Professional Documents
Culture Documents
Design of An Integrated Electrostatic Stepper Motor With Axial Field (Sensors and Actuators A - Physical, Vol. 27, Issue 1-3) (1991)
Design of An Integrated Electrostatic Stepper Motor With Axial Field (Sensors and Actuators A - Physical, Vol. 27, Issue 1-3) (1991)
E. BORNAND
Asulab S.A., R +D Laboratories of SMH Group, Passage Max-Meuron 6, CH-2001 NeuchBtel (Switzerland)
(e)
Fig. I. Simplified fabrication process sequence of the Modelling
electrostatic motor with axial field.
A geometrical motor model is shown in
Fig. 2(a). If a voltage V is applied across one
the rotation axis. The aspect ratio of the phase (one pair of diametrically opposed sta-
cross section has been modified for presenta- tor teeth), the rotor teeth tend to align to it.
tion purposes. (a) The silicon wafers (3”, 100 Continuous rotation of the rotor is obtained
orientation, p-type, resistivity = 5 -7 Szcm) by switching V to the next adjacent phase,
are oxidized, and a ground plane region act- causing the rotor to turn in the same or
ing as an electrostatic shield [6] is opened opposite direction, depending on the number
with BHF. N-doping of the region is per- and width of the stator and rotor teeth. The
599
(2)
M,,, = -sgn(cp)$
V22 (R,2 - R:) (3)
1
tric constant of the air is assumed to be one.
The Newton motion law, 1, J2~/8t2 = C M
with I, being the rotor inertia and I: M the
total torque, in our case where a constant (6)
voltage is applied (see Fig. 2) becomes where qair is the viscous coefficient for air. By
replacing eqns. (2), (3) and (5) in eqn. (I),
the general solution for the step motion q(t)
600
0 0.5 1.5
Numerical Example Air gap d (pmf
(b)
We choose R, = 100 pm. Currently de-
posited sacrificial layer thicknesses lead to air Fig. 3. Numerical motor example: (a) rotor static equi-
gaps of typically d = 1.5 pm. A reasonable librium olihet angle, (b) axial deflection equilibrium
d.c. phase voltage is V = 50 V so we have a position for a 200 pm diameter rotor.
nominal electrostatic field of 16.7 V/pm (six
times below the maximum of 100 V/pm, see in Fig. 3(a), and is 11.4” for p = 0.2 and 16.6
[2] and measurement results). We choose to for ~1=0.5. This motor does not fulfil the
have a 12:8 stator:rotor number, where the forward operating condition of eqn. ( 13)
pole width is 0, = 20”. p is assumed to be where (Pi should be <5”, nor the no-reversal
between 0.2 and 0.5. The layout design con- condition of eqn. (14) where (Pi should be
strains the minimum rotor hole radius Ri, to Q lo”. The following maximum ratings are
about 30 pm. In order to still have a good calculated for the aligned position rp = 0,
rigidity of the ring part (R, to Ri), we are hence the phase capacitance is C,,, = 3.3 fF,
bound to have Ri = 60 pm. The bushing ra- the maximum motor torque is M,,, =
dius is between Ri and &: R, = 50 pm. 11.8 pN m, the axial force is F= = 2.7 pN, the
First, the critical speed at which the vis- maximum dry friction torque is M,=
cous and dry friction torques are equal is 27.4 pN m for p = 0.2. The axial equilibrium
calculated with eqns. (3), (6), ( 11) and is deflections of the tooth extremity at 50 V and
found to be only J(P/dtc,i, = 280000 rpm, the 200 V are calculated with eqns. ( 15) and ( 16)
viscosity of the air at ambient T and P being and plotted in Fig. 3(b), the Young’s mod-
qair= 1.8 x 10e5 kg/m s. The equilibrium po- ulus for polysilicon being E = .1.7 x 10” N/
sition rp, is calculated with eqn. (12) as shown m* [7]. It can be observed that the teeth are
602
still very stiff under 50 V, and that no pull-in type motor (four bushings of 8 pm x 8 ,um
voltage seems to exist from that structure each). The sample is offset from the tum-
below a phase voltage of at least 250 V. table centre by Y= 15 mm. Two-thirds of a 3”
silicon wafer (material 2) is glued to the
turntable. The lower sample is spun from 0 to
Dry Friction Measurements 2 rpm, corresponding to a rotational speed of
the numerical example motor of 0 to
Silicon wafers have been processed in the 600 rpm. The friction force Ff causes the arm
same way as for the motor contact layers (see to bend laterally, which deformation is sensed
Fabrication). Thin films of LPCVD silicon with an induction coil. Appropriate electron-
nitride and polysilicon have been deposited, ics deliver a voltage proportional to the force,
then chips have been cut to appropriate di- which is recorded on an x-t plotter. Initially,
mensions. A pin-on-disc machine from we have a peak value due to higher static
CSEM S.A. (see Fig. 4) has been used for the friction. During rotation, the speed is in-
friction measurement. Previous measure- creased to 2 rpm. All wafer parts and chips
ments have also been made with the same have been rinsed in acetone and isopropanol,
principle [S]. A vertical force of N = 1 New- and dried with nitrogen prior to testing. We
ton is applied on a 4.7 x 4.7 mm’ silicon chip counted every run which did not exhibit vis-
(material l), corresponding to approximately ible wear scars on the chips, indicating that
double the pressure calculated for the proto- we measured proper dry friction and not
some unknown grinding. There are not
motorised “ppzr wafer upper sample
enough data to permit a better distinction
lower wafer
turntable with holder between the various materials. Nevertheless,
we can already state that one material on
itself is not a good solution. The nitride was
harder because it left scars on the other sam-
ples. Sometimes the static coefficient was
lower than the dynamic one, as indicated
clearly by the absence of stick-slip. A strong
variation of the coefficient within the same
material category, even if issued from the
same wafer, has been observed. Dynamic fric-
tion coefficients of 0.35 f 0.10 for polysilicon
on nitride, 0.40 & 0.10 for polysilicon on itself
and 0.30 + 0.10 for nitride on itself has been
measured over approximately five runs. Pre-
vious work [5] reported friction coefficient
values of 0.21 and 0.38 for polysilicon on
itself from measurements and modelling on a
side-drive VC motor. This can be in agree-
ment with the value found here: the difference
may be explained by the effect of a smoother
contact surface in the case of the motor.
F,=pN
t
Conclusions
Fig. 4. Schematics of the dry friction coefficient mea-
surement set-up using a pin-on-disc tribometer from Theoretical considerations show that for
CSEM. this electrostatic motor architecture, the dry
603