International Journal of Non-Linear Mechanics 142 (2022) 103988

Contents lists available at ScienceDirect

International Journal of Non-Linear Mechanics

journal homepage: www.elsevier.com/locate/nlm

An analytical model for the response of a piezoresistive micro-cantilever to

surface stress: Preliminary considerations for biochemical sensor design
I.M. Garnica-Palafox a , S.M.A. Jiménez b ,∗, D. Díaz-Alonso a , D.A. Fernández-Benavides a ,∗,
W. Calleja-Arriaga c , J.M. Alvarado-Orozco a
a
Center for Engineering and Industrial Development, Querétaro, Av. Pie de la Cuesta 702, 76125, Santiago de Querétaro, Mexico
b
Cátedras CONACYT - Center for Engineering and Industrial Development Airport Branch, Carretera Estatal 200, Parque Aeroespacial,
Colón, 76270, Querétaro, Mexico
c CD-MEMS INAOE, National Institute for Astrophysics Optics and Electronics, Luis Erro 1, 72840 Santa Maria Tonantzintla, Puebla, Mexico

ARTICLE INFO ABSTRACT

Keywords: In this paper, we derive an analytical model to describe the response of longitudinal and transverse
Biochemical sensors piezoresistors embedded in micro-cantilever biochemical sensors. This model estimates the relative change
Microcantilever in resistance and sensitivity of the piezoresistive elements taking into consideration the biaxial stress profile
Modeling
induced by surface stress loadings, the dimensions of the piezoresistor, and its location within the cantilever.
Piezoresistive
To demonstrate its applicability and usefulness, we utilize our model to construct an analytical model for
Surface stress
a piezoresistor with a U-shaped configuration. The effect of variation in the piezoresistor and cantilever
dimensions, as well as the piezoresistor location and Poisson’s ratio, on the relative change of resistance and
sensitivity to the applied surface stress are examined. The analytical predictions show that low aspect ratio
micro-cantilevered plates (i.e. wide cantilevers) are better suited for piezoresistors subject to surface stresses.
Moreover, the analytical model results allow us to identify a set of preliminary piezoresistor dimensions to
increase the sensitivity of p-type U-shaped piezoresistors embedded in rectangular micro-cantilevered plates.
Limitations in the microfabrication techniques of piezoresistors are also discussed in the context of our model
predictions. The model here presented can be readily extended to describe the response and sensitivity of other
piezoresistor configurations such as a streamer.

1. INTRODUCTION methods require expensive and cumbersome instruments not suitable

for portability or for use in opaque environments [11]. Therefore, their
Micro-electro-mechanical systems (MEMS) based on the micro- use is restricted to laboratory settings only. A more effective approach
cantilevers technology offer a versatile and sensitive platform for the consists in embedding piezoresistive elements in the micro-cantilever
detection of a wide variety of biochemical species. Sensors for the to convert the deformation into an electrical signal. In this manner, the
detection of pesticides [1], explosives [2], nerve agents [3], bac- detection system is compact, portable, relatively low cost, suitable for
teria [4], antibodies [5], proteins [6], DNA [7], and RNA [8] at use in opaque environments, scalable, and easy to operate [3,11,12].
concentrations as low as 10 pM have been reported in the literature. Boron doped single-crystal silicon wafers are the preferred material
The basic principle of species detection is as follows. One face of the for biochemical sensors fabrication given their higher piezoresistive
cantilever is functionalized to selectively immobilize the target species coefficients and low noise effects in comparison to polysilicon [13].
(molecules). Interactions between immobilized neighboring molecules In an effort to guide sensors design, a number of research groups
(e.g. van der Waals, chemical bonding, hydrogen-bonding, electric have developed analytical models to describe the change in resistance
charge redistribution, etc.) give rise to local forces that overall result in piezoresistors with micro-cantilever deformation. Some of these
in a uniform biaxial state of stress on the functionalized face [9]. The models have been developed in the context of narrow and long micro-
difference between the states of stress prevailing in the functionalized cantilevers subject to point forces at their free end such as those used in
and non-functionalized faces results in deformation/bending of the atomic force microscopy (AFM) applications. For instance, Harley and
micro-cantilever. Given their high resolution, optical methods have Kenny [14] derived an expression to predict the force resolution of split
become one of the most well stablished techniques to measure de- piezoresistive cantilevers in a manner that resembles a U-shaped config-
formation (deflections as small as 0.1 nm) [10]. Nevertheless, optical uration. Their model takes into consideration the cantilever length, the

∗ Corresponding authors.
E-mail addresses: miguel.a.jimenez.zapata@gmail.com (S.M.A. Jiménez), david.fernandez@cidesi.edu.mx (D.A. Fernández-Benavides).

https://doi.org/10.1016/j.ijnonlinmec.2022.103988
Received 20 October 2021; Received in revised form 17 February 2022; Accepted 20 February 2022
Available online 25 February 2022
0020-7462/© 2022 Elsevier Ltd. All rights reserved.
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

depth of the doped layer, the doping concentration, and a uniaxial state 2. Micro-cantilevered plate biochemical sensor mechanics
of stress. No explicit consideration is given to the contribution of the
transverse segment to the overall resolution of the piezoresistor. Pruitt The present section describes the simplifying assumptions made to
and coworkers [15,16] build on the work of Harley and Kenny [14] model the deformation mechanics of the micro-cantilever biochemical
to construct an improved analytical model for estimating the piezore- sensor. Consider the cantilevered plate shown in Fig. 1. The length (𝐿),
sistive cantilever force sensitivity. This improved model takes explicit width (𝑏), and thickness (ℎ) are aligned with the 𝑥, 𝑦 & 𝑧 directions
consideration of the effect of variation in dopant concentration through respectively. The cantilevered plate consists of two layers. That is, a
the piezoresistor thickness. However, the contribution of transverse top infinitesimally thin layer and a thicker (∼1 μm) structural layer. The
segments to the overall piezoresistor response was not yet considered.
structural layer is isotropic in its plate directions. The top layer consists
A situation that results in errors of 5%–35% in the predicted sensitivity
of an array of immobilized molecules. In this manner, the structural
with respect experimental measurements [17]. These models consider
layer is functionalized to selectively immobilize target molecules. For
uniaxial states of stress only so that they are not applicable to the
practical purposes, the top layer is much thinner than the structural
context of biochemical sensing where significant transverse stresses
are present [18]. In this score, Yang and Yin [19] pointed out that layer, and the cantilevered plate thickness is considered to equal that
a 2-D model for the response of piezoresistors is needed to establish of the structural layer. We now assume, as have others [19,24,26],
that interactions among molecules result in a state of homogeneous
connections between the sensitivity, the dimensions, and the surface ( )
stress of a micro-cantilever. The surface stress in micro-cantilevers is equibiaxial stress 𝜎𝑡 throughout the top functionalized layer. The
frequently related to changes in the piezoresistor resistance through action of this state of stress results in bending of the cantilevered plate
Stoney’s equation [3,20–23]. However, this equation is only accurate as schematically indicated in Fig. 1a. The bending direction (positive
for free standing cantilevered plates, it does not describe the correct or negative 𝑧 direction) is controlled by the sign of the equibiaxial
state of deformation in the vicinity of the cantilever support (region stress (𝜎𝑡 ). Upon exposure to the presence of target molecules, molecule
where piezoresistor elements are usually located) [24,25]. Rasmussen immobilization and molecule interaction processes taking place at the
et al. [26] reported an analytical model to calculate relative change functionalized layer will result in an increase in the stress level 𝜎𝑡 and
in resistance and the optimum length of an infinitesimally thin silicon further bending of the cantilevered plate [19,24,26]. In practice, the
piezoresistor. However, their predictions are not accurate in the vicinity immobilization processes at molecules scales may not occur in a ho-
of the cantilever fixed end due to uniaxial state of strain assumption mogeneous manner throughout the functionalized layers. Nevertheless,
made during model construction. Nevertheless, the predictions of their we assume that at scales much larger than a molecule characteristic
model appear in good agreement with finite element simulations in length and much smaller than the in-plane dimensions of the can-
cantilever regions sufficiently away from the fixed end. Analytical mod- tilevered plate, the immobilization and molecule interaction processes
els based on maximum or average stress criteria have been reported
yield a homogeneous state of stress. This appears to be a sensible
to estimate the sensitivity of piezoresistors to surface stress [21,27–
working assumption given that surface functionalization processes such
30]. A shortcoming of these models is that no explicit consideration
as monolayer mixed-self-assembly yield spacing in the range of 200 to
is given to the finite character of the piezoresistor dimensions or their
300 nm for some molecules of interest [36].
orientation with respect to the cantilever directions. Moreover, the
predictions of these models are restricted to straight piezoresistors With the simplifying mechanical considerations described above, we
only. Given the limitations of the available analytical models, the now write the statement of the principle of virtual work [37,38] for
design and optimization of micro-cantilever based biochemical sen- the cantilevered plate as follows
sors has been conducted through finite element analysis [18,21,29–
𝜎𝑖𝑗 𝛿𝜀𝑗𝑖 𝑑𝑉𝑠 + 𝜎𝑖𝑗 𝛿𝜀𝑗𝑖 𝑑𝑉𝑡 + 𝜎𝑖𝑗 𝛿𝜀𝑗𝑖 𝑑𝑉𝑏 = 0. (1)
32]. Notwithstanding their numerical accuracy and more detailed ge- ∫ ∫ ∫
ometric descriptions, finite element models require significantly more
Here 𝛿𝜀𝑗𝑖 &𝜎𝑖𝑗 are the virtual strain and stress tensors in the cantilevered
preparation time, and computational power than analytical models. In
plate. Note that the principle of virtual work statement considers three
addition, finite element packages with capacities to model piezoresistor ( )
contributions. Namely, the internal virtual work in the top 𝑉𝑡 and
elements are not available to everyone. More accurate analytical mod- ( )
bottom 𝑉𝑏 functionalized layers, and the internal virtual work in the
els not only will assist/expedite finite element method simulations but ( )
thicker (∼1 μm) structural layer 𝑉𝑠 . The bottom functionalized layer
also, they will increase our understanding of biochemical sensors. ( )
To the best of the author’s knowledge, there are no analytical 𝑉𝑏 is introduced here to permit the cantilevered plate to bend under
models reported in the literature specifically addressing the response the action of a mismatch between the states of stress prevailing at the
of piezoresistors embedded in micro-cantilevers for biochemical sensor top and bottom layers. We assume that the thickness of the bottom layer
applications. The objective of this paper is to construct an analytical (ℎ𝑏 ) is uniform and of the same order of magnitude as that of the top
model that describes the response of micro-cantilevers with embedded layer (ℎ𝑡 ). Nevertheless, we permit ℎ𝑏 to be different from ℎ𝑡 . If we now
piezoresistors to surface stress. We first describe the mechanics of a assume homogeneous equibiaxial states of stress in the top and bottom
micro-cantilevered plate subject to surface stress. Thereafter, we utilize layers, 𝜎𝑡 &𝜎𝑏 respectively, Eq. (1) becomes
the stress–surface stress relations developed by Tamayo et al. [24] to ( ) ( )
describe the spatial distribution of biaxial stress in the micro-cantilever. − 𝜎𝑡 ℎ 𝑡 𝛿𝜀𝑥 + 𝛿𝜀𝑦 𝑑𝑆𝑡 − 𝜎𝑏 ℎ𝑏 𝛿𝜀𝑥 + 𝛿𝜀𝑦 𝑑𝑆𝑏
∫ ∫
Next, we construct analytical expressions for the relative change in ( )
resistance for prismatic piezoresistors aligned with the longitudinal = 𝜎𝑥 𝛿𝜀𝑥 + 𝜎𝑦 𝛿𝜀𝑦 𝑑𝑉𝑠 , (2)
∫
and transverse micro-cantilever directions. These models explicitly take
into consideration the dimensions and the spatial variation of stress in where 𝑑𝑆𝑡 &𝑑𝑆𝑏 are surface differential elements on the top and bottom
the piezoresistors. Afterwards, we combine the models for longitudinal layers. Specialization of Eq. (2) for a Kirchhoff cantilevered plate with
and transverse piezoresistors to construct an analytical model for a coinciding neutral and middle planes yields
U-shaped piezoresistor configuration. We present predictions for the ( ) ( )
relative change in resistance and sensitivity of the U-shaped piezoresis- 𝛥𝜎𝑠 𝛿𝜀𝑥 + 𝛿𝜀𝑦 𝑑𝑆 = 𝜎𝑥 𝛿𝜀𝑥 + 𝜎𝑦 𝛿𝜀𝑦 𝑑𝑉𝑠 , (3)
∫ ∫
tor. Finally, we discuss some preliminary considerations for the design
of biochemical sensors. The results presented in this work support The principle of virtual work in the form of Eq. (3) clearly shows
( )
the preliminary design of a multi-spectral biosensor for the detec- that the stress difference −𝜎𝑡 ℎ𝑡 + 𝜎𝑏 ℎ𝑏 between the top and bottom
tion/quantification of pesticides such as carbaryl, methiocarb, diazinon, layers, denoted by 𝛥𝜎𝑠 , is the loading parameter controlling the stress
fenitrothion, thiabendazole, and the TCP biomarker in water, urine, and deformation states in the micro-cantilevered plate. We note that
fruit juice, and honey samples. All of these pesticides are well known residual stresses usually present in micro-cantilevers [39,40] are ac-
for their toxicity and potentially carcinogen effect in humans [33–35]. counted for in an implicit manner by the stress difference (𝛥𝜎𝑠 ). In

2
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

Fig. 1. Schematic diagram of a piezoresistive micro-cantilevered plate with an embedded U-shaped piezoresistor. (a) Geometry of the cantilevered plate subjected to an equibiaxial
surface stress (𝜎𝑡 ) on its top face. (b) Side view diagram of the piezoresistive micro-cantilever. The distance from the mid-plane of the piezoresistor to the neutral plane of the plate
is denoted as 𝑧0 . (c) Geometry of the longitudinal and transverse piezoresistive elements. (d) Schematic diagram of the stacked thin resistive layers that integrate the longitudinal
(𝑅L , layers) and transverse (𝑅T , layers) piezoresistive segments.

this fashion, 𝛥𝜎𝑠 is composed by the contributions of the residual We note that, the non-linearity of the curvatures shown in Eqs. (4)
stresses and that of the stresses arising from the interactions between arises as a consequence of the boundary conditions imposed to the
the immobilized molecules. 𝑑𝑆 in Eq. (3) is the area of a differential micro-cantilever. No large strain conditions, non-linear material behav-
surface element on the cantilevered plate neutral plane. Single-crystal ior or permanent deflection are implied by Eqs. (4).
silicon wafers are the preferred material for biochemical MEM sensors. The stresses 𝜎𝑥 &𝜎𝑦 in the cantilevered plate are related to the
Experimental investigations report no size effects on the mechanical curvatures 𝑘𝑥 &𝑘𝑦 as follows
properties and elastic behavior of single-crystal silicon nano- (charac- ( ) 𝐸𝜅 6𝛥𝜎𝑠
𝐸
teristic length > 150 nm), micro-, and millimeter scale beams [41–43]. 𝜎𝑥 = − 𝑘𝑥 + 𝑣𝑘𝑦 𝑧 = − 𝑆𝑡 𝑧 = − 𝑧
1 − 𝑣2 1−𝜈 ℎ2
This observation is attributed to the silicon crystal structure controlling ( ) 𝐸𝜅 [ 𝑥]
𝐸
material properties [42,44]. In addition, p- and n- types single-crystal 𝜎𝑦 = − 𝑣𝑘𝑥 + 𝑘𝑦 𝑧 = − 𝑆𝑡 1 − (1 − 𝜈) 𝑒−𝛼 𝑏 𝑧 (6)
1 − 𝑣2 1−𝜈
silicon are reported to exhibit linear elastic, and linear piezoresistive 6𝛥𝜎𝑠 [ 𝑥]

behaviors, at strain levels up to 0.1% (174 MPa stress) in micro- =− 𝑧 1 − (1 − 𝜈) 𝑒−𝛼 𝑏

ℎ2
cantilever bending [45] and nano-fiber stretching tests [46]. With Examination of Eqs. (6) shows that 𝜎𝑥 &𝜎𝑦 are equal to zero on the
these elements at hand, hereafter, we assume that the micro-cantilever neutral plane, and vary linearly through the cantilevered plate thick-
exhibits size independent and linear elastic/piezoresistive behaviors. ness. Moreover, 𝜎𝑥 only varies through the thickness while 𝜎𝑦 decreases
Tamayo et al. [24] derived analytical solutions for the longitudinal exponentially with 𝑥. The ratio 𝛼∕𝑏 controls the extent of cantilevered
and transverse curvatures of a cantilever plate subject to the same plate deformation influenced by the fixed support at 𝑥 = 0. The larger
loading conditions shown in Fig. 1 as follows 𝛼∕𝑏, the shorter the cantilever region influenced by the fixed support.
( 𝑥) For 𝛼 > 0, cantilever regions located at least one width away from the
𝑘𝑥 (𝑥) = 𝜅𝑆𝑡 1 + 𝑣𝑒−𝛼 𝑏
( fixed support, i.e. 𝑥 = 𝑏, the state of stress and strain is equibiaxial
𝑥) (4)
𝑘𝑦 (𝑥) = 𝜅𝑆𝑡 1 − 𝑒−𝛼 𝑏 and homogeneous (Stoney region). Eqs. (6) are the framework for the
construction of the models that describe the response of piezoresistor
The coefficient 𝛼 in the exponential terms in Eqs. (4) is a function elements to surface stress in the sections that follow.
of Poisson’s ratio given by 𝛼 (𝜈) ≅ 2.2803 + 1.0056𝜈 − 0.1762𝜈 2 . For
materials with 0 ≤ 𝜈 ≤ 0.5, the 𝛼 parameter is in the range of 3. Piezoresistor elements embedded in a micro-cantilevered plate
2.2803 ≤ 𝛼 (𝜈) ≤ 2.7392. 𝑏 is the cantilever width, and 𝜅𝑆𝑡 is the so
called Stoney curvature corresponding to the curvature of an isotropic, Consider two prismatic piezoresistor elements with uniform rectan-
free standing, cantilevered plate subject to an equibiaxial state of stress gular cross sections embedded in a micro-cantilevered plate as shown
in its top surface. The Stoney curvature 𝜅𝑆𝑡 is related to the surface in Fig. 1c. Hereafter, we will refer to these piezoresistor elements as
stress 𝛥𝜎𝑠 , plate thickness ℎ, plate Young’s modulus and Poisson’s ratio longitudinal and traverse in accordance with the alignment of their
(𝐸, 𝜈) through Eq. (5) [20,23]. axis with respect to the cantilevered plate 𝑥&𝑦 directions, respectively.
6 (1 − 𝜈) Moreover, the geometry and location of these piezoresistive elements
𝜅𝑆𝑡 = 𝛥𝜎𝑠 (5) are representative of piezoresistors doped in silicon monocrystalline
𝐸ℎ2

3
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

wafers [13]. That is, the piezoresistors lie below the surface of the where 𝑅𝑜,𝑝𝑟 = 𝜌0 𝑙𝑝𝑟 ∕𝑏𝑝𝑟 ℎ𝑝𝑟 is the resistance of the longitudinal piezore-
cantilevered plate, and their side faces are parallel to the cantilevered sistor in its unstressed-undeformed configuration, and 𝜆̃𝐿 is a dimen-
plate. We assume that the piezoresistors are uniformly doped along sionless loading parameter given by
their length 𝑙𝑝𝑟 , width 𝑏𝑝𝑟 , and thickness ℎ𝑝𝑟 . In this manner, the resistiv-
𝜆̃𝐿 = 𝑎̃𝐿 𝛥𝜎̃ 𝑠
ity 𝜌𝑜 , and piezoresistive coefficients, of each piezoresistor are constant
( )
and uniform over its volume. The distances between the piezoresistor 𝛥𝜎̃ 𝑠 = 𝛥𝜎𝑠 𝜋𝐿 + 𝜋𝑇 ∕ℎ
axis and the cantilevered plate neutral plane, and the 𝑦𝑧&𝑧𝑥 planes, [ ( )( )−1 ] (13)
̃
are denoted by 𝑧𝑜 , 𝑦𝑜 &𝑥𝑜 , respectively. Note that, in the discussions 𝑎̃𝐿 = 6 1 + (1 − 𝜈) 𝜋̃𝑇 𝑒−𝛼𝑙𝑝𝑟 − 1 𝛼 𝑙̃𝑝𝑟
that follow, we have permitted the axis of the piezoresistor elements to 𝜋𝑇
( ) 𝜋̃𝑇 =
lie above or below the neutral plane −ℎ∕2 < 𝑧𝑜 < ℎ∕2 . 𝜋𝐿 + 𝜋𝑇
We model each piezoresistive element as a composite made from
with 𝑧̃ 𝑜 , 𝑙̃𝑝𝑟 &ℎ̃ 𝑝𝑟 defining the normalized through thickness location,
infinitesimally thin and homogeneous resistive layers 𝑙𝑝𝑟 × 𝑏𝑝𝑟 × 𝑑𝑧
length and thickness, respectively, of the longitudinal piezoresistor
stacked on top each other (Fig. 1d). Given the simplifying assumptions
element as follows
above, the resistance of all layers is the same in their undeformed state.
Upon loading of the micro-cantilever, the resistance of each layer is ℎ𝑝𝑟
ℎ̃ 𝑝𝑟 =
given by ℎ
( ) 𝑧𝑜
𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 𝑧̃ 𝑜 = (14)
𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 = 𝑅𝑜,𝑙𝑎𝑦𝑒𝑟 1 + (7) ℎ
𝑅𝑜,𝑙𝑎𝑦𝑒𝑟
𝑙𝑝𝑟
𝜌 𝑙 𝑙̃𝑝𝑟 =
where 𝑅𝑜,𝑙𝑎𝑦𝑒𝑟 = 𝑏 0 𝑑𝑧 𝑝𝑟
is the resistance value in the undeformed state, 𝑏
𝑝𝑟
and 𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 is a stress induced change in resistance. The total re- With Eqs. (8) & (12) at hand, we write the change in resistance for
sistance of the piezoresistor element is given by the contribution in the longitudinal piezoresistor as follows
parallel of all its infinitesimally thin layer components (Eq. (7)). That 𝛥𝑅𝑝𝑟 𝑅𝑝𝑟 𝜆̃𝐿 ℎ̃ 𝑝𝑟
is, = −1= [ ( )] −1 (15)
𝑅𝑜,𝑝𝑟 𝑅𝑜,𝑝𝑟 1−𝜆̃𝐿 𝑧̃ 𝑜 −ℎ̃ 𝑝𝑟 ∕2
1 ln ( )
𝑅𝑝𝑟 = (8) 1−𝜆̃𝐿 𝑧̃ 𝑜 +ℎ̃ 𝑝𝑟 ∕2
𝑏𝑝𝑟 𝑧 +ℎ ∕2 𝑑𝑧
𝜌0 𝑙𝑝𝑟
∫𝑧 𝑜−ℎ 𝑝𝑟∕2 𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 Notwithstanding its non-linear character, insight regarding the re-
𝑜 𝑝𝑟 1+ 𝑅
𝑜,𝑙𝑎𝑦𝑒𝑟
sponse of the longitudinal piezoresistor can be gained from a close
The relative change in resistivity 𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 ∕𝑅𝑜,𝑙𝑎𝑦𝑒𝑟 is a function of the inspection of Eq. (15). For instance, in the limit where the loading
state of stress prevailing in the micro-cantilever, and the orientation parameter approaches zero (𝜆̃𝐿 → 0), Eq. (15) shows that the change
and dimensions of the piezoresistor element. We will derive 𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 in resistance in the piezoresistor is also equal to zero, 𝛥𝑅𝑝𝑟 = 0. A
( )
expressions for longitudinal and transverse piezoresistor elements in similar analysis shows that for thin piezoresistor conditions ℎ̃ 𝑝𝑟 → 0 ,
the subsections that follow. 𝛥𝑅𝑝𝑟
𝑅𝑜,𝑝𝑟
= −𝜆̃𝐿 𝑧̃ 𝑜 varies linearly with the loading parameter and the
piezoresistor distance from the neutral plane. This last result sug-
3.1. Longitudinal piezoresistive elements gests that for a fixed value of the loading parameter, and for thin
piezoresistor conditions, placement of the piezoresistor element at the
For conditions where an electric current flows through the axis of top or bottom cantilever surfaces maximizes the change in resistivity.
a longitudinal piezoresistive element, and for the loading conditions 𝛥𝑅 ( )
The analysis of 𝑅 𝑝𝑟 for thick ℎ̃ 𝑝𝑟 = 1&𝑧̃ 𝑜 = 0 and for moderate
described in Section 2, the change in resistivity of any section of a ( 𝑜,𝑝𝑟 )
thickness 0 < ℎ̃ 𝑝𝑟 < 1& − 0.5 < 𝑧̃ 𝑜 < 0.5 piezoresistor conditions re-
piezoresistor layer of dimensions 𝑑𝑥 × 𝑏𝑝𝑟 × 𝑑𝑧 with respect to its value
quire numerical computations that can only be conducted once the
in the undeformed configuration, 𝜌𝑜 , is given by
material properties and loading magnitude for the cantilevered plate
𝛥𝜌𝑥 are defined. Therefore, these conditions will not be addressed yet in
= 𝜎𝑥 𝜋𝐿 + 𝜎𝑦 𝜋𝑇 , (9)
𝜌𝑜 this section.
where 𝜋𝐿 &𝜋𝑇 are the longitudinal and transverse piezoresistive coeffi-
cients respectively [13,47,48]. Substitution of Eqs. (6) in Eq. (9) yields 3.2. Transverse piezoresistive elements

( )[ ] For conditions where an electric current flows through the axis

𝛥𝜌𝑥 𝛥𝜎𝑠 𝜋𝐿 + 𝜋𝑇 𝜋𝑇 𝑥
= −6𝑧 1 − (1 − 𝜈) 𝑒−𝛼 𝑏 , (10) of a transverse piezoresistive element, and for the loading conditions
𝜌𝑜 ℎ2 𝜋𝐿 + 𝜋𝑇
described in Section 2, the change in resistivity of any differential
Given Eq. (10), we now write the change in longitudinal resistance element of dimensions 𝑑𝑥 × 𝑙𝑝𝑟 × 𝑑𝑧 with respect to its value in the
𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 as follows undeformed configuration, 𝜌𝑜 , is given by [47,48]
( )
𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 𝛥𝜎𝑠 𝜋𝐿 + 𝜋𝑇 𝑙𝑝𝑟 [ 𝜋𝑇 𝑥
] 𝛥𝜌𝑦
= −6𝑧 1 − (1 − 𝜈) 𝑒−𝛼 𝑏 𝑑𝑥 = 𝜎𝑥 𝜋𝑇 + 𝜎𝑦 𝜋𝐿 (16)
𝑅𝑜,𝑙𝑎𝑦𝑒𝑟 𝑙𝑝𝑟 ℎ2 ∫0 𝜋𝐿 + 𝜋𝑇 𝜌𝑜
( )( )
𝛥𝜎𝑠 𝜋𝐿 + 𝜋𝑇 𝛼𝑙𝑝𝑟 −1 After substitution of Eqs. (6) & 𝛥𝜎̃ 𝑠 from Eq. (13), Eq. (16) becomes
= −6𝑧 [
ℎ2 𝑏 𝛥𝜌𝑦 𝑧 𝑥]
[ ( )] = −6𝛥𝜎̃ 𝑠 1 − (1 − 𝜈) 𝜋̃𝐿 𝑒−𝛼 𝑏
𝛼𝑙𝑝𝑟 𝜋𝑇 𝑙𝑝𝑟
𝜌𝑜 ℎ
× + (1 − 𝜈) 𝑒−𝛼 𝑏 − 1 (11) (17)
𝑏 𝜋𝐿 + 𝜋𝑇 𝜋𝐿
𝜋̃𝐿 =
We now substitute Eq. (11) into Eq. (8) to write the piezoresistor 𝜋𝐿 + 𝜋𝑇
resistance as Eq. (17) shows that the change in resistivity in the transverse
⎛ ⎞
−1 piezoresistive element varies along its width and thickness, as indicated
⎜ 𝑏𝑝𝑟 𝑧𝑜 +ℎ𝑝𝑟 ∕2
𝑑𝑧 ⎟ 𝜆̃𝐿 ℎ̃ 𝑝𝑟 𝑅𝑜,𝑝𝑟 by its dependence of the 𝑥&𝑧 coordinates, respectively. The resistance
𝑅𝑝𝑟 =⎜ ⎟ = ( ( )) (12) of the prism in its undeformed state is given by 𝑅𝑜,𝑝𝑟𝑖𝑠𝑚 = 𝜌𝑜 𝑙𝑝𝑟 ∕𝑑𝑥𝑑𝑧.
⎜ 𝜌 0 𝑙 ∫𝑧 −ℎ ∕2
𝛥𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟
⎟ 1−𝜆̃𝐿 𝑧̃ 𝑜 −ℎ̃ 𝑝𝑟 ∕2
𝑝𝑟 𝑜 𝑝𝑟 1+ 𝑅 ln ( )
⎝ 𝑜,𝑙𝑎𝑦𝑒𝑟 ⎠ 1−𝜆̃𝐿 𝑧̃ 𝑜 +ℎ̃ 𝑝𝑟 ∕2 We now utilize Eq. (17) to compute the resistance of a piezoresistor

4
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

( )
layer as the contributions in parallel of the differential layer elements becomes sufficiently large 𝑙̃𝑝𝑟 ≫ 1 . Eq. (21) on the other hand shows
follows that the magnitude of 𝑎̃𝑇 is controlled by the piezoresistor’s width
( ) ( )
𝜌𝑜 𝑙𝑝𝑟 𝑏̃ 𝑝𝑟 and distance 𝑥̃ 𝑜 from the cantilevered plate fixed end. In this
𝑑𝑧 manner, 𝑎̃𝑇 increases in an exponential manner as 𝑥̃ 𝑜 approaches zero,
𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 = 𝑏𝑝𝑟
𝑥𝑜 + 2 𝑑𝑥 and as 𝑏̃ 𝑝𝑟 approaches 1. Given the nonlinear nature of Eqs. (15) & (21),
∫ 𝑏𝑝𝑟
( 𝛥𝜌𝑦 )
𝑥𝑜 − 2 1+ 𝜌 the impact of increasing the loading intensity through 𝑎̃𝐿 &𝑎̃𝑇 on the
(
𝑜
) 𝛥𝑅
magnitude of 𝑅 𝑝𝑟 can only be established once numerical values for
1 − 6𝛥𝜎̃ 𝑠 ℎ𝑧 𝑅𝑜,𝑙𝑎𝑦𝑒𝑟 𝑜,𝑝𝑟
= (18) the cantilevered plate parameters are proposed.
( )
⎧ ⎡ −𝛼 𝑥̃ 𝑜 +
𝑏̃ 𝑝𝑟 ⎤ ⎫
⎪ 1−6𝛥𝜎̃ 𝑠 ℎ𝑧 ⎢1−(1−𝜈)𝜋̃𝐿 𝑒 2 ⎥⎪
( )−1 ⎪ ⎢ ⎥⎪ 4. U-shaped piezoresistor
̃ ⎣ ⎦
1 + 𝛼 𝑏𝑝𝑟 ln ⎨ (
𝑏̃ 𝑝𝑟 ) ⎬
⎡ ⎤
⎪ 1−6𝛥𝜎̃ 𝑧 ⎢1−(1−𝜈)𝜋̃ 𝑒 −𝛼 𝑥̃ 𝑜 −
2 ⎥⎪
⎪ 𝑠 ℎ ⎢ 𝐿 ⎥⎪
The Eqs. (15) & (21) derived in the previous subsections are the
⎩ ⎣ ⎦⎭
basis to construct analytical models for the response of a variety of
Here 𝑅𝑜,𝑙𝑎𝑦𝑒𝑟 = 𝜌𝑜 𝑙𝑝𝑟 ∕𝑏𝑝𝑟 𝑑𝑧 is the resistance of the layer in its unde- piezoresistor configurations to surface stress. From single longitudinal
formed configuration, and 𝑏̃ 𝑝𝑟 = 𝑏𝑝𝑟 ∕𝑏 is the width of the piezoresistor and transverse elements to U and streamer piezoresistor configurations.
(layer) normalized by the cantilevered plate width. The resistance of The present section discusses the analytical model for a U-shaped
the transverse piezoresistor is computed as the contribution in parallel piezoresistor.
of all layers through its thickness as indicated in Eq. (8). To the best A piezoresistor in a U configuration is modeled as consisting of
of the authors’ knowledge, there is not a closed form solution for the one transverse and two identical longitudinal elements as shown in
integral in the denominator of (8) when 𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 is given by Eq. (18). Fig. 1a and Fig. 1c. The dimensions of the longitudinal and transverse
In an effort to obtain an analytical expression for the resistance of elements are 𝑙̃𝑝𝑟𝐿 × 𝑏̃ 𝑝𝑟𝐿 × ℎ̃ 𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇 × 𝑏̃ 𝑝𝑟𝑇 × ℎ̃ 𝑝𝑟𝑇 , respectively with
the transverse piezoresistor, we construct a model for 𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 alterna- 𝑏̃ 𝑝𝑟𝑇 = 𝑏̃ 𝑝𝑟𝐿 = 𝑏̃ 𝑝𝑟 &ℎ̃ 𝑝𝑟𝑇 = ℎ̃ 𝑝𝑟𝐿 = ℎ̃ 𝑝𝑟 . The axes of all three piezoresistor
tive to that given in (18). First, we assume that the change in resistivity elements are located at a distance 𝑧̃ 𝑜 with respect to the neutral plane
in the piezoresistor (and its layers) is uniform along its width, and given of the cantilevered plate. The location of the transverse element axis
𝑏 𝑏
by the average value of Eq. (17) in the interval 𝑥𝑜 − 2𝑝𝑟 ≤ 𝑥 ≤ 𝑥𝑜 + 2𝑝𝑟 . with respect to the cantilever fixed end is given by the length of the
That is longitudinal piezoresistive elements. That is 𝑥̃ 𝑜 = 𝑙̃𝑝𝑟𝐿 . Moreover, the
𝑏𝑝𝑟 axes of the longitudinal elements are placed at equal and opposite
𝛥𝜌𝑦 𝑥𝑜 + 2 𝛥𝜌 ( )
1 𝑦 distances 𝑦̃𝑜 = ±𝑙̃𝑝𝑟𝑇 ∕2 from the cantilevered plate longitudinal plane.
= 𝑑𝑥
𝜌𝑜 𝑏𝑝𝑟 ∫𝑥𝑜 − 𝑏𝑝𝑟 𝜌𝑜 Given these considerations, we now write the change in resistance of a
[2 ( )]
𝑧 ( )−1 −𝛼 𝑥̃ 𝑏̃ 𝑝𝑟 𝑏̃ 𝑝𝑟 U-shaped piezoresistor as follows
= −6𝛥𝜎̃ 𝑠 1 + (1 − 𝜈) 𝜋̃𝐿 𝛼 𝑏̃ 𝑝𝑟 𝑒 𝑜 𝑒−𝛼 2 − 𝑒𝛼 2 (19)
ℎ 𝛥𝑅𝑈 𝑙̃𝑝𝑟𝐿 𝛥𝑅𝑝𝑟𝐿 𝑙̃𝑝𝑟𝑇 𝛥𝑅𝑝𝑟𝑇
=2 + (23)
where 𝑥̃ 𝑜 = 𝑥𝑜 ∕𝑏. It is not difficult to show that in the limit 𝑏̃ 𝑝𝑟 → 0, Eq. 𝑅𝑜,𝑈 𝑙̃𝑈 𝑅𝑜,𝑝𝑟𝐿 𝑙̃𝑈 𝑅𝑜,𝑝𝑟𝑇
(19) is equal to Eq. (17) evaluated at 𝑥 = 𝑥𝑜 . Combination of Eqs. (18)
where 𝑙̃𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇 are defined in Eqs. (14), and 𝑙̃𝑈 = 2𝑙̃𝑝𝑟𝐿 + 𝑙̃𝑝𝑟𝑇 is the
& (19) yields an expression for 𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 as follows
dimensionless length of the U piezoresistor. 𝑅𝑜,𝑈 = 𝜌0 𝑙𝑈 ∕𝑏𝑝𝑟 ℎ𝑝𝑟 is the
( )
𝛥𝜌𝑦 resistance of the U-shaped piezoresistor in its undeformed configura-
𝑅𝑖𝑡ℎ,𝑙𝑎𝑦𝑒𝑟 = 𝑅𝑜,𝑙𝑎𝑦𝑒𝑟 1 + (20) 𝛥𝑅 𝛥𝑅
𝜌𝑜 tion. 𝑅 𝑝𝑟𝐿 & 𝑅 𝑝𝑟𝑇 are the changes in the longitudinal and transverse
𝑜,𝑝𝑟𝐿 𝑜,𝑝𝑟𝑇
piezoresistors given by Eqs. (15) & (21), respectively.
Substitution of Eqs. (7) & (20) into Eq. (8) yields the following closed The sensitivity of the response of the U-shaped piezoresistor to
form for the change in resistance of the transverse piezoresistor variations in the loading and geometric parameters is derived from Eq.
𝛥𝑅𝑝𝑟 ̃ ̃
𝜆𝑇 ℎ𝑝𝑟 (23) as follows
= [ ( )] −1 (21) ( )
𝑅𝑜,𝑝𝑟 1−𝜆̃𝑇 𝑧̃ 𝑜 −ℎ̃ 𝑝𝑟 ∕2 𝛥𝑅𝑈
ln ( )
̃ ̃
1−𝜆𝑇 𝑧̃ 𝑜 +ℎ𝑝𝑟 ∕2 𝑆=𝑑 = 𝑆𝛥𝜎̃ 𝑠 𝑑𝛥𝜎̃ 𝑠 + 𝑆ℎ̃ 𝑝𝑟 𝑑 ℎ̃ 𝑝𝑟 + 𝑆𝑧̃0 𝑑 𝑧̃ 0
𝑅𝑜,𝑈
where 𝑅𝑜,𝑝𝑟 = 𝜌𝑜 𝑙𝑝𝑟 ∕𝑏𝑝𝑟 ℎ𝑝𝑟 and ( ) ̃ ( ) ̃
𝜕 𝛥𝑅𝑈 𝜕 𝜆𝐿 𝜕 𝛥𝑅𝑈 𝜕 𝜆𝑇
𝑆𝛥𝜎̃ 𝑠 = +
𝜆̃𝑇 = 𝛥𝜎̃ 𝑠 𝑎̃𝑇 𝜕 𝜆̃𝐿 𝑅𝑜,𝑈 𝜕𝛥𝜎̃ 𝑠 𝜕 𝜆̃𝑇 𝑅𝑜,𝑈 𝜕𝛥𝜎̃ 𝑠
[ ( ) ] ( ) ( )
𝑏̃ 𝑝𝑟 𝑏̃ 𝑝𝑟 ( )−1 𝑙̃𝑝𝑟𝐿 𝜕 𝛥𝑅𝑝𝑟𝐿 𝑙̃𝑝𝑟𝑇 𝜕 𝛥𝑅𝑝𝑟𝑇
𝑎̃𝑇 = 6 1 + (1 − 𝜈) 𝜋̃𝐿 𝑒−𝛼 𝑥̃ 𝑜 𝑒−𝛼 2 − 𝑒𝛼 2 𝛼 𝑏̃ 𝑝𝑟 = 2𝑎̃𝐿 + 𝑎̃𝑇
(22) 𝑙̃𝑈 𝜕 𝜆̃𝐿 𝑅𝑜,𝑝𝑟𝐿 𝑙̃𝑈 𝜕 𝜆̃𝑇 𝑅𝑜,𝑝𝑟𝑇
( ) ̃ ( ) ̃ ( )
ℎ𝑝𝑟 𝜕 𝛥𝑅𝑈 𝑙𝑝𝑟𝐿 𝜕 𝛥𝑅𝑝𝑟𝐿 𝑙𝑝𝑟𝑇 𝜕 𝛥𝑅𝑝𝑟𝑇
ℎ̃ 𝑝𝑟 = 𝑆ℎ̃ 𝑝𝑟 = =2 +
ℎ 𝜕 ℎ̃ 𝑝𝑟 𝑅𝑜,𝑈 𝑙̃𝑈 𝜕 ℎ̃ 𝑝𝑟 𝑅𝑜,𝑝𝑟𝐿 𝑙̃𝑈 𝜕 ℎ̃ 𝑝𝑟 𝑅𝑜,𝑝𝑟𝑇
In a similar manner to a longitudinal piezoresistor, the change ( ) ̃ ( ) ̃ ( )
𝜕 𝛥𝑅𝑈 𝑙𝑝𝑟𝐿 𝜕 𝛥𝑅𝑝𝑟𝐿 𝑙𝑝𝑟𝑇 𝜕 𝛥𝑅𝑝𝑟𝑇
in resistance in the transverse piezoresistor, i.e. Eq. (21), approaches 𝑆𝑧̃0 = =2 +
( ) 𝜕 𝑧̃ 𝑜 𝑅𝑜,𝑈 𝑙̃𝑈 𝜕 𝑧̃ 𝑜 𝑅𝑜,𝑝𝑟𝐿 𝑙̃𝑈 𝜕 𝑧̃ 𝑜 𝑅𝑜,𝑝𝑟𝑇
zero as the loading approaches zero 𝜆̃𝑇 → 0 . For thin piezoresistor
( ) 𝛥𝑅𝑝𝑟 (24)
conditions ℎ̃ 𝑝𝑟 → 0 , = −𝜆̃𝑇 𝑧̃ 𝑜 . 𝑅𝑜,𝑝𝑟
Given the assumption utilized in the construction of Eq. (19), and where 𝑆𝛥𝜎̃ 𝑠 , 𝑆ℎ̃ 𝑝𝑟 &𝑆𝑧̃𝑜 are the sensitivities to the applied stress, thick-
its similarity to Eq. (10), it is no surprise that Eqs. (15) & (21) have ness, and through thickness location of the piezoresistor with respect
the same analytical form. Nevertheless, the specific information re- to the cantilevered plate. The explicit forms for the sensitivities in Eq.
lated to the mechanics of each piezoresistor, in accordance with their (24) are given below
orientation and location in the cantilevered plate, is contained in the
𝜕𝛥𝜎̃ 𝑠 ( )
dimensionless 𝑎̃𝐿 &𝑎̃𝑇 parameters. A close inspection to Eqs. (15) & 𝑆𝛥𝜎𝑠 = 𝑆𝛥𝜎̃ 𝑠 = 𝜋𝐿 + 𝜋𝑇 𝑆𝛥𝜎̃ 𝑠 ∕ℎ
𝜕𝛥𝜎𝑠
(21) provides insight to strategies to increase the intensity of loading
parameters 𝜆̃𝐿 &𝜆̃𝑇 through judicious placement of the piezoresistors in ( ) ⎧ ( ) ⎫
𝜕 𝛥𝑅𝑝𝑟𝑖 1 ⎪ 𝑅𝑝𝑟𝑖 𝑅𝑝𝑟𝑖 2 1 ⎪
the cantilever. For instance, Eq. (15) shows that 𝑎̃𝐿 increases monoton- = ⎨ − ( ) ⎬
[ ] 𝜕 𝜆̃𝑖 𝑅𝑜,𝑝𝑟𝑖 𝜆̃𝑖 ⎪ 𝑅𝑜,𝑝𝑟𝑖 𝑅𝑜,𝑝𝑟𝑖
ically from 6 1 − (1 − 𝜈) 𝜋̃𝑇 towards 6 as the longitudinal piezoresistor 1 − 2𝑧̃ 𝑜 𝜆𝑖 + 𝜆𝑖 𝑧̃ 𝑜 − ℎ𝑝𝑟 ∕4 ⎪
̃ ̃ 2 2 ̃ 2
⎩ ⎭

5
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

Table 1 resistance for a family of U-shaped piezoresistors flushed with the can-
Elastic and piezoresistive properties for single-crystal silicon used in the analytical
tilever top surface and 𝑏̃ 𝑝𝑟 ranging from 0.05 to 0.3. Both, the exact and
computations. Values for resistivity (𝜌𝑜 ) and dopant concentration (𝑁𝐴 ) are also listed
for reference purposes only. In practice, 𝜌𝑜 &𝑁𝐴 depend on fabrication conditions and
average approximation to the transverse piezoresistor relative change
are determined after piezoresistor characterization is conducted. of resistance, are considered. The plot in Fig. 2a shows that the relative
Si (100) wafer, <110> crystal direction change in resistance varies linearly from –1.3 × 10−4 to 1.3 × 10−4 with
Young’s modulus [49,50] 𝐸 = 169 GPa
surface stress. This result indicates that, for a U-shaped piezoresistor
Poisson’s ratio [49,50] 𝑣 = 0.064 with a resistance value in its unstressed state of 𝑅𝑜,𝑈 = 5 kΩ [54], the
𝛼 (𝜈) from analytical model derived by Tamayo et al. [24] 𝛼 (𝜈) ≅ 2.34 change in resistance 𝛥𝑅𝑈 equals ±58.4 mΩ and ±642.4 mΩ for surface
P-type Si piezoresistor <110> stress levels of ∓1 mN∕m and ∓200 mN∕m, respectively. In addition,
Longitudinal piezoresistive coefficient [13] 𝜋𝐿 = 71.8 × 10−5 MPa−1 the plot in Fig. 2a also shows that there is no difference between the
Transverse piezoresistive coefficient [13] 𝜋𝑇 = −66.3 × 10−5 MPa−1 predictions yielded by the exact and average approximation to the
Resistivity [13] 𝜌𝑜 = 7.8 Ω cm transverse piezoresistor relative change of resistance. The plot in Fig. 2a
Dopant concentration [13] 𝑁𝐴 = 1.75 × 1015 cm−3
also shows that, for the mechanical properties chosen (𝑣 = 0.064), the
width 𝑏̃ 𝑝𝑟 ≤ 0.3 (𝛼 𝑏̃ 𝑝𝑟 ≤ 0.703) does not affect the relative change in
Table 2 resistance. However, a close inspection to Eq. (22) shows that the width
Normalized dimensions and properties for the longitudinal and transverse elements of 𝑏̃ 𝑝𝑟 will have a more significant in the relative change of resistance for
the U-shaped piezoresistor. conditions where 𝛼 𝑏̃ 𝑝𝑟 >1. Note that discrepancies between the change
𝑙̃𝑝𝑟 𝑏̃ 𝑝𝑟 ℎ̃ 𝑝𝑟 𝑥̃ 𝑜 𝑧̃ 𝑜 𝜋̃𝐿 𝜋̃𝑇 in resistance predictions yielded by the average and exact transverse
Longitudinal elements 0.3a
0.05–0.3 0.3
__
−0.35–0.35 13 −12 piezoresistor descriptions could arise as 𝛼 𝑏̃ 𝑝𝑟 >1. Fig. 2b shows the
Transverse element 0.2b c
influence of the through thickness location (𝑧̃ 𝑜 ) on the relative change
a 𝑙̃
𝑝𝑟 values in the range of 0.15 to 1.5 have been reported in the literature for of resistance for the piezoresistors considered in Fig. 2a. The plot in
longitudinal piezoresistors [23,27,31,32,51,52]. Fig. 2b shows that the absolute value of the slope of the relative
b̃
𝑙𝑝𝑟 values in the range of 0.1 to 1 have been reported in the literature for transverse change of resistance–surface stress curve increases from zero to 11.68
piezoresistors [19,31,32,52,53].
c
while the piezoresistor location is moved from the neutral axis towards
𝑥̃ 𝑜 is only relevant to the transverse piezoresistor. In the case of a U-shaped
piezoresistor 𝑥̃ 𝑜 is equal to the 𝑙̃𝑝𝑟 of the longitudinal piezoresistor.
the cantilever top/bottom surfaces. The relative change in resistance–
surface stress curves exhibit positive slope for 𝑧̃ 𝑜 < 0, negative slope
for 𝑧̃ 𝑜 > 0, and zero value for 𝑧̃ 𝑜 = 0. Fig. 2c shows that, for the
properties and dimensions considered, the relative change of resistance
⎡ ⎤ is not affected by the piezoresistor thickness ℎ̃ 𝑝𝑟 . Notwithstanding the
( ) ( )
𝜕 𝛥𝑅𝑝𝑟𝑖 ⎢ 𝑅𝑝𝑟𝑖 1 𝑅𝑝𝑟𝑖 2 1 − 𝑧̃ 𝑜 𝜆̃𝑖 ⎥ observations made from the results shown in Fig. 2a and Fig. 2b, it is
⎢ = − ( )⎥
𝜕 ℎ̃ 𝑝𝑟 𝑅𝑜,𝑝𝑟𝑖 ⎢ 𝑅𝑜,𝑝𝑟𝑖ℎ̃ 𝑝𝑟𝑅𝑜,𝑝𝑟𝑖 noted that the absolute change in resistance depends on 𝑏̃ 𝑝𝑟 &ℎ̃ 𝑝𝑟 given
1 − 2𝑧̃ 𝑜 𝜆̃𝑖 + 𝜆̃2𝑖 𝑧̃ 2𝑜 − ℎ̃ 2𝑝𝑟 ∕4 ⎥
⎣ ⎦ that the resistance of the piezoresistor is inversely proportional to the
( ) ( ) product 𝑏̃ 𝑝𝑟 ℎ̃ 𝑝𝑟 .
𝜕 𝛥𝑅𝑝𝑟𝑖 𝑅𝑝𝑟𝑖 2 𝜆̃𝑖
=− ( ) We note that the slopes of the curves shown in Fig. 2a and Fig. 2b
𝜕 𝑧̃ 𝑜 𝑅𝑜,𝑝𝑟𝑖 𝑅𝑜,𝑝𝑟𝑖 1 − 2𝑧̃ 𝜆̃ + 𝜆̃2 𝑧̃ 2 − ℎ̃ 2 ∕4
𝑜 𝑖 𝑖 𝑜 𝑝𝑟 correspond to the sensitivity 𝑆𝛥𝜎̃ 𝑠 , or alternatively to its counter-
(25) part 𝑆𝛥𝜎𝑠 . This sensitivity is not affected by the magnitude of the
stress level but increases as 𝑧̃ 𝑜 → 0.5. Our calculations show that
Here, the 𝑖 subscript indicates that the expression is valid for both, 𝑆𝛥𝜎̃ 𝑠 varies between ±16. A comparison between the Eqs. (24) sug-
longitudinal and transverse segments. 𝑆𝛥𝜎𝑠 is the dimensional version gests that the piezoresistor sensitivities scale as follows: 𝑆𝛥𝜎̃ 𝑠 ∕𝑆ℎ̃ 𝑝𝑟 ∼
of the sensitivity to the U-shaped piezoresistor to surface stress. Note ℎ̃ 𝑝𝑟 ∕𝛥𝜎̃ 𝑠 &𝑆𝛥𝜎̃ 𝑠 ∕𝑆𝑧̃𝑜 ∼ 1∕𝛥𝜎̃ 𝑠 . Our numerical calculations confirm the
that the sensitivity definition in Eq. (24) can be readily extended to scaling and show that 𝑆𝛥𝜎̃ 𝑠 ∕𝑆ℎ̃ 𝑝𝑟 ≈ 1010 &𝑆𝛥𝜎̃ 𝑠 ∕𝑆𝑧̃𝑜 ≈ 105 . With these
include the effects of variations in the piezoresistors lengths (𝑙̃𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇 ). results at hand, 𝑆𝛥𝜎𝑠 (𝑆𝛥𝜎̃ 𝑠 ) is the only sensitivity addressed in the
However, for the sake of brevity in this manuscript, we will not address discussions that follow. For the sake of ease in visualization and
the effects of variations 𝑙̃𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇 on sensitivity in the discussions that discussion of results, and given that the sensitivity sign changes with
follow. the sign of the applied surface stress, hereafter, the predictions for 𝑆𝛥𝜎𝑠
We consider now that the micro-cantilever is made from a (100) are presented in magnitude (positive value only).
monocrystalline silicon wafer, i.e. the most widely used wafer in MEMS Fig. 3 shows the predictions from Eq. (25) for the effect of variations
fabrication. With its length and width aligned with the ⟨110⟩ in-plane in the U-shaped piezoresistor geometric parameters on its sensitiv-
wafer directions (see Fig. 1a), the cantilever exhibits equivalent elastic ity 𝑆𝛥𝜎𝑠 . The plot in Fig. 3a shows the variation of sensitivity with
properties along the 𝑥&𝑦 directions [49]. The cantilever is doped with the length of the transverse and longitudinal piezoresistor elements,
boron on its top surface following a U-shaped pattern to form a p-type 𝑙̃𝑝𝑟𝑇 &𝑙̃𝑝𝑟𝐿 respectively. In constructing Fig. 3a, the parameters 𝑏̃ 𝑝𝑟 , ℎ̃ 𝑝𝑟 ,
piezoresistor. In this manner, the piezoresistor is effectively embedded 𝑧̃ 𝑜 , ℎ&𝛥𝜎𝑠 were held fixed at 0.1, 0.3, 0.35, 1 μm, and −200 mN∕m,
in the cantilever. The segments of the U-shaped piezoresistor are also respectively. The solid curves in Fig. 3a show that for all 𝑙̃𝑝𝑟𝑇 val-
aligned with the ⟨110⟩ wafer directions, and therefore exhibit the same ues considered, the sensitivity of the U-shaped piezoresistor increases
elastic properties as the cantilever. Table 1 lists the mechanical and in a monotonic manner with 𝑙̃𝑝𝑟𝐿 until it attains a maximum value.
piezoresistive properties of the doped and undoped (100) silicon wafer. Thereafter, the sensitivity decreases towards an asymptotic value of
In the discussions that follow we consider U-shaped piezoresistors 0.2 mm∕N. Fig. 3a also shows that the sensitivity increases at a faster
with normalized dimensions based on those of representative MEMS rate, and attains higher maximum values, as 𝑙̃𝑝𝑟𝑇 decreases from 1
piezoresistive biochemical sensors [18,27,31,32,51–53]. Table 2 lists towards zero. For a given 𝑙̃𝑝𝑟𝑇 value, the maximum sensitivity occurs
1∕2
the normalized dimensions and locations of the U-shaped piezoresistor when 𝑙̃𝑝𝑟𝐿 = +𝑙̃𝑝𝑟𝑇 as shown in the inset in Fig. 3a. The long and
elements. The magnitude of the surface stress is in the range −200 ≤ short dashed curves in Fig. 3a show the sensitivities of the longitudinal
𝛥𝜎𝑠 ≤ 200 mN∕m. In dimensionless form, this range is giving by and transverse piezoresistor elements respectively. An inspection of
−1.1 × 10−5 ≤ 𝛥𝜎̃ 𝑠 ≤ 1.1 × 10−5 for a 1 μm thick cantilever. Eq. (23), and of the short dashed curves in Fig. 3a, clearly shows that
[,belowfloat=21pt] Fig. 2 plots the predictions from Eq. (23) for the the contribution of the transverse element to the U-shaped piezoresistor
relative change in resistance with surface stress level for a variety of U- sensitivity is opposite to that of the longitudinal elements. The negative
shaped piezoresistors. The plot in Fig. 2a shows the relative change in effect of the transverse element on the sensitivity is reduced as 𝑙̃𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇

6
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

Fig. 2. Relative change in resistance with applied surface stress for a variety of U-shaped piezoresistors with 𝛼 𝑏̃ 𝑝𝑟 ≤ 0.702. (a) Influence of the piezoresistor width (𝑏̃ 𝑝𝑟 ) on the
relative change in resistance. (b) Influence of the location of the piezoresistor with respect to the cantilever neutral plane (𝑧̃ 𝑜 ). (c) Influence of the piezoresistor thickness (ℎ̃ 𝑝𝑟 ) for
𝑧̃ 𝑜 = 0.35..

Fig. 3. Influence of piezoresistor dimensions on the sensitivity of the piezoresistor to the applied surface stress. (a) Variation of the sensitivity with the length of the longitudinal
and transverse piezoresistive segments, 𝑙̃𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇 . Inset in the graph shows the recommended length for 𝑙̃𝑝𝑟𝐿 to get the maximum sensitivity if 𝑙̃𝑝𝑟𝑇 is fixed to a specific length. (b)
Effect of varying the piezoresistor thickness, ℎ̃ 𝑝𝑟 , for fixed cantilever thicknesses (ℎ𝑝𝑟 ).

7
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

Fig. 5. Effect of the Poisson’s ratio (𝑣) on the sensitivity of the piezoresistor to the
Fig. 4. Effect of variation in the cantilever dimensions (width (𝑏), and thickness (ℎ)) applied surface stress.
on the sensitivity of the piezoresistor to the applied surface stress.

the piezoresistor thickness may be more challenging depending on the

increase towards 1. However, increases in 𝑙̃𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇 also result in a doping technique available for fabrication. For instance, spin-on-dopant
reduction in the piezoresistor sensitivity. Fig. 3b shows the variation offers more limited depth control and junction uniformity at low dopant
in the sensitivity with cantilever ℎ and piezoresistor ℎ𝑝𝑟 thicknesses for concentrations (i.e. 7 × 1021 cm−3 ) than its diffusion or ion implantation
a U-shaped piezoresistor flushed with the cantilever top surface; and counterparts [56,57].
with values for 𝑙̃𝑝𝑟𝐿 , 𝑙̃𝑝𝑟𝑇 𝑏̃ 𝑝𝑟 &𝛥𝜎𝑠 fixed at 0.3, 0.2, 0.1 and −200 mN∕m, For the conditions assumed in the construction of Fig. 4, and with
respectively. The curves in Fig. 3b show that, for a fixed ℎ𝑝𝑟 value, thickness ℎ = 1 μm, the curves in Fig. 5 shows that the sensitivity of
the sensitivity first increases as ℎ decreases from ∞ to 2ℎ𝑝𝑟 (𝑖.𝑒. ℎ̃ 𝑝𝑟 the U-shaped piezoresistor decreases in a quasi-linear manner as the
increases from 0 to 0.5). For all the ℎ𝑝𝑟 values considered, the sensitivity Poisson’s ratio increases from zero to 0.5. All the U-shaped piezoresistor
reaches a maximum value at ℎ̃ 𝑝𝑟 = 0.5. Afterwards, the sensitivity sensitivity curves in Fig. 5 exhibit similar reduction rates with Poisson’s
decreases monotonically towards zero as ℎ decreases from 2ℎ𝑝𝑟 to ℎ𝑝𝑟 ratio (∼ −0.6, −0.58 and −0.5 mm∕N for 𝑏 values of 200, 100 and 50 μm,
(i.e. ℎ̃ 𝑝𝑟 increases from 0.5 to 1). Finally, the curves in Fig. 3b also respectively). In practice, a change in Poisson’s ratio is achieved by
show that for the conditions given, the sensitivity increases rapidly as changing the cantilever material or by using a wafer with a different
ℎ𝑝𝑟 decreases from 0.9 to 0.3 μm. This observation suggests that thin orientation to that chosen in this work. We note that, in clear contrast
piezoresistors close to the cantilever top/bottom surfaces, are more with the assumptions made to plot Fig. 5, a change in the Poisson’s
sensitive than those that are thick and cross the cantilever neutral ratio is likely accompanied by changes in the piezoresistive coefficients
plane. (𝜋𝐿 &𝜋𝑇 ) as well.
Fig. 4 shows the effect of variation in the cantilever dimensions 𝛥𝑅
Fig. 6 shows the relative change in resistance ( 𝑅 𝑈 ) for a variety
𝑜,𝑈
on the sensitivity of the piezoresistor to the applied surface stress. of U-shaped piezoresistors in terms of the loading parameter mismatch
The piezoresistor is flushed with the upper cantilever surface with (𝜆̃𝐿 − 𝜆̃𝑇 ) and the location with respect to the cantilever neutral plane
dimensions 𝑙𝑝𝑟𝐿 , 𝑙𝑝𝑟𝑇 , 𝑏𝑝𝑟 &ℎ𝑝𝑟 as well as the surface stress level 𝛥𝜎𝑠 , (𝑧̃ 𝑜 ). The plot in Fig. 6 is constructed for conditions where ℎ̃ 𝑝𝑟 , 𝑏̃ 𝑝𝑟 ,
fixed at 30 μm, 20 μm, 10 μm, 0.3 μm, and −200 mN∕m, respectively. On 𝑙̃𝑝𝑟𝐿 &𝑙̃𝑝𝑟𝑇 are held fix at 0.3, 0.1, 0.3 and 0.2, respectively. The contour
the other hand, the cantilever width and thickness are permitted to vary lines in Fig. 6 show that the relative change in resistances varies in
in the ranges of 40 ≤ 𝑏 ≤ 300 μm and 1 ≤ ℎ ≤ 3 μm. The curves in Fig. 4 𝛥𝑅
the range of −1.21 × 10−4 ≤ 𝑅 𝑈 ≤ 1.21 × 10−4 . In addition, Fig. 6
𝑜,𝑈
show that, for all thickness values considered, the sensitivity increases 𝛥𝑅𝑈
shows that the magnitude of increases as the piezoresistor moves
in a slow manner and approaches asymptotic values as the cantilever 𝑅𝑜,𝑈
width 𝑏 increases from 40 to 300 μm. No significant improvements in away from the neutral plane (𝑧̃ 𝑜 → ±0.35). Increasing the 𝜆̃𝐿 − 𝜆̃𝑇 also
𝛥𝑅 𝛥𝑅
sensitivity are achieved by extending 𝑏 beyond 150 μm. That is, for ℎ = improves the magnitude of 𝑅 𝑈 . The highest 𝑅 𝑈 values are obtained
𝑜,𝑈 𝑜,𝑈
1 μm, increasing the width from 40 to 150 μm increases the sensitivity for piezoresistors near the top or bottom surfaces, and for high values
in approximately 30%. Only a 5% increase in sensitivity is achieved by ̃ ̃
of 𝜆𝐿 − 𝜆𝑇 .
increasing the width beyond 150 μm. Note that the maximum sensitivity Strategies to increase the loading parameter mismatch (𝜆̃𝐿 − 𝜆̃𝑇 ),
𝛥𝑅
level achieved at 𝑏 ≥ 150 μm increases as ℎ decreases from 3 to 1 μm. and in this manner increase the magnitude of 𝑅 𝑈 , are identified from
𝑜,𝑈
In fact, Eq. (25) shows that the sensitivity varies with ℎ in an inversely the contour plot in Fig. 7. This plot is presented in terms of the surface
proportional manner. Practical considerations concerning the design of stress (𝛥𝜎̃ 𝑠 ) and longitudinal piezoresistor length (𝑙̃𝑝𝑟𝐿 ), for conditions
biochemical sensors may require cantilevers with widths larger than where 𝑏̃ 𝑝𝑟 , 𝑣, 𝜋𝐿 &𝜋𝑇 are held fix at 0.1, 0.064, 71.8 × 10−5 MPa−1
150 μm. For instance, ensuring a minimum surface area to immobilize and −66.3 × 10−5 MPa−1 , respectively. Fig. 7 shows that the loading
a desired number of receptor molecules without compromising the parameter mismatch varies between ±10 × 10−4 for −1.1 × 10−5 ≤
cantilever mechanical stability (short cantilevers). Molecules such as 𝛥𝜎̃ 𝑠 ≤ 1.1 × 10−5 and 0 ≤ 𝑙̃𝑝𝑟1 ≤ 5. The loading mismatch is only
the Carbaryl pesticide in a functionalized layer are spaced between zero when 𝛥𝜎̃ 𝑠 = 0. A first strategy to increase the loading parameter
200 to 300 nm [31,41]. Accommodation of a million molecules will mismatch consists in utilizing short longitudinal piezoresistors near the
require cantilever areas in the order of 40 000 μm2 . Cantilever surface cantilever fixed support. That is, 𝑙̃𝑝𝑟𝐿 ≤ 0.2. Increasing 𝑙̃𝑝𝑟𝐿 beyond
areas as low as 5 × 5 μm2 are readily achieved with current laser direct 0.2 will result in a fast decay in the loading parameter mismatch
writing micro pattern generators [55]. On the other hand, controlling (see Fig. 7). A second strategy consists in increasing the magnitude

8
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

𝛥𝑅𝑈
Fig. 6. Relative change of resistance ( 𝑅 ) for a variety of U-shaped piezoresistors in terms of the loading parameter mismatch (𝜆̃𝐿 − 𝜆̃𝑇 ) and its location with respect to the
𝑜,𝑈
cantilever neutral plane (𝑧̃ 𝑜 ).

Fig. 7. Influence of the surface stress (𝛥𝜎̃ 𝑠 ) and the longitudinal piezoresistor length (𝑙̃𝑝𝑟𝐿 ) on the loading parameter mismatch (𝜆̃𝐿 − 𝜆̃𝑇 ).

of the dimensionless surface stress (𝛥𝜎̃ 𝑠 ). An inspection of Eq. (13) The results in Fig. 2 show that the relative change in resistance is
suggests three approaches. Namely, increasing the surface stress (𝛥𝜎𝑠 ), insensitive to changes in ℎ̃ 𝑝𝑟 &𝑏̃ 𝑝𝑟 . Therefore, the results presented in
increasing the piezoresistive parameter (𝜋𝐿 + 𝜋𝑇 ), and reducing the Fig. 6 and Fig. 7 are also representative of piezoresistors with thickness
cantilever thickness (ℎ). The two first approaches may not be very and width in the ranges of 0 ≤ ℎ̃ 𝑝𝑟 ≤ 0.3&0 ≤ 𝑏̃ 𝑝𝑟 ≤ 0.3, respectively.
practical given that they depend on the target molecule, and doping ele-
ment. Microfabrication on the other hand, could provide the designer a 5. Conclusions
mean to control the cantilever thickness. A 10% reduction in thickness
will result in a 10% increment in 𝛥𝜎̃ 𝑠 . Caution should be observed We presented an analytical model for the response of piezoresis-
when reducing thickness as it could lead to elastic instability during tive micro-cantilevers subject to surface stress. With this model at
fabrication or operation of the cantilever. hand, we have discussed some preliminary guidelines relevant for the

9
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

design of monocrystalline p-type doped silicon biochemical sensors. Acknowledgments

Our model estimates relative changes in resistance taking into consid-
eration the dimensions and location of piezoresistive elements within The present research work was funded by Mexico’s National Council
the micro-cantilever. for Science and Technology (CONACYT) through FORDECYT-
The preliminary guidelines identified in this work for the design PRONACES Frontiers of Science project number 618306/2020. I.M.
of U-shaped piezoresistors are summarized next. The length of the Garnica-Palafox would like to thank CONACYT for the fellowship re-
two longitudinal segments of the U-shaped piezoresistor should not ceived to conduct her postdoctoral research at CIDESI. S.M.A. Jiménez
be greater than 20% of the width of the cantilever (i.e. 𝑙𝑝𝑟𝐿 ∕𝑏 = 0.2) would like to thank the Cátedras CONACYT program (México) for the
whereas the length of the transverse segment (𝑙𝑝𝑟𝑇 ) should be less than support to conduct the present research work.
60% of the cantilever width. The width (𝑏𝑝𝑟 ) and thickness (ℎ𝑝𝑟 ) of
the piezoresistor do not affect the relative change in resistance. Never- References
theless, they control the piezoresistor resistance in its unstressed state.
[1] M. Alvarez, et al., Development of nanomechanical biosensors for detection of
Also, the piezoresistor must be located as close to the micro-cantilever the pesticide DDT, Biosens. Bioelectron. 18 (5) (2003) 649–653, http://dx.doi.
surfaces as the fabrication technique permits it. The sensitivity of org/10.1016/S0956-5663(03)00035-6.
the piezoresistor increases with the micro-cantilever width. This is of [2] P. Ray, S. Pandey, V.R. Rao, Development of graphene nanoplatelet embedded
polymer microcantilever for vapour phase explosive detection applications, J.
great advantage for design conditions where it is required to increase
Appl. Phys. 116 (12) (2014) 124902, http://dx.doi.org/10.1063/1.4896255.
the biochemical sensor surface area without compromising sensibility [3] R. Zhao, et al., Cantilever-based aptasensor for trace level detection of nerve
or elastic stability. Our calculations also show that the presence of agent simulant in aqueous matrices, Sensors Actuators B 238 (2017) 1231–1239,
the transverse piezoresistors reduces the net change in resistance of http://dx.doi.org/10.1016/j.snb.2016.09.089.
[4] A. Mader, et al., Discrimination of escherichia coli strains using glycan cantilever
the U-shaped piezoresistor. Shortening of the transverse segment or
array sensors, Nano Lett. 12 (1) (2012) 420–423, http://dx.doi.org/10.1021/
substitution by a wire bridge offer an opportunity to increase the U- nl203736.
shaped piezoresistor response. However, this substitution may not be [5] V. Dauksaite, et al., Antibody-based protein detection using piezoresistive can-
feasible for all applications. In addition, the change of resistance in tilever arrays, Nanotechnology 18 (12) (2007) 125503, http://dx.doi.org/10.
1088/0957-4484/18/12/125503.
the U-shaped piezoresistor is increased through the reduction of the
[6] A.M. Moulin, S.J. O’Shea, M.E. Welland, Microcantilever-based biosensors, Ul-
cantilever thickness. tramicroscopy 82 (1) (2000) 23–31, http://dx.doi.org/10.1016/S0304-3991(99)
The sensitivity of the micro-cantilever biochemical sensor can be 00145-X.
increased through the introduction of piezoresistors in meander shaped [7] R.M.R. Pinto, V. Chu, J.P. Conde, Label-free biosensing of DNA in microfluidics
using amorphous silicon capacitive micro-cantilevers, IEEE Sens. J. 20 (16)
configurations. Our analytical model can be readily extended for this
(2020) 9018–9028, http://dx.doi.org/10.1109/JSEN.2020.2986497.
purpose in a manner similar to that presented in this paper for the case [8] J. Zhang, et al., Rapid and label-free nanomechanical detection of biomarker
of a U-shaped piezoresistor. transcripts in human RNA, Nature Nanotechnol. 1 (3) (2006) 214–220, http:
The model here presented can also be extended to include dopant //dx.doi.org/10.1038/nnano.2006.134.
[9] S. Sang, et al., Surface stress-based biosensors, Biosens. Bioelectron. 51 (2014)
concentration and piezoresistive coefficients depth profiles. However,
124–135, http://dx.doi.org/10.1016/j.bios.2013.07.033.
the change in resistance for longitudinal and transverse piezoresistors [10] Z. Wang, et al., A theoretical model for surface-stress piezoresistive microcan-
will no longer have the simple analytical closed-form that helped us tilever biosensors with discontinuous layers, Sensors Actuators B 138 (2) (2009)
gain insight regarding biochemical sensor design. Our model offers 598–606, http://dx.doi.org/10.1016/j.snb.2009.02.046.
[11] R. Mathew, A. Ravi Sankar, A review on surface stress-based miniaturized
two characteristics of interest to biochemical sensors designers. These
piezoresistive SU-8 polymeric cantilever sensors, Nano-Micro Lett. 10 (2) (2018)
characteristics are: 35, http://dx.doi.org/10.1007/s40820-018-0189-1.
(i) Notwithstanding its non-linear form, the model is amenable for [12] J. Bausells, Piezoresistive cantilevers for nanomechanical sensing, Microelectron.
analytic and numerical interrogation. Eng. 145 (2015) 9–20, http://dx.doi.org/10.1016/j.mee.2015.02.010.
[13] W.T. Barlian, et al., Review: Semiconductor piezoresistance for microsystems,
(ii) Given a surface stress level and a set of geometric or material
(in Eng), Proc. IEEE Inst. Electr. Electron. Eng. 97 (3) (2009) 513–552, http:
properties restrictions, the model is very useful to determine, in a //dx.doi.org/10.1109/JPROC.2009.2013612.
preliminary and expedite manner, the micro-cantilever dimensions that [14] J.A. Harley, T.W. Kenny, 1/f noise considerations for the design and process
would yield the maximum/desired estimated response. This is before optimization of piezoresistive cantilevers, J. Microelectromech. Syst. 9 (2) (2000)
226–235, http://dx.doi.org/10.1109/84.846703.
any finite element simulation or experimental characterization takes
[15] J.C. Doll, et al., Design optimization of piezoresistive cantilevers for force
place. sensing in air and water, (in Eng), J. Appl. Phys. 106 (6) (2009) 64310,
http://dx.doi.org/10.1063/1.3224965.
[16] S.J. Park, J.C. Doll, B.L. Pruitt, Piezoresistive cantilever performance—Part
CRediT authorship contribution statement
I: Analytical model for sensitivity, J. Microelectromech. Syst. 19 (1) (2010)
137–148, http://dx.doi.org/10.1109/JMEMS.2009.2036581.
I.M. Garnica-Palafox: Literature research, Methodology develop- [17] J.C. Doll, B.L. Pruitt, Piezorresistor Design and Applications, first ed., in:
ment, Analytical model construction, Numerical predictions and anal- Microsystems and Nanosystems, Springer, New York, NY, 2013, p. 245.
[18] F.T. Goericke, W.P. King, Modeling piezoresistive microcantilever sensor response
ysis of results, Manuscript preparation. S.M.A. Jiménez: Analytical
to surface stress for biochemical sensors, IEEE Sens. J. 8 (8) (2008) 1404–1410,
model construction, Methodology development, Analytical model dis- http://dx.doi.org/10.1109/JSEN.2008.920706.
cussions, Numerical predictions and analysis of results, Manuscript [19] S.M. Yang, T.I. Yin, Design and analysis of piezoresistive microcantilever for
preparation. D. Díaz-Alonso: Project administration, Results discus- surface stress measurement in biochemical sensor, Sensors Actuators B 120 (2)
(2007) 736–744, http://dx.doi.org/10.1016/j.snb.2006.03.053.
sions, Writing – review & editing. D.A. Fernández-Benavides: Funding
[20] G.G. Stoney, The tension of metallic films deposited by electrolysis, Proc. R. Soc.
acquisition, Project administration, Supervision, Writing – review & Lond. Ser. A Contain. Pap. A Math. Phys. Charact. 82 (553) (1909) 172–175,
editing. W. Calleja-Arriaga: Funding acquisition, Writing – review & [Online]. Available: http://www.jstor.org/stable/92886.
editing. J.M. Alvarado-Orozco: Writing – review & editing, Project [21] S. Kassegne, et al., Design issues in SOI-based high-sensitivity piezoresistive
cantilever devices, in: SPIE’s 9th Annual International Symposium on Smart
administration, Project support.
Structures and Materials, SPIE, 2002.
[22] S.M. Yang, C. Chang, A piezoresistive bridge-microcantilever biosensor by CMOS
Declaration of competing interest process for surface stress measurement, Sensors Actuators B 145 (1) (2010)
405–410, http://dx.doi.org/10.1016/j.snb.2009.12.037.
[23] G.C.A.M. Janssen, et al., Celebrating the 100th anniversary of the stoney equation
The authors declare that they have no known competing finan- for film stress: Developments from polycrystalline steel strips to single crystal
cial interests or personal relationships that could have appeared to silicon wafers, Thin Solid Films 517 (2009) 1858–1867, http://dx.doi.org/10.
influence the work reported in this paper. 1016/j.tsf.2008.07.014.

10
I.M. Garnica-Palafox, S.M.A. Jiménez, D. Díaz-Alonso et al. International Journal of Non-Linear Mechanics 142 (2022) 103988

[24] J. Tamayo, et al., Quantification of the surface stress in microcantilever biosen- [41] T. Yi, L. Li, C.-J. Kim, Microscale material testing of single crystalline silicon:
sors: Revisiting Stoney’s equation, Nanotechnology 23 (2012) 475702, http: process effects on surface morphology and tensile strength, Sensors Actuators A
//dx.doi.org/10.1088/0957-4484/23/47/475702. 83 (1) (2000) 172–178, http://dx.doi.org/10.1016/S0924-4247(00)00350-2.
[25] Y. Zhang, Q. Ren, Y. Zhao, Modelling analysis of surface stress on a rectangular [42] T. Namazu, Y. Isono, T. Tanaka, Evaluation of size effect on mechanical
cantilever beam, J. Phys. D: Appl. Phys. 37 (15) (2004) 2140–2145, http: properties of single crystal silicon by nanoscale bending test using AFM, J.
//dx.doi.org/10.1088/0022-3727/37/15/014. Microelectromech. Syst. 9 (4) (2000) 450–459, http://dx.doi.org/10.1109/84.
[26] P.A. Rasmussen, O. Hansen, A. Boisen, Cantilever surface stress sensors with 896765.
single-crystalline silicon piezoresistors, Appl. Phys. Lett. 86 (20) (2005) 203502, [43] H. Sadeghian, et al., Effects of size and defects on the elasticity of silicon
http://dx.doi.org/10.1063/1.1900299. nanocantilevers, J. Micromech. Microeng. 20 (6) (2010) 064012, http://dx.doi.
[27] A. Venkatasubramanian, et al., MOF @ MEMS: Design optimization for high org/10.1088/0960-1317/20/6/064012.
sensitivity chemical detection, Sensors Actuators B 168 (2012) 256–262, http: [44] T. Yi, C.-J. Kim, Measurement of mechanical properties for MEMS materials,
//dx.doi.org/10.1016/j.snb.2012.04.019. Meas. Sci. Technol. 10 (8) (1999) 706–716, http://dx.doi.org/10.1088/0957-
[28] A. Choudhury, et al., A piezoresistive microcantilever array for surface stress 0233/10/8/305.
measurement: curvature model and fabrication, J. Micromech. Microeng. 17 (10) [45] K. Matsuda, et al., Nonlinear piezoresistance effects in silicon, J. Appl. Phys. 73
(2007) 2065–2076, http://dx.doi.org/10.1088/0960-1317/17/10/019. (4) (1993) 1838–1847, http://dx.doi.org/10.1063/1.353169.
[29] R. Mathew, A. Ravi Sankar, Design of a triangular platform piezoresistive [46] J.M. Chen, N.C. MacDonald, Measuring the nonlinearity of silicon piezoresistance
affinity microcantilever sensor for biochemical sensing applications, J. Phys. D: by tensile loading of a submicron diameter fiber using a microinstrument, Rev.
Appl. Phys. 48 (20) (2015) 205402, http://dx.doi.org/10.1088/0022-3727/48/
Sci. Instrum. 75 (1) (2004) 276–278, http://dx.doi.org/10.1063/1.1633005.
20/205402.
[47] Y. Kanda, Piezoresistance effect of silicon, Sensors Actuators A 28 (2) (1991)
[30] M. Joshi, et al., Modeling, simulation, and design guidelines for piezoresistive
83–91, http://dx.doi.org/10.1016/0924-4247(91)85017-I.
affinity cantilevers, J. Microelectromech. Syst. 20 (3) (2011) 774–784, http:
[48] G. Yang, H. Xie, Mechanical derivation of the longitudinal and transverse
//dx.doi.org/10.1109/JMEMS.2011.2140353.
piezoresistive coefficient on piezoresistive pressure sensor, Procedia Eng. 29
[31] S. Mondal, D. Arya, M.Z. Ansari, High sensitive MEMS piezoresistive microcan-
(2012) 1612–1617, http://dx.doi.org/10.1016/j.proeng.2012.01.182.
tilever sensor, Procedia Comput. Sci. 133 (2018) 793–798, http://dx.doi.org/10.
[49] M.A. Hopcroft, W.D. Nix, T.W. Kenny, What is the Young’s modulus of silicon? J.
1016/j.procs.2018.07.117.
Microelectromech. Syst. 19 (2) (2010) 229–238, http://dx.doi.org/10.1109/
[32] D. Rotake, A. Darji, J. Singh, Ultrasensitive multi-arm-microcantilever-based
JMEMS.2009.2039697.
piezoresistive sensor for Biomems application, in: 2020 24th International
[50] D. McCarter, R. Paquin, Isotropic Behavior of an Anisotropic Material: Single
Symposium on VLSI Design and Test (VDAT), 23-25 2020, 2020, pp. 1–6,
Crystal Silicon (SPIE Optical Engineering Applications), SPIE, 2013.
http://dx.doi.org/10.1109/VDAT50263.2020.9190249.
[33] D.A. Fernández-Benavides, et al., A novel bismuth-based lead-free piezoelectric [51] M. Mohamad, N. Soin, F. Ibrahim, Design of a high sensitivity MEMS piezoresis-
transducer immunosensor for carbaryl quantification, Sensors Actuators B 285 tive intracranial pressure sensor using three turns meander shaped piezoresistors,
(2019) 423–430, http://dx.doi.org/10.1016/j.snb.2019.01.081. in: 2016 International Conference on Bio-Engineering for Smart Technolo-
[34] Chemicals Evaluated for Carcinogenic, US Environmental Protection Agency gies (BioSMART), 4-7 Dec. 2016, 2016, pp. 1–4, http://dx.doi.org/10.1109/
Office of Pesticide Programs, United States, 2020. BIOSMART.2016.7835596.
[35] D. Aguilera Márquez, et al., in: F. Bejarano González (Ed.), Los Plaguicidas [52] R. Mathew, A. Ravi Sankar, In silico modeling and investigation of self-heating
Altamente Peligrosos En México, first ed., Red de Acción sobre Plaguicidas y effects in composite nano cantilever biosensors with integrated piezoresistors,
Alternativas en México, A.C., México, 2017. AIP Adv. 7 (3) (2017) 035108, http://dx.doi.org/10.1063/1.4977827.
[36] M. Jaramillo, et al., Advances in the development of a piezoelectric im- [53] R. Mathew, A. Ravi Sankar, Influence of surface layer properties on the thermo-
munosensor for the detection of a tuberculosis biomarker, in: 2014 IEEE 9th electro-mechanical characteristics of a MEMS/NEMS piezoresistive cantilever
IberoAmerican Congress on Sensors, 15-18 Oct. 2014, 2014, pp. 1–4, http: surface stress sensor, Mater. Res. Express 6 (8) (2019) 086304, http://dx.doi.
//dx.doi.org/10.1109/IBERSENSOR.2014.6995535. org/10.1088/2053-1591/ab1c18.
[37] L.E. Malvern, Introduction to the Mechanics of a Continuous Medium, [54] Y. Tian, et al., A flexible PI/Si/SiO2 piezoresistive microcantilever for trace-
Prentice-Hall, 1969. level detection of aflatoxin B1, Sensors 21 (4) (2021) http://dx.doi.org/10.3390/
[38] J.N. Reddy, Mechanics of Laminated Composite Plates and Shells: Theory and s21041118.
Analysis, second ed., CRC Press, 2003. [55] 𝜇PG 101 Micro pattern generator, user guide, DWL-HI-021, Heidelberg-
[39] W. Fang, Determination of the elastic modulus of thin film materials using instruments, 2015.
self-deformed micromachined cantilevers, J. Micromech. Microeng. 9 (3) (1999) [56] C. Barri, et al., Engineering of the spin on dopant process on silicon on insulator
230–235, http://dx.doi.org/10.1088/0960-1317/9/3/303. substrate, Nanotechnology 32 (2) (2020) 025303, http://dx.doi.org/10.1088/
[40] F.J. Quiñones-N, et al., Mechanical characterization of polysilicon cantilevers 1361-6528/abbdda.
using a thermo-mechanical test chip fabricated with a combined bulk/surface [57] J. Plaza Castillo, Very shallow boron junctions in Si by implantation and SOD
micromachining technique, Results Phys. 4 (2014) 119–120, http://dx.doi.org/ diffusion obtained by RTP, Microelectron. J. 39 (3–4) (2008) 678–681, http:
10.1016/j.rinp.2014.07.007. //dx.doi.org/10.1016/j.mejo.2007.07.058.

11