Toward mechano-spintronics: Nanostructured magnetic multilayers for the realization of microcantilever sensors featuring wireless actuation for liquid environments

Abstract

Silicon microcantilevers are realized and tested using different ferromagnetic thin films as active actuators. The exploited design is optimized for operating the sensor in a liquid environment. Different magnetic materials are used as actuator elements: a soft layer of face-centered cubic Co, a hard layer of Co₈₀Cr₂₀ (subscript: atomic composition in percentage) and a (Co⁵Cu¹⁰)₅ multilayer (superscript: thickness; subscript: number of repetitions). The thin film magnetizations are characterized both in the film plane and out of it. We characterize the devices in air and in water comparing piezoelectric and magnetic actuation, confirming that nanostructured magnetic multilayers represent a new and promising route to optimize the actuation of magnetic microcantilevers. Complete sets of dynamical measures, consisting of stability plots, are discussed. Finite element simulations performed with a commercial code and inherent to a static analysis of different magnetic microcantilevers are commented, casting more light on the importance of having a nanostructured actuator for a high-efficiency energy transfer. This opens the route to new challenging devices, where the spin arrangement at the nanoscale is used to induce either mechanical deformations or movements by effect of an electromagnetic field.

Keywords

Actuator ferromagnetic thin films nanomagnetism

Introduction

Microcantilevers (MCs) are used as sensor devices in a variety of applications (Singamaneni et al., 2008). To operate in a liquid environment for biodetection purposes, a wireless actuation is recommended mainly because forests of peaks develop in the oscillating structure when it is excited by means of piezoelectric disk (Kokavecz and Mechler, 2007). Dipolar magnetic actuation, eddy current mechanism, and Lorentz force actuation have been addressed as viable solutions (Baù et al., 2008; Han et al., 1996; Vančura et al., 2007).

Resonant MC sensors are employed in a variety of physical, chemical, and biochemical sensing applications (Eom et al., 2011). Their working principle is based on the measure of a frequency shift as a result of the interaction with an analyte (Eom et al., 2011). Through this shift, it is possible to estimate the mass of the target substance, thus realizing a microbalance with an extremely high mass resolution (Eom et al., 2011).

When liquid environment operation is required, particular care should be dedicated to the choice of both the design and the actuation technique. Plate-like structures are demonstrated (Ricciardi et al., 2010; Vančura et al., 2007) to perform better in liquid than beam-like structures, because of the higher ratio between the fluid-induced inertial and damping forces. This fact results in an increase of the device Q factor, a parameter directly connected to the sensitivity and detection limit (Nickolay et al., 2004) and usually quite critical for vibration in highly viscous fluids like liquids (Van Eysden and Sader, 2007).

The most common actuation scheme, that is, the mechanical one, realized by means of a piezoelectric disk put in contact with the cantilever support, shows some difficulties when applied to liquid environment. Because of the low signal-to-noise ratio, a phenomenon called “forests of peaks” (Kokavecz and Mechler, 2007), tending to hinder the cantilever resonance, appears in the output signal, thus seriously compromising the reliability of measurements.

Magnetic actuation, on the other hand, being a contactless method, overcomes this difficulty and also appears convenient for integration and miniaturization purpose. Two different techniques, both using an external magnetic field, are possible: actuation by means of the Lorentz force (Nannini et al., 2006; Nickolay et al., 2004; Temnykh and Lovelace, 2002) or magnetically via a magnetic film deposited on the structure (Judy and Muller, 1996; Lee et al., 2007).

We here propose a new solution for cantilever-based sensors vibrating in liquid, where a microplate structure is coated with a thin (around 50 nm) nanostructured magnetic layer and put in vibration by means of an external sinusoidal varying magnetic field, while the local spin arrangement at the nanoscale is used to induce a mechanical response tunable by an appropriate material choice. A thicker magnetic layer, featuring thicknesses in the micrometer range, would result in magnetic noise and poorer response caused by multiple magnetic domains forming across the layer.

Different layers were deposited and tested as actuator elements: a soft 50-nm layer of face-centered cubic Co, a hard 50-nm layer of Co₈₀Cr₂₀ (subscript: atomic composition in percentage), and a (Co⁵Cu¹⁰)₅ multilayer (superscript: thickness; subscript: number of repetitions). The first choice for magnetic actuation, a thin Co layer, meets simplicity requirements and cost-effectiveness (Chiolerio et al., 2007). A typical magnetization arrangement is sketched in Figure 1(a): the magnetic thin film, represented in a cross section along the main axis of the cantilever, exhibits a predominant in-plane magnetization due to shape anisotropy; however, a random distribution of local easy magnetization axes is expected in the different grains originating from the growth method, resulting in mesoscopic regions of tilted magnetization, which can be assimilated to magnetic domains (Chiolerio et al., 2007). Therefore, at remanence, a small out-of-plane component of the magnetization is present. The hard material Co₈₀Cr₂₀ (CoCr in the following text) was chosen for its perpendicular anisotropy, well known and exploited for a number of applications (Chiolerio et al., 2009); in fact, a significant component of the magnetization is oriented out of the plane of the film even in the in-plane remanence (IPR) state (Fisher et al., 1984; Svedberg et al., 2002). In contrast, the (Co⁵Cu¹⁰)₅ multilayer structure (CuCo in the following text) was chosen because of the strong shape anisotropy typical of very thin ferromagnetic layers which favors the parallel alignment of the magnetization of each Co layer; dipolar coupling among different layers may play a role too in keeping the magnetization parallel to the film plane (Fruchart and Thiaville, 2005). As a result, it is very difficult to rotate the magnetization vector out of the plane of the cantilever and align it to the external driving field; consequently, the IPR state should give an optimal resonance efficiency with very good operating stability. Furthermore, each Co layer is magnetically homogeneous, since the energy required for the occurrence of Bloch domain walls with perpendicular magnetization is far from being reached during the sensor operation, hence increasing the actuation efficiency. This configuration is sketched in Figure 1(b), where the magnetization is represented in full parallel alignment throughout the Co layers. As well known, the alignment of magnetization in adjacent layers at the remanence can be either parallel or antiparallel (Figure 1(c)), the arrangement being dictated by interlayer exchange coupling (IEC): ultrathin metallic spacers allow the Co layers to interact and spontaneously arrange themselves in order to minimize the total magnetic energy (Korkin and Rosei, 2008). Antiparallel alignment of layers at the remanence is particularly interesting for spintronic devices such as spin valves (Chiolerio et al., 2007). However, this specific arrangement is such that the torque generated by a magnetic field perpendicular to such cantilever axis would be near to zero.

The thickness of Cu interlayers in the (Co⁵Cu¹⁰)₅ multilayer investigated here was selected in order to give rise to parallel alignment of the magnetization of all layers.

Experimental

Starting from silicon on insulator substrates, we realized freestanding MCs through a bulk and surface micromachining process consisting of a wet and dry etching for the handle and device layer, respectively (Then et al., 2006). Our microdevices were fabricated with a reduced aspect ratio (length/width) with respect to standard MCs since such geometry is expected to show the highest Q factor (resonance frequency/amplitude width at 3 dB) in a liquid environment (Canavese et al., 2007; Ricciardi et al., 2010). Typical dimensions are as follows: 1000 μm in width, 1550 μm in length, and 6.5 μm in thickness.

Different magnetic materials were radio frequency (RF) sputtered as actuator elements: a 50-nm face-centered cubic Co layer, a Co₈₀Cr₂₀ layer (referred to as CoCr), and a (Co⁵Cu¹⁰)₅ multilayer structure (referred to as CuCo). A 20-nm-thick Au layer was thermally evaporated on the structures to prevent oxidation and chemical instability, in particular because of the liquid environment.

We also realized a structure nominally characterized by antiparallel magnetization arrangement (Figure 1(c)), corresponding to a (Co⁵Cu^1.6)₅ multilayer, in order to proof the mechano-spintronic concept. The nanometric granular structure of deposited metals was investigated by a field effect scanning electron microscope (FESEM). Their magnetic properties were measured at room temperature up to 18 kOe by means of an alternating gradient force magnetometer (AGFM), both in the plane of the films and out of it.

The experimental magnetic microcantilever (MMC) actuation setup is composed of a solenoid (500 turns, 2 cm in diameter, 10 cm in length) placed at 1 mm from the MC chip, in which a steady direct current (DC; from 0.1 to 3 A) is rippled by a high-frequency relay; an alternative actuation system makes use of a commercial piezoelectric device (PI Instruments). A position sensitive diode (PSD) is used to monitor the motion of the cantilever, thanks to an optical lever readout.

The samples were brought to either the IPR or out-of-plane remanence (OPR) before the characterization, using an SmCo permanent magnet. Each resonance measurement consisted in collecting oscillation phase and amplitude. Stability plots consisted in the sequential acquisition of 50 measures. Devices were characterized while vibrating in air and in liquid (deionized (DI) water) environment.

Results

Magnetic thin film characterization

The morphology of the three magnetic layers prior to the cap layer evaporation is shown in Figure 2. Sub-20-nm grains are evidenced in each panel: having a fine and homogeneous structure at the nanoscale is necessary to avoid magnetic dissipation, which may increase the cantilever temperature and eventually modify the functional layers. The magnetic properties were measured at room temperature, applying a magnetic field in the plane of the thin film, and are summarized in Figure 3: starting from hysteresis cycles, the average coercivity (field at zero magnetization) and remanence (magnetization at zero field) to saturation (maximum magnetization) ratios were extracted. What is remarkable is the very low in-plane coercivity of the nanostructured multilayer (CuCo) if compared to the Co simple layer, meaning that it can be easily magnetized by the application of a magnetic field directed along the plane of the film, much more than its single-layer counterpart, showing the effects of shape anisotropy and interlayer dipolar coupling (Korkin and Rosei, 2008). The ratio between the out-of-plane coercivity and the in-plane one found for this material is about 10 times lower than that of the two single layers (Table 1). This means that in the case of the multilayered film, a magnetic field applied perpendicularly to the film/cantilever plane is about 10 times less effective in modifying the magnetic state (i.e. the magnetic domain structure) at the remanence with respect to the two single-layer films. This is particularly interesting for actuation purposes, where the magnetic field is applied perpendicular to the ferromagnetic film in order to actuate the cantilever oscillation without however changing its magnetic response and the associated magnetomechanical coupling; this allows for a longer operational stability of the MMC.

Table 1.

Input materials’ properties used in the FEM simulation. Notice that the magnetic permeability tensor is approximated, having zero nondiagonal terms. Furthermore, the elastic properties are that of bulk matter.

Material	Young’s modulus (GPa)	Poisson’s ratio	Density (kg m⁻³)	Thickness (nm)	Magnetic permeability (H m⁻¹)			Out-of-plane over in-plane coercivity ratio	Remanence (10⁵ A m⁻¹)
					χ_xx	χ_yy	χ_zz		M _x	M _y	M _z
Co	200	0.31	8900	50	209	209	7.60	2.1	7.00	0	11.2
CoCr	215	0.29	8500	50	155	155	1.21	2.5	6.07	0	9.70
CuCo	150	0.33	8900	75	92	92	2.29	23	9.34	0	0.00

FEM: finite element method.

Numerical simulations

Numerical simulations with the finite element method (FEM) were performed using the commercial code Comsol Multiphysics^™ 4.2. The model, as shown in Figure 3, consists of two parallelepiped-shaped blocks representing the cantilever and a portion of the fluid domain surrounding it, respectively. Magnetic and mechanical properties are reported in Table 1.

In the static case, the software solves the following equations

{\begin{matrix} \nabla • (μ_{0} \bar{\bar{μ_{r}}} \bar{H}) = 0 \\ \bar{H} = - \nabla V_{m} \end{matrix}

(1)

- \nabla • \bar{\bar{σ}} = F_{v}

(2)

Equation (1) is introduced to characterize the magnetostatics of the problem, while equation (2) describes the mechanical behavior of the actuated cantilever. The quantities $μ_{0}$ , $\bar{\bar{μ_{r}}}$ , $\bar{H}$ , $V_{m}$ , $\bar{\bar{σ}}$ , and $F_{v}$ represent the vacuum permeability, the relative permeability tensor, the magnetic field vector, the magnetic potential, the stress tensor, and the volume force vector, respectively.

The presence of the active sputtered film on one MMC side has been taken into account both from a magnetic and a mechanical point of view by means of a “thin film” approach (Figure 4). The constitutive relation of the layer $\bar{B} = μ_{0} {\bar{\bar{μ}}}_{r} \bar{H} + {\bar{M}}_{rem}$ was implemented through a magnetic shielding boundary condition, while its mechanical behavior is described by the standard shell interface (COMSOL AB, 2010).

It is worth noting that the term ${\bar{M}}_{rem}$ represents a remanent magnetization (see Table 1) and it is a nonstandard contribution added by the user. A normal magnetic field directed upward with the intensity equal to 5 kA m⁻¹ was assigned at the bottom surface of the fluid domain to account for the presence of the solenoid. A magnetic insulation condition was imposed to all the other surfaces of the model. The cantilever is considered as clamped at its left end (Figure 4). Symmetry boundary conditions are applied on all the surfaces lying on the xz plane.

Finally, to couple equations (1) and (2) on the upper surface of the cantilever, representing the sputtered film, the following equation was set

\bar{t} = - 0.5 \bar{n} (\bar{H} • \bar{B}) + (\bar{n} • \bar{H}) • {\bar{B}}^{T}

(3)

where $\bar{t}$ represents the mechanical traction and $\bar{n}$ is the normal unit vector; the second term of equation (3) derives from the definition of Maxwell stress tensor.

The 3D model was meshed with tetrahedral quadratic elements, resulting in about 10⁵ degrees of freedom solved for. The following three different studies were performed: (a) a static mechanical study for evaluating the impact of the magnetic film on the stiffness of the whole structure, (b) an eigenfrequency study for calculating the first-mode undamped resonance frequency, and (c) a static multiphysics study, involving structural mechanics and magnetostatic equations, in order to evaluate the static response of the whole structure to an external magnetic field. In all the three studies, the mesh was refined until the error between two consecutive runs was lower than 1%. The default direct solver MUltifrontal Massively Parallel Sparse (MUMPS) was employed in all the calculations.

Table 2 summarizes all the results of the simulations. As one can see, the structure with the CuCo layer is the one which displays the highest stiffness and the lower resonance frequency. This is due to the combination of the effects of the thickness and the density of the multilayer, the former leading to an increase of the static stiffness and the latter leading to a higher inertial contribution. In all the cases, the cantilever subjected to an external magnetic field bends upward in agreement with the formula for the torque

\bar{T} = \bar{m} \times \bar{B}

(4)

where $\bar{m}$ represents the magnetic moment and $\bar{B}$ the magnetic induction (Temnykh and Lovelace, 2002).

Table 2.

Output characteristics of the different magnetic actuators.

Magnetic layer	Stiffness (N m⁻¹)	First-mode resonance frequency (Hz)	MC displacement (nm)
Co	3.338	3821	47.4
CoCr	3.344	3827	41.9
CuCo	3.349	3801	65.2

MC: microcantilever.

The cantilever covered by the CuCo layer is also the one that performs best in an external magnetic field, showing the largest displacement. This fact is due to the highest magnetic moment of the film available along the lever axis, thanks to the extremely low in-plane coercive field of the system (see Figure 2) and to the fact that saturation is easily reached under normal operation conditions.

It is possible therefore to conclude that the structure covered by the CuCo layer is the one which responds better to an external magnetic field. As reported in the following section, this finding is also confirmed by the experimental results.

Characterization of MC resonance curves

Typical measurements of first resonance mode in air environment at room temperature are shown in Figure 5. Each graph compares the absolute flexural oscillation amplitude of the three investigated materials (Co, CoCr, and CuCo in left, middle, and right panels, respectively) when excited with the piezoelectric (P) and with the magnetic (OPR and IPR states) actuation system. All the curves span a 600-Hz-wide region. Resonance frequency and Q factor stability in air (50 measurements taken every 20 s) are shown in Figure 6 (left and right panel, respectively): for clarity, just Co and CuCo samples were compared. Figure 7 shows a comparison between resonance curves obtained with piezoelectric (P) and with magnetic (IPR state) actuation when the CuCo-MMC is vibrating in liquid environment. Monitored here was the second flexural mode, since the first one proved too noisy. Measurements were performed at room temperature with a frequency span of 3 kHz.

We also measured the multilayered structure nominally characterized by the antiparallel magnetization arrangement in adjacent layers; no mechanical response was found by driving the MMC by means of alternating current (AC) magnetic field, as indeed expected for this arrangement (data not shown).

Table 3 presents a collection of useful parameters of the microdevices tested in different working conditions: the resonance frequency (as determined by the Levenberg–Marquardt best fit with a Lorentzian curve), the Q factor, and the R² of the fit (coefficient of determination (COD)).

Table 3.

Selected parameters describing the MMC in air, vacuum, and liquid (DI water).

Magnetic layer	Environment	Excitation	Resonance frequency (Hz)	Q factor	R² (COD)
Co	Air	P 10.0 V	3642.6 ± 0.2	84.6	0.9874
Co	Air	IPR 3.0 A	3642.14 ± 0.09	86.8	0.9963
Co	Air	OPR 3.0 A	3642.47 ± 0.09	86.1	0.9959
CoCr	Air	P 10.0 V	3708.86 ± 0.03	81.2	0.9989
CoCr	Air	IPR 3.0 A	3713.3 ± 0.1	80.4	0.9954
CoCr	Air	OPR 3.0 A	3718.7 ± 0.1	79.0	0.9955
CuCo	Air	P 10.0 V	3551.1 ± 0.3	47.9	0.9606
CuCo	Air	IPR 3.0 A	3554.4 ± 0.1	87.9	0.9904
CuCo	Air	OPR 3.0 A	3550.54 ± 0.04	85.1	0.9979
CuCo	Vacuum	P 50 mV	3626.896 ± 0.001	7895	0.9753
CuCo	Vacuum	IPR 0.5 A	3621.022 ± 0.001	9394	0.9884
CuCo	Liquid (second mode)	P 10.0 V	5960 ± 5	5	Multiple peaks
CuCo	Liquid (second mode)	IPR 3.0 A	5429 ± 1	16	0.9411

MMC: magnetic microcantilever; DI: deionized; P: piezoelectric; IPR: in-plane remanence; OPR: out-of-plane remanence; COD: coefficient of determination.

For what concerns data in vacuum and air environment, the deviation of the experimental frequencies from the FE results has to be ascribed to the unavoidable geometrical imperfections of the samples as well as to the damping induced by air.

Discussion

Resonance curves of Co-deposited MMCs show that the magnetic actuation is less efficient than the piezoelectric one (smaller oscillation amplitude with respect to the “P” curve, Figure 5) and that the OPR configuration is more efficient than the IPR state.

The stability curves, which require a long sampling time, were recorded setting the ripple source to 2.0 A to avoid coil and MMC overheating. A comparison with the piezoelectric actuation reveals a stronger noise but a higher Q factor together with a slightly improved stability over time of the resonance frequency (Figure 6).

The CoCr-deposited MMCs show an improved efficiency actuation of the magnetic system over the piezoelectric one, the oscillation amplitude being bigger for both the IPR and the OPR configurations with respect to the piezo (Figure 5). The stability (not shown) is similar to that of Co-deposited MMCs, while the Q factor relative to the magnetic actuation is comparable to the one obtained through the piezoelectric excitation.

The CuCo-deposited MMCs finally show the best Q factor, stability, and efficiency, exceeding the piezoelectric actuation (Figures 5 and 6). It is worth to note that the highest efficiency is also confirmed by the FE data (section 2.1). In addition, the IPR configuration, for this case, behaves better than the OPR one, as expected.

In vacuum measurements, magnetic actuation leads to better performance than the piezoelectric one for all the proposed microdevices. Finally, the amplitude plot of the CuCo-deposited MMC in liquid is shown in Figure 7, where the magnetic actuation is compared to the piezoelectric one, in the case of the second normal mode of vibration. Magnetic actuation is observed to generate a noisier but simpler response featuring a single peak; on the contrary, piezoelectric actuation gives a response which is the convolution of two peaks, resulting in a lower Q factor. In fact, the full width at half maximum (FWHM) moves from 340 ± 10 to 1195 ± 5 Hz, which is more than 3.5 times. The experimental resonance properties of different microdevices are summarized in Table 3.

Our experimental findings open the route to mechano-spintronics: tailoring the local spin arrangement by means of microfabrication (the magnetization lies along the cantilever beam) and nanofabrication (each Co layer is homogeneously magnetized in parallel arrangement with respect to the others), it is possible to explore magnetomechanical couplings. We foresee that by exploiting IEC (Korkin and Rosei, 2008), it could be even possible to measure mechanical torques during the growth of layer-by-layer magnetic multilayers with a strong sensitivity to the spacer thickness. In fact, when the thickness of Cu spacer is reduced below the mean free electron path, a quantum arrangement of spins is realized and an oscillatory coupling is obtained, so that depending on the actual Cu thickness, one can produce either a parallel (Figure 1(b)) or an antiparallel (Figure 1(c)) alignment, the first one resulting in a measurable mechanical torque and the second one in zero torque.

Conclusion

We fabricated MMCs optimized for working in a liquid environment, experimenting with different actuation solutions. In particular, simple single layers, such as Co and Co₈₀Cr₂₀, and more complex multilayered structures based on nanostructuring of the ferromagnetic layers (structure: (Co⁵Cu¹⁰)₅) were realized. Film morphology and room temperature magnetic properties were characterized and compared. Characterization of the microdevice actuation efficiency in air, vacuum, and liquid environment and the analysis of their resonance properties lead us to conclude that the active layer based on nanostructured multilayers represents the best compromise and gives the lowest noise and maximum Q factors when actuated by an AC magnetic field. In conclusion, we indicate this class of materials as a good candidate for MMCs actuation and we foresee interesting applications for the exploitation of mechano-spintronic coupled systems.

Footnotes

Acknowledgements

The authors wish to thank Dr R. Castagna, Dr P. Pandolfi, and Dr S. Guastella for their precious help.

Funding

This work was supported in part by Regione Piemonte (Namatech grant) and MIUR (FIRB2003 LATEMAR grant and PRIN2007 grant). G.C. was supported by Fondazione CRT (Progetto Lagrange).

References

Baù

Ferrari

Marioli

. (2008) Contactless electromagnetic excitation of resonant sensors made of conductive miniaturized structures. Sensors and Actuators A: Physical 148: 44–50.

Canavese

Marasso

Quaglio

. (2007) Polymeric mask protection for alternative KOH silicon wet etching. Journal of Micromechanics and Microengineering 17: 1387–1393.

Chiolerio

Allia

Chiodoni

. (2007) Thermally evaporated Cu-Co top spin valve with random exchange bias. Journal of Applied Physics 101: 123915-1–123915-6.

Chiolerio

Martino

Coïsson

. (2009) Vector magnetisation measurements on thermally evaporated CoCr multilayers and solid solutions for spintronic applications. Journal of Magnetism and Magnetic Materials 321: 3099–3103.

COMSOL AB (2010) Comsol Multiphysics 4.2 User Manuals. COMSOL AB, Burlington, MA 01803 USA.

Eom

Park

Yoon

. (2011) Nanomechanical resonators and their applications in biological/chemical detection: nanomechanics principles. Physics Reports 503: 115–163.

Fisher

Au-Yeung

Sabo

(1984) Perpendicular anisotropy constants and anisotropy energy of oriented cobalt-chromium alloys. IEEE Transactions on Magnetics 20: 806–808.

Fruchart

Thiaville

(2005) Magnetism in reduced dimensions. Comptes Rendus Physique 6: 921–933.

Han

Lindsay

Jing

(1996) A magnetically driven oscillating probe microscope for operation in liquids. Applied Physics Letters 69: 4111–4113.

10.

Judy

Muller

(1996) Magnetic microactuation of torsional polysilicon structures. Sensors and Actuators A: Physical 53: 392–397.

11.

Kokavecz

Mechler

(2007) Investigation of fluid cell resonances in intermittent contact mode atomic force microscopy. Applied Physics Letters 91: 023113.

12.

Korkin

Rosei

(eds) (2008) Nanoelectronics and Photonics. Heidelberg: Springer.

13.

Lee

Kang

(2007) Comparison of etching characteristics of SiO₂ with ArF photoresist in C₄F₆ and C₄F₈ based dual-frequency superimposed capacitively coupled plasmas. Microelectronic Engineering 84: 1719–1723.

14.

Nannini

Pacia

Pieri

. (2006) A CMOS—compatible bulk technology for the fabrication of magnetically actuated microbalances for chemical sensing. Sensors and Actuators B: Chemical 118: 343–348.

15.

Nickolay

Michael

Panos

(2004) Cantilever transducers as a platform for chemical and biological sensors. Review of Scientific Instruments 75: 2229–2253.

16.

Ricciardi

Canavese

Castagna

. (2010) Integration of microfluidic and cantilever technology for biosensing application in liquid environment. Biosensors and Bioelectronics 26: 1565–1570.

17.

Singamaneni

LeMieux

Lang

. (2008) Bimaterial microcantilevers as a hybrid sensing platform. Advanced Materials 20: 653–680.

18.

Svedberg

Khizroev

Chang

. (2002) Signal-to-noise degradation in perpendicular media under thermal aging and aging in magnetic field. Journal of Applied Physics 92(11): 6714–6720.

19.

Temnykh

Lovelace

RVE

(2002) Electro-mechanical resonant magnetic field sensor. Nuclear Instruments and Methods 484: 95–101.

20.

Then

Vidic

Ziegler

(2006) A highly sensitive self-oscillating cantilever array for the quantitative and qualitative analysis of organic vapor mixtures. Sensors and Actuators B: Chemical 117: 1–9.

21.

Van Eysden

Sader

(2007) Frequency response of cantilever beams immersed in a viscous fluid with applications to the atomic force microscope: arbitrary mode order. Journal of Applied Physics 101: 044908.

22.

Vančura

Lichtenberg

. (2007) Liquid-phase chemical and biochemical detection using fully integrated magnetically actuated complementary metal oxide semiconductor resonant cantilever sensor systems. Analytical Chemistry 79: 1646–1654.