Introduction

The development of low-frequency weak magnetic detection is driving higher demands for magnetic sensors, including high sensitivity, low power consumption, miniaturization, and high stability. Magnetoresistive sensors offer a wide frequency band and good linearity, but the direction of the magnetic field limits their performance1,2. Fluxgate sensors are known for their pT level sensitivity and simple structures; however, they tend to be large and power-hungry3. Superconducting quantum interference devices provide extremely high sensitivity for biomagnetic detection, but necessitate low temperatures and entail significant costs4. Optically pumped magnetometers exhibit high precision with no zero shift; nevertheless, they are relatively large and environmentally sensitive5. Magnetoelectric (ME) sensors composed of magnetostrictive and piezoelectric materials, offer low power consumption and multifunctionality. Their detection limits are typically in the 1–10 pT range, with some devices achieving sub-pT sensitivity6,7,8. These characteristics make them highly promising for applications in detecting weak magnetic fields at low frequencies9,10. However, the magnetic detection limit remains below the theoretical value (~fT)11,12.

The optimization of ME coupling research has so far focused on material properties13, geometric enhancements14, and coupling methods15. In 2011, Gao et al. successfully developed the Metglas/PMN-PT laminated push-pull ME composites with 1100 V/Oe·cm and a DC detection limit of 1 nT16. In 2022, Xu et al. developed the Metglas/PZT-based compass with a magnetoelectric response of 274 V/Oe·cm and an angle sensitivity within ±0.02° through optimizing geometric dimensions17.

Meanwhile, diverse annealing methods show their potential to alter the magnetic or electrical characteristics of ME materials18,19,20, by improving the mechanical mass factor8,11,21,22,23. In 2013, Li et al. achieved a magnetoelectric coupling coefficient as high as 380 V/Oe·cm in a self-based annealed Metglas/PZT/Metglas laminate with a (2-2) mode multi-push-pull structure24. In 2021, Peddigari et al. engineered a multilayer structure consisting of piezoelectric single crystal macro fiber composites and flash photon annealed (FPA) Metglas, which enhanced the amorphous nature of Metglas and achieved a detection threshold of 0.5 pT25. In 2023, Palneedi et al. employed intense pulsed light(IPL) pulses for annealing PZT piezoelectric thick films on Metglas substrates, which led to non-resonant coupling coefficients reaching 20 V/Oe·cm26. In 2024, Li et al. used laser annealing to control the transformation of sheet-like nanocrystalline FSAC32 alloy from α-Fe to DO3, thereby improving the magnetic permeability of the alloy powder. Notably, due to the differing magnetization difficulties of the various phases, they observed asymmetric hysteresis loops27. In 2017, Dong’s group reported a heightened magnetoelectric response (7000 V/Oe·cm) through the targeted laser treatments on Metglas fibers within (1-1) laminated ME composites15,28. However, these annealing treatments may lead to the fragmentation of magnetic domain structures19,29,30, thereby restricting the improvement of the mechanical mass factor. The optimal annealing treatment is needed to achieve optimal device performance.

This paper introduces the femtosecond laser annealing technique as an effective means for tuning the magnetic properties of FeSiB in a (2-2) type sandwich-structure magnetoelectric laminates, leveraging its precise control over rapid melting and cooling dynamics. The laser-induced microstructure and its magnetic properties are characterized using Transmission Electron Microscope (TEM), Vibrating Sample Magnetometer (VSM), and Magneto-optical Kerr effect (MOKE). The effects of laser treatment were systematically analyzed in terms of the magnetoelectric coefficient, piezomagnetic coefficient, and mechanical quality factor.

Preparation

Femtosecond laser processing

The surface modification of Metglas foils was achieved using a femtosecond laser pulse scanning technique (model FemtoYL-4) to enhance magnetic properties and mitigate inherent surface oxidation. The parameters were set as follows: line spacing of 0.01 mm, Z-type filling, processing speed of 5000 mm/s, and spot diameter of 0.03 mm. The key factors affecting surface reactions are laser power, scanning times, and pulse line spacing. Generally, once the laser powers are greater than 20 W, it easily results in the formation of pronounced crystalline structures and leads to a significant reduction in the piezomagnetic coefficient. Therefore, a laser power of 7 W was selected to prevent excessive high-temperature crystallization and control the rate of magnetic decline. The annealing power was set to 7 W, with a frequency of 50 kHz. A filling scan was chosen to fill the gaps between laser spots. Scanning was performed 300 and 600 times, respectively, to control the growth of nanocrystalline grains and form various crystalline structures.

Magnetoelectric material fabrication

Firstly, 8 layers of Laser-treated Metglas foils, with 100 mm in length and 2.5 mm in width, were laminated using epoxy in a bespoke 3D mold for curing. Secondly, a PMN-PT single crystal (10 mm (L) × 2 mm (W) × 0.2 mm (T)) was pre-polarized in silicone oil at 60 °C. Here, flexible electrodes were positioned at the geometric centers of both piezoelectric and magnetostrictive layers to measure the voltage output induced by magnetoelastic strain. To optimize the amount of epoxy resin and enhance interface bonding, a specially designed T-shaped fixed 3D shell was utilized. This facilitated multi-layer curing under a high pressure of 20 kN for twenty-four hours. This curing process is critical to ensure the maximal removal of excess flow colloids from each layer while maintaining the flexibility of the multi-layer plate structure, and the effective transfer of magnetostrictive strain. The Metglas foils, in conjunction with a PMN-PT single crystal, were utilized as the magnetostrictive and piezoelectric materials laminated in a (2-2) connectivity pattern. This configuration operated in the longitudinal-transverse (L-T) mode, as depicted in Fig. 1a. The magnetoelectric coupling coefficient of ME Laminates, as shown in Fig. 1b, characterized by enhanced interface bonding, exhibits an approximately 1.5-fold increase. Compared to conventional connection processes, the 3D uniform reinforcement process prevents delamination under prolonged operational conditions, thereby extending the service life of the samples.

Fig. 1: Fabrication and performance of laser-annealed ME composites.
figure 1

a Schematic illustration of the fabrication process and layered architecture of the ME composite. b Interface bonding enhancement. c Diagram of the magnetic fields system for measuring ME composites

Full size image

Magnetic material performance and ME coupling testing

The surface morphologies of Metglas samples were observed by using the high-precision Gemini SEM500 model of Scanning Electron Microscopy (SEM). X-ray diffraction (XRD) was utilized to detect the presence of crystalline structures. The Thermo Fisher Talos-F200X Transmission Electron Microscope (TEM) model was utilized to examine the internal structure and material composition. Vibrating Sample Magnetometer (VSM) and Magneto-Optical Kerr Effect (MOKE) were employed to assess variations in hysteresis loops and surface magnetic anisotropy. The applied excitation DC magnetic fields were from 0 to 50 Oe. The Lorentz field emission transmission electron microscope (FE-TEM, Fresnel mode LTEM) was also utilized for observing domain flipping and domain wall movement. The magnetic field gradually changes from a zero state to full magnetization, corresponding to a range of ±3% of the objective lens’s magnetic field capacity.

The magnetoelectric performance testing platform was put up and shown in Fig. 1b. The test was performed at room temperature at a frequency of 0 to 25 kHz, with a DC excitation providing a −40 to 40 Oe magnetic field. Additionally, a LabVIEW-based testing system controlled the signal generator and lock-in amplifier for various tests, including magnetoelectric coupling coefficient measurement, optimal DC bias field testing, and AC/DC detection limit testing.

Results and discussion

Nanostructure analysis

The nanostructure is a crucial factor in determining the performance and functionality of materials. Here, one with laser-treated Metglas foils and the other with untreated counterparts were utilized to observe the surface morphology differences. As shown in Fig. 2a, b, it reveals irregular point-like regions in the surface of the laser-annealed Metglas by SEM, culminating in the formation of stacked stripes in Fig. 2b. In marked contrast, the untreated Metglas surface (Fig. 2c), exhibits a remarkably smooth texture. More importantly, Fig. 2d displays a new nanostructure in treated Metglas, which is bifurcated into two layers, the 40 nm thick of upper Layer 1 and the 61 nm thick of lower Layer 2. Here, Layer 3 is composed of less altered Metglas. Moreover, Fig. 2e demonstrates the heterogeneous nucleation of a nanocrystalline phase within Layer 2, approximately 5 nm in scale, resulting in short-range ordered clusters. The results indicate that nanocrystals are growing during the laser scanning process. The diffraction pattern (Fig. 2f) reveals that the predominant nanocrystalline phase is Fe3O4 corresponding to the (222) and (022) planes of Fe3O4, which contrasts with the formation of α-Fe (Si) nanocrystals observed in high-power laser annealing19,31,32. As shown in Fig. 2g, Layer 1 predominantly features silicon oxide, carbide, and a notable absence of iron, while there is a noticeable decrease in silicon content and an increase in iron content in Layer 2. Notably, in addition to a small amount of uniformly distributed amorphous iron oxide, there are regions with extremely high iron content, characterized by circular areas shown in Fig. 2l. In addition, XRD spectra show a reduced peak around 45° in the laser-treated foils than that of untreated foils in Fig. 2h, suggesting a reduction of ordered clusters within the surface microstructure33,34.

Fig. 2: Laser-annealed Metglas surface and cross-sectional morphology.
figure 2

a Surface topography of laser-treated Metglas. b Carbonized fringe observed on the laser-treated Metglas. c Surface topography of Metglas. d Stratified distribution of tripartite layer sections of laser-treated Metglas. e An ordered crystalline structure within the second layer. f Fourier transform of HRTEM image of Fe3O4 crystal. g Cross-sectional eds line scanning analysis. h XRD analytical comparison. i Silicon elemental distribution analysis. j Oxygen elemental distribution analysis. k Carbon elemental distribution analysis. l Iron elemental distribution analysis

Full size image

Figure 3 shows the surface chemical reconstruction process with a three-layer nano-magnetic structure, consisting of an oxidation isolation layer, a magnetic dipole field layer, and a soft magnetic layer33. Due to the instantaneous temperature effects, silicon (Si) or boron (B) in FeSiB cannot directly form α-FeSi or α-FeB with iron (Fe)35. The bombardment causes boron to be expelled to the surface, while silicon escapes to the surface and reacts with air to form a silicon-rich oxide film36. It provides the Metglas foil with an effective oxidation isolation layer, enhancing its corrosion resistance37. In Layer 2, oxygen thoroughly integrates with iron to form new amorphous iron oxides, carbon oxides, and silicon oxides at the overlap scanning spot. Simultaneously, a small amount of iron forms Fe3O4 crystalline structures embedded within the new amorphous material. This results in long-range magnetic dipole field interactions.

Fig. 3: Schematic of laser-annealed Metglas composition and microstructure.
figure 3

a Surface chemical reconstruction. b Three-layer magnetic structure

Full size image

Magnetic analysis

Magnetic properties are closely related to the nanostructures of magnetic materials, Fig. 4 shows the changes in magnetic properties of laser-treated and untreated Metglas samples, respectively. The magnetic permeability exhibits minimal variation as the layer number increases in Fig. 4a, whereas the saturation magnetization demonstrates a progressive increase. Untreated Metglas exhibits a coercivity of roughly 0.02 Oe and a strong soft magnetism. After the laser-treated process (600 scans), it shows a significant decrease in saturation magnetization, due to the amorphous-to-ordered structural transition. The irregularity of the oxide layer and the magnetic dipole layer serving as the pins of magnetic domain walls can induce impurity fields on the tape surface and reduce the initial magnetic permeability. Thus, the new nanostructure results in a noticeable shift and the emergence of self-bias in the initially symmetrical hysteresis loops. Additionally, the coercivity increases to 0.25 Oe after 600 scans, which is 12 times greater than that before annealing in Fig. 4b. Thus, 600 scan cycles, which provide stronger self-bias feedback, are selected as the processing parameter for a single layer of Metglas foil.

Fig. 4: Magnetic properties of laser-annealed vs. unprocessed Metglas.
figure 4

a Hysteresis loop of Metglas under a high magnetic field. b Hysteresis loop under a low magnetic field. c Untreated Metglas: Kerr effect-induced magnetic domain flipping process. d Treated Metglas: Backside magnetic domain flipping via Kerr effect. e Untreated Metglas Lorentz lens: Initial magnetic wall state under zero field. f Untreated Metglas Lorentz lens: pre-saturation magnetic wall motion. g Treated-Metglas Lorentz lens: Initial magnetic wall state under zero field. h Treated-Metglas Lorentz lens: pre-saturation magnetic wall motion

Full size image

Meanwhile, the domain realignment process during magnetization for both untreated and treated samples is depicted by MOKE. For the untreated samples, irregularly shaped magnetic reversal regions are observed as the magnetic field increases, as shown in Fig. 4c. The magnetic domains show a complete and rapid reversal. For the laser-treated, a periodic pattern of white and black striped domains of uniform width is shown in Fig. 4d, and a more rapid domain flipping activity is noted under the same dynamic magnetic field conditions. Nanostructure results in a significant refinement of the magnetic domain structure19. Fig. 4g shows a Bloch-type38 orientation in the vertical plane after laser treated process, while untreated Metglas exhibits magnetic domains with an iconic double-Y configuration39 (Fig. 4e). However, the parallel domain regions migrated towards nanostructure (Fig. 4h), opposing to the bidirectional divergence observed in unprocessed cross-domain walls39,40,41,42,43.

The transition from 45° magnetic domains to parallel domains indicates the enhancement of magnetic anisotropy,38,44 which changes the direction of the effective field. As the laser thermal energy dissipates into the soft magnetic layer, the effective field slowly returns to its original position, and the magnetic moment will damp oscillate towards the direction of the effective field, inducing a gradual shift in the magnetic wall45. Although the laser destroys the paramagnetic properties of the amorphous crystal, its heat energy is injected46. Due to the insulating layer, the spin majority cannot penetrate, resulting in the accumulation of spin majority in the soft magnetic layer (Layer 3), which may strengthen the magnetic moment of FeSiB in Layer 3 and decrease the overall magnetic loss47,48,49.

ME coupling enhancing

For comparative analysis, magnetoelectric (ME) and self-biased magnetoelectric (SME)50 laminates were assembled under the same flow. Under optimal DC biasing, the SME laminate demonstrated a magnetoelectric coupling coefficient of 2033.40 V/Oe·cm at 23.3 kHz in Fig. 5a, which is approximately 7.6 times higher than the ME laminate (266.76 V/Oe·cm at Hbias). Additionally, both the limit of detection (LoD)51 and power spectral density (PSD)52 for the SME laminate were quantified under optimal DC bias at the resonance frequency (Fig. 5b). The SME laminate exhibits exceptional sensitivity in the detection of ultra-weak AC signals, with measured values of 383.98 fT and 1.04 × 10−7 V/Hz1/2, respectively. Notably, the ME laminate shows a pronounced dependency on DC bias at 21.7 kHz in Fig. 5c across a −40 Oe to 40 Oe range. Remarkably, the SME laminate exhibits exceptional linearity and high DC sensitivity at 23.3 kHz. The direct magneto-electric effect (DME) exhibits a distinctive double peak phenomenon, with an initial peak observed at approximately 3.5 Oe, thereby demonstrating remarkable linear DC sensitivity within the ±3.5 Oe range. The second spike (approximately 38 Oe) achieves a peak height that is twice as high as the first one. The double-peak phenomenon was also found in FeSiB/Terfenol-D/PZT and Ni/PZT/Terfenol-D multiferroic composites53,54,55. Meanwhile, in Fig. 5f, the hysteresis behaviors of the SME laminate exhibit an offset of 0.97 V/Oe near the zero point of the DC magnetic field, whereas the ME laminate shows zero at 0 Oe. The observed modification in the hysteresis pattern of laser-treated Metglas can be attributed to the self-biased effects induced by its nanostructure. The SME laminate (Fig. 5d) exhibits a significant magnetoelectric response of 67.45 V/Oe·cm and a DC LoD of 22 nT, with noise levels around 9.42µV (refer to Fig. 5e), all achieved without a DC bias magnetic field.

Fig. 5: Magnetoelectric response of laser-annealed vs. unprocessed ME laminate.
figure 5

a Magnetoelectric coupling coefficients: comparative analysis of self-biased and ME laminates under optimal DC bias conditions. b Sensitivity metrics: limit of detection (LoD) and power spectral density (PSD) for the self-biased laminate. c Hysteresis behavior: DC response curves for both SME and ME laminates. d Unbiased coupling efficiency: Magnetoelectric coupling in SME and ME laminates without DC bias application. e Noise Characterization: Direct current noise levels associated with the SME laminate. f Near-Zero DC Hysteresis: Comparative hysteresis curves of SME and ME laminates in proximity to DC zero

Full size image

Magnetic-mechanical-electrical coupling

Magnetic-mechanical-electrical coupling model

By establishing an equivalent circuit model of magnetic-mechanical-electrical coupling56,57,58,59,60 as shown in Fig. 6a, the influences of laser processing on the magnetic-mechanical-electrical conversion efficiency of the magnetic sensor were evaluated. The expression for the total mechanical impedance is presented here56,60:

$${{\rm{Z}}}_{{\rm{mech}}}=\frac{\bar{Z}}{j2\tan \left(\frac{\bar{k}{l}_{p-m}}{2}\right)}+{Z}_{m}\left(\frac{1}{j2\tan \left({k}_{m}{l}_{m}\right)}-\frac{1}{j\sin \left({2k}_{m}{l}_{m}\right)}\right)$$
(1)
Fig. 6: Magnetic-mechanical-electrical coupling performance.
figure 6

a Simplified magnetic-mechanical-electrical coupling model56,57,58,59,60. b Predicted SME laminate frequency response curves. c Predicted ME laminate frequency response curves. d Predicted DC sensitivity of SME model vs ME model. e Comparative piezomagnetic coefficients. f Equivalent quality factor analysis. g Compliance coefficient comparison. h Young’s modulus (\({{\Delta }}\)E coefficient) comparison

Full size image

The magnetoelectric coupling coefficient (V/Oe) is expressed as60

$${{\rm{\alpha }}}_{{\rm{ME}}}=\frac{V}{{H}_{3}}=|\frac{{\varphi }_{m}{\varphi }_{p}}{j\omega {C}_{0}{Z}_{mech}}|$$
(2)

Based on the equivalent circuit theory, the frequency response data of the magnetoelectric sensors under various bias magnetic fields were initially fitted in Figs. 6b, c. In the fitting results, a DC bias of 15 Oe yields the optimal magnetoelectric response at 23.0 kHz, which was in good agreement with the experimental findings from the SME laminate. Under a specific amplitude of the AC magnetic field, the perturbation of the DC magnetic field induces a change in the output voltage of the magnetoelectric sensor. Utilizing extracted material property parameters, interpolation calculations are performed to determine the DC magnetic sensitivity under different working conditions (bias magnetic field and driving frequency), defining it as56

$$\,{{\rm{S}}}_{{\rm{DC}}}=\frac{\Delta {\rm{V}}}{{{\rm{H}}}_{{\rm{ac}}}{\Delta {\rm{H}}}_{{\rm{dc}}}}$$
(3)

The experimental findings reveal a DC sensitivity of 13.6 V/Oe/Oe at 23.0 kHz, achieved with a DC bias magnetic field of 3.5 Oe, which is five times higher than the sensitivity obtained without undergoing the annealing treatment. It is noteworthy that laser treatment not only enhances the AC performance of the ME laminate but also significantly amplifies its sensitivity for DC detection.

Magneto-mechanical characteristics of the ME laminates

By employing an equivalent model, magneto-mechanical characteristics, including the piezomagnetic coefficient (dm), compliance coefficient (sm), and the equivalent mechanical quality factor (Qm) were correlated with the DC bias magnetic field, refining the data through model fitting. The piezomagnetic coefficient of SME laminates decreases obviously under ±20 Oe DC bias, whereas that of ME increases with the strengthening of the excitation magnetic field. The comparative analysis in Fig. 6f reveals that the Qm value in the SME laminate increases to a range of 350–500, approximately 2.5 times higher than that of ME laminates. In addition, Fig. 6g, h demonstrates a reduced compliance factor and an increased coefficient (ΔE) for the SME laminate. The alteration in material structure results in a slight reduction in both the piezomagnetic coefficient and compliance coefficient, consequently affecting the overall magnetic properties which become more rigid. This subsequently impacts the magnetoelectric response of the device at lower frequencies. However, it is worth noting that there is a significant breakthrough enhancement in magnetoelectric response at resonant frequency and LoD of SME laminate at resonant frequency due to its exponentially increasing Qm and ΔE.

Additionally, the performance indices of (2-2) ME laminates, obtained through a comparative analysis of various Metglas annealing studies, are presented in Table 1. The utilization of low-powered laser annealing significantly enhances the magnetoelectric response by approximately sevenfold for both self-biased and non-self-biased configurations. Our sensor achieves AC detection at fT levels and DC detection at nT levels, with a quality factor reaching 380 at the peak magnetoelectric coefficient. This represents a substantial improvement compared to previous findings.

Table 1 Comparison of ME performance with different annealing methods
Full size table

Conclusion

A novel laser-induced nano-structure of FeSiB Metglas efficiently increased the ME coupling coefficient and high sensitivity to weak magnetic detection, which influenced the equilibrium between the piezomagnetic coefficient and magneto-mechanical loss. This approach not only enhances magnetic-mechanical coupling at high frequencies but also introduces the self-biasing in Metglas, promising a pathway for advancing sensor technology in vector detection applications.