1. Introduction
Ferromagnetic-ferroelectric composites have attracted interests in recent years for studies on the nature of magneto-electric interactions (ME) and for use as sensors, memory devices, and for signal processing [
1,
2,
3,
4,
5,
6]. The coupling between the two subsystems is mediated by mechanical strain. An applied AC field
H produces a magnetostrictive strain in the ferromagnetic layer, leading to a voltage response
V in the ferroelectric layer. The ME voltage coefficient (MEVC) = V/(t·H) is a measure of the strength of ME coupling, where
t is the thickness of the ferroelectric layer. Multiferroic composites studied so far include ferrites, manganites, or transition metals/alloys for the ferromagnetic phase and barium titanate, PZT, or PMN-PT for the ferroelectric phase [
6]. A giant low-frequency DME effect was observed in several layered composites [
1,
2,
3,
4,
5,
6,
7,
8,
9,
10]. A related ME phenomenon of fundamental interests is the coupling at bending resonance or electromechanical resonance (EMR) modes in the composite [
6]. When the AC field is tuned to these modes, MEVC increases by orders of magnitude.
A new generation of magnetic field sensors based on layered composites of ferromagnetic and ferroelectric/piezoelectric phases has been reported in recent years [
10,
11,
12,
13,
14,
15]. Such sensors generally require a bias magnet for operation since strong ME interactions are achieved only when a DC bias
Hb is present. In our previous studies we were able to demonstrate the elimination of the need for bias magnetic field with the use of functionally-graded ferromagnetic layer in the composite. In composites of PZT and magnetization (
M) graded ferromagnetic layer consisting of Ni (with 4
πM = 6 kG) and Metglas (4
πM = 21 kG) a strong ME coupling was measured at zero-bias (
Hb = 0) [
16,
17]. The coupling is due to the interaction of out-of-plane internal magnetic field arising from grading in
M and the AC magnetic field
H. Sensors based on these functionally graded multiferroics were shown to have sensitivities somewhat smaller than those sensors operating under a DC bias magnetic field [
11,
12,
13,
14,
15,
16,
17]. The multiferroic magnetic sensors are potentially useful for medical imaging and security-related applications [
18,
19]. At present there are practical difficulties for large scale use of biomedical imaging techniques such as magneto-cardiography due to the need for expensive superconducting quantum interference device (SQUID) sensors that also have the size disadvantage. There is critical need for passive, room-temperature sensors such as ME sensors that could replace SQUID sensors.
This report is on novel pT-magnetic sensors in which a permanent magnet proof mass is used and, therefore, eliminating the need for a bias magnetic field [
5]. Two different types of sensors are discussed: (i) a sensor based on ME coupling in a PZT bimorph with a permanent magnet proof mass and has the advantage of not requiring a ferromagnetic-ferroelectric composite for operation and (ii) a multiferroic sensor of layered PZT-Ni-Metglas. The PZT-bimorph sensor consisted of epoxy bonded two oppositely poled PZT platelets and NdFeB permanent magnet proof mass. A giant magneto-electric effect with MEVC of ~28 V/cm·Oe at low frequencies and enhancement to ~500 V/cm·Oe at bending resonance have been measured for the sensor. The measured equivalent magnetic noise is on the order of 100 pT/√Hz to 10 nT/√Hz at 1–10 Hz. When the AC magnetic field is applied at the bending resonance for the bimorph the measured equivalent magnetic noise is ≈700 pT/√Hz.
In the case of multiferroic sensor with neodymium (NdFeB) magnet proof mass, the ME coupling arises due to magnetostriction in an AC field and also interaction between applied AC magnetic field and M of proof mass. The sensor has been characterized in terms of low frequency and resonance ME effects as a function of the proof mass and noise. It is shown that the use of active proof mass enhances the ME sensitivity at low frequency by an order of magnitude and a corresponding decrease in the magnetic noise. Models have been developed for both types of sensors and are based on equations for the strain and electric displacement of piezoelectric bimorph or ME composite due to interaction between H and M. For finding the low frequency and resonance ME voltage coefficients, we solve elastostatic and electrostatic equations in PZT, taking into account boundary conditions. The MEVC has been estimated as a function of frequency and is found to be in very good agreement with the data. In the sections to follow we discuss the sensor fabrication and characterization and models for the sensors.
Due to its relatively high sensitivity, the ME sensor discussed in this work can be potentially used in cases where measurement of pT-magnetic field is necessary. For instance, it could be a potential candidate for medical imaging applications such as magneto-cardiography in which magnetic fields involved in cardiology events are sensed and dynamic images are reconstructed for disease diagnosis. Compared with other sensors, the ME sensors in this work feature small size, low cost (compared to SQUID), high sensitivity (compared to Hall sensors), and possible miniaturization if MEMS integration technology is utilized.
2. Experiment
The PZT-bimorph sensor studied here is schematically shown in
Figure 1 and consisted of a cantilever of two oppositely poled piezoelectric layers of length 50 mm, width 10 mm, and thickness 0.15 mm. We used commercial PZT (#850 obtained from APC International, PA) platelets that were poled by heating to 400 K and cooling to room temperature in a field of 30 kV/cm. The PZT platelets were bonded to each other with a 2 μm thick (West Systems) epoxy. Similarly, for the multiferroic sensors we used 0.16 mm thick Ni foil (99.8% pure and annealed) and 25 μm thick Metglas ribbons (2605SA1, Metglas, Inc., USA) for the ferromagnetic layer and the same samples of PZT as for the bimorph (# 850, APC) were used. Composites of PZT-Metglas-Ni were made by bonding a 6 cm × 1 cm × 0.03 cm PZT to a single layer of Ni and three layers of 25 μm thick Metglas of similar lateral dimensions. The PZT with silver electrodes was initially poled in an electric field and then was bonded to Ni and Metglas with 2 μm thick epoxy layer.
The PZT-bimorph or PZT-Metglas-Ni composite was clamped at one end and a magnet assembly of two NdFeB magnets was epoxy bonded to top and bottom of the bimorph at the free end as shown in
Figure 1. The use of NdFeB magnet is due to its high output magnetic flux-to-mass ratio, commercial availability and various shapes for convenient applications. The magnets were discs of diameter 5 mm, 10 mm in height, and mass of 2.5 g each. The remnant magnetization
M of NdFeB magnet assembly (along direction
3) was measured to be 15 kG. An AC magnetic field
H generated by a pair of Helmholtz coils was applied parallel to the sample length (direction
1) so that interaction with
M gives rise to a strain in PZT resulting in a voltage
V across the thickness. In the bimorph, since the PZT platelets are poled in opposite directions and the strain produced is compressive in one of them and tensile in the other, the ME voltage in PZT layers (measured across the thickness along direction
3) will be of opposite sign so that the overall ME response is enhanced with the use of a bimorph [
5].
Measurements of ME sensitivity and magnetic noise were carried out by placing the sample in a plexiglass holder in magnetically shielded μ-metal chamber surrounded by an acoustic shield. The sample clamped at one end was subjected to an AC magnetic field
H produced by a pair of Helmholtz coils powered by a constant current source (Keithley, model 6221). The ME voltage generated across the thickness of the bimorph was measured with a signal analyzer (Stanford Research Systems, model SR780). Since the ME voltage across PZT is non-uniform along the length of the bimorph, we measured the ME voltage V close to the clamped end where one expects maximum value [
20]. The ME sensitivity
S =
V/H and the ME voltage coefficient MEVC =
S/t (
t is the PZT thickness) were measured as a function of frequency and at room temperature. Measurements of sensor noise were performed with the signal analyzer and was converted to equivalent magnetic noise.
In the case of the multiferroic sensor, measurements of the ME voltage coefficient as a function of frequency and bias magnetic field and equivalent magnetic noise were first carried out without the proof mass and then with the proof mass.
4. Theory and Discussion
A model for the magneto-electric response of the PZT-bimorph and the multiferroic composite with permanent magnet proof mass is considered next. We first discuss the case of PZT-bimorph and then the composite of PZT-Metglass-Ni. The specific focus is on low-frequency ME response and MEVC versus frequency characteristics around the bending resonance frequency. A cantilever with PZT layers in (x,y) or (1,2) plane as in
Figure 1 is assumed with one end clamped and the permanent magnet assembly on the free end. The thickness of PZT along
z-direction (direction 3) is assumed to be small compared to its length or width. The interaction between the AC magnetic field along direction 1 and remnant magnetization of the magnet along direction 3 gives rise to a piezoelectric strain in PZT. Based on equations of bending vibrations [
20,
25], the general expression for displacement
w in
z direction perpendicular to the sample plane can written as:
The wave number
k is defined by expression:
where
ω is circular frequency,
t is thickness of each layer,
ρ is density, and
D is cylindrical stiffness of cantilever. The integration constant in Equation (1) should be determined from boundary conditions that have the following form for the cantilever with an attached permanent magnet at the free end:
where
m,
v,
I, and
J are mass, volume, moment of inertia of magnet with respect to axis that is positioned in the middle plane along
y axis, and remanent magnetization, respectively;
My is the torque moment relative to
y-axis produced by internal stresses in bilayer per unite width;
Vy is the transverse force per unite width;
H is applied ac magnetic field; and
L is the sample length.
Induced electric field can be found from the open circuit condition
where
1,2D3 is electric induction in first and second piezoelectric layers,
G is the cross-section of sample normal to the
z-axis, and
. Here
d31 and
ε33 are piezoelectric coupling coefficient and permittivity of piezoelectric and
1,2E is internal electric field in layers. The stress components
1,2T can be expressed in terms of strain components
1,2S from elasticity equations
where Y is the modulus of elasticity of piezoelectric component at constant
E and
(
z1,2 is distance of current point of first or second layer from the middle plane). Low frequency MEVC is reduced to the following expression for the 1D case:
where
t,
b,
v, and
J are thickness and width of each piezoelectric layer, volume, and remanent magnetization of magnet, correspondingly.
Finally, one can get the following expression for ME voltage coefficient at EMR with the assumption that the moment of inertia of magnet is negligibly small and
:
where
It can be inferred from Equations (4) and (5) that ME coefficient is substantially determined by product of piezoelectric coefficient and remanent magnetization of magnet. Taking into account the magnet mass results in some variation of ME coefficient. EMR frequencies are determined by roots of denominator. The influence of magnet mass on resonance frequencies is specified by the ratio of tip mass to bilayer mass,
m/m0 and vanishes when this parameter is much less then unity. A dramatic decrease in EMR frequency occur when the proof mass is of the order of bilayer mass. Expanding the denominator of Equation 5 into series in
kL to sixth order and solving this equation for
kL leads to the approximate expression for fundamental EMR frequency:
The peak ME voltage coefficient at bending mode frequency can be estimated as:
with
Q denoting quality factor for bending resonance. Taking the total magnetic moment as proportional to magnet mass, expression for peak ME voltage coefficient can be reduced to form:
where σ is mass magnetization of magnet. Here Q is the quality factor for bending resonance. Resonance losses are taken into account by a using a complex frequency
ω +
iω’ with
ω’
/ω =
1/Q, and Q was estimated from observed resonance profiles
. The following material parameters were used for the calculations:
Y = 0.65 × 10
11 N/m
2, density of PZT
ρ = 7.7 × 10
3 kg/m
3,
d31 = −1750 × 10
−12 m/V,
ε33/
ε0 = 1750,
m = 5 g and
μ0J = 1.5 T.
Theoretical estimates of MEVC
vs. frequency are shown in
Figure 7a. Measured values (in
Figure 2) are also shown for comparison. One observes a very good agreement between theoretical MEVC
vs. f profile and the data. Both the values of MEVC and the bending mode frequency are within 2% of the measured value. Calculated values of the bending mode frequency are plotted as a function of the mass of the permanent magnet in
Figure 7b. One infers the following from the results in
Figure 7. (i) The cantilever arrangement facilitates electromechanical resonance at low-frequencies compared to longitudinal or thickness acoustic modes; (ii) It is possible to control the resonance frequency with proper choice for the mass of the permanent magnet; and (iii) assuming a linear increase in M with the magnet mass, any decrease in the resonance frequency with increasing m will be accompanied by an increase in the peak MEVC.
Similarly, we derive expression for the MEVC for trilayer of Ni, Metglas, and PZT with active tip mass. A cantilever with Ni, Metglas, and PZT layers in (
x,y) or (
1,
2) plane is assumed with one end clamped and the permanent magnet assembly on the free end. The thickness of the trilayer is assumed to be small compared to its length or width. As opposed to above calculation for PZT bimorph, the expression for the wave numbed takes on the form:
The torque moment
My and transverse force per unite width
Vy that enter into the boundary conditions are calculated over the total volume of the trilayer. Finally, assuming the moment of inertia of the magnet to be negligibly small, and electromechanical and magnetomechanical coupling coefficients to be small compared to unity, we obtain expression for MEVC:
with dimensionless factors:
Assuming the total magnetic moment to be proportional to magnet mass, expression for ME voltage coefficient can be reduced to form:
Equations (9) and (10) show that the contribution of magnetization of the proof mass to ME coupling is similar to that of the piezomagnetic components. The expression enclosed in first square brackets of numerator includes the terms corresponding to the proof mass magnetization and piezomagnetic coupling in two magnetostrictive layers. For zero tip mass, ME response is stipulated by the product of piezoelectric and piezomagnetic coefficients. A non-zero passive load (J = 0) results in an increase in the total ME response. This increase is a linear function of m/m0. When employing an active tip mass with J ≠ 0, the term quadratic in m is added to the product of piezoelectric and piezomagnetic coefficients. In addition, the load mass gives rise to a decrease in EMR frequency.
The approximate expression for fundamental EMR frequency and peak ME voltage coefficient at bending mode frequency can be estimated as:
where
and t are the average density and total thickness of the sample.
The model developed here, therefore, provides several avenues for tailoring the bending resonance frequency and MEVC to achieve desired sensitivity and reduce noise. The sensor of magnetic fields discussed in this report has several unique advantages over traditional sensors such as the Hall effect sensors, SQUID sensors, or magnetoelectric composite sensors. The proposed sensor is passive, does not require a bias magnet for operation, operates at room temperature, and has cost advantages as well. Although temperature certainly will affect the sensor response, it is not the focus of the discussion in this manuscript. Moreover, at this exploratory stage, the external magnetic field is usually generated using a pair of Helmhotz coils, there is no room for a temperature chamber for temperature characterization. Future research will certainly include comprehensive study of sensor responses including the temperature effects.