Magnetoelectric gradiometer with enhanced vibration

Magnetoelectric gradiometer with enhanced vibration rejection efficiency under H-field modulation Junran Xu, Xin Zhuang, Chung Ming Leung, Margo Staruch, Peter Finkel, Jiefang Li, and D. Viehland

Citation: Journal of Applied Physics 123, 104501 (2018); doi: 10.1063/1.5017726 View online: https://doi.org/10.1063/1.5017726 View Table of Contents: http://aip.scitation.org/toc/jap/123/10 Published by the American Institute of Physics

Articles you may be interested in Band-offset-induced lateral shift of valley electrons in ferromagnetic MoS2/WS2 planar heterojunctions Journal of Applied Physics 123, 104301 (2018); 10.1063/1.5012775

JOURNAL OF APPLIED PHYSICS 123, 104501 (2018)

Magnetoelectric gradiometer with enhanced vibration rejection efficiency under H-field modulation Junran Xu,1,a) Xin Zhuang,1 Chung Ming Leung,1 Margo Staruch,2 Peter Finkel,2 Jiefang Li,1 and D. Viehland1 1

Department of Materials Science and Engineering, Virginia Tech, Blacksburg, Virginia 24060, USA The Naval Research Laboratory, 4555 Overlook Ave. SW, Washington, DC 20375, USA

2

(Received 29 November 2017; accepted 22 February 2018; published online 9 March 2018) A magnetoelectric (ME) gradiometer consisting of two Metglas/Pb(Zr,Ti)O3 fiber-based sensors has been developed. The equivalent magnetic noise of both sensors was first determined to be about 60 pT/冑Hz while using an H-field modulation technique. The common mode rejection ratio of a gradiometer based on these two sensors was determined to be 74. The gradiometer response curve was then measured, which provided the dependence of the gradiometer output as a function of the source-gradiometer-normalized distance. Investigations in the presence of vibration noise revealed that a ME gradiometer consisting of two ME magnetometers working under H-field modulation was capable of significant vibration rejection. The results were compared to similar studies of ME gradiometers operated in a passive working mode. Our findings demonstrate that this active gradiometer has a good vibration rejection capability in the presence of both magnetic signals and vibration noise/interferences by using two magnetoelectric sensors operated under H-field modulation. Published by AIP Publishing. https://doi.org/10.1063/1.5017726

I. INTRODUCTION

Magnetoelectricity (ME) is a property that results in an energy/power conversion between electrical and magnetic forms. Generally, ME materials exhibit an electrical property change induced by the application of a magnetic field or vice versa. In the 1960s, the ME effect was first found in Cr2O3 single-crystals;1 however, the ME effect in single phase materials is very weak. In order to realize much higher ME properties, two-phase composites of piezoelectric and magnetostrictive materials have been developed which exhibit a ME product tensor property.2 The magnetic flux sensors with strong ME effect have been developed, while an input magnetic flux signal is converted to an output electrical charge.3 The underlying principle is that a magnetic signal induces a magnetization change in the magnetostrictive layer which results in an induced shape change, and this shape change is then transmitted across a bonded interface to the piezoelectric layer, which in turn converts the strain to a voltage/ charge output with the help of the piezoelectric effect.4 Laminate composites of piezoelectric fibers and Metglas foils have been reported to have extremely high ME properties.5,6 A longitudinal-longitudinal (LL) multi-push-pull structure was developed using these two layers and interdigital electrodes on both the top and bottom sides of piezoelectric fibers. Because of a high in-plane piezoelectric coupling k33,p, a giant ME coefficient of 52 V/(cm Oe) was found at sub-resonant frequencies, which was increased to about 1000 V/(cm Oe) at the mechanical resonant frequency.7,8 As a consequence of such giant ME gain coefficients, optimized passive magnetoelectric sensors can achieve an extremely

a)

Author to whom correspondence should be addressed: [email protected]

0021-8979/2018/123(10)/104501/8/$30.00

low equivalent magnetic noise (EMN) which was reported as 5.1 pT/冑Hz at 1 Hz.7 Gradiometers have been widely used in magnetic field/ flux detection.9–12 Based on the above mentioned advances of ME sensors, magnetometers have been developed, which consist of a pair of ME sensors. Previously, gradiometers based on two passive ME sensors have been studied for low frequency sensing applications.10 Passive mode ME sensors are capable of detecting low frequency signals. Coupling between piezoelectric and magnetostrictive layers is achieved by a mediating mechanical stress/strain. Since the piezoelectric layers are inherently sensitive to mechanical stimuli, the ME composite also partially picks up microphonic noise. At low frequencies, this noise can be substantial. This is one of the major limitations of passive ME sensors. Thus, it is important that gradiometers be capable of rejecting microphonics. Previously, Xing et al. showed that microphonics could in part be rejected by built-in structures that effectively included an acoustic gradiometer incorporated into a magnetic sensor.13 The differential output was shown to double the ME gain coefficient while providing some rejection of microphonics. The built-in approach was that the gradiometer was operated simply by adding/subtracting signals in an analog manner. Higher rejection efficiencies would require more precise matching of the sensor sizes and signal outputs or computational analysis. To overcome these limitations, Zhuang et al. developed an active mode ME sensor driven under H-field modulation, which used a frequency shift modulation technique.14 The ME composites were excited by an external sinusoidal signal: when this signal was a harmonic of the magnetic field one, it is designated as an H-field modulation.15 An applied low frequency magnetic signal then modulates the excitation signal via the nonlinearity of the magnetostrictive layer of

123, 104501-1

Published by AIP Publishing.

104501-2

Xu et al.

the laminate, producing two side-band signal peaks around the excitation signal frequency in the spectrum similar to a classic amplitude modulation (AM) method.16,17 It is known that there are two principle noise sources that dominate the contributions to the total equivalent magnetic noise spectral density level.8 The first noise source is in the piezoelectric layer, which is a Johnson noise that results from free electric charge random motions. The second is the dielectric loss noise related to the electric dissipation from ferroelectric domain wall motion. These two noise sources belong to thermo-electric dissipations.18 They both are intrinsic sources of loss. It is thus difficult to lower the equivalent magnetic noise spectral density level by only considering the present piezoelectric layer construction and passive detection. Accordingly, modulation techniques were developed to overcome these limitations. Using H-field modulation, a low frequency magnetic field to be sensed modulates an excitation carrier signal that is applied on the ME laminate at a higher frequency.14 The low frequency output signals can then be separated from the carrier by means of classic demodulation techniques.19 Accordingly, the only expected noise is the one near the frequency of the excitation carriers, and thus, the noise contribution from the low frequency thermo-electric dissipations is removed. Furthermore, this enables the H-field modulation to also partially reject low frequency vibration noises. The EMN is converted from the output electric noise spectral density by dividing with the transfer function of the ME sensor. On the other hand, there are some limitations for H-field modulation. First, its power consumption is significantly higher than the passive mode. It also requires continuous drive, and its associated detection circuitry is much more complicated than the passive mode one. To date, a gradiometric pair of ME sensors operated under H-field modulation has not been reported. It is an important way by which the rejection efficiency of microphonics may be increased. Here, such a gradiometer consisting of a pair of ME sensors operating in an active mode by H-field modulation was constructed and tested. Investigations focused on the common mode rejection ratio (CMRR) of this ME gradiometer, and it will be shown that this pair of H-field modulation sensors has enhanced vibration rejection efficiency, compared to a single ME sensor in a passive or a modulation working mode.

J. Appl. Phys. 123, 104501 (2018)

FIG. 1. Ideal gradiometer response curve.

a given value of the baseline d, the working regime is between a start point Dstart and an end point Dend and limited by the value of 1/CMRR. A. Common mode rejection ratio

The common mode rejection ratio of a magnetic gradiometer is defined as the ratio of the amplitudes of differential mode outputs over a common mode output in a homogenous field, that is to say, the ratio between the differential mode and the common mode gains. By applying a magnetic field H on the gradiometer, the detected magnetic flux density can be expressed as B ¼ l0H. This magnetic signal is amplified by the first and second magnetometers, producing the two output voltage signals V1(¼Tr1  B1) and V2(¼Tr2  B2), as shown in Fig. 2. The output voltage of the gradiometer can then be given as Vout ¼ T~r1  B1 þ T~r2  B2 ~ ¼ T~ r d ðB1  B2 Þ þ Tr cm



 B1 þ B2 ; 2

II. THEORETICAL ANALYSIS

In this section, several benchmarks of gradiometers will be introduced, such as differential and common mode gains, the common mode rejection ratio, and the spatial transfer function. In addition, the basic theoretical analysis on these benchmarks will be given by demonstrations of basic parameters of the proposed gradiometer. The principal working mode of a gradiometer is based on a detection of the gradient output DV/V1 as a function of the normalized distance D/d, as shown in Fig. 1. DV is the difference between the voltage output of the ME sensor 1 (V1) and the voltage output of the ME sensor 2 (V2). D is the distance between magnetic moment and the gradiometer, and d is the baseline of gradiometer. For

FIG. 2. Scheme of two ME sensors with a subtractor.

(1)

104501-3

Xu et al.

J. Appl. Phys. 123, 104501 (2018)

where we used a popular formula that expresses the output T~r1 T~r2 and a comvoltage as a differential mode gain T~ r d ¼ 2 mon mode gain Tr~cm ¼ T~r1 þ T~r2 . Arrows indicate that the symbols have both amplitude and phase contributions. We introduced a phase shift of 180 in the following equations. In practice, there are differences between the first and second magnetometers with regard to their phases and gains, which gives ( DTr ¼ jTr1  Tr2 j (2) u ¼ ur1  ur2 : Thus, the differential mode and common mode gains can be given as 8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi > > ðTr1 Þ2 þ ðTr2 Þ2  2Tr1 Tr2 cos ðp  uÞ > < Tr d ¼ (3) 2 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi > > > : Tr cm ¼ ðTr1 Þ2 þ ðTr2 Þ2 þ 2Tr1 Tr2 cos ðp  uÞ: Subsequently, the CMRR can be calculated by qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðTr1 Þ2 þ ðTr2 Þ2 þ 2Tr1 Tr2 cos ðuÞ Tr d CMRR ¼ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : Tr cm 2 ðT Þ2 þ ðT Þ2  2T T cos u ð Þ r1 r2 r1 r2 (4) The experimental data showed that the phase shift between the two active mode ME sensors was nearly zero. There is a component of our LabView program, which is the phase difference between channel 1 and channel 2. The phase difference value each second for a period of 5 sequential seconds was measured. The values at seconds 1–5 were 1.43 , 1.02 , 358.31 , 0.81 and 0.83 . All of them are nestle 0 or 360 . Thus, the CMRR can be simplified to CMRR ¼

Tr d Tr1 þ Tr2 Tr1 Tr1 ¼ ¼ :  Tr cm 2ðTr1  Tr2 Þ DTr DTr

(5)

By placing the pair of sensors in a common magnetic field, the CMRR can also be measured according to the following expression: CMRR ¼

Tr1 B1 V1 V1 : ¼ ¼ DTr B1 V1  V2 DV

(6)

B. Spatial transfer function

The spatial transfer function (STF) defines the gradient response of a magnetic gradiometer as a function of the distance between the gradiometer and a magnetic target. An axial gradiometer consists of two sensors aligned with a base line d, as shown in Fig. 3. In the following, we will show the theoretical formula of a gradient response for an axial gradiometer and compare the simulations using this formula with measured data. The absolute value of the magnetic flux density generated by a magnetic moment with a distance D along the perpendicular direction can be written as

FIG. 3. Scheme of a ME gradiometer and a magnetic moment.

B¼

l0 m ½T; 2p D3

(7)

where m is the absolute value of a magnetic moment and D is the distance between the magnetic moment and the gradiometer. The output voltage values of the first and second sensors, and the difference between the two sensors, can be given as 8 l m l m > >  Tr1 ¼  ðTr2 þ DTr Þ V1 ¼ B1  Tr1 ¼ > 3 > 2p D1 2p D31 > > > > l m <  Tr2 V2 ¼ B2  Tr2 ¼ 2p D32 > " ! # > > > lm 1 1 DTr > > V ¼ V2  V1 ¼  Tr2 þ 3 ; > > : 3 2p D1 D31 ðD1 þ dÞ3 (8) where D1 is the distance between the magnetic source and the first ME sensor, D2 is the distance between the magnetic source and the second ME sensor, and DTr (¼Tr2  Tr1) is the difference of the transfer function between the two sensors. Thus, the spatial transfer function of the gradiometer can be derived as " ! #  lm 1 1 DTr    Tr2 þ 3     2p D1  D31 ðD1 þ dÞ3 V 3    STF ¼   ¼   lm Tr1 V1     3 2p D1   ! # "  D31 DTr   1 þ  3 Tr2   ðD1 þ dÞ : ¼   DTr   1þ   Tr2  

(9)

When the magnetic moment is close to the gradiometer, the STF is unity; however, when the magnetic moment is far enough from the magnetic gradiometer (D1  d), the STF becomes

104501-4

Xu et al.

    DTr  DTr     ¼ 1 : STF ¼  ¼  Tr1 Tr2  CMRR

J. Appl. Phys. 123, 104501 (2018)

(10)

In practice, another factor which limits the detection performance is the non-common noise, such as the intrinsic noise of the sensors. Using a gradiometer with a sufficiently enhanced CMRR value, the sensors’ intrinsic noise levels determine the ultimate performance of the gradiometer for measuring the gradient in an unknown magnetic environment. III. STRUCTURE AND WORKING PRINCIPLE

The proposed ME gradiometer is composed of two ME laminate sensors that have similar sizes, weights, structures, and equivalent magnetic noise levels. A Pb(Zr,Ti)O3 (PZT) piezoelectric layer made of five 40 mm long macro-fibers in parallel was bonded between two 80 mm long Metglas layers using epoxy, where each layer consisted of three Metglas foils that were stacked together by epoxy. The PZT fibers were poled in the longitudinal direction with the help of interdigital electrodes on both top and bottom sides of these fibers. The intrinsic equivalent magnetic noise (EMN) of the two ME sensors was 60 pT/冑Hz and 65 pT/冑Hz at 5 Hz, respectively, when using two independent drive coils with sections of 730 mm2. The center-to-center distance between the two ME sensors was defined as the baseline of the gradiometer. By varying the values of the baseline, we investigated the CMRR of the proposed gradiometer, which is the capacity to reject common mode signals/noise. In Fig. 1, the baseline of the ME gradiometer determined the starting point Dstart, which was the position at which the ME gradiometer is able to detect the gradient signals. The smaller the baseline, the closer the position at which the gradiometer can start to work. The closest baseline is determined by considering the sensor geometries and the interferences between the two sensors. In this paper, the baseline was chosen to have values of 15 cm, 20 cm, and 25 cm. In all of these cases, negligible crosstalk effects were found. The ending point Dend

of the gradiometer curve is the farthest distance that the ME gradiometer can detect away from a magnetic target that causes a magnetic field disturbance in the environment before the signal falls below the noise floor. The farthest detection distance theoretically is determined by both the CMRR of the gradiometer and the EMNs of the ME sensors in the gradiometer. The CMRR was measured by applying a homogenous magnetic field simultaneously to both active ME sensors. The two ME sensors were located in parallel within a Helmholtz coil that provided an AC magnetic signal of 110 nT at 5 Hz, shown in Fig. 4(a). According to Eq. (6) in the analysis in the theoretical section, the CMRR can be calculated as the amplitude of the output signal over the difference between the two sensors, yielding CMRR ¼

V1 V1 : ¼ V1  V2 DV

(11)

Then, the predicated maximum CMRR for our sensor pair was measured to be 74, which was calculated from the data in Fig. 4(b). The farthest distance that the ME gradiometer can detect a signal depends on the minimum difference between the two sensors in amplitude and in phase, presented as the value of 1/CMRR. Furthermore, a higher intrinsic noise floor of the two individual ME sensors may limit their detection performance in the case of small amplitude magnetic fields. Therefore, there are different external noise levels in different environments. In order to efficiently reject the common mode external noise, a minimum CMRR is required. By dividing the noise floor of the ME sensors measured outside of a shielding chamber over the one in the chamber (intrinsic EMN of ME sensors), one can then estimate the minimum required value of the CMRR for the proposed ME gradiometer working in this environment, which gives External noise level CMRRrequired ¼ Sensors ’ intrinsic noise level. The noise floors of our two ME sensors measured in a shielding chamber were determined to be 60 pT/冑Hz and

FIG. 4. (a) Experimental setup for the CMRR measurement and (b) outputs from two ME sensors and their difference.

104501-5

Xu et al.

J. Appl. Phys. 123, 104501 (2018)

65 pT/冑Hz at 5 Hz, respectively. The noise floors of these same two ME sensors were then simultaneously measured out of the chamber to be 189 pT/冑Hz and 314 pT/冑Hz at 5 Hz. Accordingly, the required minimum CMRR of our ME gradiometer was determined to be a ratio of 4.8 in this noise environment. The gradient sensitivity is defined as a ratio of the output of a gradiometer over the gradient of the magnetic signals detected by the gradiometer. Taking into account that the transfer function of the two sensors are similar (i.e., Tr2  Tr1), the formula of the gradient sensitivity can be given as Tr

Gradient

V2  V1 ðB2  B1 Þ=d Tr1 ðB2  B1 Þ ¼ Tr1 d ½V=ðT=mÞ : ¼ ðB2  B1 Þ=d

¼

(12)

Thus, the detection performance can be predicated to have a value around 260–433 pT/m/冑Hz using the baselines in our proposed active ME gradiometer, where these values were derived from the equation of The intrisic noise floor of gradiometer . Figure 5(a) shows the experiThe baseline of gradiometer mental configuration that we reused to measure the gradiometer spatial response curve, where d was the baseline and D was the distance between the gradiometer and the magnetic moment. A small coil with a size of 16 mm in diameter and a thickness of 11 mm served as the magnetic moment. The intensity of the magnetic source used in the measurements had a value of about 50 lT. The expected curves based on (9) are given as dashed lines by using different values of the baseline. During measurements of the gradiometric response curve, the magnetic source was gradually moved further

away from the ME gradiometer. The voltage outputs from the two sensors were then simultaneously recorded by a data logger (National Instrument) which was controlled using a LabView program. Figures 5(b) and 5(c) show the gradiometric response curves with baselines of 15 cm, 20 cm, and 30 cm, where the difference between the outputs of the two sensors is along the vertical axis. The horizontal axis in (b) is the distance over which the gradiometer signal detection was measured, whereas in (c), the horizontal axis is the ratio of the distance D normalized over the baseline of the gradiometer. In Fig. 5(b), it is shown that the longer the baseline, the further the distance that the gradiometer can detect a magnetic signal, and conversely, the shorter the baseline, the closer the beginning point at which the gradiometer can begin to work in its gradient regime, as shown in Fig. 5(c). The three curves which were normalized to D/d nearly overlapped, where the blue line is the 3 dB point when it crosses the gradiometer response curves is used to guide eyes. This intersection on the horizontal axis occurred near D/d ¼ 1, consistent with predictions that our ME gradiometer behaves similar to a first order gradiometer.20 The limiting factor for the gradiometer performance is set by when the response curves cross the EMN floor. Sensors with a lower EMN will allow for a longer detection distance. The farthest distance is set by the CMRR and the intrinsic noise of this gradiometer. IV. VIBRATION REJECTION EFFICIENCY

Magnetoelectric gradiometer measurements using passive mode ME sensors were previously reported by Shen et al.10 Their ME gradiometer was stationary during magnetic source detection, thus reducing the introduction of low frequency mechanical vibrations. Here, a ME gradiometer

FIG. 5. (a) Experimental setup for the gradiometer response curve measurement, (b) gradiometer response curve as a function of the distance between the ME gradiometer and the magnetic source, and (c) gradiometer response curve as a function of the distance between the ME gradiometer and the magnetic source normalized to the baseline.

104501-6

Xu et al.

J. Appl. Phys. 123, 104501 (2018)

working under H-field modulation is compared to one working in a passive mode. We found that the vibration rejection of a ME gradiometer working under H-field modulation was significantly higher than that of a passive mode one. Two experiments were performed, whose results are presented in the following paragraphs. The first experiment measured the vibration rejection ratio (VRR) of the ME gradiometer working under H-field modulation. The second was to determine how a ME sensor working under H-field modulation compared to a similar one working in a passive mode, which enabled the measurement and comparison of their CMRR values. Figure 6 shows the experimental setup for the vibration rejection measurements which we designate as E1. The gradiometer was located on a stiff plastic board for both parallel (solid line) and serial (dash line) sensor configurations. In both cases, the baselines were 15 cm, which is the same as shown in Fig. 5(a). The plastic board with the ME gradiometer was bonded stably on one end of a rigid aluminum bar, where the other end of the bar was connected to a shaker. The shaker was then excited by a common mode 5 Hz vibration through the bar. Additionally, two pieces of high l metal were placed next to the shaker to avoid the effect of DC magnetic fields generated from the shaker. Using the experimental setup of E1, vibrations of three different amplitudes were applied incident on the gradiometer from the shaker. This was done for both the parallel and serial configurations of the sensors. These vibrations were characterized as an equivalent magnetic flux density (EMFD) by using a fluxgate magnetometer. This allowed all output voltages from the ME gradiometers that were generated by the vibration sources applied to the gradiometer to be converted to an EMFD value. The EMFD was calculated by EMFD ¼

Fluxgate output voltage ½T : Fluxgate’s transfer function

(13)

The vibration rejection ratio (VRR) was determined by using (11). The measured results are subsequently shown in Figs. 7 and 8. In Fig. 7, the vibration of the three different amplitudes that were applied to the gradiometer resulted in EMFD values of 15 nT, 10 nT, and 7.8 nT for a gradiometer with a parallel sensor configuration. The corresponding vibration rejection ratios for these three vibration amplitudes were 17.4, 13.4, and 10.7 respectively. Figure 8 shows the vibration rejection ratios that were measured by using a ME gradiometer with sensors in a serial configuration. Similar vibrations were applied, as for the parallel sensors, which resulted in EMFD values of 25 nT, 19 nT, and 15 nT. The corresponding vibration rejection ratios were determined to be 10.9, 12.3, and 9.3 respectively. Overall, the vibration rejection ratio of a ME gradiometer working under H-field modulation was around 12: this is the mean value of all vibration rejection ratios measured. The rigid plastic board for the serial configuration of sensors was notably longer than the one used for the parallel one (see Fig. 6). Thus, the serial mode also had bending vibrations under a 5 Hz mechanical drive. This is the reason why even when larger amplitude vibrations were applied to the serial configuration, lower vibration rejection ratios were measured. A second type of experiment (E2) was done to compare the vibration rejection ratio between two identical ME sensors working under an H-field modulation and working in a passive mode. The experiment was divided into two parts: under H-field modulation (Part 1) and in a passive mode (Part 2). To confirm that the ME sensors (in Parts 1 and 2) were working under the same vibration strength, a fluxgate magnetometer was used as a reference to determine EMFD. The ME sensor working under modulation and the fluxgate magnetometer were located in a parallel configuration using the same experimental setup shown in Fig. 6. The value of EMFD was determined using the fluxgate magnetometer to

FIG. 6. Experimental setup for the vibration rejection ratio measurement of the ME gradiometer in the parallel mode and the line mode.

104501-7

Xu et al.

FIG. 7. Outputs of the two ME sensors and their corresponding subtractions under different vibrations for the parallel mode. (a) Vibration amplitude equivalent to 15 nT magnetic flux density. (b) Vibration amplitude equivalent to 10 nT magnetic flux density. (c) Vibration amplitude equivalent to 7.8 nT magnetic flux density.

be 19.6 nT, and the output from the ME sensor was 65.4 mV. Using the transfer function (2.5  106 V/T) of the ME sensor working under H-field modulation, the value of EMFD for the ME gradiometer operated under modulation was determined to be 26 nT. In order to insure that the applied vibration in Part 2 of the experiment was the same as in Part 1, the output from the fluxgate magnetometer was adjusted manually to match that from Part 1. Then, the EMFD measured using the fluxgate magnetometer in Part 2 was found to be 19.5 nT, which was close to that for Part 1. The output of 196 mV from the ME sensor in the E2 measurement was divided by the transfer function (2.5  106 V/T) of the sensor

J. Appl. Phys. 123, 104501 (2018)

FIG. 8. Outputs of two ME sensors and their corresponding subtractions under different vibrations for the line mode. (a) Vibration amplitude equivalent to 25 nT magnetic flux density. (b) Vibration amplitude equivalent to 19 nT magnetic flux density. (c) Vibration amplitude equivalent to 15 nT magnetic flux density.

working in a passive mode, and the EMFD of the passive mode was then calculated to be 78 nT. All the data measured in E1 and E2 are listed in Table I. According to the data shown above, the value of EMFD measured by the ME sensor working under H-field modulation was 30% larger than that measured using the fluxgate magnetometer. Considering that the vibration rejection of the fluxgate magnetometer is quite high,21 this shows that the vibration rejection of the ME sensor working under Hfield modulation is also quite high. Fluxgate magnetometers can hardly detect vibration noises,21 but there remain small noise outputs due to environmental magnetic induction (EMI).

104501-8

Xu et al.

J. Appl. Phys. 123, 104501 (2018)

TABLE I. The data measured in E1 and E2. E1

Parallel mode Line mode

EMFD (nT) VRR EMFD (nT) VRR

15 17.4 25 10.9

10 13.4 19 12.3

The outputs from the ME sensors working under modulation techniques and in a passive mode had contributions from EMI and vibrations generated by the shaker. The EMFD value measured for the ME sensor working in a passive mode was 3 times higher than that measured by an identical ME sensor working under modulation techniques. Clearly, we confirm that the vibration rejection of ME sensors working under modulation techniques is considerably higher than that of the passive mode. Since the intrinsic noise floor of ME sensors is the bottleneck to further enhance the performance of a ME gradiometer, efforts going forward need to focus on noise reduction, by improving both the excitation coils and the ME composites. V. SUMMARY

In this paper, ME gradiometers whose individual sensors operated under H-field modulation were constructed for the first time and characterized. The ME gradiometer response curve was measured for this ME active gradiometer and shown capable of detecting a source out to a distance normalized to the baseline of D/d ¼ 13. Also, the vibration rejection ratio was determined. We found that the vibration rejection of a ME gradiometer working in the H-field modulation was at least 3 larger than that of a passive mode ME gradiometer but yet only 30% smaller than a fluxgate gradiometer. This is in addition to the fact the H-field modulation ME sensors have a significantly lower noise floor than a corresponding passive sensor. ACKNOWLEDGMENTS

This work was supported by the Office of Naval Research under Grant N00014-15-1-2457. 1

C.-W. Nan, M. I. Bichurin, S. Dong, D. Viehland, and G. Srinivasan, “Multiferroic magnetoelectric composites: Historical perspective, status, and future directions,” J. Appl. Phys. 103, 031101 (2008). 2 M. Fiebig, “Revival of the magnetoelectric effect,” J. Phys. D: Appl. Phys. 38(8), R123–R152 (2005). 3 W. Eerenstein, N. D. Mathur, and J. F. Scott, “Multiferroic and magnetoelectric materials,” Nature 442, 759–765 (2006). 4 J. Das, J. Gao, Z. Xing, J. F. Li, and D. Viehland, “Enhancement in the field sensitivity of magnetoelectric laminate heterostructures,” Appl. Phys. Lett. 95(9), 092501 (2009).

7.8 10.7 15 9.3

5

E2 Fluxgate Active mode ME sensor Passive mode ME sensor

EMFD (nT) 19.6 19.5 26 N/A N/A 78

Y. J. Wang, J. Q. Gao, M. H. Li, Y. Shen, D. Hasanyan, J. F. Li, and D. Viehland, “A review on equivalent magnetic noise of magnetoelectric laminate sensors,” Philos. Trans. R. Soc. A 372, 20120455 (2012). 6 S. Dong, J. Zhai, J. Li, and D. Viehland, “Near-ideal magnetoelectricity in high-permeability magnetostrictive/piezofiber laminates with a (2-1) connectivity,” Appl. Phys. Lett. 89, 252904 (2006). 7 Y. Wang, D. Gray, D. Berry, J. Gao, M. Li, J. Li, and D. Viehland, “An extremely low equivalent magnetic noise magnetoelectric sensor,” Adv. Mater. 23(35), 4111–4114 (2011). 8 X. Zhuang, M. Lam Chok Sing, S. Saez, C. Cordier, C. Dolabdjian, J. Gao, J. Li, and D. Viehland, “Theoretical analysis of the intrinsic magnetic noise spectral density of magnetostrictive-piezoelectric laminated composites,” J. Appl. Phys. 109, 124512 (2011). 9 T. H. A. Salem, J. K. Asahina, and K. Ushijima, “Detection of unexploded ordnance (UXO) using marine magnetic gradiometer data,” Explor. Geophys. 36(1), 97–103 (2005). 10 Y. Shen, K. L. McLaughlin, J. Gao, D. Gray, D. Hasanyan, Y. Wang, J. Li, and D. Viehland, “Metglas/Pb(Mg1/3Nb2/3)O3–PbTiO3 magnetoelectric gradiometric sensor with high detection sensitivity,” IEEE Trans. Ultrason., Ferroelectr. Freq. Control 59(9), 2111–2115 (2012). 11 H. Xi, X. Qian, M.-C. Lu, L. Mei, S. Rupprecht, Q. X. Yang, and Q. M. Zhang, “A room temperature ultrasensitive magnetoelectric susceptometer for quantitative tissue iron detection,” Sci. Rep. 6, 29740 (2016). 12 H. Xi, M.-C. Lu, X. Qian, and Q. M. Zhang, “An ultrasensitive magnetoelectric sensor system for the quantitative detection of liver iron,” in 2016 IEEE Sensors, 30 October–3 November 2016. 13 Z. Xing, J. Zhai, J. Li, and D. Viehland, “Investigation of external noise and its rejection in magnetoelectric sensor design,” J. Appl. Phys. 106(2), 024512 (2009). 14 X. Zhuang, M. Lam Chok Sing, C. Cordier, S. Saez, C. Dolabdjian, L. Shen, J. Li, M. Li, and D. Viehland, “Evaluation of applied axial field modulation technique on ME sensor input equivalent magnetic noise rejection,” IEEE Sens. J. 11, 2266–2272 (2011). 15 P. Ripka, “Review of fluxgate sensors,” Sens. Actuators, A 33, 129–141 (1992). 16 S. Dong, J. Li, and D. Viehland, “Magnetoelectric couplings, efficient, and voltage gain effect in piezoelectric-piezomagnetic laminate composites,” J. Mater. Sci. 41, 97–106 (2006). 17 L. Shen, M. Li, J. Gao, Y. Shen, J. Li, D. Viehland, X. Zhuang, M. Lam Chok Sing, C. Cordier, S. Saez, and C. Dolabdjian, “Magnetoelectric nonlinearity in magnetoelectric laminate sensors,” J. Appl. Phys. 110, 114510 (2011). 18 H. B. Callen and T. A. Welton, “Irreversibility and generalized noise,” Phys. Rev. 83, 34–40 (1951). 19 X. Zhuang, M. Lam Chok Sing, C. Dolabdjian, Y. Wang, P. Finkel, J. Li, and D. Viehland, “Mechanical noise limit of a strain-coupled magneto(elasto)electric sensor operating under a magnetic or an electric field modulation,” IEEE Sens. J. 15, 1575–1587 (2015). 20 L. Mei, S. Rupprecht, Q. Yang, and Q. M. Zhang, “A first order gradiometer based low noise magnetoelectric sensor system,” in 2013 71st Annual Device Research Conference (DRC) (2013). 21 See http://magnetometer.com/ for Billingsley Aerospace & Defense, 2600 Brighton Dam Road, Brookeville, MD 20833 USA.