing of the composite is another domain rotation. Our magnetization measurements key factor that determines k. In some recent works, yielded an initial permeability of 20 for NFO vs 3.5 the bonding between magnetostrictive and for CFO. Figure 17 shows the composition piezoelectric media was accomplished with the use of dependence of the initial permeability µi for CZFO silver epoxy [30]. The understanding of interface and NZFO [33]. With increasing Zn concentration in coupling in such cases is further complicated by the CZFO-PZT, µi increases by a factor of 30 to a introduction of a foreign material. In our sintered maximum value for x = 0.5 in agreement with composites it is difficult to quantify the strength of anticipated reduction in the anisotropy. For NZFO, mechanical bonding. Simple peel tests are useful for µ shows an order of magnitude increase for the such information, for example, in metal-polymer i composition range of importance. Even though µ samples. We are not aware of such tests for sintered i and α tracks each other in CZFO-PZT, a similar composites. One also expects k to be influenced by E,31 behavior is absent in NZFO-PZT. It is clear that the the possible presence of chemical inhomogeneities at important factor is not just µi, but the magneto- the interface. An observation of importance in this 1/2 regard is the deterioration in ME coupling in CZFO- mechanical coupling km given by km = (4πλ′µr/E) PZT samples sintered at the high end of the where λ′ is the dynamic magnetostrictive constant temperature range 1375-1475 K. Although x-ray and µr is the reversible permeability, parameters diffraction data did not indicate any detectable levels analogous to q=(q11+q12) and µi, respectively, and E of impurities, a “fused” interface and sample warping is the Young’s modulus [34]. In Fig.18, we compare indicate microscopic chemical inhomogeneities in maximum values of αE,31 and the product µiq as a CZFO-PZT that will have adverse impact on k. function of x for both systems. The factors rise and Finally we address the anticipated effects of fall in tandem and the maxima occur around the same electromagnetic parameters on k. For answers, one composition, i.e., x = 0.2-0.3 for NZFO-PZT and 0.3- needs to focus on the effects of magnetostriction and 0.4 for CZFO- PZT. Thus it is logical to associate magnetoelastic coupling for the following reasons. the strong interface coupling for the entire series of PZT is the only piezoelectric phase used in all the NZFO-PZT and Zn-rich CZFO-PZT with efficient composites and its lattice is strained in both systems. magneto-mechanical coupling. The selection of composites and measurements are directed toward examining the role of ferrites/manganites on (i) ME voltage and (ii) the 11

350 35 that result in favorable magneto-mechanical coupling

300 30 in the composites. References i i

250 25 µ

y 1. J. van Suchtelen, “Product properties: A new t ty i l ili i b b

200 20 a application of composite materials,” Philips Res. a e e m r

rm Rep., 27, 28 (1972). e 150 15 Pe l P l

a 2. J. van den Boomgaard, D. R. Terrell, and R. A. J. i ia it

it CZFO

NZFO n n 100 10 I I Born, “An in situ grown eutectic magnetoelectric composite material: Part 1: Composition and 50 5 unidirectional solidification,” J. Mater. Sci. 9, 1705 0 0 (1974). 00.20.40.6 3. A. M. J. G. van Run, D. R. Terrell, and J. H. Zn Concentration x Fig.17: Composition dependence of the initial Scholing, “An in situ grown eutectic magnetoelectric composite material: Part 2: Physical Properties,” J. permeability µ (from Ref.33) for CZFO and NZFO. i Mater. Sci. 9, 1710 (1974). The lines are guide to the eye. 4. J. van den Boomgaard, A. M. J. G. van Run, and

700 10 J. van Suchtelen, “ Piezoelectric-piezomagnetic composites with magnetoelectric effect,”

600 e Ferroelectrics 14, 727 91976. g ) 8 a e lt O o 5. J. van den Boomgard and R. A. J. Born, “A 500 )

V NZFO-PZT e m

c E O / / sintered magnetoelectric composite material BaTiO3- V M u 6 400 se (m Ni(Co,Mn)Fe O ,”J. Mater. Sci. 13, 1538 (1978). r 1 em 2 4 3 e -6 E, v 0 α

6. G. Harshe, Magnetoelectric effect in t (1

300 i ans

4 µ r

11 piezoelectric-magnetostrictive composites, PhD T ci i q f 200 ef thesis, The Pennsylvania State University, College o mum i C x CZFO-PZT 2

a Park, PA, 1991; G. Harshe, J. P. Dougherty and R. E.

M 100 Newnham, “Theoretical modelling of multilayer 0 0 magnetoelectric composites,” Int. J. Appl. 0 0.10.20.30.40.50.6 Electromag. Mater. 4, 145 (1993); M. Avellaneda and Zn Concentration x G. Harshe, “Magnetoelectric effect in Fig.18: Comparison of Zn substitution dependence of piezoelectric/magnetostrictive multilayer (2- maximum values of αE,31 and the product qµi for 2)composites,” J. Intell. Mater. Sys. Struc. 5, 501 NZFO-PZT and CZFO-PZT. The parameter q = (1994). q11+q12 is the net piezomagnetic coupling constant 7. G. Srinivasan, E. T. Rasmussen, B. J. Levin, and estimated from the slope of λ vs H data. R. Hayes, “Magnetoelectric effects in bilayers and multilayers of magnetostrictive and piezoelectric 5. CONCLUSIONS perovskite oxides,” Phys. Rev. B 65, 134402 (2002). Studies on layered samples of lanthanum 8. G. Srinivasan, E. T. Rasmussen, J. Gallegos, R. manganite-PZT and pure and zinc substituted ferrites Srinivasan, Yu. I. Bokhan, and V. M. Laletin, - PZT show evidence for strong ME interactions. Of “Magnetoelectric bilayer and multilayer structures of particular interest is the giant ME voltage coefficients magnetostrictive and piezoelectric oxides,” Phys. in nickel ferrite-PZT. The voltage coefficient αE Rev. B 64, 214408 (2001). show an overall increase with increasing Zn 9. G. Srinivasan, E. T. Rasmussen and R. Hayes, concentration x in CZFO-PZT and NZFO-PZT. A “Magnetoelectric effects in ferrite-lead zirconate maximum in αE occurs for x = 0.2-0.4, depending on titanate layered composites: Studies on the influence the ferrite. Analysis of the data using our model for a of zinc substitution in ferrites,” submitted to Phys. bilayer reveals ideal interface conditions for NZFO- Rev. B. PZT. The data implies poor coupling in other ferrite- 10. G. Srinivasan, E. T. Rasmussen, A. A. Bush, K. PZT and manganite-PZT. The Zn assisted E. Kamentsev, V. F. Meshcheryakov, and Y. K. enhancement in ME coefficient is primarily due to Fetisov, “Structural and magnetoelectric properties of low anisotropy and high permeability for the ferrites 12

MFe2O4-PZT (M=Ni,Co) and Lax(Ca,Sr)1-xMnO3- 22.D. N. Astrov, “Magnetoelectric effect in PZT multilayer composites,” Appl. Phys. A. (2003) chromium oxide,” Soviet phys. JETP 13, 729 11. G. Srinivasan, R. Hayes, E. T. Rasmussen, D. (1961). Feckel, V. M. Laletin, N. Paddubnaya, “Giant 23. G. T. Rado and V. J. Folen, “Observation of Magnetoelectric Effects in Layered Composites of magnetically induced magnetoelectric effect and Nickel Zinc Ferrite and Lead Zirconate Titanate,” evidence for antiferromagnetic domains,” Phys. Rev. Solid State Commun. 124, 373 (2002). Lett. 7, 310 (1961). 12. M. I. Bichurin, V. M. Petrov, and G. Srinivasan, 24. S. Foner and M. Hanabusa, “Magnetoelectric “Theory of low-frequency magnetoelectric effects in effects in Cr2O3 and (Cr2O3)0.8 (Al2O3)0.2,” J. Appl. ferromagnetic-ferroelectric layered composites,” J. Phys. 34, 1246 (1963). Appl. Phys. 92, 7681 (2002). 25. G. Srinivasan, R. Hayes, C. P. De Vreugd, U. 13. M. I. Bichurin, V. M. Petrov, and G. Srinivasan, Laletsin, and N. Pudubnaya, “Dynamic “Theory of low-frequency magnetoelectric coupling magnetoelectric effects in bulk and layered in magnetostrictive-piezoelectric bilayers,” Phys. composites of cobalt zinc ferrite and lead zirconate Rev. B. 68, 054402 (2003). titanate,” in press, Appl. Phys. A. 14. M. I. Bichurin, I. A. Kornev, V. M. Petrov, A. S. 26.J. Ryu, A. V. Carazo, K. Uchino, and H. Kim, Tatarenko, Yu. V. Kiliba, and G. Srinivasan, Phys. “Piezoelectric and magnetoelectric properties of lead Rev. B 64, 094409 (2001). zirconate titanate/Ni-ferrite particulate composites,” 15. M. I. Bichurin, V. M. Petrov, Yu. V. Kiliba, G. J. Electroceramics, 7, 17 (2001). Srinivasan, “Magnetic and magnetoelectric 27.R. E. Mistler and E. R. Twiname, Tape Casting: susceptibilities of a ferroelectric/ferromagnetic Theory and Practice, The American Ceramics composite at microwave frequencies,” Phys. Rev. B Society, Westerville, Ohio (2000). 66, 134404 (2002). 28. A. P. Remirez, “Colossal magnetoresistance,” J. 16. M. I. Bichurin, D. A. Fillipov, V. M. Petrov, U. Phys.:Condens. Matter 9, 8171 (1997). Laletsin, and G. Srinivasan, “Resonance 29.T. G. Lupeiko, I. V. Lisnevskaya, M. D. magnetoelectric effects in layered magnetostrictive- Chkheidze, and B. I. Zvyagintsev, “Laminated piezoelectric composites,” in press, Phys. Rev. B. magnetoelectric composites based on nickel ferrite (2003). and PZT materials,” Inorganic Mater. 31, 1139 17. M. I. Bichurin, V. M. Petrov, I. A. Kornev, (1995). A. S. Tatarenko, Yu. V. Kiliba, N. A. Konstantinov, 30. J. Ryu, A. V. Carazo, K. Uchino, and H. Kim, and G. Srinivasan, “Resonance magnetoelectric effect “Magnetoelectric properties in piezoelectric and in multilayer composites,” Ferroelectrics 280, 187 magnetostrictive laminate composites,” Jpn. J. Appl. (2002). Phys. 40, 4948 (2001). 18. M. I. Bichurin, V. M. Petrov, and G. 31. K. Mori and M. Wuttig, “Magnetoelectric Srinivasan, “Modeling of magnetoelectric effect in coupling in terfenol-D/polyvinylidenedifluoride ferromagnetic/piezoelectric multilayer composite,” composites,” Appl. Phys. Lett. 81, 100 (2002). Ferroelectrics 280, 165 (2002). 32. “Low frequency and microwave 19. V. M. Laletin and G. Srinivasan, magnetoelectric effects in thick film heterostructures “Magnetoelectric effects in composites nickel ferrite of lithium ferrite and lead zirconate titanate,” G. and barium lead zirconate titanate,” Ferroelectrics Srinivasan, R. Hayes and M. I. Bichurin, in press, 280, 177 (2002). Solid State Commun. 20. V. E. Wood and A. E. Austin, “Possible 33. Landolt-Bornstein; Numerical data and applications for magnetoelectric materials,” Proc. functional relationships in science and technology, Symposium on Magnetoelectric Interaction Group III, Crystal and Solid State Physics, vol 4(b), Phenomena in Crystals, Seattle, May 21-24, 1973, Magnetic and Other Properties of Oxides, eds. K.-H. eds. A. J. Freeman and H. Schmid, Gordon and Hellwege and A. M. Springer, Springer-Verlag, New Breach Science Publishers, New York, (1975), p.181. York (1970). 21.L. D. Landau and E. M. Lifshitz, Electrodynamics 34. C. M. Van der Burgt, Philips Res. Reports 8, 91 of Continuous Media, Pergamon Press, Oxford (1953). (1960) p.119 (Translation of Russian Edition, 1958)