Online proceedings - EDA Publishing Association
Online proceedings - EDA Publishing Association
Online proceedings - EDA Publishing Association
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
11-13 <br />
May 2011, Aix-en-Provence, France<br />
<br />
Reduced Order Modelling of MEMS Dynamics<br />
Stefano Mariani 1 , Saeed Eftekhar Azam 1 , Aldo Ghisi 1 , Alberto Corigliano 1 , Barbara Simoni 2<br />
1 Politecnico di Milano, Dipartimento di Ingegneria Strutturale<br />
Piazza Leonardo da Vinci 32, 20133 - Milano (ITALY)<br />
2<br />
STMicroelectronics, MSH Division<br />
Via Tolomeo 1, 20010 Cornaredo (ITALY)<br />
Abstract- The dynamics of a uniaxial micro-accelerometer<br />
subjected to accidental drop events is studied by means of a<br />
reduced order model. A two degrees of freedom model is built,<br />
which carefully reproduces the MEMS response under high<br />
acceleration events. The results of the reduced order model are<br />
compared to those obtained with a three-dimensional finite<br />
element model, in terms of accuracy of the results and<br />
simulation speed-up<br />
I. INTRODUCTION<br />
During accidental drop events, polysilicon MEMS<br />
sensors are often exposed to high-g loadings because of<br />
their extremely small mass and, therefore, inertia [1-3].<br />
This fact can cause mechanical failure due to cracking in<br />
high stressed regions. In a series of papers [4-8] we<br />
recently proposed a numerical approach to accurately link<br />
the features of the shock-inducing cause (like, e.g. a drop)<br />
to the effects at the MEMS level. Because of the several<br />
length-scales affecting the dynamics of the whole device<br />
when subjected to such shocks, a multi-scale frame was<br />
adopted. We then showed that macro-scale (at device level)<br />
and meso-scale (at sensor level) analyses can be routinely<br />
investigated making use of commercial finite element<br />
codes, since the features of the polycrystalline film<br />
constituting the movable parts of the MEMS have a<br />
marginal impact. A different situation characterizes microscale<br />
(at polysilicon film level) analyses, which turn out to<br />
be extremely complicated and time demanding, in case high<br />
accuracy of the results is mandatory.<br />
A possible way to drastically reduce the computing time<br />
is to make use of reduced order models for the whole<br />
MEMS sensor, built in an accurate and micro-mechanically<br />
informed way. Reduced models would allow to avoid<br />
running analyses at the micro-scale, keeping a similar<br />
accuracy in the results. This issue was partially addressed<br />
in previous works [9-10].<br />
In the present work we go further on in the use of reduced<br />
models built on the basis of purely mechanical<br />
considerations, routed by the investigated details of the<br />
MEMS dynamics. A simple two degrees of freedom<br />
reduced model is built for a commercial micro<br />
accelerometer (Fig. 1) which measures the acceleration<br />
orthogonal to the device substrate. The dynamics of<br />
accidental drop events characterized by two acceleration<br />
levels, is numerically studied by means of the reduce model<br />
and compared with the outcome of a fully 3D finite element<br />
model.<br />
More sophisticated reduced order modelling techniques<br />
could be used, e.g. based on the proper orthogonal<br />
decomposition (POD) [11] and compared with the approach<br />
presented in this paper; this issue will be the subject of<br />
forthcoming works.<br />
II. MODELLING IMPACTS IN MEMS ACCELEROMETERS<br />
The accelerometer was assumed to be subjected to a low-g<br />
acceleration input directed along the Z-axis (see Fig. 2),<br />
whose maximum is about 90 g, and to a high-g input (see<br />
Fig. 3), with a maximum acceleration peak of about 5,500 g.<br />
As for the boundary conditions applied to the model, the<br />
accelerometer is anchored at its center with two slender<br />
suspension springs. As a reference solution a finite element<br />
model of the accelerometer, featuring 34,000 nodes and<br />
26,000 elements, i.e. about 100,000 degrees of freedom (dof)<br />
has been considered. In parallel, a reduced, two-dof model<br />
has been envisaged: if we impose a rigid behavior for the<br />
plate, only dof reproducing the vibration modes #1 and #5<br />
from Fig. 6 need to be considered. The spectral content of<br />
the two input accelerations, as shown in Fig. 4-5 through<br />
their energy spectral density, confirms that the relevant,<br />
excited dynamics pertains the relative rotation of the plate<br />
−=Δ<br />
θθθ<br />
and its relative translation −=Δ<br />
www<br />
at the<br />
spring clamped end, as shown in Fig. 1.<br />
The motion of the device in this two-dof framework can<br />
be described as follows:<br />
− aMuKu<br />
=++ (1)<br />
where the vector u=[Δw Δθ ] T collects the two dof,<br />
superposed dots indicate time derivative. If we assume h to be<br />
the plate thickness, L the plate length in the x direction, ρ the<br />
polysilicon mass density (reduced to take into account for the<br />
holed plate), l the spring length, and V the plate volume, then<br />
the mass matrix collects the plate translational mass<br />
53