Printable anisotropic magnetoresistance sensors for highly compliant electronics

Printed electronics are attractive due to their low-cost and large-area processing features, which have been successfully extended to magnetoresistive sensors and devices. Here, we introduce and characterize a new kind of magnetoresistive paste based on the anisotropic magnetoresistive (AMR) effect. The paste is a composite of 100-nm-thick permalloy/tantalum flakes embedded in an elastomer matrix, which promotes the formation of appropriately conductive percolation networks. Sensors printed with this paste showed stable magnetoresistive properties upon mechanical bending. The AMR value of this sensor is 0.34%\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$0.34\%$$\end{document} in the field of 400 mT. Still, the response is stable and allows to resolve sub-mT field steps. When printed on ultra-thin 2.5-μm\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu \hbox {m}$$\end{document}-thick Mylar foil, the sensor can be completely folded without losing magnetoresistive performance and mechanically withstand 20μm\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$20\, \upmu {\hbox {m}}$$\end{document} bending radius. The developed compliant printed AMR sensor would be attractive to implement on curved and/or dynamic bendable surfaces for on-skin applications and interactive printed electronics.


Introduction
Printing technologies are low-cost ubiquitous techniques that are in the manufacturing process of household goods and devices [1][2][3][4][5]. Apart from the impression of product labels or the printed version of this journal, printing is now being explored to obtain multifunctional electronic devices as RFID (radio-frequency identification) antennas [6], solar cells [7], ferroelectric memories [8], and magnetic field sensors [9]. To arrive to the widespread of printed systems, we need to address the main challenges of printable electronics which are related to the instability, noise, and the low conductivity of the inks or pastes [10,11]. Overcoming these issues will allow us to produce on-demand functional parts with printed utility systems for diverse applications as implants [12], smart devices [13], and sensors [14].
Printed magnetic field sensors have been underrepresented, mainly due to the complexity to obtain printable, well-conductive magnetosensitive composites with stable resistance readout [9,15]. The introduction of magnetic particles into elastomer matrices is one approach to obtain magnetoresponsive compounds [16][17][18][19][20]. Magnetic nanoparticles embedded in conductive agglomerates can be concatenated using an external magnetic field during the curing process of the binder (e.g. Polydimethylsiloxane (PDMS)) [17,21,22]. The resulting composite is a magnetoresistive compound that can measure the magnetic field components in the direction of the aligned particles [17]. After the development of these flexible magnetoresistive composites, examples of solution-processable magnetic sensors with gel-like binders and elastomers appeared [23]. These systems achieved large resistance change signal, yet typically at the expense of rather high resistance and noise. A paradigm change arrived with the fabrication of pastes that were not based on nanoparticles, but on magnetic flakes obtained from thin film deposition [24,25]. Flakes-based GMR (Giant magnetoresistance) printed sensors have achieved 37% resistance change, with good temperature stability up to 95 • C, and flexed down to 12 mm bending radii as a result of the incorporation of polymer binders like polyepichlorohydrin (PCH) [25].
By now, printed magnetoelectronics were based on GMR sensors [9,[24][25][26]. Although they provide a large magnetoresistive effect, this technology requires involved fabrication due to the necessity to control the thickness of individual layers down to sub-nm [27,28]. In this regard, the use of AMR films is attractive because the complexity to obtain a magnetosensitive material is reduced [29,30].
In this work, we fabricate and investigate the sensing performance of a printable sensor based on AMR flakes of permalloy (Py). We demonstrate that despite having a lower magnetoresistance than GMR sensors, the signal stability is enough to resolve sub-mT range fields. In addition, the obtained printable paste has low resistance, can be printed in ultrathin polymer foils (2.5 μm), and withstands mechanical deformations down to 20 mm bending radii which makes it attractive for bendable and compliant electronics for on-skin applications [30][31][32][33].

Preparation of AMR powder
A 4-inch Si/SiO 2 wafer was spin-coated with AZ1505 photoresist (Microchemicals GmbH, Germany) as a sacrificial layer. Then, a thin film of [Ta (5 nm)/Py (100 nm)] was deposited on top of the covered wafer by magnetron sputtering deposition at room temperature (deposition rate = 0.14/0.18 (Ta/Py) nm/s , base pressure = 1 × 10 −7 mbar, Ar sputter pressure = 9 × 10 −4 mbar). The flakes were formed after dissolving the photoresist in acetone bath with an ultrasound-assisted lift-off procedure. The Py/Ta flakes were then recovered with a permanent magnet and rinsed with isopropanol. The remaining solvent was evaporated in an oven at 70 • C for 1 h until a dry powder was obtained.

Paste formulation and printing
The Py/Ta powder was incorporated in poly(styrenebutadiene-styrene) copolymer (SBS). First, the SBS was dissolved in xylol to a 0.15 g/mL concentration with magnetic stirring for 8 h. Afterwards, the magnetic flakes were added until we achieved a concentration of 50 mg/mL in the SBS solution. The obtained paste is then printed by pipetting onto the desired substrate and dried for 2 h at room temperature.

AMR characterization
The resistance was measured in a four-point configuration with a Tensormeter setup (HZDR Innovation, Germany) [34,35]. The amplitude of the applied voltage in the external contacts was 200 mV to induce a current of around 10 mA flowing through the sensor. A computer-controlled electromagnet applied magnetic field sweeps in a ± 400 mT range. The AMR response of the sensor was calculated as where H sat refers to the field when the resistance value saturates and H to each static field value.

Signal-to-noise analysis
The time evolution of the four-point resistance of the sensor was measured with a Tensormeter to determine the noise characteristics of the sensor. The four-point resistance in time was assessed with 5 ms integration time data points and then applying a Fast Fourier Transform (FFT) algorithm to obtain the noise spectral density of the signal.

AMR response upon bending
The sensor was mounted onto a bending setup to measure the AMR response at different bending radii. The bending setup allowed us to curve the sample to predefined fixed radius varying from flat state down to 2 mm. The magnetic field was applied in plane perpendicular to the bending axis. In this way, we ensure that the sensor was exposed to a homogeneous magnetic field of ±300 mT during the experiment. Figure 1a shows a schematic representation of the fabrication process of the Py/Ta flakes. A wafer was spin coated with a sacrificial layer and deposited with a 105-nm-thick Py/Ta AMR film. The AMR performance was measured on as prepared samples, revealing an AMR ratio of 1.9% and a saturation field of 5 mT (Fig. 1b). Then, the thin layer of Py/Ta was particulated dissolving the sacrificial resist in acetone with ultrasound assistance. After solvent evaporation, a magnetosensitive powder consisting of Py/ Ta flakes was obtained, and this powder mainly consists of tubular flakes (Fig. 1c) with a log-normal distribution for length with l mean = 85 μm and width with w mean = 26 μm (Fig. 1d). At this stage, the AMR flakes conform a dry powder which is employed in subsequent processing steps as the solid phase of the magnetic composite. The incorporation of a binder is necessary to promote both good contact between particles and conformal adhesion to a target surface. For printed flexible applications, it is necessary to choose an elastomer binder with low curing temperature and good adhesion properties. In this work, we use the SBS triblock copolymer since it promotes highly interconnected electrically conductive pathways which provide stable electrical signal even upon mechanical deformation [36]. Once the flakes are homogeneously incorporated into the SBS matrix, the resulting paste was applied with a pipette on a substrate (schematics The SEM micrograph shows that the sensor can comply to a highly irregular surface. g The changes in resistance produced by an external magnetic field are measured in a 4-point configuration. This layout allows measuring the h resistance during the drying process of the paste after printing, and the i AMR response of the sensor when an in-plane (parallel/perpendicular with respect to the current flow) magnetic field is applied. j Inset showing the AMR response at low fields ± 5 mT in Fig 1a). We used a Mylar foil ( 2.5 μm ) with patterned Au electrodes as a mechanically imperceptible flexible target substrate for printing (Fig. 1e). The ultrathin flexible Mylar substrate allows the sensor to fully comply to uneven surfaces with creases below 1 mm (Fig. 1f). To exclude the resistance contribution of the contacts from the sensing signal, the Au electrodes are patterned in a four-point configuration (Fig. 1g). After printing, we assessed the drying time necessary to have a stable sensor response. The paste started to show a measurable signal at 16 min (Fig. 1h). After 125 min, the paste resistance decreased five orders of magnitude and converge to a stable value. The resistance value stabilization was the sign that the solvent was completely evaporated.

Magnetoresistive performance of printed AMR sensors
To assess the AMR response of the printed sensor, we performed a magnetoresistive characterization (Fig. 1i). During the measurement, the field was applied in-plane parallel and perpendicular to the current direction to assess the anisotropic response of the sensor. In the parallel direction, the sensor demonstrated a 0.34% response; in the case when the applied field was perpendicular to the current flow direction, the AMR response was 0.28% . The sensor also demonstrated to have two different sensitivity regions, for magnetic flux densities above 5 mT the max slope was 12% T −1 ; meanwhile, the sensitivity reached 190% T −1 below 5 mT (Fig. 1j).

Bending performance
The bending properties are important to define the possibility to incorporate AMR printable sensors in, for instance, wearable applications. For this, the printed sensor was subjected to an analysis of its performance at different bending radii. We started with the sensor in flat layout and measured the AMR response upon bending, to predetermined radii of 6, 4, and 2 mm (Fig. 2a). During the experiment, the sensor showed a stable AMR response of around 0.28% (Fig. 2b).
A good percolation in a flakes system with a high number of interconnections allows to the network to compensate the connection breaks that may occur during the bending. In a system with sufficiently large amount of flakes, the current is able to pass through different paths in the network and maintain similar properties as the initial system. Despite having a relatively high concentration of conductive magnetoresistive flakes, the printed composite can withstand mechanically the bending up to 20 μm bending radius (Fig. 2c) without any sign of flake delamination. The smart combination of thin foils and flexible materials allowed us to achieve this stable mechanical performance that demonstrate the attractiveness to use this ultrathin foils as a substrate for highly compliant printed sensors.

Detection at low magnetic fields
The World Health Organization has established 40 mT as the safe continuous exposure limit for humans [37]. In this respect, the possibility to measure low fields is important to comply with this safety standard. To quantify the sensor performance, we first analyzed the stability of the resistance in time. We consider a window of 60 s to measure the resistance level variation of the magnetoresistive sensor without stimuli in time. When sampled at 60 S/s the fourpoint resistance of the sensor with nominal resistance of R 0 = 25.693 Ω had a peak fluctuation of ±2 mΩ , this was consistent among different iterations, as shown in Fig. 3a. When applying FFT algorithm, the noise spectra of these signals (Fig. 3b) showed a 1∕ √ f trend with noise density at 1 Hz, n 1 Hz = 6.1 × 10 −4 Ω∕ √ Hz . Considering these parameters for a 3-sigma ( 3 ) confidence detection, the expected resolution of the sensor implementation is 60 μT in the region of highest sensitivity. This confirm the capabilities of the sensor to differentiate sub-mT magnetic fields with high level of confidence. To confirm the high stability of the sensor, we tracked the sensor response from 0 up to only 300 μT . The sensor showed a linear trend with Fig. 2 a Photographs revealing the bending state of the sensor and b the respective AMR response measured at ±400 mT . c SEM micrograph of the printed AMR sensor upon complete folding and d an inset showing good adhesion of the AMR flakes down to 20 μm bending radius 32 mΩ∕mT sensitivity and the data distribution illustrated with 3-sigma error bars was in accordance with a sub-mT resolution (Fig. 3c).

Conclusions
Here, we report printable AMR sensors, which can be used to assess sub-mT fields and can be bent to 2 mm bending radius without sacrificing the sensor magnetoresistive performance. Also, we demonstrated the advantages of using SBS to prepare AMR-based sensors. The proposed formulation permits printing of sensors on ultrathin foils with stable mechanical performance even if the sensor is folded (20 μ m bending radius). The equivalent floor noise of n 1 Hz = 6.1 × 10 −4 Ω∕ √ Hz shown by the printed sensors allowed us to resolve fields below 1 mT which complies with the safety requirements of continuous magnetic field exposure from the WHO. The good performance at low fields and the high compliancy show the suitability to use this system for interactive wearable and printed electronics as switches, and sensors for physical activity. This technology can be also extended to additive manufacturing techniques, as 3D printing, for direct integration of magnetic sensors into ondemand functional parts.