APPLIED PHYSICS LETTERS 98, 013502 共2011兲

Active microcantilevers based on piezoresistive ferromagnetic thin films

Harish Bhaskaran,1 Mo Li,1 Daniel Garcia-Sanchez,1 Peng Zhao,2 Ichiro Takeuchi,2 and
Hong X. Tang1,a兲
1
Department of Electrical Engineering, Yale University, New Haven, Connecticut 06511, USA
2
Department of Materials Science and Engineering, University of Maryland, College Park,
Maryland 20742, USA
共Received 23 October 2010; accepted 6 December 2010; published online 3 January 2011兲
We report the piezoresistivity in magnetic thin films of Fe0.7Ga0.3 and their use for fabricating
self-transducing microcantilevers. The actuation occurs as a consequence of both the ferromagnetic
and magnetostrictive properties of Fe0.7Ga0.3 thin films, while the deflection readout is achieved by
exploiting the piezoresistivity of these films. This self-sensing self-actuating micromechanical
system involves a very simple bilayer structure, which eliminates the need for the more complex
piezoelectric stack that is commonly used in active cantilevers. Thus, it potentially opens
opportunities for remotely actuated cantilever-based sensors. © 2011 American Institute of Physics.
关doi:10.1063/1.3533390兴

Microcantilevers have been at the heart of atomic force demonstrate the piezoresistivity in metallic Fe0.7Ga0.3 thin
microscopy and a range of chemical and biological sensing films and its use for both magnetic actuation and piezoresis-
applications. The transduction of microcantilevers often in- tive sensing in micromechanical devices. Our fabrication
volves off-chip actuators such as a piezoshaker, as well as process starts with micromachined membranes of high-stress
off-chip displacement readout, such as optical deflection. Ac- silicon nitride 共330 nm thick兲 on a silicon wafer made by
tive cantilevers that contain integrated actuation and sensing anisotropic KOH etch. A thin layer of Fe0.7Ga0.3 共100 nm兲 is
elements do not require alignment of actuating and sensing sputtered on the membranes 共for details of the material, see
components. Therefore they offer a higher level of system Zhao et al.7兲. The cantilever structures are then patterned
integration and are more suitable for portable applications. using photolithography. Ion milling is used to etch the
Active cantilevers can be constructed by integrating Fe0.7Ga0.3 layer. This is followed by etching Si3N4 in a reac-
piezoelectric1–3 or magnetic thin films4,5 to exert driving tive ion etcher using CHF3 to pattern and release the canti-
forces. Using magnetic forces to actuate micromechanical lever beams. The cantilevers are designed to have a stress
devices is advantageous for applications that are susceptible concentration region at the base in order to optimize the pi-
to perturbations, as magnetic fields can be generated re- ezoresistive response 关Fig. 1共a兲兴. Based on finite element
motely. To build active magnetic cantilevers, on-chip readout simulations of stress distribution in the cantilever, we esti-
of the mechanical motion is also desired. One scheme is to mate that the stress concentration region accounts for ⬃85%
integrate a layer of piezoelectric material, so that a potential of the measured resistance in the fabricated devices, thus
difference is generated in the piezoelectric layer when the contributing to the bulk of the piezoresistive response. A
device is stressed by magnetic force. This so-called magne- scanning electron micrograph image of a fabricated cantile-
toelectric effect 共ME兲 has been utilized in sensors using a ver of dimensions 100 共length兲 ⫻ 20 共width兲 ␮m2 is shown
variety of magnetostrictive materials.6–9 However, when de- in Fig. 1共a兲. We note that due to the intrinsic stress in the
vices are scaled to smaller dimensions, the use of a thick deposited Fe0.7Ga0.3 layer, bending of the cantilever struc-
multilayer structural stack that is necessary for harvesting tures is observed. This bending causes imperfect alignment
ME effects is often undesirable. In this letter, instead of using of magnetic field with the cantilever beam. As a result, the
the ME effect, we utilize the intrinsic piezoresistivity in a magnetic actuation is a combined effect of magnetic torque
thin film Fe0.7Ga0.3 共Ref. 7兲 to detect and magnetically actu- and magnetostriction of Fe0.7Ga0.3 thin films. As we demon-
ate micromechanical motion. This method has significant ad- strate later, a qualitative comparison of the relative magni-
vantages over the ME effect in that it permits further scaling tude of each of these effects can be obtained.
down of device dimensions, meanwhile eliminating the need We first investigate the piezoresistivity of the Fe0.7Ga0.3
for integration of piezoelectric materials and electrodes film by directly driving a piezodisk mounted below the
which involves extensive processing. sample holder. The nominal resistance of the cantilevers is
Although piezoresistivity in metals is a well-known phe- around 1.7 k⍀. All measurements are carried out in high
nomenon, this property is rarely applied as transducing ele- vacuum at the resonance frequency of the cantilever in the
ments in microdevices due to the nominally low gauge factor dynamic regime. A schematic for the piezoresistive measure-
␥ 关defined as 共dR / R兲 / 共dL / L兲, where L and R are the length ment is shown in Fig. 1共b兲. The device resistance Rcant
and resistance of a strain gauge, respectively兴. Recently, gold changes as a function of strain induced by the bending of the
thin films have been shown to be efficient piezoresistive sen- cantilever. A dc bias voltage Vcant = 1 V is supplied to the
sors on nanoscale mechanical devices.10,11 In this letter, we cantilever to convert the resistance variation to voltage sig-
nal, which is amplified with a preamplifier. The frequency
a兲
Author to whom correspondence should be addressed. Electronic mail: response of the cantilever measured with a network analyzer
[email protected]. is shown in Fig. 1共c兲. Fe0.7Ga0.3 cantilevers are observed to

0003-6951/2011/98共1兲/013502/3/$30.00 98, 013502-1 © 2011 American Institute of Physics

Downloaded 03 Jan 2011 to 128.36.45.227. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
 013502-2 Bhaskaran et al. Appl. Phys. Lett. 98, 013502 共2011兲

FIG. 2. Frequency response due to magnetically driven actuation of two

different cantilevers, one aligned in parallel 共longitudinal兲 to the applied
field 共left兲, while the other along the orthogonal 共transverse兲 directions
共right兲.

resonance frequencies of the two cantilevers is due to fabri-

cation variation.
In order to ascertain the mechanism of actuation, we
note that the magnetostrictive effect is different from a
simple torque effect. In thin sputtered films of nanocrystal-
line materials 共as in the present case兲, the film is a multido-
main with no dominant magnetic easy axis. Thus, a magnetic
field applied in the plane of the cantilever and in a direction
perpendicular to the cantilever’s length 共in-plane transverse
drive兲 would produce a pure magnetostrictive effect, whereas
FIG. 1. 共Color online兲 共a兲 Scanning electron micrograph of a fabricated
Fe0.7Ga0.3 microcantilever. Bright areas indicate Fe0.7Ga0.3 on silicon nitride. the torque effect is absent since the film magnetization is
共b兲 Schematic of circuit used for the measurement. A piezodisk shaker or ac always aligned with the field.12 For the magnetic field ap-
coils are used to drive the cantilever and 共c兲 piezoresisitive readout of reso- plied parallel to the cantilever length 共longitudinal drive兲,
nances of a gold 共left兲 and an Fe0.7Ga0.3 microcantilever 共right兲 of the same although every effort is made in this experiment to keep the
geometry, showing the relative amplitude of resonance peaks for a measure
of relative piezoresistive sensing efficiency. Responses at two drive ampli-
cantilever length well aligned to the magnetic fields, there is
tudes 共to the piezoelectric disk drive兲 are displayed. an experimental uncertainty as to the presence of residual
misalignment. This misalignment introduces additional mag-
netic torque actuation of the cantilever. As a result, the ob-
resonate at their fundamental modes in the frequency range served cantilever vibration amplitude is higher with longitu-
of 16–24 kHz with quality factors ranging from 3000 to 6000 dinal drive than with transverse drive. To isolate the relative
in high vacuum 共⬍1 ⫻ 10−4 Torr兲. This compares well with effect of each and directly access the magnetoelastic stress
the expected resonance frequency of 21 kHz for such canti- effect, we monitor the resonance frequency shift of the can-
levers, as computed using finite element methods 共modeled tilever when subjected to dc magnetic fields in two normal
as a no-stress silicon nitride cantilever with Fe0.7Ga0.3 in-plane directions. The magnetic torque effect is expected to
film on top兲. We compare the piezoresistive effect of the be present only when the field is applied with the longitudi-
Fe0.7Ga0.3 film with that in a gold film of the same thickness nal drive configuration especially for cantilevers that are bent
of 100 nm, as shown in Fig. 1共c兲. Accounting for frequency due to intrinsic stress, whereas it would be significantly
smaller with the transverse drive configuration 关inset in Fig.
differences between the two cantilevers 共explained by the
3共a兲兴.
difference in density and Young’s modulus of Au and
The frequency response of the cantilevers is monitored
Fe0.7Ga0.3兲 and biasing conditions, we conclude that the
with a phase lock loop while driving the cantilever using the
gauge factor of Fe0.7Ga0.3 thin films is only ⬃50% of the piezodisk. The frequency variation of one cantilever is
gauge factor of gold films of the similar thickness. shown in Fig. 3共a兲, whereby the transverse dc magnetic field
In a second experiment, we remotely actuate the cantile- 共H兲 is swept from ⫺270 to +270 Oe. As the cantilever is
vers using an external magnetic field, without the use of a now driven by the piezodisk with constant power, the vibra-
piezoactuator. Two cantilevers, one parallel to and the other tion amplitude remains nominally the same, i.e., there is no
perpendicular to the applied magnetic field, are excited using amplitude modulation due to applied magnetic fields. The
an ac magnetic field with rms amplitude of ⬃0.5 Oe. As frequency is observed to follow a hysteresis curve similar to
shown in Fig. 2, the fundamental mode of the cantilevers can that expected in magnetic susceptibility of ferromagnetic ma-
be driven using a magnetic field for both cantilevers, terials, with a saturation field of ⬃120 Oe. This saturation
whereas the piezoresistive cantilevers made from gold films behavior rules out the contribution of magnetic torque effect
fail to show such a response. Thus, this demonstrates that the as the torque 共L = M ⫻ B兲 is proportional to the biasing field
cantilevers with Fe0.7Ga0.3 films can be self-sensing and be and thus will not saturate at high magnetic fields. On the
actuated using remote magnetic fields. The difference in the contrary, magnetrostrictive stress along the length due
Downloaded 03 Jan 2011 to 128.36.45.227. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions
 013502-3 Bhaskaran et al. Appl. Phys. Lett. 98, 013502 共2011兲

oriented cantilevers 共cantilevers 2 and 3兲 whose resonances

closely match those in Fig. 2. The ac magnetic field is
coaligned with the dc bias field. The frequency variation for
both sets is shown in Figs. 3共b兲 and 3共c兲. As shown in Fig.
3共b兲, cantilever 2 that is driven using a magnetic field trans-
verse to its length exhibits a behavior similar to Fig. 3共a兲 of
cantilever 1, albeit noisier, due to the reduced vibration am-
plitudes. However, for cantilever 3 driven with longitudinal
magnetic fields, the frequency shows no saturation for higher
values of H 关Fig. 3共c兲兴. This indicates that the effect is
mostly a torque effect in this configuration. An analysis simi-
lar to the one described in Ref. 13 shows that, in this case,
because of the cantilever’s orientation, there exists a compo-
nent of the magnetic field in the direction perpendicular to
the cantilever plane, and the torque effect would modify the
frequency in the observed manner. Furthermore, as the effect
is more pronounced in this case, this gives a qualitative in-
dication that the torque effect is a greater effect in this con-
figuration. We expect that further device optimization, in par-
ticular stress compensation during growth process, could
alleviate the cantilever bending and thus allow us to fully
decouple the magnetostrictive effect and torque effect in our
device system.
The results here indicate that Fe0.7Ga0.3, a magnetostric-
tive material, can be deposited on surfaces of microelectro-
mechanical systems 共MEMS兲 and nanoelectromechanical
systems 共NEMS兲 devices as sensing and actuation elements.
Such a mechanical resonator can be actuated by a remote
magnetic field. At the same time, self-sensing is achieved by
exploiting the intrinsic piezoresistivity of Fe0.7Ga0.3. Al-
though these experiments suggest that the transduction is a
FIG. 3. 共a兲 Variation of resonant frequency of an Fe0.7Ga0.3 microcantilever combination of magnetostriction and torque effects, careful
共cantilever 1兲 driven with a piezoshaker and subject to a varying magnetic design of mechanical resonators can exploit only one effect
field 共H兲 along an in-plane direction perpendicular to the length of the or the other in future. The total thickness of the sensor-
cantilever 共inset兲, and frequency variation of magnetically actuated cantile- actuator layer is 100 nm, compared to several microns in
vers with 共b兲 transverse 共cantilever 2, directions indicated in inset兲 and 共c兲
parallel 共cantilever 3兲 fields. Arrows on data indicate direction in which dc
devices that use the ME effect, thus significantly shrinking
magnetic field H is swept. device dimensions. Such a combination of two effects in the
same material would find use in applications that require low
complexity, high robustness, and reduced dimensions.
to a transverse magnetic field can be written as13
S = −2K␭EFeGa共1 + ␯FeGa兲, where K is a constant that adjusts The authors acknowledge funding from DARPA/DSO’s
for nondirectionality of the sputtered nanocrystals, ␭ is the HUMS project under Grant No. FA8650-09-C-1-7944.
coefficient of magnetostriction, EFeGa is Young’s modulus,
and ␯FeGa is the Poisson ratio of Fe0.7Ga0.3. The effective 1
B. Rogers, D. York, N. Whisman, M. Jones, K. Murray, J. D. Adams, T.
spring constant for the bilayered cantilever can be expressed Sulchek, and S. C. Minne, Rev. Sci. Instrum. 73, 3242 共2002兲.
2
as keff = EeffGk, where Gk is the geometrical constant, which S. C. Masmanidis, R. B. Karabalin, I. De Vlaminck, G. Borghs, M. R.
Freeman, and M. L. Roukes, Science 317, 780 共2007兲.
depends on the cantilever geometry, and Eeff is the effective 3
S. C. Masmanidis, H. X. Tang, E. B. Myers, M. Li, K. De Greve, G.
Young’s modulus.12 The effective spring constant is a func- Vermeulen, W. Van Roy, and M. L. Roukes, Phys. Rev. Lett. 95, 187206
tion of Young’s modulus of the Fe0.7Ga0.3 film as well as that 共2005兲.
4
of the underlying silicon nitride. Given that the magnetostric- W. Han, S. M. Lindsay, and T. Jing, Appl. Phys. Lett. 69, 4111 共1996兲.
5
tion is only a property of Fe0.7Ga0.3, this implies that Eeff is J. P. Davis, D. Vick, D. C. Fortin, J. A. J. Burgess, W. K. Hiebert, and M.
R. Freeman, Appl. Phys. Lett. 96, 072513 共2010兲.
modulated only due to a magnetostrictive effect in the 6
A. Edelstein, J. Phys.: Condens. Matter 19, 165217 共2007兲.
Fe0.7Ga0.3 layer. The frequency ␻0 is then given by ␻0 7
P. Zhao, Z. Zhao, D. Hunter, R. Suchoski, C. Gao, S. Mathews, M. Wuttig,
= 共keff / m兲−1/2, where m is the cantilever modal mass. Thus, a and I. Takeuchi, Appl. Phys. Lett. 94, 243507 共2009兲.
8
change in EFeGa due to S would result in a change in Eeff and S. Dong, J. Zhai, F. Bai, J.-F. Li, and D. Viehland, Appl. Phys. Lett. 87,
062502 共2005兲.
corresponding variation in the cantilever frequency. Thus, it 9
H. Greve, E. Woltermann, H.-J. Quenzer, B. Wagner, and E. Quandt,
follows that our observed frequency change hysteresis pro- Appl. Phys. Lett. 96, 182501 共2010兲.
10
vides evidence that the effect is predominantly magnetostric- M. Li, H. X. Tang, and M. L. Roukes, Nat. Nanotechnol. 2, 114 共2007兲.
11
tive. M. Li, E. B. Myers, H. X. Tang, S. J. Aldridge, H. C. McCaig, J. J.
We repeat the measurements shown in Fig. 3共a兲, but re- Whiting, R. J. Simonson, N. S. Lewis, and M. L. Roukes, Nano Lett. 10,
3899 共2010兲.
configure our setup to drive the cantilevers remotely with an 12
R. Koch, A. K. Das, H. Yamaguchi, C. Pampuch, and A. Ney, J. Appl.
ac magnetic field 共as opposed to driving the cantilever using Phys. 96, 2773 共2004兲.
13
the piezo disk as in the previous case兲, on two orthogonally E. Klokholm, IEEE Trans. Magn. 12, 819 共1976兲.

Downloaded 03 Jan 2011 to 128.36.45.227. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions