|
By Kim Tuck and Matt Ellis, Zyvex Instruments
Download a printable copy here.
The ability to power MEMS devices inside
a scanning electron microscope (SEM) is a new featured application
of the Zyvex S100 Nanomanipulator System. This allows for
greater characterization of MEMS devices than existing alternative
methods. While many researchers have examined the performance
of MEMS devices in ambient conditions, very little has been
done in vacuum. Because MEMS devices often operate in hermetically
sealed packages, characterizing their properties in vacuum
presents a more accurate representation of the device properties
in the field.
Probe station characterization in ambient
environments are limited both by the resolution of the optics,
as well as the relatively uncontrolled conditions which introduce
unknown variable (such as dust and temperature fluctuations)
into the test. In contrast, a scanning electron microscope
environment not only eliminates the particle variable, but
also offers very high resolution photographic capability.
The S100 system is equipped with four
quick-change positioners so that probes and other end effectors
can be used for testing a large variety of devices. Complex
MEMS structures require at least two probes, while most active
electronic components such as transistors require at least
three probes for in situ testing. The S100 system
was designed for four point probing, a capability which extends
the use of the system from device characterization to materials
evaluations.
This application note will discuss techniques
used for activating and evaluating MEMS devices using S100
system.
This document will discuss the following:
1. Measuring Current and Voltage
2. Experimental Set-up
3. Powering Devices
4. Observing Backlash and Creep
5. Measuring Plastic Deformation
6. Measuring Device Lifetime
7. Dynamic Behavior of a Device
8. Powering an Electrostatic Actuator
9. Powering an Electrothermal Actuator
Various MEMS devices have been powered
and tested in the S100. From the data obtained comparisons
can be made between testing done in air verses in vacuum.
During the experiment the voltage and current
were monitored using a 2-channel oscilloscope as shown in
Figure 1. The current is measured across
a resistor of known value. The voltage is measured on another
channel of the oscilloscope. It is shown that quadrant “positioners”
4 and 1 are used in the experiment to drive the actuator.
This will vary depending on which quadrant is in use to power
the device.
Figure 1: Power data collection set-up
An electrothermal actuator can be powered in
the SEM using the S100 system. The user must be careful to
track which positioner is being powered, and which pin associated
with that positioner is connected to the probe. Figure
2 shows the electrical head. Figure 3
shows a close view of the 5 electrical connections on the
head where the probe can be placed.
It is a good idea to make note of the probe
positions for all quadrants before placing the S100 system
in the SEM. But sometimes it is not obvious which quadrant
will be best for powering the device until the S100 is in
the SEM and the image is in view. In such cases, the powered
probe can be identified by applying a square wave with a voltage
greater than 1V. This will cause the contrast to change and
the probe will change contrast with the signal.
Figure 2: S100 electrical head
Figure 3: Head connector close-up view 5 pins
The Power Data Collection set-up shown in Figure
1 is connected to the S100 patch panel. The S100
Patch panel interface is shown below in Figure 4
. There are separate panels for each positioner and each BNC
connection corresponds to one of the 5 pin connections on
the electrical head shown above in Figure 3.
Figure 4:S100 patch panel interface
The power on and off positions can be captured
in the SEM, giving a double exposure effect which makes it
useful for taking measurements. The SEM was put into an averaging
mode to create this effect. Videos of the device were also
taken. Various useful measurements can be taken using the
SEM tools. Figure 5 shows the electrothermal
actuator designed with the on and off states
superimposed in one image. This actuator was powered with
a 15 V peak to peak square wave to minimize the time between
the on and off states of the device. It
was noticed that one state (on or off) was
brighter than the other. By modifying the duty cycle of the
input signal the contrast can be adjusted so that both states
have the same brightness. The driving frequency was varied
between 5Hz and 30Hz and adjusted manually until one could
clearly see the on and off position of the
actuator on the screen.
Backlash has been observed while the device
is in the powered (forward moving) state. This is shown below
with the zoom view of the edge of the device (Figure
5). The device overshoots slightly at first and the
backlash is observed as the device moves backwards. Creep
can be observed if the device moves slowly while being powered
by a DC signal. The amount of creep can be measured using
the measure tool in the SEM.
Figure 5: Electrothermal ganged bimorph actuator
The MUMPS bimorph shown in Figure 6
has been purposely overdriven and the resulting plastic deformation
is shown in Figure 7.
Figure 6: SEM image MUMPS bimorph
The two pictures on the right in Figure
7 show the plastically deformed overdriven condition.
The pictures on the right show the normal rest position of
the bimorph. It is observed that the final rest position of
the bimorph is now several microns backwards based on the
vernier scale. The SEM measure tool can also be used to quantify
the displacement. This is demonstrated by comparing the bottom
two images in Figure 7. This plastic deformation
of the bimorph does not reduce the total range of motion of
the bimorph.
Figure 7: Before and after overdriving the
actuator
Square waves of sufficient amplitude drive the
thermal bimorph about its equilibrium position. Figure
8 shows a 2 micron bimorph driven by a 358 Hz square
wave with a 2 Volt amplitude. The driving frequency was chosen
so that the image could easily record the two extreme positions
of the thermal bimorph’s motion. The leftmost position
of the thermal bimorph corresponds to the voltage off position.
In Figure 8, the off position is
different from the fabricated position, probably due to the
higher thermal mass of the cold arm. When driving the bimorph
with a square wave, both the hot arm and the cold arm reach
a steady state temperature distribution if the length of the
square wave pulse exceeds the thermal time constant of the
actuator. When the voltage returns to zero, the hot arm cools
down faster than the cold arm since it has less thermal mass.
The cold arm also takes longer than the hot arm during the
power off part of the square wave cycle. This produces motion
about the power off equilibrium position.
Figure 8: Thermal bimorph 358Hz
Extended testing of the thermal bimorphs showed
that they could be operated for long periods of time under
the appropriate conditions. Figure 9 shows
two images of a 2 micron thermal bimorph in the power off
position. The image on the left was taken prior to operation.
The image on the right was taken after the thermal bimorph
had completed about 30 million cycles. The driving frequency
during the test was approximately 357 Hz and the amplitude
of the square wave was about 1.5 volts. No difference can
be seen in the equilibrium position.
Figure 9: Before and after lifetime testing
Qualitative data regarding the performance of
MUMPS thermal bimorphs has been obtained by testing several
thermal bimorphs from three different dies. The current versus
displacement is easily measured. This is done by measuring
the voltage across the resistor as shown in Figure
1. The voltage is divided by the resistance value
of the resistor used. The displacement is measured using the
SEM measure tool. Current, voltage and resistance versus displacement
can be taken with the set-up shown earlier in Figure
1.
Dynamic behavior was measured by sweeping the
frequency of 1 volt peak to peak sine wave from zero to 120kHz.
The dynamic behavior of thermal bimorph actuators operating
in vacuum is of interest for determining the maximum operating
frequency of the actuator. For applications, the thermal time
constant of the device determines the maximum operating frequency
regardless of the resonant frequency. This time constant is
a measure of how quickly the actuator can cool down after
actuation. For ambient operation, heat transfer out of the
device can be convective through the surrounding air, and
conductive through the anchors connecting the actuator to
the substrate and radiative. In vacuum the heat transfer only
occurs via conduction and radiation; thus the thermal time
constant in vacuum should be longer and the maximum operational
frequency should be lower. This was confirmed from the data
obtained.
The electrostatic device tested is shown in
Figure 10. This electrostatic device requires
higher voltage than electrothermal devices (between 40V and
150V). Therefore some external amplifiers are required to
boost the signal. The power was monitored in a similar fashion
as shown above in Figure 1. Again the contrast
of the image was modified (on and off states) by adjusting
the duty cycle on the square wave signal from the signal generator.
Figure 10: SEM image of an electrostatic
actuator
The electrothermal “hula”
actuator has been tested in the SEM because this device actually
changes form after it has been powered. The pictures show
distinct before and after orientation in Figure 11
(before) and in Figure 12 (after). The device
plastically deforms and its shape remains so after it has
been powered. This device has been specifically designed utilizing
the mismatch in thermal expansion coefficient of metal and
polysilicon to bend the structure out of plane like an accordion.
The “before and after” actuation pictures of the
device can be taken at one time in the SEM due to the ability
to power MEMS devices in the SEM using the S100.
Figure 11: Hula actuator before power
is applied
Figure 12: Hula actuator after power
is applied
Various MEMS devices can be easily characterized
and analyzed in the SEM using the S100. The method of testing
is very similar to that done under an optical microscope at
a probe station, yet with the added resolution of SEM.
back to top
| ©
Copyright 2008, Zyvex Instruments. All Rights Reserved. |
|
